Professional Documents
Culture Documents
J T. Black
Steve L. Hunter
Dearborn, Michigan
Copyright © 2003 by Society of Manufacturing Engineers
98765432
All rights reserved, including those of translation. This book, or parts thereof, may not be
reproduced by any means, including photocopying, recording or microfilming, or by any information
storage and retrieval system, without permission in writing of the copyright owners.
No liability is assumed by the publisher with respect to use of information contained herein. While
every precaution has been taken in the preparation of this book, the publisher assumes no respon-
sibility for errors or omissions. Publication of any data in this book does not constitute a recom-
mendation or endorsement of any patent, proprietary right, or product that may be involved.
studied and understood how the system of rates. The materials were pulled through the
mass production functioned. However, he rec- factory by consumption of parts in final
ognized that the Ford system was designed to assembly. An empty cart returned to the manu-
handle large volumes of the same parts with no facturing point was the signal to make more
variety. Over time, a new system invented by parts. Ohno quickly discovered that it was
Ohno emerged. This new manufacturing more productive to move machines closer
system evolved into a hybrid of the flow shop together so that the workers could make one,
and the job shop designs. The Ohno system was check one, and pass it on to the next machine.
designed to handle large or small volumes of a Thus, manufacturing cells evolved through the
variety of parts using the same economies of factory-wide desire to eliminate waste. There is
volume as the Ford system. Many people now no record of Toyota using group technology, a
say that Ford invented lean production. This is well-known method for finding families of
simply not so; the inventor of lean manufac- parts around which cells can be designed.
turing was Ohno, and he should be so rec- Many decision-makers have resisted con-
ognized. verting to lean manufacturing because it is not
Ohno got the idea for his kanban system high-tech or they perceive that it may be dif-
after visiting an American supermarket, where ficult to implement. However, lean production
he observed people pulling goods from the may severely impact the political and social
shelves to fill their shopping carts. The empty structure of the entire company. It takes many
space on the shelf was the signal for the stock reasons for a company to undertake such an
person to restock by reordering cases of the effort and many excuses for not doing it.
consumed item. In effect, the shopper provided However, this much is clear: companies that
a totally flexible final assembly for the custom have the courage to undertake this change will
order. In applying this idea to manufacturing, be survivors—the factories with a future. Those
Ohno developed a system whereby downstream who resist will become history.
use of parts dictated upstream production
Preface..................................................................................................................................xv
Acknowledgments................................................................................................................xvii
Reliability...........................................................................................................................192
The Second Shift.................................................................................................................195
Total Productive Maintenance Implementation.........................................................................197
Summary ...........................................................................................................................197
References ..........................................................................................................................327
Index ..................................................................................................................................333
Layout Functional layout Product layout One-piece flow Linked assembly of large
via Linked-cells modules or
subassemblies
Enabling Power for machines Moving final assembly line U-shaped cells; Virtual reality/simulation
technology Steel production Division of labor Kanban 3D design using low-cost
(physical Railroad for Standardization leading to Rapid die exchange High-performance
implementation) transportation true interchangeability Zero defects computers
Automatic material Total preventive
handling maintenance
*System developed by Taiichi Ohno who called it the Toyota production system (TPS)
Toyota production system, is now referred to as tion began with the advent of powered machine
lean production. The Toyota production system tools, the creation of factories, and a movement
uses a system design often called the linked- of people from farms to towns and cities. This
cell manufacturing system. This system is sim- mass movement to metropolitan areas was a
pler, produces lower cost, higher quality parts, megatrend of the first industrial revolution.
and is more flexible than the first two manu- Now 200 years later, only about 2–3% of
facturing industrial revolutions’ manufactur-
Americans work directly on farms, but the pro-
ing system designs. The third industrial
ductivity and yield of farms continues to
revolution’s critical control functions are inte-
grated into the design of the basic building increase. Large, often special-purpose farming
blocks of the Toyota production system. The equipment, decreases the number of direct
basic building blocks of the Toyota production laborers (the number of people working in the
system are manufacturing cells. fields). A similar trend occurs in the number of
direct laborers in factories. Modern factories
First Industrial Revolution are continually decreasing the amount of
In 1800, approximately 80% of Americans required labor due to improved factory designs
worked on farms. The first industrial revolu- and better equipment.
Water wheel
Grinders
need to power machine tools was used to
River
increase productivity by increasing cutting and
Main
forming speeds. Therefore, the method needed power
to drive or power the machines developed into shaft Lathes
a functional design.
The first factories were built on the banks of Overhead shafts in
departments
rivers and streams so that the machines and
processes could be powered by water pressure
turning waterwheels. Waterwheels drove Gears in gear house
shafts that ran into the factories (see Figure Return
1-1). Machines and processes were aligned
underneath appropriate power shafts, since
individual shafts turned at speeds required to
drive particular sets of machines. For example,
Figure 1-1.Early factories used water power and a system of
lathes were collected under their own power shafts to drive machines.
shaft, likewise, milling machines and presses.
Belts transferred power off the shaft to deliver
a single, set speed to a particular group of
machines. Later, waterwheels were replaced by Lathe department Milling department Drilling department
steam engines, which allowed factories to be
located beyond riverbanks where spring floods L L M M D D
were often a problem. Eventually, large electric
motors replaced steam engines. Finally, indi- L L M M D D
vidual electric motors replaced large electric
motors. Nevertheless, the job-shop design con- L L M M
G G
tinued and the functional design held on blind-
ly to the original system design (see Figure L L Assembly
1–2). This era became known to historians as G G
A A
the American Armory System because most
early factories manufactured weapons such as Receiving and A A G G
shipping
muskets. The industry leaders of the world Grinding department
came to view the revolutionary American
Armory System and it was disseminated Figure 1-2. Layout of a small job shop where processes are
throughout the industrial world. functionally gathered into departments.
Large batches
duction system, or economic-order-quantity 37
36
equation. 35
34
Figure 1-5 shows a schematic diagram of a 33
32
mass production system with a job shop, flow 31
shop, and final assembly line. The job shop is 30
Conveyor
Lines
handlers and forklift operators the next stop
where parts are to go. Production control has
the responsibility of determining when (tim-
ing), where (which machine), and how many
(lot size), so work on parts is carried out in a
supposedly orderly and timely manner.
Figure 1-5 details a schematic of a final
assembly line—a type of flow-line manufactur- Parts storage
ing system design. A modern-day automobile
assembly plant with a moving assembly line is
Job shop
a good example of a flow line. The moving
assembly line for automobile production is bro-
ken into hundreds of stations sequenced to sys- Lathe Mill Drill
tematically build an automobile. This requires
work at each station to be balanced, or equal-
ized, so each task takes about the same amount Receiving
Assembly Grind
of time. This system requires interchangeable Shipping
parts with the repeatability of each manufac-
turing process based on precise measurement Raw materials
standards. The moving assembly line produces
vehicles one at a time, in a method sometimes Figure 1-5. The mass-production system for automobiles is a
called one-piece flow. moving assembly line (stations 30-37 shown).
The transfer line is an extensively automat-
ed flow line for machining products in large
volumes. The enabling technology of the trans- ly justify expensive capital investments.
fer line is the use of repeat-cycle automatic Transfer-line systems were not only expensive,
machines. This system also requires inter- but they were inflexible. Automatic repeat-
changeable parts based on precise measure- cycle machine tools on transfer lines were
ment standards. The transfer line was designed replaced with computer numerical control
to produce large volumes of identical goods, a machines or flexible manufacturing systems.
process that was often required to economical- Today, flexible manufacturing systems are
External customer
are most often produced by job shops in large A manufacturing system is Outputs
lots. Component parts are often held in inven- a complex arrangement ‡
Inputs of physical elements* Feedback
tory for long periods of time and eventually characterized by
brought to the transfer line where a particular measurable parameters#
product is being assembled. The line produces
one product during a long run and then switch- *Physical elements: #
Measurable system
es to another product. The changeover setup • Machine tools for processing parameters:
can take days or even weeks. • Material handling • Throughput time
Transfer lines were in place during World equipment • Production rate/cycle time
• People (internal customers) • Work-in-process inventory
War II and were clearly responsible for • Tools (tooling refers to ‡
Complex arrangement
America having the military equipment and workholding devices, cutting
• Design
weapons to win the war. tools, and dies)
• Layout
This massive production system thrived
after World War II, permitting automobile pro- Figure 1-6. The manufacturing system with inputs and outputs,
ducers to manufacture vehicles in large vol- the portion that actually makes the products.(Black,1990)
umes using economy of scale. Just when it
appeared that nothing could top this manufac- ment of the unique cellular manufacturing sys-
turing system, a new player, Toyota, devised a tem, however, is what made Toyota the top car
system design that thrust the manufacturing maker in the world. The new system linked cell
world into the third industrial revolution.
manufacturing, known as the Toyota produc-
tion system, the Just-in-time (JIT) system, or
Third Industrial Revolution
world-class manufacturing system. See Table
At this writing, the third industrial revolu- 1-2 for an extensive listing of linked-cell manu-
tion is over 40 years old. The third industrial facturing system names. In 1990, the system
revolution is not based on hardware or any par-
was called lean production. Lean production
ticular process, but on the design of the manu-
was coined by John Krafcik, an engineer work-
facturing system—the complex arrangement of
ing in the international motor vehicle program
physical elements characterized by measurable
at MIT (Womack, et al 1991).
parameters. Figure 1-6 illustrates this manu-
facturing system. Toyota was the first to devel- What was different about this linked-cell
op this design scheme. Among the Japanese manufacturing system was the development
manufacturers, Toyota Motor Company distin- of manufacturing and assembly cells designed
guished itself as the best. Many factors made to be linked to final assembly. Figure 1-7
this company the top automobile manufacturer shows cell-to-cell and cell-to-other system-
in the world. There was a time when the world component relationships. The fundamental
looked to Japan and observed a tiny nation. linking function of this manufacturing philos-
Now this small country has become a giant in ophy produces a functionally integrated sys-
the global manufacturing arena. tem for inventory and production control. The
A new manufacturing system design vaulted results of this system are low unit costs, high
Colt and Remington to the manufacturing fore- productivity, superior quality, and on-time,
front in the first industrial revolution; Ford, in every-time delivery of unique products from a
the second industrial revolution. The develop- flexible system.
M L
L L In
Milling D G
M department M Out
Manufacturing cell 1
M M
L M D
D D D
D D G
D D D
Manufacturing cell 2
L M D
G G
G D G
Manufacturing cell 3
Figure 1-8. The job-shop portion of a factory requires a systems-level conversion to reconfigure it into manufacturing cells.
In Out
1 Passed on
conveyor 1 Coil 1 1 Coil 1 1 Coil 1 3 Coil 3 4 Coil 4 4 Coil 4
After:
U-Shape layout
1 Coil 1 2 Coil 2 1 Coil 1 2 Coil 2
3 Coil 3 3 Coil 3
In In
Out Out
Measurable parameters
Before After
Output 1,105 units/shifts 1,000 units/shift
In-process inventory 180 units /line 5 units/cell
Personnel 6 or 7 operators 6 operators
Daily output per person 145 units 180 units
Cycle time 0.44 minute 0.83 minute
Figure 1-9. Subassembly flow lines redesigned into U-shaped cells using walking workers. (Sekine 1990)
work on the factory floor itself and more will be plant floor will be seen. The workers remaining
involved in the production of goods. Value will be will be better educated, more involved in solv-
added and wealth created only through the con- ing daily production problems, work to improve
version of raw materials by manufacturing. the entire system, and make decisions about
There is a general trend toward more labor how to improve industrial jobs and manufac-
saving automation in factories, but this turing systems.
should not be an undue concern. Factory
automation creates more jobs every year than The Evolution of Lean Manufacturing
it eliminates. Tomorrow’s factory will require What spurred the third revolution was the
greater levels of worker knowledge and more competition of the Japanese in a wide variety of
effective modes of transferring information on products and markets. Japanese prices were
the quality and quantity of goods manufactured. competitive and the quality superior. The coun-
The knowledge base of factory workers must be try financed its remarkable transformation
increased to improve productivity. Knowledge through low-interest government and private
has a market value, particularly technical
loans. More significantly, it developed lean pro-
knowledge, thus, one can expect to pay factory
duction techniques to reduce inventory levels.
floor workers more in the future.
Better cash flow resulted because of the vastly
The United States tends to give technical
increased frequency of inventory turnovers.
knowledge away or sell it to the lowest bidder.
Americans have theorized that Japanese
For example, Japan purchased much of the tech-
manufacturing success can be attributed to
nical knowledge it needed to build vehicles, elec-
their dumping of excess products in United
tronics, machine tools, and even electron
microscopes. The early Japanese transmission States markets; governmental support given to
electron microscopes were essentially duplicates targeted industries; or cultural differences. The
of the precision-made Siemens instruments. A Japanese are known to be hardworking and
common thread of high, or sophisticated, tech- industrious citizens, like many of their col-
nology binds Japanese product areas. Precision leagues worldwide. The theory that the success
machining of magnetic lenses, along with the of the Japanese is simply the result of working
fabrication of high-voltage electronics, are key to long hours for substandard wages does not ring
the construction of quality electron microscopes. true because of the complex, quality products
The factory with a future needs superior being built. The Japanese developed a new
information systems and people who can pro- manufacturing system design, specifically
gram, analyze, and otherwise deal with the Toyota, that is functionally and operationally
information flowing to and from the factory different from any other manufacturing sys-
floor. Unlike energy, knowledge and informa- tem—lean production. At the outset, it is
tion do not follow the laws of conservation. important to understand this new system so
Rather, they are synergistic, each promoting that these principles can be implemented into
growth in the other. As the factory worker factories.
becomes better educated and knowledgeable Lean production is flexible, yet still delivers
about how the entire manufacturing produc- products on time, at the lowest possible cost,
tion system works, the system subsequently and on a continuing basis. They educated their
begins to function better. work force and placed their best engineering
Manufacturing processes and systems must talent on the production floor, rather than in
become simpler with automation. Often prod- the design room. Rather than invent a new
ucts are redesigned so that they can be auto- mousetrap, the Japanese developed a better
matically processed or assembled. In the 21st way to make mousetraps of superior quality at
Century, significantly fewer workers on the a lower cost.
The creative process technology of the single mind, strongly focused on making the
Japanese spurred the manufacturing cell con- system better everyday.
cept. Suppliers to final-assembly plants have
captured much of this creative process technol- Cultural Change
ogy. This is one reason why lean manufacturers Coupled with the redesign of manufactur-
today opt for single-source suppliers, where the ing and production systems is a required cul-
proprietary-process technology is captured in tural change in the operating philosophy
one place. Toyota and many other Japanese within a company. Employee involvement and
manufacturers do not write about their tech- teamwork are rooted in the idea that no one
nology or manufacturing systems (not even in employee is better than any other. All employ-
their technical journals). The Japanese are sen- ees are addressed as associates. There is no
sitive to the fact that they have received much executive lunchroom—the CEO eats lunch
of their knowledge by picking up an American with supervisors and workers. There are no
journal and reading about it. Japanese organi- preferred parking places, except for the asso-
zations sometimes make it very difficult for ciate of the month.
outsiders to get access to proprietary-process The old system of management telling
technology. workers what to do and how to do it must
The Toyota Production System was designed change. However, this change usually requires
to produce superior quality products. Toyota that the CEO have the courage to shift a por-
has taught the concept of total quality control tion of the decision-making responsibilities
to everyone, from company presidents to pro- from management to the factory floor.
duction workers. Japan has changed from a Restructuring manufacturing and production
country that made junk to a nation that gives systems helps this most difficult transition.
customers high-quality, robust, and reliable The biggest and perhaps most difficult
products. The Japanese accomplished this change is cultural. Employees’ mentalities
world-class manufacturing status through must change along with the factory floor
manufacturing system design. design. Workers must be convinced that what
W. Edwards Deming, the famous quality they think and feel about the manufacturing
expert, said that the reduction of variation is system is important to the success of the
the key to success (Deming 1982). If this is so, organization. For this to work, achievement
then reducing variation is the goal for continu- must be tracked closely and rewarded. People
ous improvement, a cornerstone of the lean will “give it an extra mile” if there is a strong
production philosophy. Types of variation reward system. Rewards do not necessarily
important to a manufacturing organization need to be monetary; there is strong evidence
are: that verbal recognition and certificates work
equally well.
• quality (defects/million);
• output (parts/day); Introduction to Lean Manufacuring
• throughput time (hours/part); and
Lean manufacturing implementation requires
• cost (dollars/part).
a systems-level change for the factory—a change
All of these can be reduced through the imple- that will impact every segment of the company
mentation of lean manufacturing philosophies from accounting to shipping. Table 1-3 presents
and methodologies. Additionally, continuous an outline for a 10-step methodology that con-
change and redesign of the manufacturing sys- verts a manufacturing system from mass to lean.
tem design are key to the lean production phi- This is a design task, so if machines are not
losophy. Here, the entire organization is like a being relocated, lean production is not being
implemented. The design and implementation and usually not elective. Changing the manu-
of manufacturing cells is absolutely critical for facturing system, especially the culture, is
a manufacturing organization to convert to often difficult.
lean production. It is the manufacturing system that pro-
Before a factory embarks on the 10 steps, duces goods that the customers demand. There
everyone must be educated on the concepts of are two groups of people the manufacturing
lean production. Top management must be system must satisfy: those who use the prod-
committed to the venture. If the top manager is ucts and those who use the system that makes
not convinced and giving 100% support to a fac- the products. The manufacturing system’s
tory conversion, then the prospects of lean pro- external customers buy or use products made
duction success drop significantly. Every by internal customers. The internal customers
employee must be involved, motivated, and are workers who use the system to manufac-
ultimately committed to the change process— ture goods. Internal customers are usually a
this is a top-down initiative. smaller group of people than the external cus-
The conversion to linked cells is a system- tomers. The manufacturing system must be
level change affecting the factory environment restructured for the benefit of internal cus-
and functional and cultural relationships with- tomers, the users of the manufacturing system.
in it. Lean production and implementation of External customers of the manufacturing sys-
its methodologies are long-term strategies. tem also must be satisfied. External customers
Changing the manufacturing system is equiva- demand competitive prices and goods that are
lent to heart transplant surgery. It is major high in quality, robust, and delivered on time.
in one cell that makes racks for the Honda The average US labor cost in manufacturing
Accord. was $14.85 per hour for hourly workers in
Materials, people, and equipment are interre- 2001. Thus, reductions in direct labor had only
lated factors in manufacturing that must be a marginal effect on overall costs. A systems
combined properly to achieve low cost, superior approach, or taking into account all factors,
quality, and on-time delivery of goods. Typically, must be used when searching for waste-cutting
as shown in Figure 1-10, 40% of the selling measures. This requires a sound understand-
price of a product is its manufacturing cost. ing by decision-makers of materials, processes,
Since the market determines the selling price, and equipment, followed by an understanding
maintaining a profit often depends on reducing of the manufacturing systems.
manufacturing costs. Direct labor is usually the
target of automation and accounts for about Changing World Competition
5–12% of manufacturing costs, even though
many view it as the main factor in increasing In recent years, major changes in the world
productivity. of goods manufacturing have occurred. Some of
The lean manufacturing strategy attacks these changes are:
material, general administration, and labor • worldwide competition by world-class
costs. Material costs include capital costs and companies (global companies);
the costs of storing and handling materials • advanced manufacturing-process technol-
within the factory. The lean strategy eliminates ogy; and
material handling, except as needed to make • new manufacturing-system structures,
products. Eliminating material handling is not strategies, and management.
the same as automating it.
Total expenses equal the selling price less Worldwide competition is now a fact of manu-
the profit. Of those expenses, approximately facturing life. In automobile manufacturing, for-
68% of the dollars are spent on people. This eign competition is relocating on American soil,
includes about 15% on engineers, 25% on mar- demonstrating that the new lean production sys-
keting, sales, and general management people, tem works just as well in the United States as it
5% for direct labor, and 10% for indirect labor. does in Brazil or Japan. The future will bring
R&D
cost 5% Profit 20% Direct labor
12%
Figure 1-10. Manufacturing cost is typically the largest cost in the selling price.
more, not fewer, world-class competitors into the teams use “pro-T” sets and “I-formations,” but
manufacturing world. Everyone with enough many US factories are still using the archaic
capital has access to new manufacturing-process single-wing version called the job shop.
technology because technology can always be The job shop is a functionally designed manu-
purchased. Thus, technology alone does not facturing system where like processes are typi-
ensure manufacturing competitiveness. The cally put together into departments. The same
secret to success in manufacturing is to build a organization occurs in football, with all the line-
company that can deliver on time (short through- men segregated from the backs. Coaches are
put time), superior-quality products to the cus- equivalent to foremen, and the head coach is the
tomer at the lowest possible cost (with the least supervisor.
amount of waste), with the ability to be flexible In the football analogy, the production system
and robust. would be the athletic department that sells tick-
ets, runs the training room (machine mainte-
PRODUCTION AND nance and repair), raises operating capital,
MANUFACTURING SYSTEMS arranges material handling (travel), and does
whatever is needed to help keep the manufac-
Manufacturing is the economic term for turing system operating. The production system
making goods and services that will be avail- does not perform the actual manufacturing
able to satisfy human wants and needs. processes, just as no one in the athletic depart-
Manufacturing is creating value by applying ment actually plays in the football games.
useful mental and/or physical labor and, thus, Production system personnel are more indirect,
converting raw materials into useful products such as managerial and staff employees. In the
demanded by consumers. plant, the production system personnel service
Manufacturing processes are combined to the manufacturing system. This division is called
form manufacturing systems. Manufacturing staff, while people who work in direct manufac-
systems take input and produce product for cus- turing are called line workers. The production
tomers. Production systems include and service system includes the manufacturing system per-
manufacturing systems. Thus, a production sys- sonnel, plus other people in functional areas of
tem refers to the total company and includes the plant who provide information, design,
manufacturing systems as shown in Figure 1- analysis, and control. When the manufacturing
11. The following football analogy helps to dis- system is a job shop, the production system is
tinguish between manufacturing and functionally designed.
production systems. The subsystems are connected to one anoth-
Imagine that college football is an example of er to produce goods, services, or both. Goods
a service industry. Football players would be refer to material items. Services are nonmater-
the equivalent to the machine tools (see Figure ial. Customers buy services to satisfy wants,
1-12). The actions that the players perform needs, and desires. Service production systems
such as punting, passing, running, tackling, include transportation, banking, finance, insur-
and blocking could be equated to manufactur- ance, utilities, health care, education, commu-
ing operations. Different players, or tools, have nication, entertainment, sporting events, etc.
different functions, and some perform better These services involve labor that does not direct-
than others. ly manufacture a product or create wealth.
The arrangement of machines (often called Production ranking has a definite rank of
the factory layout) defines the basic design of the importance. This is simple and easily under-
manufacturing system within the company. In stood, for example confusing system with
football, this arrangement is called an offensive process is similar to mistaking a CEO for the
alignment or defensive formation. Modern floor sweeper. Knowledge of rank is necessary
Customer sales
Distribution and forecasts
Inspect Predict
products demand
(Q)
Inspection
(quality Finance
Recommended
ility
Production
% of scrap
Pro
Finished budget
products lost
Drawings,
Production
ia
specifications,
ls
Material Purchasing or
planning and and standards
procurement
delivery control
Inventory
control Material
schedules
Material
requisitions R&D
Vendors
Figure 1-11. Functions and systems of the production system, which includes the manufacturing system.
Manufacturing process
Football players
or machine tools
Operations Operations
Run, punt, pass, block Turning, drilling, boring,
tackle, catch tapping
Complex design or
layout of the
manufacturing system
Functional layout
Product layout
Designing the
plays or Process layout
formations One-piece-flow layout
Figure 1-12. Analogy of football to manufacturing and production system terminology. (Black, 1990)
(see Table 1-4). The terms tend to overlap Much of the information given for the pro-
because of the inconsistencies of popular usage. duction system is relevant to the service pro-
duction system, which, by design, is usually a
Production System job shop. The job shop, which contains elements
The highest-ranking area in the manufac- of the flow shop when volume is large enough
turing hierarchy is the production system. A to justify special-purpose equipment, is a com-
production system includes people, financing, mon system worldwide. The job shop with flow-
funding, equipment, materials, supplies, mar- shop elements is called the production job shop.
kets, management, and the manufacturing sys- Most manufacturing systems require a service
tem. In effect, all aspects of commerce production system for the most effective prod-
(manufacturing, sales, advertising, profit, and uct sales. This is particularly true in industries
distribution) are involved. such as the restaurant industry where cus-
Manufacturing process A specific piece of equipment designed Spot welder, milling machine, lathe,
(machine or machine tool) to accomplish specific processes, often drill press, forge, drop hammer, die
called a machine tool; machine tools caster—manufacturing process
gathered together to make a technology.
manufacturing system.
Operation or task A specific action or treatment, the col- Drill, ream, bend, solder, turn, face,
(sometimes called a process) lection of which makes up the job of a mill, extrude.
worker.
Tools or tooling The implements used to cut, shape, or Grinding wheel, drill bit, tap,
(cutting tools, workholders) form the work materials, called cutting end-milling cutter, die, mold, clamp,
tools if referring to machining; also three-jaw vise, plate jib.
refers to jigs and fixtures used for
workholding and punches and dies in
metal forming.
typical job shop spend only 5% of total manufac- However, because these machines still are in a
turing time on machines and the rest waiting or job-shop environment, there remains a high
moving from one functional area to the next. percentage of waiting and delay time.
Carter studied the machines themselves to deter- The percentage of time a machine spends mak-
mine what happens once a part is on a machine. ing chips (40%) multiplied by the percentage of
He found that as a part is being processed, value time on the machine (5%), yields a 2% productive
is added only 30–40% of the time by changing its time for an eight-hour shift. Suppose it is recom-
geometry (see Figures 1-13a and b). The remain- mended that the process be upgraded, increasing
der of the time the part is being loaded, unloaded, the percentage of time making chips to 50%
inspected, etc. The advent of numerical control (using a faster machine), or even 60%? Such an
machines has increased the percentage of pro- improvement might cost $300,000. If no change
ductive machine time because tool movements in the manufacturing system takes place, the
are programmed and the machine can automati- effect on the bottom line will be small, in the
cally change tools or load and unload parts. neighborhood of 1%. If the manufacturing system
Figure 1-13b. Typical utilization of production time in metal-removal operations with conventional tool handling, workpiece
loading, setups, and inspection.
is restructured to raise the time on the machine companies is that managers seem to have an
to 50%, the process will result in a significant aversion to understanding the industrial revolu-
20% overall improvement in productivity. tion’s effect on their own companies’ manufac-
turing technologies. This may be related to the
UNDERSTANDING argument presented in the previous section: why
A COMPANY’S BUSINESS worry about process improvements when they
have such a small impact on the bottom line?
Understanding process technology is impor- Failure to understand the company’s busi-
tant for everyone in the company. Manufacturing ness, especially its fundamental process tech-
technology affects the design of the product and nology, can lead to failure. The way for a CEO
manufacturing systems, the way the manufac- to overcome an aversion to manufacturing
turing system can be controlled, the type of per- technology and acquire a true understanding of
sonnel employed, and the materials that can be it is to go to the factory floor and run the actu-
processed. Table 1-5 outlines process technology al process. Suppose a job on an assembly line is
characteristics. One valid criticism of American to tighten bolts to a specific torque. How will a
worker know if the right amount of torque has unique product features. As the competitive
been applied? Only someone who has run a focus shifts during the different stages of the
drill press can understand the sensitive rela- product life cycle, the requirements placed on
tionships between feed rates, drill torque, and manufacturing cost, quality, flexibility, and
thrust. All processes require skill. The CEO delivery dependability also change. The stage
who spends time learning the processes by of the product life cycle affects product design
working on the plant floor will be well on the changes and the commonality of components—
way to being the head of a successful company. all of which have implications for the manufac-
It is equally important for those who run the turing processes and system. The first line of
processes to have a part in the decision-making Figure 1-14 shows that the manufacturing sys-
process on any issues concerning the factory tem is changed during the product’s life cycle.
processes and manufacturing systems. The product life-cycle concept provides a
framework for thinking about both the product
PRODUCT LIFE CYCLES evolution through time, and the types of mar-
ket segments likely to develop. The different
Products have a life cycle (birth, growth, designs of manufacturing systems reflect a
maturation, decline and death) and, therefore company’s ability to manufacture at various
they are dynamic and change with time. A volumes while decreasing the cost per unit over
company’s products may change, the volume of time. The general trend has been that product
product they produce may change, or the actu- life cycles are becoming shorter, leading to two
al process may change. There is relationship consequences. First, the time to develop new
between a product’s life cycle and the kind of products must be shorter since the old products
manufacturing system that produces it. Figure are being phased out quicker. Second, the man-
1-14 simplifies the life cycle into these steps: ufacturing system must be designed to be flex-
1. The startup step involves a new product or ible enough to deal with new products as they
company, low volume, and a small compa- emerge from the development process.
ny. The linked-cell approach has changed the
2. The rapid-growth step occurs when prod- classic life-cycle concept. Linked cells enable a
ucts become standardized and volume company to decrease cost per unit significantly,
increases rapidly. The company’s ability to while maintaining flexibility and, concurrently,
meet demand often stresses its capacity. making smooth transitions from low- to high-
3. The maturation step evolves when stan- volume manufacturing. That is, the same flexi-
ble system can accommodate large changes in
dard designs emerge. Process development
volume without needing to make major
is important here.
changes in the design of the company’s manu-
4. The commodity step includes the long life,
facturing system.
standard-of-the-industry product.
Toyota and Honda are examples of compa-
5. The decline-and-death step occurs when a
nies with years of experience in the develop-
product is slowly replaced by an improved
ment of linked cells. They have been able to
product and then dies. introduce new models (designs, styles) every
In Figure 1-14, the horizontal axis is time, three years, and both believe a two-year styling
usually in years. Preceding the startup step are change is within reach.
the initial design and prototyping steps that
may take more years. The maturation of a A NEW MANUFACTURING SYSTEM
product in the marketplace generally leads to Many countries have achieved an equal
fewer competitors, with competition based level of process and machine tool develop-
more on price and on-time delivery than ment. Much of today’s technology was devel-
el
od
Man Commodity
rm
ufac
Volume increasing
turin
pe
g
Saturation
e
um
De
ol
cli
v
ne
Cos
es
t pe
al
r un
s
it
al
nu
Time (years) An
Time
Traditional Job shop Production Production
manufacturing job shop job shop with flow lines More flow, mass produce
systems with some flow and assembly lines
Product
Form of Product quality, Price and quality with Price with consistent
competition characteristics cost, and reliability quality
availability
Figure 1-14. Traditional relationship between product life cycle and manufacturing system development/evolution.
oped in England, Germany, the United States, lean production system permits functional
and Japan. Now, Taiwan and Korea are also integration of critical subsystems into the
making great inroads into American markets, manufacturing system. In future years, this
particularly in the automotive and electronics new system based on linked cells will take its
industries. What many American managers place with the Taylor system of scientific
and engineers have failed to recognize is that management and the Ford system of mass
methodology is as powerful as technology. The production. The original working model for
Operations in the cell include metal cutting, heat seconds. (See the breakdown of time in Figure
treating, inspection, and assembly. Grinding and 2-2). The cycle time for the cell is 110 seconds.
superfinishing are also completed in the same
This can be found by:
cell. By grouping similar components into part
families, a group or set of processes can be col- 1
lected to manufacture parts families. Arrange- CT = (2-1)
PR
ment of machines in a cell is defined by the
sequence of manufacturing processes performed where:
on the parts.
Many companies begin with a pilot cell so that CT = cycle time, seconds
everyone can see how cells function. It requires PR = production rate
time and effort to train operators. In addition,
they may need time to adjust to cellular opera- Machines in a cell are usually single-cycle
tions. Selecting a product or group of products for automatics; they can complete the desired pro-
a cell is a simple, but risky, design method. cessing untended and turn off automatically
Cell operators should be involved in the after the processing cycle. If an operator comes
early stage of designing the cell if possible. to a machine that has just finished its produc-
Otherwise, they may not take ownership of the tion cycle, he or she will unload, check, and load
operation. The pilot operation shows everyone the next part into the machine. The operator
how cells operate and reduces the setup time then starts the machining cycle by hitting a
on each machine or process. Some cell process- walkaway switch before stepping to the next
es or machines may not be used 100% of the machine. The cell usually includes all of the
time. The machine-utilization rate may or may processing equipment and machines required
not improve in an archaic job shop. Process uti- for a completed part or subassembly.
lization is likely to be higher in a functional Figure 2-2 shows the average manual time
system where overproduction is allowed. and walking time. Times for the machining
Overproduction is considered a high form of cycle are given as machine time. The cell is
waste in the lean production philosophy. The designed so the machining time for any part in
objective in a manned linked-cell manufacturing the family on any machine in the cell is less
system is to utilize people and non-depreciable than the necessary cycle time. That is, machin-
assets fully, enlarging and enriching jobs, and ing time is less than cycle time. Thus, machin-
allowing workers to become multifunctional. ing times are uncoupled from cycle time.
Workers learn to operate many different kinds However, note that the machining time for the
of machines and perform many functions includ- third operation is greater than the necessary
ing quality control, machine tool maintenance, cycle time. That is, 180 seconds is greater than
setup reduction, and continuous improvement. 110 seconds. Therefore, this operation must be
In unmanned cells and systems, the application duplicated and the worker must alternate
of equipment is important. This is because the between the two lathes, visiting each lathe
most flexible and intelligent resource in the cell, every other trip around the cell.
the worker, is removed and replaced by a robot. This makes the average machining time:
Manned cells are designed in a U shape so
workers can move from machine to machine to 180
load and unload parts with the shortest walk- = 90 seconds
2
ing distance.
Figure 2-2 shows an example of a simple- Therefore, the turning process stays below the
manned cell. The cell has one worker who can required 110 seconds; that is, 90 seconds is less
make a walking loop around the cell in 110 than 110 seconds.
L2
L1 HT WT MT
Work Human Walk Machining
Name of operation time, time,
sequence time, sec
sec sec
3A 3B
2 Drill hole DP 15 5 20
DP 2
3A or 3B Turn, bore L1 or L2 13 5 5 180
4
Shaft
cell Mill flats HM2 12 20
4 5 5
13 5 30
1 5 Mill steps VM
5 VM 15 5
6 Final inspect
HM1
Totals 80 30 280
Parts 6
Finished Final
parts
Cart inspection
Another example of lean manufacturing con- in an assembly cell are usually manual. An oper-
version (Sekine 1990) shows the marked ator must stay at each workstation until a task is
improvement in productivity that cells can completed. As with manufacturing cells, long
achieve. In some instances, functional job-shop or setup times typical in flow lines must be vigor-
flow-line conversion is impractical. Therefore, ously attacked and reduced so flow lines can be
flow lines and flow-line conversions are used. In changed quickly from making one product to
Sekine’s example, it is shown that at the same another. This makes them flexible and compatible
time manufacturing cells are being designed, flow with the manufacturing cells and subassembly
lines using a conveyor are being reconfigured into and final assembly lines. Again, cells are designed
U-shaped cells (Figure 1-9). The conversion to manufacture specific groups or families of
allows these newly arrayed flow-line systems to parts. Hence, manufacturing and assembly cells
operate on a one-piece flow basis. The operations are the first choice. In the U-shaped layout that
was shown in Figure 1-9, Workers 2, 3, and 4 and processes. It is necessary only that no
cover multiple operations. Notice that Workers 3 machining time be greater than the required
and 4 share Operation 7. Workers 1 and 2 are cov- cycle time. The machining speeds and feeds can
ering Operations 1, 11, and 12 using the rabbit be relaxed to extend the tool life of cutting tools
chase (Suzaki 1987) method. and reduce wear and tear on machines, as long
The need to line balance is necessary in the as the machining time does not equal or exceed
flow line, but is eliminated in the cellular the cell’s cycle time.
design methodology. This is accomplished by Fixtures on cell machines are designed to hold
using standing and walking workers who are a family of parts, so rapid changeover from one
capable of performing multiple operations. part to another is possible within the parts fami-
Cells have many features that make them dif- ly. Fixturing is designed with error proofing, or
ferent from other manufacturing systems. Parts poka-yoke functions, in mind to prevent errors
move from machine to machine and one at a time and stop defective parts from getting into the sys-
within the cell. Small, highly controlled batches tem. Fixtures are designed for correct and easy
move from cell to cell, from cell to subassembly, loading and unloading and do not allow defective
or sometimes straight to final assembly. parts to be loaded.
For material processing, machines are typi- In some cells, decouplers are placed between
cally capable of completing a cycle initiated by processes, operations, or machines to provide
a worker. These machines are called single- flexibility, part transportation, inspection for
cycle automatics. defect prevention (poka-yoke), and quality con-
The U-shape layout puts the start and finish trol. The decoupler inspects a part for a critical
points of the cell next to each other. Every time dimension and feeds back adjustments to the
an operator completes a walking trip around machine to prevent it from making oversize
the cell, a part is completed. The cell is parts as the process cutter wears. A process-
designed so cycle time is equal to or slightly delay decoupler delays the part so it can cool,
less than that for final assembly. This is the heat, cure, or undergo whatever process is nec-
necessary cycle or takt time. Machining time essary for a time period greater than the cycle
for each machine is less than the time it takes time required by the cell. Decouplers and flexi-
for an operator to complete a walking trip ble fixtures are vital parts of both manned and
around the cell. Thus, the machining time can unmanned cells.
be altered without changing the production
schedule, as long as the machining time stays Step 2: Setup Reduction/Elimination
within the confines of the cell’s cycle time. When cells are formed to make a family of
The cell is designed to make parts as parts, the problems of process and machine
demanded by downstream processes and oper- changeover from one part to another must be
ations. There is no overproduction allowed—a addressed. Therefore, everyone on the plant
major form of manufacturing waste. floor must be taught how to reduce setup time
Overproduction results in the need to store using single-minute exchange-of-die tech-
parts, transport them to storage and retrieve niques. A setup-reduction team acts to facili-
when needed, account for them, purchase bas- tate the single-minute exchange-of-die process
kets for storage, and acquire forklifts. This for production workers and foremen. The team
requires people and costs money, but adds no demonstrates the methodology on a project;
value. In assembly cells, cycle time may be usually the plant’s worst setup problem.
throttled up or down by adding or removing Reducing setup time is critical to reducing lot
workers. size. Dr. Shigeo Shingo, father of the single-
In manufacturing cells, there is no need to minute exchange-of-die system, designed its
balance the machining time for the machines four stages to be sequential (Figure 2-3).
Last Remove
Load in First
Old Die Make
good "A" New Die Check good "B"
Parts Adjust Redo product
product Part
Last 1st
Good Internal Good
Cut = 90%
product product
A B
Figure 2-3. The four conceptual stages of single-minute exchange of dies (Shingo 1989).
The lean production approach to manufac- The similarities in part geometry and/or
turing means running small lots. This is impos- processes in the parts family allow setup time
sible if machine setups take hours to to be reduced or even eliminated. Initially,
accomplish. The economic order quantity setup time should be less than 10 minutes. As
(EOQ) formula determines the quantity to be the cell matures, the setup time is continually
manufactured to cost-justify a long and costly reduced. A final goal is to reduce setup time to
setup time. However, it is a faulty approach around 10–15 seconds, commonly called one-
that accepts long setup times as a given. Setup touch exchange of dies.
times can be reduced, often drastically. This Setup time for a cell should be reduced until
action results in a direct reduction in lot sizes. it is equal or less than the typical manufactur-
Successful setup reduction is easily achieved ing-cell cycle time of one or two minutes. This is
when approached from a methods-engineering usually accomplishable and provides signifi-
perspective. Much of the initial work in this cant initial reduction in lot size. The next goal
area has been done through time-and-motion is to reduce the setup time to less than manual
studies, as well as Shingo’s single-minute tech- time, or the time a worker needs to load,
niques for the rapid exchange of dies. Setup- unload, inspect, deburr, etc., during a process.
time reduction occurs in four stages. The initial After each production setup, defect-free prod-
stage determines what is being done for setup ucts should be made from the start. Ultimately,
in the operation. The current setup operation is the ideal condition is to eliminate setup; this is
usually videotaped and everyone reviews the called no touch exchange of dies.
tape to determine its elemental steps. A In summary, a savings in setup time decreas-
detailed list of the process is compiled from the es lot size and increases the frequency of lots
video. produced. The smaller the lot sizes, the lower the
The next stage is to separate setup activities
inventory a system is carrying, making for
on the list into two categories, internal and
shorter throughput time and improving quality.
external. Internal elements are those done only
when a machine is inoperative, while external Step 3: Integrate Quality Control
elements are done while a machine is opera-
tive. This elemental division usually shortens A multiprocess worker is capable of operating
lead time considerably. more than one kind of process. A multifunc-
The third and fourth stages focus on reduc- tional worker can perform tasks other than
ing internal time. The key is for workers to operating processes or machines. A multifunc-
learn to reduce setup time by applying simple tional worker is also an inspector who under-
single-minute exchange-of-die principles. If a stands process capability, quality control, and
company must wait for setup-reduction work- process improvement. In lean production, every
ers to examine every process, a lean manufac- worker has the responsibility and authority to
turing system never will be achieved. make a product right the first time and every
Therefore, it is absolutely necessary that cellu- time, and to stop an operation when something
lar design is at least concurrent with setup goes wrong. This line-stop authority for work-
reduction, if not ahead of it. ers is critical to the success of lean production
In the last stages, it may be necessary to in a factory. While the integration of quality
invest capital to lower setup time to less than a control into a manufacturing system markedly
minute. Intermediate jigs and fixtures and reduces defects and eliminates the need for
duplicate workholders represent the typical inspectors, cells provide a natural environment
kinds of hardware needed. In many cases, long for integration of quality control. The funda-
setup times can be reduced to less than 15 sec- mental idea is to inspect to prevent a defect
onds in relatively short order. from occurring, and never allow a defective
product to leave the manufacturing cell and get Through redesign, cells produce parts one at
into the system. a time, like assembly lines. This is called one-
When management and production workers piece flow. The process allows all production
trust each other, then and only then, is it pos- entities and internal customers to operate a
sible to implement an integrated quality-con- make-one, check-one, move-one-on operation.
trol program. The operator checks what a previous process
Japan started on the road to superior quali- produced to assure it is correct 100% of the
ty by the visits of Dr. W. Edwards Deming in time. Pull cords are installed on assembly lines
the late 1940s and early 1950s. The Japanese to stop them if anything goes wrong. For exam-
were desperate to learn about quality. The sta- ple, if production is going too fast (according to
tistical quality-control techniques offered by the quantity needed for the day), or if workers
Deming and others were readily accepted. find a safety hazard, then they are obligated to
Many Japanese believed that everyone in pull the cord and stop the line. Once a problem
America used statistical process-control tech- is discovered, it is rectified immediately.
niques. This, of course, was, and is not today, Meanwhile, other workers on the line who are
the case. However, the Japanese eagerly stopped go about other required tasks such as
accepted quality-control concepts, and then maintaining equipment, changing tools, sweep-
they did something not even being done by ing the floor, or practicing setups. A line does
American companies using statistical quality not move until a problem is solved. This is a
control. They taught factory workers quality fundamental difference between lean produc-
techniques and methods. The Japanese even tion and other manufacturing systems.
For manual work on assembly lines, a sys-
had a journal on the subject.
tem for tracking defective work is called an
When every worker is responsible for quality
andon. An andon is actually an electric light
and performs the seven basic tools of quality
board that hangs high above conveyor assem-
control (Figure 2-4), the number of inspectors
bly lines where everyone can see it. When
on a factory floor is markedly reduced.
everything goes according to plan, the board’s
Products failing to conform to specifications are lights are all green. When a worker on a line
immediately discovered because they are needs help, he or she can turn on a yellow light.
checked or used immediately. Nearby multifunctional workers who have fin-
Under the leadership of Vice President of ished their tasks within an allotted cycle time
Manufacturing Taiichi Ohno, a new idea took see the yellow signal and move to assist the
hold at Toyota. Ohno’s ideas on quality were worker having a problem. If a problem cannot
quite different in concept from the inspection be solved within the cycle time, a red light
philosophies of the time. First, every worker comes on and the line stops automatically until
was an inspector responsible for quality at it is solved. Music usually plays to let everyone
his/her workstation. Every worker on an know there is a line problem.
assembly line could stop the line if they found In most cases, red lights go off within 10 sec-
something wrong. Everyone’s attention was onds. The next cycle begins when a green light
focused on problems delaying production. comes on and all processes start at the same
Problems were resolved first and permanently, time. The name for this system is yo-i-don,
so a line was not stopped again by them. which literally means “ready, set, go.”
Inspection to prevent defects from occurring, The stages on an assembly line are synchro-
rather than inspecting to find defects after they nized. Such systems are built through team-
had occurred, became the mode of operation in work and a cooperative spirit among workers,
cells. Ultimately, the concept of autonomation fostered by a management philosophy based on
evolved (see Step 9). harmony and trust. Contrast this with the way
MT MT
Number of occurances
L
Frequency
C
FA Inspect
1 Unit
F C
SA
RM C
Measurements Elements or events
Maps the process Analyzes the data on individual parts Ranks highest to lowest
Scatter diagram Fishbone diagram
Manpower Methods
.. .
Variable A
.. . .
.
Cause Effect
. . .
. .
. ..
Materials Machines
Variable B
Shows correlation between Relates cause to effect
variables
Control chart Run chart
-USL
UCL
Measurements
_
X
LCL -LSL
Sample no.
Time
UCL= Upper Control Limit USL= upper specification limit
Figure 2-4. Seven tools of quality control used by workers to control the process.
things often operate in the functional job shop. should be put away, so it is ready to use
How long does it take to find a problem, con- the next time.
vince somebody that it is a problem, get it • Each worker must be responsible for the
solved, and the solution implemented? How cleanliness of the workplace and equip-
many defective parts are produced in the ment. The 5s strategy is adopted—sort,
meantime? Line shutdowns in lean factories straighten, shine, standardize, and sustain.
are encouraged to protect quality. Management Naturally, processes still need attention from
must have confidence and trust in individual experts in the maintenance department, just as
workers to give them line “stop” authority. an aircraft is taken out of service periodically
for engine overhaul and maintenance. This
Step 4: Integrate Preventive vital service must be carried out, but must not
Maintenance overtly impact production. One alternative is to
Making machines operate reliably begins switch to two eight-hour shifts separated by
with the installation of an integrated preven- two four-hour time blocks for machine mainte-
tive maintenance program that provides work- nance, tooling changes, restocking, long setups,
ers with training and the tools they need to overtime, early time, and other functions. This
maintain equipment properly. Excess process- is called the 8-4-8-4 scheme.
ing capacity obtained by reducing setup times Processes can decrease variability, but they
allows operators to reduce equipment speeds or must be reliable and dependable. Generally
feed, and to run at less-than-full capacity. said, smaller machines are simpler and easier
Reducing pressure on workers and processes to to maintain, and therefore, more reliable.
produce a given quantity fosters a drive to pro- Multiple copies of small machines add to the
duce perfect quality. flexibility of the overall system as well. The
Multifunctional operators are trained to per- linked-cell system permits certain machines in
form routine low-level process maintenance. cells to be slowed and, like the long-distance
Low-level tasks include adding lubricants, runner, to run longer and easier without a
checking for wear and tear, replacing damaged breakdown. Many observers of the Just-in-time
nuts and bolts, routinely changing and tighten- manufacturing system come away with the
ing belts and bolts, and listening for telltale feeling that machines are “babied.” In reality,
noises that signify pending failures. These and they are being run at the pace required to meet
other tasks can do wonders for machine tool takt time demand.
reliability, while giving workers a sense of Advanced lean manufacturers build and
pride in ownership. The maintenance depart- modify their manufacturing process technolo-
ment must instruct workers on performing gy. It is what makes them unique from their
these tasks and assist them with preparing competitors. In addition, they try to build
routine checklists for process maintenance. equipment in multiple copies so that there are
Workers are also responsible for keeping their multiple sets producing similar products.
areas of the factory clean and neat. Thus, addi- Suppose there is a manufacturing cell produc-
tional functions that are integrated into the ing racks for a rack-and-pinion steering-gear
lean manufacturing system are maintenance assembly. It makes six different racks for the
Honda Accord. In the same factory, a cell for
and housekeeping. ® ®
Toyota makes racks for the Camry and Avalon
Typically, the following housekeeping rules
models; the cell can make 10 different racks. The
are implemented:
two rack cells are similar in design. In the event
• There is a place for everything and every- of a machine failure, a machine from another
thing should be in its place. Everything cell could be borrowed. Also, processing capacity
Table 2-1. Mixed model final assembly line example that determines model cycle time
Subassembly cell
Figure 2-5. The arrival of an empty container at the manufacturing cell is the signal in the kanban system to produce more parts.
achieve the correct balance of inventory. Within Suppose there are 10 carts in the kanban link
a manufacturing cell, parts are already han- and each cart holds 20 parts. So, the maximum
dled one at a time, just as they are in assembly inventory in the frame area is 200 parts. The
lines. Therefore, cells should already be operat- foreman goes to the stock area outside the cell
ing with minimum inventory. Buffer inventory and picks up the kanban cards (one withdrawal
in cells is a signal there are problems that and one production-ordering), which takes one
full cart of parts out of the system. The (maxi-
should be addressed.
mum) inventory level is now nine carts, times 20
The level of work-in-process between the
parts per cart, or 180 parts. The foreman waits
stand-alone process, cells, subassembly, and until a problem appears. When it appears, the
final assembly is actually controlled by fore- foreman immediately restores the kanban cards,
men in various departments. Control is inte- which brings inventory to its previous level. The
grated and performed at the point of use. The cause of the problem may or may not be identi-
following is an example of how the methodolo- fied by the restoration of inventory, but the prob-
gy works for a frame area. lem condition is relaxed until a solution can be
Manufacturing manager
Manufacturing
cell
Final
assembly
Subassembly
Subassembly cell
ly
emb
ss
ba
Su
Kanban link
Figure 2-6. In the linked-cell manufacturing system, cells are linked with controllable inventory buffers called kanbans.
found. Then the cell team works at finding the each cart will hold only 10 parts, instead of the
root cause of the problem. Once the problem is former 20. Thus, up to 100 parts are in the kan-
solved, the foreman repeats this procedure. If no ban link. If everything works smoothly with the
other problems arise, then the foreman tries to new reduced work-in-process lot size, the fore-
reduce inventory to 8 multiplied by 20, which is man will then remove a cart to see what hap-
equal to 160 parts. This procedure is repeated pens. More than likely, some setup times will
daily in all kanban links in the factory. After a need to be reduced. In this way, inventory in the
few months, the foreman in the frame area may linked-cell system is continually reduced, expos-
be down to five carts, with 20 parts each. Over ing problems. Problems are solved one by one in
the next weekend, the system will be restored to all areas of the factory. Teams work on exposing
10 carts between the two points; but this time, and solving problems. This is the method used
Figure 2-7. In the linked-cell manufacturing system, cells are linked with controllable inventory buffers called Kanban links.
for continuous improvement in the lean produc- company’s work. Normally the issue is the pur-
tion factory. chasing department’s claim that pitting one ven-
The minimum level of inventory that can be dor against another will give the company a
achieved is a function of the quality level, prob- competitive advantage and lower-cost parts.
ability of a machine breakdown, length of the This may have a bit of truth in it, but the
setups, variability in the manual operation, method is shortsighted in the long run.
number of workers in the cell, parts shortages, The lean manufacturing system handles its
transportation distance, and other factors. It is vendor program differently. Just-in-time pur-
determined that the minimum number of carts chasing is a program of continual long-term
is three, and the minimum lot size is one. A improvements. The buyer and vendor work
very significant point is that inventory together to reduce lead times, lot sizes, and
becomes a controllable independent variable, inventory levels. Thus, both companies become
rather than an uncontrollable variable depend- more competitive in the world marketplace
ent on system-user demands. because of this teaming relationship.
In the lean production environment, longer-
Step 8: Integrate Suppliers
term (18–24 months), flexible contracts are
In lean production, one tries to reduce the drawn up at the outset, with three-or four-week
number of suppliers and cultivate a single delivery lead times. The buyer supplies updat-
source for each purchased component or sub- ed forecasts every month, which are good for 12
assembly. This is a primary tenet of lean pro- months, commits to long-term quality, and per-
duction. Vendors and suppliers are educated haps even promises to buy back excess materi-
and encouraged to develop their own lean pro- als. Exact delivery times are specified by
duction systems for superior quality, low cost, midmonth for the next month. Frequent com-
and rapid on-time delivery of a product to the munication between the buyer and vendor is
lean production factory. They must be able to
the norm. The kanban subsystem controls the
deliver perfect parts to their customers when
material movement between the vendor and
and where they are needed, and without incom-
buyer. The vendor is considered a remote cell.
ing inspection. The linked-cell network ulti-
Long-range forecasting encompasses six
mately should include every supplier. The goal
months to a year. As soon as the buyer sees a
is for suppliers to become remote cells in the
linked-cell manufacturing system. change, the vendor is informed. This knowledge
In the traditional job-shop environment, the gives the vendor better visibility, instead of a
purchasing department permits its vendors to limited lead-time view. The vendor gets build-
make weekly, monthly, and semiannual deliv- schedule stability and is not jerked up and
eries with long lead times (weeks and months down by the build schedule.
are common). A large safety stock is kept in The lean manufacturing buyer moves toward
case something goes wrong. Quantity vari- fewer vendors, often going to local, sole sources.
ances are large. This, coupled with normal Frequent visits are made to the vendor by the
late and early deliveries, can cause the vendor lean production buyer, who may supply engi-
supply situation to be chaotic. The situation neering aid (quality, automation, setup reduc-
leads to expediting, another form of manufac- tion, packaging, etc.) to help the vendor become
turing-system waste, since it does not add more knowledgeable on how to deliver on time,
value to the product. the right quantity of parts that require no
As a hedge against vendor problems in the incoming inspection. This is true technology
functional job-shop manufacturing system, mul- transfer. The vendors learn from the lean pro-
tiple sources are developed. Problems may occur duction customer. In this lean production sys-
because one vendor cannot handle all of the tem, the buyer and seller must build a bond of
trust, be willing to work together to solve prob- raise its quality to superior levels to penetrate
lems, and share cost savings. the world automotive market. He wanted every
The advantage of using a single source is worker to be personally responsible for the
that resources can be focused on selecting, quality of the piece part or product that he/she
developing, and monitoring one source, produced.
instead of many. When tooling dollars are con- The need for automation simply reflects the
centrated in one source, money is saved. The gradual transition of the factory from manual
higher volume should lead to lower costs. The to automated functions. Some people think of
vendor is now more inclined to go the extra this as computer-integrated manufacturing.
mile or exceed expectations for the buyer. The Others recognize that people are the most
buyer and vendor learn to trust one another, important and flexible resource in the company
cultivating the relationship over time. An and see the computer as just another tool in the
additional benefit to the lean production process, rather than the heart of the system.
buyer is that quality is more consistent and Hence, these companies are moving toward
easier to monitor when there is a single human-integrated manufacturing, where a cre-
source. ative, motivated work force is seen as the key to
Finally, there is the aspect of proprietary lean production.
processes. Toyota, Honda, and other users of
Quite often, inspection devices are placed in
lean manufacturing have published little about
machines or the factory uses source inspection
their manufacturing and assembly cells. Unique
to maintain quality. Other times, devices called
processes exist that give companies like Toyota
decouplers are placed between machines or
and Honda the edge in manufacturing strate-
gies. They develop machine tools and the processes so the inspection may be performed
processes in-house, rather than buying them automatically without stopping the process to
from a vendor. Lean manufacturers gain their do internal inspection. This attempt prevents
competitive edge by developing unique manu- the defect from occurring, rather than attempt-
facturing process technology in-house and keep- ing to inspect after a part is made. Inspection
ing it locked in the cells. by a machine can be faster, easier, and more
Chapter 4 expands this concept of in-house repeatable than inspection by a human. This is
development of machine tools with real exam- called in-process control inspection.
ples. Table 2-2 summarizes the key points in Autonomation means inspection becomes part
managing the lean production plant. This list of the production process. It does not involve
was obtained from the plant manager at a first- a separate location or person. Parts are 100%
tier supplier to Toyota. inspected by devices that either stop the
process if a defect is found or correct it before
Step 9: Autonomation the defect occurs. The latter requires an elec-
Autonomation is defined as the autonomous tronic feedback loop to the process controller.
control of quality and quantity. Generally, in A machine may be set to shut off automati-
the lean production factory this relates to stop- cally when a problem arises. This function
ping a line immediately when something goes can prevent mass production of defective
wrong. Specifically, this means workers control parts. The machine may also shut off auto-
the quality at the source, instead of using matically when the required volume parts
inspectors to find the problem. The workers in have been produced; this prevents overpro-
the lean factory inspect each other’s work. This duction—a major and expensive form of
method is called successive checking. Taiichi waste. The prevention of overproduction is a
Ohno, Toyota’s former vice president of manu- vital function of the lean production invento-
facturing, was convinced that Toyota had to ry-control methodology.
5S (seiri, seiton, seiketsu, seiso, shitsuke) The 5S strategy is used to sort, straighten, shine, standardize, and
(see Step 4) sustain the work environment.
Standard operation in The manufacturing cells are used to combine people and processes. The
manufacturing cells (see Step 1) components of standard operation include cycle time, work sequence,
and standard stock on hand in the cells.
Morning meeting A daily meeting is held for the purpose of sharing production and safety
information, quite often by a quality circle.
Key points: process sheets The process sheets, which are visually posted at each workstation, detail
the work sequence and most critical points for performing the tasks.
Tooling parts: changeover and setup Machine setup is performed when an assembly line changes tooling to
(see Step 2) accommodate a different product.
Seven tools of quality control (see Step 3) The seven tools to quality are Pareto diagrams, check sheets, histograms,
cause-and-effect diagrams, run charts for individuals, control charts for
samples, and scatter diagrams.
Production behavior Rules include information on personal safety, safety equipment, clothing,
restricted areas, vehicle safety, equipment safety, and housekeeping.
Visual management Each line in the plant has a complete set of charts, graphs, or other
devices, like andons, for reporting the status and progress of the area.
design, engineering, production planning and designs, readily adapt to product design
control, inventories and their control, purchas- changes, and cope with volume fluctuations.
ing, quality control, inspection, the production The critical control functions of production
workers, supervisors, middle managers, and and inventory are designed right into the man-
top management. Such a conversion cannot ufacturing system. Quality control and process
take place overnight and must be viewed as a reliability are tied to process technology and
long-term transformation. Reorganization can designed into the process.
be traumatic for the business arm of the com- Suppose a cell has six operations. The num-
pany and it usually has a negative impact on ber two machine may have a workholding
worker morale. device that checks to make sure that the first
However, Step 10 recognizes the company’s machine produced the correct part geometry
need to reorganize and adopt lean production before it performs its processing steps.
philosophies and methodologies. This effort Sometimes the checking occurs in the part-
often begins by forming product realization holding transporting device, or the decoupler,
teams to bring new products to the market- between the first and second machines. The
place faster. In the automotive industry, these decoupler may simply check a dimension, or it
groups are called platform teams. Platform may provide feedback to the first machine to
teams are composed of people from design, make process corrections. The worker is critical
engineering, manufacturing, marketing, sales, in this process and handles every part and
finance, etc. As the philosophy of team building checks every process.
spreads and the lean manufacturing system is
implemented and established, it is only natural
SUMMARY
that the production system follows suit.
Unfortunately, many companies are restructur- In manufacturing cells, workers are consid-
ing the business part of the company without ered to be the company’s most important non-
having completed Steps 1–8 to convert the depreciable resource, a point of view not
existing manufacturing system into a lean and traditionally held in United States’ factories
productive one. Reengineering the production where managers often think of labor as unstable
system, without simplifying and redesigning and a costly input to the manufacturing system.
the manufacturing system, can lead to very dif- Information systems, or computers, are com-
ficult times for the enterprise. mon on the factory floor in this modern age.
The design of manufacturing process tech- However, these information systems cannot
nology must be done early in the product devel- control the factory. They supply information for
opment process. Manufacturing process people to act upon in the decision-making
technology within cells must be part of a well- process. People are much more flexible than
designed, integrated system. Flexibility in the computers. Therefore, in the lean manufactur-
design of an integrated manufacturing system ing scheme, computers should be viewed as
means it can readily accept new product only a useful tool in the process.
• Green manufacturing strategies continue trol of a system as well; for without control,
to help protect the earth from further there is no optimization.
pollution. Each company is unique, due to differences
in subsystem combinations, people, product
These trends require the following types of
designs, and materials. Different interactions
manufacturing system responses:
within social, political, and business environ-
• Continuous improvement, or continuous ments also individualize each company’s man-
redesign and improvement of manufactur- ufacturing, production, or service systems. In
ing systems, is an ongoing goal. short, each company has its own set of prob-
• Systems must produce superior products lems. Clearly, there is a danger in grouping
with reduced per-unit costs and on-time companies on a functional basis or calling them
delivery in response to customer demands. all job shops. However, the functional job shop
• All manufacturing costs can be reduced. is the most common design.
• Systems must be designed to offer flexibil-
ity and reliability. Flexibility refers to the EVOLUTION OF FUNCTIONAL STRUCTURE
ability to change rapidly with customers’ People invented and developed basic machine
demands, volume, product mix, and design tools in the first industrial revolution. With
changes. Reliability refers to systems’ these tools came the first levels of mechaniza-
robustness; that is, the ability to produce tion and automation. Factories developed along
high-quality products without break- with manufacturing processes. These factories
downs or quality problems. focused resources (materials, workers, and
• Redesigned systems capturing system pro- processes) at sites where power was available.
cessing into manufacturing cells need For the most part, waterpower was used,
product design engineers with design explaining why early factories were placed near
envelopes that are readily visible, under- streams. Water turned waterwheels that drove
standable, and adaptable to new product overhead shafts running the length of factories.
designs. The simpler, lean production sys- Belts from the main shafts powered machines
tem design is easily understood, con- on manufacturing process lines.
trolled, and automated. Grouping like machines that operated at
approximately the same speeds was logical and
OPTIMIZATION OF THE expedient. Factories were laid out functionally,
according to types of machines used. Machines
MANUFACTURING SYSTEM
were extensions of human capabilities or attrib-
In general, a manufacturing system should utes. Machinists developed skills that were dif-
be an integrated whole. The system should be ferent from leather, iron, or foundry workers.
composed of integrated subsystems, each inter- Therefore, processes were divided according to
acting with the whole system. A system should the kinds of skills that were needed to operate
have a number of objectives that optimize the particular processes.
whole process, as opposed to refining only por- When steam engines and, later, electric
tions of a system, such as certain processes or motors, replaced other types of power, early
subsystems. manufacturing systems achieved greater flexi-
A system’s daily operation requires informa- bility. Still, the functional arrangement persist-
tion gathering, communication, and decision- ed and became known as the job shop. As
making processes to be integrated. Answering product complexity increased and the factory
questions such as: what, when, where, and how grew larger, separate functional departments
are critical to achieving a healthy company. evolved for product design, accounting, and mar-
These questions are related to the overall con- keting. Later in the scientific management era
of F.W. Taylor and Frank and Lillian Gilbreth, • A system’s behavior in response to excita-
departments for production planning, work tions or disturbances from the environ-
scheduling, and methods improvements were ment must be predictable through system
added to help control the manufacturing system. parameters.
These functions were initially developed for the
In general, models describe how a system
job shop and were performed external to the fac-
works or behaves. Mathematical models for con-
tory floor. Taylor and Frank and Lillian Gilbreth
trol purposes generally require a theory or set of
are generally recognized as founders of the
equations describing a system’s boundaries and
industrial engineering profession. Scientific
behaviors through input parameters. In short, if
management evolved to manage the factory, and
no theory exists, the model is not viable or
systems for production control were developed.
robust and a system is uncontrollable. This is
why 3D/virtual-reality simulation is such a
SYSTEMS DEFINED great tool for factory design and analysis. The
The word system abstractly defines a relative- use of high-level simulation gives managers and
ly complex assembly (or arrangement) of physi- engineers greater control. They visualize situa-
cal elements characterized by measurable tions and then take immediate action on the fac-
parameters (Rubinstein 1995). This definition is tory floor.
quite appropriate for manufacturing systems Manufacturing systems are complex, thus dif-
(Figure 3-1). The important physical elements for ficult to model. Therefore, their design, analysis,
all manufacturing systems are people, processes, and control are difficult processes as well. There
and material holding and handling equipment. are several reasons for this difficulty:
Raw materials and products are inputs, in-
• System size and complexity inhibit model-
process materials, and outputs of the system.
ing and control, due to implied-time
Some more common measurable parameters for
expenditures.
manufacturing systems are throughput time and
• Systems are dynamic and unstable. The
cycle time. Cycle time is the reciprocal of the pro-
environment can change the system and
duction rate. If a system can make one vehicle
vice versa.
per minute, the cycle time is equal to one
• Relationships may be awkward to express
minute/vehicle. Throughput time reflects the
in analytical terms, and interactions may
time that a product spends moving through the
be nonlinear. Thus, well-behaved functions
factory, while having value added to it via pro-
often do not apply.
cessing. Work in process reflects the volume of in-
• Data or information may be difficult to
process inventory in the factory. The greater the
secure, inaccurate, conflicting, missing, or
volume of work in progress, the longer the
too abundant to digest.
throughput time. Cycle time, throughput time,
• Objectives may be difficult to define, par-
and work-in-progress parameters for the manu-
ticularly in systems that have an impact
facturing system are very different from those
on social and political issues. Goals may
used for individual machines. These parameters
conflict.
are productivity measures during the implemen-
• The act of observing and trying to control a
tation stages of lean manufacturing. An effective
system changes the behavior of the system.
manufacturing system satisfies customers, both
• Analysis and control algorithms for systems
internal and external. Conflicts between these
can be subject to errors of omission and com-
two groups of customers must be resolved.
mission. Some errors will be related to
To model and control the system:
breakdowns or delays in feedback elements.
• A system’s boundaries or constraints must This is because manufacturing systems
be defined. include people in information loops.
Disturbances to
manufacturing system
External customer
Demand
Physical elements are: Service to
Social, political Machine tools, tools and tooling, customer
environment material handling equipment,
and people (internal customers)
Defectives and scrap
Information to
system on demand
Because of these difficulties, the technique of some other process. These outputs are called
digital simulation is not widely used for model- produced goods.
ing and analysis of manufacturing systems. The production system services the manu-
However, more companies are simulating por- facturing system. An analogy would be the
tions of their factories. human heart, representing the manufacturing
system. The manufacturing system (the heart)
MANUFACTURING VERSUS pumps blood (the material flow). The produc-
tion system supplies oxygen and food and
PRODUCTION SYSTEMS checks blood pressure, volume (work in
In general terms, a manufacturing system process), and pulse rate (cycle time). Material-
inputs materials that can be measured by control functions are critical to performance of
machines and people. Materials are processed the manufacturing system. A linked-cell manu-
and gain in value. Manufacturing system out- facturing system permits, even invites, integra-
puts may be either consumer goods or inputs to tion of critical control functions. Continuing
with the heart analogy, these critical control The Job Shop
functions include: The most common manufacturing system
• quality control (no blood pathogens or worldwide is the job shop. It is characterized by
clogged arteries restricting material flow); large varieties of components, general-purpose
• production control (good blood pressure or machines, and a functional layout (Figure 3-4).
movement in the system); The job shop’s distinguishing feature is that it
• inventory control (optimum blood supply can manufacture a wide variety of products as a
or amount of work in process); and result of manufacturing small, often one-of-a-
• machine tool reliability (no blood clots and kind, lot sizes. Job shop manufacturing is com-
material flowing well). mon for specific customer orders; but in truth,
Management cannot fully control all inputs many job shops produce material to fill finished-
into the manufacturing system. Manipulating goods inventories. Because job shops must per-
controllable inputs or the system itself must form a wide variety of manufacturing processes,
counteract the effect of disturbances. For exam- general-purpose-manufacturing equipment is
ple, controlling material-availability problems required. Workers must have relatively high
such as shortages or demand fluctuations are skill levels to perform the range of different
both difficult tasks. National and international work assignments. Job shop products include
economic climates cause shifts in the business space vehicles, aircraft equipment, machines,
environment that can seriously change any of tools, all types of equipment, and a long list of
these inputs. All manufacturing systems differ other products. Distribution of the total factory
in structure or physical arrangement. In addi- capacity for a job shop is shown in Figure 3-5.
tion, all manufacturing systems are serviced by The 6%-production fraction is equivalent to the
a production system. Because the oldest and 36% value shown earlier in Figure 1-13b, except
most common manufacturing system is the that the calculation in Figure 3-5 assumes a
functionally organized job shop, most produc- theoretical 100% capacity that equals 365 days
tion systems are also functionally organized multiplied by 24 hours per day. Though depress-
into departments. Walls usually separate peo- ing sounding, these figures clearly demonstrate
ple in functional areas from other areas. Thus, that a productivity problem exists in making a
communication breakdowns are common. Also, manufacturing system (not individual process-
long lags in feedback loops result in additional es) more productive. An obvious conclusion is
manufacturing system problems. that in order to make the manufacturing system
more productive, one must redesign it.
CLASSIFICATIONS OF In the job shop, machine tools are function-
MANUFACTURING SYSTEM DESIGNS ally grouped according to the general type of
Five manufacturing system designs can be manufacturing process: lathes in one depart-
identified: the job shop, the project shop, con- ment, drill presses in another, plastic molding
tinuous process the flow shop, and the linked- in still another, etc. This layout’s advantage is
cell manufacturing system. The continuous its ability to produce a wide variety of products.
process design deals primarily with liquids, Each different product or component requires
powders, or gases (such as an oil refinery), its own unique sequence of operations so it can
rather than discrete parts. The other four man- be routed through required departments in the
ufacturing systems manufacture discrete proper order. Routing sheets are used to control
parts. Figure 3-2 shows the four traditional movement of material through the factory.
systems and Figure 3-3 shows the linked-cell Most of the time forklifts and handcarts are
manufacturing system. used to move materials between machines.
Milling Drilling
Lathe dept. dept.
dept.
Sub- Grinding
assembly dept.
Receiving &
shipping dept.
Flow shop
Continuous process
Raw materials Energy
Workstations
Raw
materials
Process I
Gas By-products
Oil
Finished
product
Figure 3-2. Schematic layouts of four out of five classical manufacturing systems.
If a company grows, the job shop often duced at regular or irregular intervals. The
evolves into a production job shop. The produc- purpose of batch production is to satisfy con-
tion job shop is basically a large job shop with tinuous customer demands for a product.
some flow lines and computer numerical con- A job shop system’s production rate can often
trol machines. The production job shop manu- exceed the customer demand rate. The job shop
facturing system builds large volumes of builds an inventory of Product A, and then
products, but still builds in lots or batches, usu- changes to Product B to fill other orders.
ally medium-sized lots of 50-200 units. An item Changing products involves breaking down
may be produced only once, or it may be pro- setups on many machines for Product A, and
= Manufacturing cell
k = Final or
subassembly
Final
(Sub A) cells
assembly
Figure 3-3. The lean shop uses a linked-cell manufacturing system design.
resetting them for Product B. When stock for Industrial equipment, furniture, textbooks,
the first item is depleted, machines are set up and piece parts for many assembled consumer
again for Product A and inventory for Product products (household appliances, lawn mowers,
A is replenished. etc.) are made in production job shops. The pro-
History has shown that manufacturing duction job shop system is sometimes called a
equipment can be designed for higher produc- machine shop, foundry, plastic molding factory,
tion rates. For example, automatic lathes capa- or pressworking shop. It is estimated that as
ble of holding many cutting tools and many as 75% of all piece-part manufacturing is
automatically loading a new piece of stock are in lot sizes of 50 pieces or fewer. This estima-
now used, rather than older, manual engine tion indicates that production job shops are an
lathes. Machine tools are often equipped with important segment of today’s total manufactur-
specially designed workholding devices called ing. Along with job and flow shops, the use of
jigs and fixtures, which increase process-output the production job shop is common in the
rates, precision, accuracy, and repeatability. United States.
Manufacturing manager
Mill Drill
Injection
Seat assembly molding Lathe
department
Subassembly
Assembly line Assembly line
Grinding
Receiving
and
shipping
Figure 3-4. Factory layout for functional design referred to as the job shop.
6% total productive
fraction, or
525.6 hrs/yr,
100% used, or about
2 hrs/day.
Figure 3-5. Distribution of total factory capacity of machine tools in a job shop (courtesy Charlie Carter).
ing industry, ships, locomotives, large-machine system, making all components for the large,
tools, and large airplanes are examples of proj- complex project; thus, it has a functional-pro-
ect shops. The number of end items produced is duction system.
usually not very large, but lot sizes of compo-
nent parts going into the end item can vary Continuous Process
from small to large. The job shop usually sup-
plies parts and subassemblies to the project The continuous process system means that
shop in small lots. the product physically flows. Oil refineries,
Fixed-position manufacturing, a variant of chemical processing plants, and food processing
the project shop, is used for construction jobs operations are examples of this system, which
such as buildings and roads, where the prod- is sometimes called continuous flow production
uct is large and/or the site confined. when referring to the manufacture of either
Therefore, construction equipment and man- high volumes of parts or of a single part.
power go to it. When a job is complete, con- Beverage-canning operations or assembled
struction workers and equipment are removed products such as televisions are not continuous
from the manufacturing site. The project shop processes, but high-volume flow lines for dis-
invariably has a job/flow shop manufacturing crete parts or assemblies. In continuous
Workstations
Raw materials
Finished
products
Figure 3-6. Examples of the flow shop include the assembly line and the transfer line.
rial-handling devices (conveyors, moving belts, Since the advent of mass production, various
transfer devices, etc.). Items move through approaches and techniques have led to develop-
operations one at a time. ment of machine tools that are highly effective
All flow lines have the same inherent prob- in large-scale manufacturing. Their effective-
lem of line balancing. Total tasks required to ness is closely related to product-design stan-
assemble (or machine) a unit are divided into dardization and the length of time permitted
smaller tasks. This is the division-of-labor con- between design changes. A machine producing a
cept. Tasks are arranged so that the time a com- part with a minimal amount of skilled labor can
ponent or assembly spends at each station or be developed if the part can be highly standard-
location is fixed and equal (thus balanced). Line ized. A part can then be manufactured in large
balancing means that the amount of work at quantities. An automatic screw machine (a com-
each station is approximately the same to plex lathe) is a good example of a machine for
reduce idle time at the station. Lines are the manufacture of small parts. An automated-
designed and set up to operate at the fastest transfer machine that makes V-8 engine blocks
possible speed, regardless of a system’s needs. (at a rate of 100 per hour) is an example of a
Flow shop layouts are typically either continu- super machine for mass production of large
ous or interrupted. A continuous flow line basi- parts. Figure 3-6 is an example of a transfer
cally produces one complex item in great machine. Transfer machines are specialized,
quantity and nothing else. A transfer line pro- expensive to design and build, and usually not
ducing an engine block is a typical example. An capable of making another product. These
interrupted flow line manufactures large lots, machines must be operated for long periods of
time to spread the cost of the initial investment-
but periodically changes over to run similar
typically $20-40 million over many units.
(but different) components. Changing over a
Although highly efficient, they exclusively
complex flow line may take hours or even days.
make products in large volume. Desired design
In a flow line manufacturing systems, facili-
changes in a product must be avoided or
ties are arranged according to a product’s
delayed, because it would be too costly to scrap
sequence of operation. A line is organized by the the machines. Such systems are clearly not flex-
processing sequence needed to make a single or ible enough for product or process-design
regular mix of products. A hybrid form of the changes. Smaller versions of transfer machines
flow line produces a batch of products moving developed for smaller-sized products made in
through clusters of workstations or processes large volumes are shown in Figure 3-7. They
organized by product flow. This is called batch range in cost from $2-3 million.
flow. Garment or apparel manufacturing is tra- The development of the numerical-control
ditionally done this way. For instance, a batch of machine tool in the late 1950s and early 1960s
shirts moves through a sequence of different permitted programmable control of the position
sewing operations. Usually, setup times to of cutting tools in relationship to the work-
change from one product to another are long, piece. By the late 1960s, automatic tool chang-
and often the process is complicated. ers had been added to the numerical-control
Most factories are mixtures of job shop and machine, marking the birth of the machining
flow line systems. Demand for products can center. Computers and workpiece changers
precipitate a shift from batch to high-volume were added next. Today, the computer-numeri-
production, and much of production is guided by cal-control machine tool is readily available to
that steady demand. Subassembly and final- all manufacturers (Figure 3-8).
assembly lines are further extensions of the Products manufactured to meet demands of
flow line, with the former usually being more the free economy and today’s mass-consump-
labor-intensive. tion markets must include changes in design
Drum type
Dial-table type
Cutting tool
Controller
Load
Workpiece
Unload Unload
Inline type
er
chang
le t ex
Pal
Figure 3-7. Examples of transfer machines. Figure 3-8. Computer-numerical-control machining center with
many different operations performed in a single setup.
for improved product performance, as well as scheduled for each part it produces. It can be as
style. Therefore, hard-automation systems complex as the job shop it replaces.
must be as flexible as possible, while retaining The modern computer-numerical-control
the ability to mass produce material. The machine can be programmed to automatically
recognition of this fact in the late 1960s lead to change tools, workpieces, and cutting parame-
a coupling of the transfer line with the numer- ters. It seems logical that the versatile machine
ical-control machine and the flexible manufac- be joined with the transfer line to expand man-
turing system was born. ufactured-part variety. Most flexible manufac-
The primary components of the flexible man- turing system installations have a system
ufacturing system are computer-numerical- manager supervising, primarily by monitoring
the system. The system manager is responsible
control machine tools, material-handling
for supervising workers, including material
systems, cutting tools, machine-workholding
handlers who perform loading/unloading tasks,
devices (pallets), and computer-control net-
a roving operator who presets tools and reacts
works. Today, machine tools are computer to unscheduled machine stops, and a mechani-
numerical control, usually horizontal or verti- cal/hydraulic technician who repairs transfer
cal spindle-milling machines. The flexible man- devices, machines, workholders, and pallet
ufacturing-system design (Figure 3-9) has changers.
eight four-axis, computer-numerical-control, Much has been written about flexible manu-
machining centers, each equipped with a 90- facturing systems and research continues. The
cutter-tool magazine and a parts-pallet-chang- systems are expensive to design and often
ing system. The system designers claim to be require months or years to implement. They
capable of making more than 500 different are complex to program, analyze, and control.
parts. Such a system must be programmed and By the end of the 1990s, it would appear that
Inspection
Tools Machine
T1 H1 H2 H3 H4 T3
Cart Cart
1 2
T2 Load/Unload station T4
(parts bin) (parts bin)
SAC
Host
computer
Figure 3-9. Design of flexible manufacturing system used at Detroit Diesel Allison in Indianapolis, Indiana.
fewer than 1,000 such systems existed world- Linked-cell Manufacturing System
wide. In the United States, most of these sys- The linked-cell manufacturing system is the
tems are found in large companies that can newest manufacturing system. It is composed of
afford large capital outlays or that receive gov- manufacturing and assembly cells linked by a
ernmental backing, such as military defense pull system for production control. In cells, oper-
contracts. The flexible manufacturing system ations and processes are grouped according to
represents the super-machine philosophy at its the manufacturing sequence needed to make a
ultimate. Fundamentally, it is an attempt to group of products or a product family. This
blend the job shop’s flexibility with productivity. design looks similar to a flow shop design, but is
Parts usually require two or three passes created for flexibility. The manufacturing cell
through a flexible manufacturing system. often is configured in a U-shape, allowing work-
Fixturing in the flexible manufacturing system ers to move from machine to machine with the
is costly and complex. The system’s control com- shortest walking distance while loading and
puter must control the conveyor, maintain the unloading parts. Figure 3-10 shows an example
computer numerical control library of programs, of a simple manned-manufacturing cell.
download these to the machines, handle sched- Machines in a cell are usually single-cycle auto-
uling of the system, track tool maintenance, matics, so they can complete a machining cycle
track performance of the system, and print man- untended, and have it turn off automatically
agement reports. Not surprisingly, the system’s when it is finished with a processing cycle. The
software development often proves to be a major cell usually includes processing needed for a fin-
limiting factor. ished part. Because machines in a cell are sin-
Almost no work in
process within the cell
Machines—small,
inexpensive,
Machine 3
dedicated,
single-cycle
M
ac
hin
e4
Machine 2
La
st
pr
oc
One part at es
s
a time and
multiple-machine
Finished goods
handling
One part
between Path of worker
machines
Out
Ma
chin
e1
Raw material In
Firs
t pr
oce
ss
Figure 3-10. Small manned cell with four machines and one walking worker.
gle-cycle automatics, they can be making chips machine), and starts the machine by hitting the
while an operator is doing other tasks in the cell. walkaway switch. The analogy here is similar to
When an operator comes to a machine to unload what one does when preparing breakfast. One
a part, the machine completes a portion of the walks to the toaster with bread, loads the bread
component and stops with its safety door open. in, and hits a lever to start the toasting process.
The operator unloads and checks the part, loads One does not stand there and watch, but walks
a new part (which has just been unloaded and away to get orange juice or make coffee. The
brought by the worker from the previous toaster produces a signal once it completes its
toasting cycle. When unloading the toaster, one systems are composed of directly or indirectly
checks to see if the bread has toasted properly. connected manufacturing cells. They utilize
These rules are the basic operational rules for physical inventory and information control
manned-manufacturing cells, or the make-one, kanban for the linking process. Savvy manu-
check-one, move-one-on methodology. facturing managers know they must examine
the job shop system for redesign to improve
STANDARDIZED WORK productivity. Manufacturing companies are
Standardized work is a methodology for converting batch-oriented job shops into linked
maintaining productivity, quality, and safety cells. One way to form a cell is by using group
at high levels. It is a consistent framework for technology. Most lean companies, however, do
performing work at designated takt times. not use this proven technology, preferring to
Standardized work illuminates opportunities simply form cells based on product families.
for making improvements in work procedures. There are two basic types of cells that feed final
There are three elements in structuring assembly: manufacturing cells, where most
standardized work: processes and machines are single-cycle auto-
matics (i.e., complete the processing cycle
• takt time—the cycle time of final assembly; untended); and assembly cells, where most, if
• working sequence—dictates cell design; not all, operations require an operator to be
and present to do tasks.
• standard in-cell stock—called stock-on-hand.
Takt time, as described earlier, should GROUP TECHNOLOGY
reflect the production rate of the final assem- Most group technology methods often ignore
bly line, which, in turn, reflects the pace of the worker and simply find machines that will
marketplace sales. process groups of parts. Group technology is a
The working sequence is a series of steps that philosophy of grouping similar parts into part
are the best ways to carry out a task. Standard families. Parts of similar size and geometry can
in-process stock is the minimum number of work- often be processed by a single set of processes.
pieces needed to maintain a smooth flow of work. A part family based on manufacturing-process
Standardized work provides detailed, step-by- type would have the same sequence of manu-
step guidelines for every job in the Toyota pro- facturing processes. The set of processes is
duction system. Team leaders and their teams arrayed to form a cell. Thus, with group tech-
determine the most productive work sequence nology, job shops can be systematically restruc-
and make continuing improvements in that tured into production cells, with each cell
sequence. Kaizen, a form of continuous improve- specializing in a particular family of parts. The
ment, is useful to produce new patterns of stan- job shop shown in Figure 3-4 is redesigned into
dardized work. Figure 3-11 illustrates an three cells to accommodate products that pre-
example of a lean manufacturing cell where the viously were manufactured in a job shop
takt time is 51.4 seconds (the work sequence and (Figure 3-12). Notice that no new machines are
stock on hand are indicated). The work in the cell needed. Cellular conversion gives a marketing
in Figure 3-11 is designed to have a cycle time department an opportunity to find a new prod-
that is slightly less than the manufacturing sys- uct line for excess processes and floor space
tem takt time. freed by cellular conversions. The machines
have at least the same utilization as in the job
LINKED-CELL SYSTEMS shop, but products spend far less time getting
Manufacturing cells are key building blocks through processes. Parts are handled less,
in linked-cell manufacturing systems. These machine-setup time is shorter, in-process
Pallet removal
Assembly
Rotator
Reverser
bench
ZK- AM- ZY- AM- AM- ZK- TS- TS-
6468 1288 1417 828 1313 6522 2190 2182
1 2 3 4 5
6 7
12 11 10 9 8
Aluminum cover
Aluminum cover
Intermediate
ZY- ZY- ZY- ZY- ZY- ZK- ZK- ZK- ZK-
3412 3413 2413 3410 3409 6482 5325 1431 3408
11 51.4 49
Figure 3-11. Standardized work chart for a lean manufacturing cell (courtesy Toyota Motor Company).
inventory is lower, throughput time is greatly • Unique technology is often developed in-
reduced, and workers are more effective. house.
The key points for manufacturing cells are:
Cells are typically manned, but unmanned
• Machines are arranged in a process cells are beginning to emerge with robots
sequence. replacing workers. A robotic-cell design is
• The cell is designed in a U-shape. shown in Figure 3-13 with one robot and three
• One part at a time is made within a cell CNC machines. For the cell to operate
(one-piece flow). autonomously, machines must have adaptive-
• Workers attend to more than one process. control capabilities and use decouplers.
• Takt time dictates cell-production rate. Cells are designed to manufacture specific
• Workers perform their jobs while standing groups or families of parts. Cells are linked
and walking. either to each other or subassembly points by a
• Slower, smaller, dedicated, and less-expen- pull system of material control called kanban.
sive machines are used. Kanban is the primary production and
Figure 3-12. Classical job shop requiring system-level conversions redesigned into manufacturing cells. Chapter 5 will include
additional details about Cell 3.
inventory control system for linked cells. setup times typical of flow lines must be vigor-
Production control is one of the critical con- ously attacked. It is imperative that flow lines
trol functions in manufacturing system be able to change over quickly from the manu-
designs. facture of one product line to another. The need
In addition to manufacturing cells, flow-shop to line-balance a flow line every time it changes
elements within a factory are redesigned to to another part must be eliminated or drastical-
make them operate like cells. To do this, long ly reduced. This can be accomplished with
decouplers. Both cells and flow lines make piece queues were viewed as necessities that per-
parts for subassembly and final-assembly lines. mitted downstream operations to continue,
It is important for flow-line behavior to mimic when there was actually a problem with the
manufacturing cells to optimize system flexibil- upstream or feeding operations. The linked-
ity and productivity. cell manufacturing system approach recog-
nizes the job shop design as the fundamental
DESIGNED FOR FLEXIBILITY problem. Better management by constraints
Flexibility is a premier design feature for only results in small improvements in produc-
lean production and cellular manufacturing tivity and quality in the job shop. It is the
systems. Once implemented, cellular systems manufacturing system that must be
can react quickly to changes in customer redesigned.
demand, product design, or product mix. Cells Figure 3-14 summarizes manufacturing
in a factory are linked directly to subassembly systems by comparing different methodologies
cells or lines. Cells and flow lines are on final based on production rates and flexibility, or
assembly lines at the point of use, or they are the amount of different parts the system can
connected by kanban links. This makes a fac- handle. Project shops and continuous process
tory product oriented and flexible in: systems are not shown. Cells provide a wide
middle ground between job shops and dedicat-
• its operation of equipment, ed mass-production flow lines. Figure 3-14 is
• changeover, derived from factory-floor data. It shows that
• process, and the widely publicized flexible manufacturing
• capacity or volume. system lies between the job shop and the
To effectively operate equipment in a lean transfer line. This is expected, since the flexi-
environment, rapid tool change is a must. ble manufacturing system was developed from
There should not be any adjustments, and a merger of these two systems.
automatic-error detection must take place. Like in a job shop, it is necessary to sched-
Changeover must be easy, with the speed of ule parts and machines within a flexible man-
the setup and exchange of tooling and dies ufacturing system. This function, however,
smooth and rapid. makes flexible designs difficult to link to other
Design criteria in process consideration manufacturing systems. The flexible system
looks at differences in operations and process- often becomes an island of automation within
es for different parts. It also looks at the abil- a job shop—a characteristic of super
ity to handle a different mix, a different order machines. Inventory accumulates in the flexi-
in the mix, and a different volume in the mix. ble manufacturing system.
In a flexible, lean environment, the ability In contrast, manufacturing cells produce
to increase or decrease production output, parts one at a time in a flexible manner. Cell
rate, and volume is significant. capacity and cycle time can be quickly altered
to respond to customer demands for change.
For manufacturing cells, the cycle time does
COMPARING LEAN PRODUCTION not depend on machining time.
TO OTHER SYSTEMS Families of parts with similar designs, flexi-
Table 3-1 provides a brief comparison of ble-workholding devices, and tool changes in
lean manufacturing and the job shop philoso- programmable machines allow rapid change-
phy. A popular idea a few years ago was to find over from one component to another. Rapid
the functional job shop system’s bottlenecks or changeover means quick or one-touch setup,
constraints and work to eliminate them. Part often like flipping a light switch. Significant
of this misguided idea was that material inventory reduction between cells is possible
Micro computer
Computer 2
1
Robot
mill
ute e
mp pindl
r
ll
Mi
CNC
C al s
tic
Horizontal
o
ver
axis
LAN
Final
3 Inspection
Robot and cell
Out
controller
In
computer
Micro computer
Key:
=Decoupler
Part flow
with rapid changeover and when the inventory machining center is similar to a cell consisting of
level can be directly controlled. The operator one machine. However, a machining center is
controls quality in a cell. Equipment is main- not a cell. It is not as flexible or productive as a
tained routinely by the worker. real cell, which uses multiple, simple machines.
For robotic (automated) cells, robots typically A machining center cannot be considered a cell
load and unload parts using up to five computer because there is no overlapping of machining
numerical-control machine tools, but this num- times. Cellular layouts facilitate the integration
ber may be increased if robots become mobile. A of critical production functions, while maintain-
Table 3-1
How lean manufacturing philosophy differs from that of a typical US Company
Factors Lean Manufacturing Typical job shop / flow shop
Inventory Inventory is wasteful. It hides problems. It is a liability. Inventory is a necessary asset. It protects against fore-
Every effort must be extended to minimize inventory. cast errors, machine problems, late vendor deliveries.
More inventory is "safer" and necessary.
Lot sizes Keep reducing lot sizes. The smallest quantity is desired Formulas. Keep revising the optimum lot size with
for both manufactured and purchased parts. some formula based on the trade-off between the cost
of carrying inventories and the setup costs.
Setups Eliminate/reduce them by extremely rapid changeover Low priority. Maximum output is the usual goal.
to minimize the impact. Fast changeover permits small Rarely does similar thought and effort go into achiev-
lot sizes and allows a wide variety of parts to be made ing quick changeover. Use EOQ to determine lot size.
frequently.
Vendors Procure from a single source. Vendors are remote cells, Adversaries. Multiple sources are the rule, and it is
part of the team. Daily, multiple deliveries of active typical to play suppliers against each other to get
items are expected. The vendor takes care of the needs lower costs but multiple vendors increase the variability
of the customer, and the customer treats the vendor as in the components.
an extension of the factory.
Quality Zero defects. If quality is not perfect, then improve- It costs money to make high quality products. Tolerate
ments can be made. Continuous improvement in peo- some scrap. Track what the actual scrap has been
ple and process is the goal. and develop formula for predicting it. Plan extra
quantity to cover scrap losses.
Equipment Constant and effective. Machine breakdown and tool As required. Not critical because inventory is available.
maintenance failure are eliminated or reduced by routine maintenance.
Lead times Keep them short. This simplifies the job of marketing, The longer the better. Most foremen and purchasing
purchasing, and manufacturing as it reduces the need agents want more lead time, not less.
for expediting.
Workers The internal customers are the experts. Changes are Engineers provide ideas and are the experts.
not made until consensus is reaches. Employee involve- Management is by edict. New systems are installed in
ment is critical, especially in the design of the cells. spite of the workers, not thanks to the workers.
Managers are coaches who serve workers in teams. Measurements are used to determine whether or not
workers are doing as directed.
Cost reduction Cost reduction comes by non-stop like water through the Cost reduction comes by driving labor out of the prod-
pipe type manufacturing, thus reducing the TPT. uct and by having high machine utilization.
Production Material should be "pulled" through the factory, using Material should be coordinated by MRP and "pushed"
control kanban. out into the factory.
Overhead Any function that does not directly add value to the Overhead functions are essential.
product is waste.
Accounting's Labor is a fixed cost. The internal customers are one of Labor is a variable cost.
view of labor the system's primary resources.
Equipment Machines are distance runners, slow but steady and Machines are sprinters, and pulled hamstrings are to
maintenance always ready to run. be expected.
Automation Autonomation is valued because it facilitates consistent Automation is valued because it drives labor out of the
quality and prevents overproduction. product.
Expediting Expediting is a manufacturing sin. Expediting is a way of life.
Cleanliness Housekeeping is everyone's responsibility. Work means getting your hands dirty.
Evaluation Multiple performance criteria based on cost, quality, on- Evaluation is based on quantified direct cost.
(measurable time delivery, and flexibility.
parameters)
1000
Transfer lines
Manned and
unmanned cells
100
PR
Production Cells
rates
Parts
per hour 10
CNC or
Job shop
NC stand No families
Low volume
alone of parts,
stand-alone
1
machines
Table 3-2
How eight manufacturing functions are controlled
Functions Categories L-CMS Job shop/flow shop
How many to make per day Families of products Leveling the manufacturing Production plan-orders plus
system forecast
What mix of products to Finished goods for
make each day make-to-stock, customer orders Master production schedule Master production schedule
for make-to-stock
Getting materials required to Components - both Pull system WLK cards Push system - material
the right place at the right manufactured and purchased requirement planning (MRP)
time
Capacity of the system Output for key work centers Controlled by number of Capacity requirement
and vendors workers planning (CRP)
Executing capacity plans Producing enough output to Meet downstream needs Input/output controls,
satisfy plans route sheets
Executing material plans - Working on right priorities in POK cards-pull system Dispatching reports, route sheets
manufactured items factory
Executing material plans - Bringing in right items from Kanban cards and unofficial Purchasing reports, invoices
purchased items vendors at the right time orders
Feedback information What cannot be executed due Immediate and automatic Anticipated delay reports
to problems
Constraints
• Customer
needs
• Expectations
• Specifications
• Bounds
• Laws
* With regard to manufacturing systems design, there are two customers of manufacturing system: the
external customer who buys the products the system produces and the internal customers who build the products.
mapped to the physical domain; and design satisfy the requirement for true axioms is
parameters are mapped to the process domain assessed by trial and error. Axioms are further
(in terms of process variables). analyzed, redefined, and refined, until the
Design axioms have two fundamental char- process converges on a set of comprehensive
acteristics:
axioms. Here are two axioms, stated as direc-
• They cannot be proven. tives rather than observations, applied to the
• They are general truths; no violations or design of manufacturing processes/systems,
contrasting examples can be observed. rather than products:
These characteristics suggest the use of a
• Axiom 1 is the independence axiom. It
heuristic approach to develop the axioms. In
this approach, an initial set of axioms is estab- maintains the independence of functional
lished and published. Axioms are then subject- requirements.
ed to trial and evaluation in manufacturing • Axiom 2 is the information axiom. It mini-
situations. The extent that hypothetical axioms mizes the information content of a design.
ZIG
FR11 FR12
FR21 FR22 FR23 ZAG DP11 DP12
ZAG
ZIG
ZIG ZAG
• How can the store and its elements be • Should carts be stored in the same place?
designed to satisfy the functional require-
A typical supermarket design can be exam-
ments?
ined in terms of two axioms. The first axiom
• How should the store be laid out (arrange-
states that functional requirements must not
ment of shelves, aisles, and cases, and
be coupled by the proposed design. This is the
stocking shelves with goods)?
case with the three functional requirements,
• How should the shopping carts and check-
which are independent. What about the design
out lines be designed to make the store
of a typical supermarket? Are functional
profitable, usable, and desirable for exter-
requirements coupled by the typical design
nal and internal customers?
suggested? The placement of the milk and
Designers have no control over local product bread maximizes the walking distance for the
pricing or daily staffing, that is, how designs customer in the store, and therefore maximizes
are actually used or misused. The same is true time spent in the store. Thus, according to the
for products. Designers cannot control product first axiom, the typical supermarket design,
misuse by customers. In summary, designers which is not designed by functional require-
have no control over how design solutions are ments, may not be rational, since the two func-
implemented. tional requirements are coupled by the
Suppose a designer elects to design a store so customer action of taking the longest path to
customers have to walk the maximum distance, get the milk and bread. The designer for the
the full length of the store, to buy items most typical supermarket uses another set of func-
often purchased, such as milk and bread. The tional requirements, so that the suggested
design may require that these items be placed design is not coupled. If the two functional
at opposite corners in the back of the store to requirements are not coupled by a design, the
force customers to walk by and purchase the design is an uncoupled one. Clearly, some func-
maximum number of products. The entrances tional requirements reflect what is desired in
and exits are at the same location, in the front the layout or what the design needs to achieve.
of the store. Trucks bringing new stock are The relationship between the functional
unloaded at the rear of the store. requirements and measurable design parame-
The following questions should be asked: ters is established by the design. Design
• Does the design violate the functional parameters represent the plan to achieve func-
requirements? tional requirements through specific system
• Why did the designer place the produce designs. For the supermarket, measurable
section in the front right corner, near the parameters may be throughput time and
entrance of the store? Is this layout money spent per trip. This income can be tied
because produce is one of the most perish- to profit through cost. Thus, an equation can be
able foods and the designer wanted to written that ties the functional requirements
maximize exposure to it to increase to the design parameter(s).
turnover? Throughput time, for example, is a measura-
ble design parameter for the second functional
What if the designer installs multiple paral- requirement. It is the sum of: walking time to
lel checkout lines, automatic laser scanners, get from the vehicle to the store, shopping time
and shopping carts that are easily unloaded to to collect the groceries, check-out time, and
minimize the time customers spend checking return time to the car.
out? The following questions should be asked:
• How many checkout lines and carts should Σmin (WT + ST + CT + RT) = Tm (4-4)
be in the store?
• the internal customer, or the user of the could have dreamed that any Japanese com-
Honda manufacturing system; and pany would become an international manu-
• the external customer, or the buyer/user of facturing leader, a phoenix rising from the
the Honda product. ashes of war?
The factors given in Table 4-1 are important
At Honda, the ultimate manufacturing engi-
for external customer satisfaction.
neering goal is to implement the necessary
technologies to enable manufacturing products
that satisfy the external customer, while recog- Table 4-1
nizing internal customer needs. Honda’s man- Factors to achieve External Customer Satisfaction
agement does not imitate the manufacturing
technologies of competitive companies. It does Factor Manufacturing Engineering Requirements
not believe it can make an exceptional product
using the technology of other manufacturers. Attractiveness New Technology
Flexibility
Therefore, Honda builds almost 100% of its Agility to meet changes
manufacturing systems around in-house-built
machines and equipment, thus arraying its Quality High accuracy
manufacturing and assembly lines with unique Durability
process technology. Reliability
Today, Honda’s manufacturing engineering
Price / Cost High operating efficiency
group is responsible for in-house development,
Low investment / low first cost
including the design and manufacture of: Constant improvement and innovation
• 100% of body welding systems, excluding
Delivery Mixed model production
conveyors; Rapid model startup
• 100% of dies and molds used for stamping
skin panels, major engine castings, instru-
ment panels, and bumper fascia; and
• 100% of main machining lines for engine The factors (functional requirements) need-
and transmission parts such as gears, ed to satisfy internal customers, or the people
who use the equipment, are given in Table 4-2.
cylinder blocks, heads, etc.
It shows how conflicts between two customers’
In addition to two plants in Marysville priorities can occur on any factor. If all factors
where the Accord® model and certain motorcy- are equally weighted, manufacturing costs will
cles are built, Honda also built a new plant in rapidly get out of hand. Priorities must be
Ohio to produce Civic® automobiles and anoth- established. This means that some factors can-
er to produce engines and transmissions. Its not be fully implemented into every system.
manufacturing engineering group provides Fortunately, these factors are not of equal
70% of new paint application equipment and importance.
50% of automated assembly equipment for The biggest dilemmas for a manufacturing
these new plants. system are cost and quality. Regardless of how
Today, the words “global” and “customer attractive products may be, or how first-class
satisfaction” are clichés; everyone is looking manufacturing systems are, products must sell
for ways to become more productive and effi- at reasonable prices, while maintaining superi-
cient. However, at the time Honda was for- or quality. Auto manufacturers’ key priorities
mulating its company principles, industries for achieving external customer satisfaction
in Japan were struggling to survive amidst are minimizing initial costs, providing a car
the ruins of World War II. At the time, who that offers a safe, comfortable ride, and produc-
Minimizing equipment size typically reduces This reduces the number of different machines
cost and space required for the facility. producing the same item and moves the com-
Flexibility within these principal factory ele- pany toward lean production. Ultimately, nec-
ments has implications beyond versatility and essary daily quantities can be manufactured
agility. It means that a system can adapt to within a single-serial system with one-piece-
changes in external customer demand in terms flow. Duplicate machines are eventually elimi-
of volume (vehicles/month) or mix of products nated to reduce product variability.
(more two-door and fewer four-door autos). Honda’s manufacturing engineers think a
Controllability refers to the critical control manufacturing system should have the follow-
functions of production, inventory, quality con- ing characteristics:
trol, and machine reliability. In other words, a
system is designed to produce the right product • highest speed at each step (fastest
at the right time, yet at a reasonable price. This throughput time);
begins the first step in the methodology out- • a minimum number of steps in each
lined in the next section. process;
• one step, one machine (one-piece flow);
HONDA’S UNIQUENESS REQUIREMENT • multiple functions at one step, if no inter-
action; and
Honda develops manufacturing equipment,
• minimum equipment size.
processes, and systems by placing priority on
factors that affect the external customer. The
Process and System Design
company adapts its manufacturing system
design to meet the highest-priority require- Figure 4-3a shows a typical welding machine
ments of its internal customers. Therefore, when designed to hold four body panels (two sides,
Honda’s managers and engineers plan a manu- and one floor and roof) for spot welding. Honda
facturing system, they establish (in conjunction engineers shrank the spot-welding process
with the manufacturing department) the neces- down to a small robot arm (Figure 4-3b). They
sary criteria for each factor. Design and imple- mounted over 70 of these programmable arms
mentation teams set priorities and work to on large box-like structures, which hold and
achieve those priorities by any means. By fol- simultaneously tack-weld the sheet metal for
lowing this procedure, Honda tries to achieve an Accord automobile body. Every Accord man-
the best possible control of quality, inventory, ufactured in the U.S. is welded on this one
manufacturing, and cost, while solving discrep- computer numerical control (CNC) machine.
ancies between customer requirements. This ensures that every Accord body is identi-
The uniqueness approach requires new cal. The machine can be changed from two- to
methods and processes that are free from con- four-door bodies in less than one minute. In the
ventional thinking. Hence, Honda engineers 1980s, Honda was making its Accord and Civic
and production workers struggle to develop models on the same line. These were different-
manufacturing systems that meet the criteria sized vehicles, so the same computer numerical
and follow function and design requirements. control welding box could not accommodate
The goal is to develop the shortest, most com- both cars. A solution was found by designing
pact, and one-piece productive flow manufac- and building two auto-welding boxes. The
turing system. As a system matures and Accord box rapidly slid off the line before the
inventory is continuously removed from the Civic box slid on. Accord or Civic models could
links, the system becomes increasingly com- be welded using the same technology. In paint
pact. As processes are improved and quality rooms, the same technology was introduced to
goes from good to superior, the production rate interchange paint booths. Minimizing the num-
at each step can be throttled up as required. ber of steps and equipment sizes reduced costs
Figure 4-3a. The general welding machine holds the roof, bottom, and side panels in the correct location for spot welding in the
body shop at Honda, Marysville, OH.
and the required floor space. This is a typical DESIGN PROCESS HIERARCHY
example of where less part handling reduced The discussion of supermarket design was
quality problems. restricted to the most important, first-order
Another example of unique process technolo- requirements of the system. The designer did
gy is found in a machine that integrates the not worry, for example, about the specific meth-
processes of boring and honing (Figure 4-4). ods used by customers to traverse the store or
These processes usually occur sequentially. where items were stored on shelves. This is
Combining them into one tool eliminates the always the case in design endeavors. That is,
variables caused by the fixturing of parts functional requirements and corresponding
between two processes, as well as wasted trans- physical solutions can be decomposed and pri-
fer time. Honda has also developed a CNC oritized. It is fortunate that there is a hierar-
gear-grinding machine that incorporates cubic chy in design, in both functional and physical
boron nitride grinding wheels and induction- domains. It is fortunate as well that functional
heat-treatment. This one-step, serial-gear- requirements can be decomposed. Because of
grinding machine is five times faster than this hierarchy, only a limited set of functional
conventional methods. requirements need to be concurrently consid-
Figure 4-3b. Spot welding at Honda went from (a) manual, to (b) large robots, to (c) small computer numerical control arms.
functional requirements at each level of the is, all functional requirements must be inde-
hierarchical tree by eliminating secondary fac- pendent from one another.
tors from consideration. Uninformed designers In summary:
often consider all functional requirements of all
• There is a hierarchy in both the functional
levels simultaneously, rather than making use
and physical domains.
of their hierarchical nature. With this approach,
• There is correspondence between each level
every design problem will appear to be too com-
of the functional and physical hierarchies.
plex and formidable to solve.
• To decompose functional requirements, a
FUNCTIONAL REQUIREMENTS physical solution for each element at each
level of the hierarchy must be conceived.
As noted, perceived needs for a product or
That is, the complete FR hierarchical tree
process must be reduced to a set of independent
cannot be established by conceiving physi-
functional requirements. Functional require-
cal solutions at every corresponding level of
ments and independence have specific mean-
the physical hierarchy. (Cochran et al 2000)
ings in the context of design axioms. Functional
requirements are independent; that is, each CONSTRAINTS
functional requirement is independent of all
other functional requirements. And, as such, Constraints, in the context of axiomatic
each can be stated without consideration of any design, are defined as required limitations of
other functional requirement. acceptable solutions. They may be classified as
Functional requirements are what a design- either input constraints, which limit design
er wishes to achieve through a design. Design specifications, and/or system constraints, which
parameters state how a designer hopes to limit the manufacturing system. Input con-
achieve them. Therefore, the best and most straints usually are expressed as limitations on
impartial method of defining functional size, weight, materials, and cost, whereas sys-
requirements is in a solution-neutral environ- tem constraints are based on the capacity of
ment. That is, functional requirements are machines, available skills for manufacturing,
defined without any preconceived physical and even the laws of nature. A typical design
solution in mind. Otherwise, functional constraint for a supermarket would be a store’s
requirements may simply reflect a designer’s size (square footage) or the number of checkout
bias or the attributes of an existing design. lines. An input constraint might be the number
An acceptable set of functional requirements of pushcarts or maximum speed of electric
is not necessarily unique. For example, func- carts in the store.
tional requirements for a supermarket are not By definition, a constraint is different from a
unique to a supermarket. Also, another design- functional requirement. A constraint does not
er may choose a different set of functional have to be independent of functional require-
requirements, depending upon his or her judg- ments and other design constraints. All function-
ment of perceived needs for the supermarket, al requirements are independent of one another.
such as the need to eat. A designer is free to In the supermarket example, profitability, acces-
choose any arbitrary set of functional require- sibility, and variability, such as selection, are
ments as long as they are consistent with per- independent functional requirements. Therefore,
ceived needs. The physical solution will be the requirement that a functional requirement
different with a different set of functional be satisfied should not in any manner compro-
requirements. However, the set of functional mise or affect the functional requirement. Time,
requirements must be consistent and minimal, a constraint, may be affected by any change in
in the sense that they are not redundant; that functional requirements.
system is better than another because it makes reduce information content. They reduce it fur-
more than is necessary, is incorrect. A design ther if the design permits generous tolerances.
should fulfill precise needs defined by function- Corollary 5. This is the use of symmetrical
al requirements—no more or less. In a lean shapes and/or arrangements consistent with
production system, the right amount is avail- the functional requirements and constraints.
able at the proven place at the required time. For example, right- and left-handed parts can
Overproduction is not an advantage. Similarly, be made together before separating them.
a process or system that fulfills more functions Symmetrical parts require less information.
than specified will be more difficult to operate They are typically easier to manufacture and
and maintain than one that meets only the orient in assembly. Not only should a shape be
stated functional requirements. Reliability also symmetrical wherever possible, hole locations
may decrease when a machine or system ful- and other features should minimize the infor-
fills more functional requirements than the mation required during manufacture and use.
increased complexity required. Symmetrical parts promote symmetry in the
Corollary 3. This is the process of integrat- manufacturing process.
ing design features into a single, physical Corollary 6. This is the use of the largest-
process, device, or system when functional allowable tolerance for products and their com-
requirements can be independently satisfied in ponent parts when specifying functional
the proposed solution. The design should move requirements. Using the largest possible toler-
toward the idea of a one-step process, or one ances and finishes on parts reduces cost.
machine with multiple operations at each step. Tolerances on surface roughness and dimen-
In the lean production scheme, be cautious sions play an important role in the final
when considering multiple (serial) operations achievement of a simple manufacturing sys-
on one machine. The number of physical tem. Reducing tolerances increases costs and
processes should be reduced through integra- the difficulty of manufacturing a product. More
tion of parts, without coupling functional information is required to produce parts with
requirements. However, mere physical integra- tight tolerances. On the other hand, if a toler-
tion is not desirable if it increases information ance is too large (that is, too loose), errors in
content or couples functional requirements. assembly accumulate, and other functional
Good examples of physical integration consis- requirements cannot be satisfied. Therefore,
tent with Corollary 3 are found in the Honda tolerance specification should be made as large
plant that developed the boring-honing as possible, but should remain consistent with
machine and shortened its engine transfer line the likelihood of producing functionally accept-
from 16 to eight steps. able parts. The correct tolerance band mini-
Corollary 4. This is the use of standardized mizes overall information content. When the
or interchangeable processes and operations tolerance band is too narrow, information con-
when they are consistent with functional tent increases, since subsequent manufactur-
requirements and constraints. Use standard ing processes require more information. Excess
parts, methods, operations, and routes to tolerances reduce reliability and, thus, increase
reduce inventory and minimize information the need for maintenance, which increases
required for material routing, manufacture, information content.
and assembly. Special parts should be mini- Corollary 7. An uncoupled design requires
mized to decrease inventory costs and simplify less information than coupled designs when
inventory management, as per Corollary 3. satisfying a set of functional requirements. In
Interchangeable parts allow for inventory lean manufacturing, most of inspection is car-
reduction, as well as simplification of manufac- ried out manually by operators when parts are
turing and service operations; that is, they produced in cells. In such a situation, the num-
ber of critical dimensions should be minimized. ufacturing system design. The linked-cell man-
This corollary is a consequence of Axioms 1 and ufacturing system is designed to automate the
2. Its implication is that if a designer proposes routine aspects of decision-making where pos-
an uncoupled design that has more information sible and leave individuals free to solve prob-
content than a coupled design, the designer lems to improve the system. To summarize the
should return to the drawing board to develop big picture:
another uncoupled or decoupled design with • Design the manufacturing system with the
less information content. The linked-cell manu- characteristics of flexibility, uniqueness,
facturing system represents a system that can efficiency (cost), and controllability (quali-
operate with minimal paperwork and informa- ty, materials, movement, and machines).
• Integrate critical control functions into the
tion. It is the model for a paperless factory with
system including quality control, produc-
a future.
tion control, inventory control, and
machine-tool reliability.
SUMMARY • Integrate vendors (the supply chain).
To get better, faster, and more competitive, a • Design products that can be made in this
company needs to continually improve its man- system (reverse concurrent engineering).
Frame manager
Foreman Foreman
subassembly subassembly
Foreman
and frame area
assembly
Manufacturing cells
Housing cell
Shaft cell
Subassembly cell
Final
assembly
In
Out
In
Kanban link
Figure 5-1. The linked-cell manufacturing system is product oriented. The cells are linked (see dashed line) to the subassembly cells
or final assembly by kanban.
VM
HM
X W3 X
X VM
X
L
W4
W2
X G
L
X
Final
X inspection
W1
S X Finished
part cart
In
Raw Out
material
cart
Key:
S = Saw
L = Lathe Paths of workers moving in the cell
HM = Horizontal milling machine
Material movement paths within cell
VM = Vertical milling machine
G = Grinder
X = Worker position Active decoupler
= Decoupler
Figure 5-2. Manned manufacturing cell for a family of parts with seven machines operated by four workers.
Figure 5-4. Process-planning sheet for pinion 101A as produced in the job shop.
VM
HM
X X
VM
X
L X
X
G
L
X
X Final
inspection
X Finished
S
part cart
In Raw Out
material
cart
Key:
S = Saw Path(s) of worker(s)
moving in cell
L = Lathe
HM = Horizontal milling machine Material movement paths
within cell
VM = Vertical milling machine
Active decoupler
G = Grinder
X = Worker positions Inactive decoupler
Figure 5-5. (a) Manned manufacturing cell, with a family of parts produced on seven machines operated by one worker.
VM
HM
X X
X VM
Worker 1
X
L
X G
L
X Worker 2
X Final
inspection
S X Finished
part cart
In
Raw Out
material
cart
Key:
S = Saw Path(s) of worker(s)
L = Lathe moving in cell
HM = Horizontal milling machine Material movement paths
VM = Vertical milling machine within cell
G = Grinder
Active decoupler
X = Worker Positions
Inactive decoupler
= Inactive decoupler
Figure 5-5. (b) Manned manufacturing cell, two workers connected by decouplers. Shaded decoupler is an active kanban square.
VM
HM
X X
Worke VM
r3 X
L
X
Worker 2 X
G
L
X
X Final
inspection
Worker 1
X Finished
S
part cart
Drop off
In Raw finished part Out
material Pick up raw
cart material
Key:
S = Saw Path(s) of worker(s)
L = Lathe moving in cell
HM = Horizontal milling machine Material movement paths
within cell
VM = Vertical milling machine
Kanban square
G = Grinder
(Decoupler)
X = Worker positions
Active decoupler
Table 5–1
List of operations and respective times for manufacturing cell
Machine Tool Operation or Process Processing Time (minutes) Operator manual time (minutes) Walk time (minutes)
Saw Cut bar to length 0.30 0.25 0.05
Lathe 1 Rough turn 0.40 0.25 0.05
Lathe 2 Finish turn 0.40 0.25 0.05
Horizontal mill Mill step 0.45 0.25 0.05
Vertical mill 1 Mill slot 0.45 0.25 0.05
Vertical mill 2 Drill and tap holes 0.30 0.25 0.05
Surface grinder Grind slot 0.45 0.25 0.05
Final inspect Final inspection 0.25 0.25 0.05
The approximate throughput time for a part What is the throughput time for the three-
moving through a cell (using one worker) can worker cell?
be calculated using Equation 5-3: TT = 54 seconds/part × 12 = 648 sec. per part
TT = 144 seconds/part × 8 transfers = 1,152 or 10.8 minutes/part.
seconds/part or 19.2 minutes Suppose the machining time on the grinder
What is the cycle time for the cell manned by must be increased to improve the surface fin-
two workers? Assume for the calculation that ish. The new machining time is 50 seconds
Worker 1 leaves a part in Decoupler 1, for What is the impact of increasing the processing
Worker 2, who leaves a part in Decoupler 2 for time on the cycle time for the cell with one, two,
Worker 1 and the workers take the path or three workers? There is no effect on cycle
sketched in Figure 5-5(b). time, because all processing times are less than
So, the cell cycle time for two workers is: cycle times. What is the impact on throughput
CT = (0.25 minutes × 4) + (3 seconds × 4) = 1 time for a cell? There is no effect on the
minute, 12 seconds/part = 72 seconds/part throughput time, because the processing times
Number of parts produced are all less than the cycle time.
3, 600 seconds/ hour Suppose the manufacturing system needs
= = 50 parts/hour more output from the cell. What should be rec-
72 seconds/part
ommended? As currently designed, what is the
The throughput time for the cell manned by maximum hourly output from the cell, regard-
two workers is: less of the number of workers; that is, what
TT = 72 seconds/part × 10 transfers = 720 constrains output? Should the addition of a
seconds or 12 minutes (approximately) fourth worker be recommended? With four
Note that two extra transfers are added for workers, each worker would do two operations
the two decouplers. and the cycle time would be about 36 sec-
The cycle time for the cell manned by three onds/part. Note: 0.45 60 < 36 sec/part.
workers is: A plot of the outputs (for one to four workers)
CT = (0.25 minutes × 3) + (3 seconds × 3) = 54 should be linear (Figure 5-6). The concept of
seconds/part flexibility in manufacturing-system design is
Number of parts produced based on the ability to throttle up or down by
3, 600 seconds/ hour the addition or subtraction of workers, with no
= = 67 parts/hour changes to the actual process times required.
54 seconds/part
cell. Decouplers can also provide functional Suppose there is a worker at every station
quality control and process delay. A process- and the cell is operating at maximum capacity,
delay decoupler would delay part movement to but there is a serious problem. Balancing work
allow the part to cool, heat, cure, etc., for longer at each station so each station-task time is
than the required cycle time. Except for process equal is typically next to impossible to accom-
delay, the decoupler does not act as a buffer. It plish. Lean-production cells are designed to be
holds only one part, completely processed to operated at less-than-full-capacity. If there are
that point in the cell sequence. two workers in a cell (Figure 5-7[b]), removal of
material from the kanban square at Station 2
WORK-IN-PROCESS VERSUS by Worker 2 (who is responsible for Stations 3
through 6) is a signal for the worker at Station
STOCK-ON-HAND
3 to attach the fan wire and install the fan. At
In Figure 5-5(b), parts arrive at the input this time, the worker at Station 3 removes
side of the cell in a cart. A different cart con- material from the kanban square at Station 2,
nects the output side of the cell to the place signaling Station 2 to install the printed circuit
where parts are to be used. Carts in the two- board. If the kanban square at the station has
card-kanban system between cells are part of material in it, then the worker must wait until
the withdrawal-kanban links. it is removed before beginning the next assem-
Downstream cells and final-assembly lines bly. In this scheme, the operation times for
are protected against problems in upstream Stations 3 through 6 are balanced with those in
cells by the presence of inventory in withdraw- Stations 1, 2, 7, and 8.
al-kanban links between cells. Material in cells Kanban squares are examples of a visual pro-
is called stock-on-hand. Material between cells duction-control system that is fast and accurate.
is work-in-process inventory. Workers can easily see the signal and (when a
Decouplers and fixtures that accommodate a part has been removed) quickly react to make
variety of parts make flow lines more flexible one to replace the one that has been removed.
and compatible with cells. Both cells and flow The U-shaped-subassembly cell becomes part of
lines make piece parts and subassemblies for the make-one, check-one, and move-one-on, seri-
subassembly and final-assembly lines. al-flow manufacturing system.
Suppose a company tries to operate at peak
HOW AN ASSEMBLY CELL WORKS capacity using eight operators. The manual
work at each station would need to be reconfig-
The primary difference between manufac- ured to require the same amount of average
turing and assembly cells is that the processes time for each operator. This is called line bal-
and operations in assembly cells are usually ancing. Any variance would cause delays and
entirely manual. That is, workers must stay at disruptions in throughput times and produc-
stations for the duration of a task. Workers tion rates. Extensive redesign of the station or
cannot simply load parts into a machine, start its hardware and methods would be required to
the machine, and have it complete metal-cut- achieve good line balance any time demand
ting or heat-treating operations while the changes. This flow-line design greatly reduces
worker moves elsewhere in the cell. flexibility.
The U-shaped-assembly cell shown in Figure A lean manufacturing cell design can be used
5-7(a) has eight stations and can be operated by to eliminate line balancing in the normal flow-
one to four workers, depending on the required line sense. Suppose a cell has only two workers.
output for the cell. Kanban squares in the sys- One worker performs the first, second, seventh,
tem pull material through the cell. These and eighth operations by walking from one sta-
squares are production-control decouplers. tion to the next. The other worker does the third,
Station 4 Station 5
Direction Station 7
Station 2
of material and Test power cord,
single operator attach label
Install printed circuit board
Station 1 Station 8
Package, add
Place disk drives count at video
into bottoms display terminal
"Problem" display
Agenda easel Pallet of
cartons
Figure 5-7. (a) Assembly cell for disk drives was designed by workers at Hewlett-Packard Greeley Division.
#4 #5
#6
#3
Worker 2
#7
#2
Worker 1
#8
#1
fourth, fifth, and sixth operations. The two shad- responsible for three tasks, and the other, five
ed decouplers illustrated in Figure 5-7(b) link the tasks, to make the loop times balance. The fore-
two workers. Only the total time it takes for man and workers can do this rebalancing quick-
workers to make half loops must be balanced. ly and almost naturally.
Obviously any two decouplers can be used to Alternatively, two workers could simply fol-
achieve this objective. Thus, one worker is low each other around the cell, with roughly
1 Assembly using
1 2 3 4 5 conveyors
Polarity Solder Solder Attach Set 3 One day's in-process
check repair transistor frame press jig inventory
Out
IN
1 2 3 Attach
Eliminate Parts
4 and
conveyor raw Polarity Solder Measure adjust 5
materials At
check repair resistor t
fra ach
me
9 Attach 8 7
12 11
Finished lid
External
parts cart Package
inspect
Main
Attach measure Caulk
10 label
shift
Output, units/shaft 700 1,056
In-process inventory, units 750 8
operators
Personnel, number of operations 10 8
Daily output per person, units 70 132
Cycle time, minutes 0.60 0.43
Figure 5-9. Example of a conveyor assembly line redesigned as a U-shaped cell (Sekine 1990).
machine-processing times. Cycle time can be on each trip through the cell. Duplicated
altered by adding or removing workers, or in machines must do identical work. This may be
some cases, fractions of a worker’s time when two checked with downstream decouplers, which
cells are close and a worker may be assigned to inspect output from twin machines to ensure
both cells. output is the same.
In the cellular-manufacturing system there
is no need to balance machining time for CYCLE TIME FOR A CELL
machines. It is only necessary that machining
BASED ON TAKT TIME
time is no greater than the required cycle time.
Since machining time for turn-bore is greater Requirements of the manufacturing system’s
than cycle time (180 110), the process is takt time dictate the cycle time for a cell. Like
duplicated and the worker alternates machines all other cells in the plant, the shaft cell is
V = cutting speed, in./min (mm/min) First, to be flexible, the process (or manufac-
T = tool life, minutes turing subsystem) must be able to handle all
n and C = empirical constants product-design changes. Engineering-design
changes are routine facts of life. This applies
Typical values for the exponent, n, are 0.14- equally to new product designs. The latter is
0.40, so a modest reduction in speed results in called concurrent design and reflects the abili-
a large increase in tool life. Also, this relaxation ty of a company to bring new products to the
in cutting speed greatly reduces tool wear, market quickly.
which can improve surface finish. Dull tools Second, to be flexible, the manufacturing
cause cutting forces to increase. Increasing cut- system must be able to reconfigure, that is,
ting forces can lead to chatter and vibration in redesign, easily. The design should not be
the process. Tool wear can affect part specifica- viewed as fixed and unchangeable, as is often
tions, hence quality. In short, when reliability the case in the job or flow shop. Design of a
of a process increases, the probability of a manufacturing system dictates material flow,
breakdown or production defect is reduced. and this must be done efficiently.
For cells, if the cutting process can be run a Third, to be flexible, an existing manufactur-
full eight-hour shift without changing any ing system should be able to adapt to changes
tools, quality is more controllable and consis- in the product mix, as well as existing volume,
tent. No computer analysis is needed to find or changes in customer demand for the prod-
the process within a cell responsible for a bot- uct. These are two different kinds of flexibility.
tleneck; it is the machine with the longest The first type is volume or demand flexibility,
machining time. Everyone in a manufacturing where adding or subtracting workers changes
system can see and understand how a cell func- production rates for a cell. This idea is in use at
tions, the process most likely to delay cycle Wendy’s® and many other fast-food restau-
rants, or manufacturing cells for hamburgers.
time, and where critical resources need to be
At lunchtime, many fast-food workers are in a
directed to obtain the greatest productivity
cell, so the production rate is high. By mid-
gains. Therefore, in cellular manufacturing, the afternoon, when business is slow and fewer
overall system design must be simple with workers are in a cell, there is a low production
built-in flexibility. rate. Both production rates are carefully
designed to meet customer demand in pre-
FLEXIBILITY IN CELL DESIGN scribed amounts of time, while maintaining
The key function-design requirement for a quality. Manufacturing and assembly cells
manufacturing or assembly cell is flexibility. need to adapt monthly, or perhaps biweekly, to
Manufacturing cells should not be confused demand changes, rather than on a daily basis.
with flexible manufacturing systems. Cells Presented here is another lean production,
are operationally different. Manufacturing linked-cell, manufacturing-system design
cells have several types of flexibility, whereas, where flexibility is incorporated. Figure 5-11(a)
flexible manufacturing cells have little real shows a detail of the frame area from Figure 5-
flexibility. 1. This area contains six manufacturing cells
Manufacturing flexibility considers: and one subassembly cell. All directly feed the
mixed-model, final-assembly line. These seven
• adaptability to change of product design; manned cells are linked directly to a nearby
• ability to reconfigure a manufacturing cell by kanban or point of use. An example of
system or subsystem easily; and this idea is a subassembly line directly linked
• ability to adapt to product mix and volume via point of use in a final assembly line. Cell E
change. is directly linked to Cell F. Cell D is feeding
parts to Cell H, using kanban links. Cell H workers are allocated to seven cells, resulting
withdraws parts from Cell D as needed. In the in a decrease in production rate and increase in
linked-cell system, work-in-process between cycle time to 165 seconds. Notice that all cells
two cells is controlled by a withdrawal kanban, have the same cycle time. This cell system is
a function of the two-card-kanban system. The well balanced. For a cycle time of 120 seconds,
inventory within cells is called stock-on-hand. each worker is tending nine or 10 processes.
When cells are close to each other, they make a Workers in these cells spend 10-15 seconds at a
high percentage of perfect parts, with no machine, and about five seconds walking to the
machine breakdowns and virtually no setups. next machine.
The kanban link often can be replaced with a The frame-area foreman tries to allocate the
direct link. minimum number of workers needed to keep
Figure 5-11(b) shows two different alloca- the area running problem-free. Notice that
tions of workers for a frame area. In the upper some frame-area cells are completely operated
position, eight workers are tending machines by one worker. Other cells have two workers
in seven cells. The cycle time is 120 seconds, per sharing tasks and machines, so only a part of
unit. The next month, a longer cycle time is the worker’s time is used to operate a cell. If
needed because demand has decreased. Six demand increases, more workers can be added
Figure 5-11a. Frame area has seven manned cells. Cells C and E are directly linked. The remaining cells are linked by kanban (K).
Figure 5-11b. Number of workers can be reduced as demand decreases, thereby increasing cycle time.
and two workers might be operating one cell. dles in several smaller machines. In the manu-
Decouplers between machines tie workpiece facturing cell, if machining time for one process
movements within a cell. is greater than the required-cycle time for the
Manufacturing and assembly cells use walk- cell, the following alternatives are available:
ing workers, similar to fast-food restaurants
• The process is duplicated, as shown in
with hamburger-manufacturing cells. These
Figure 5-10, effectively splitting machine
cells employ multifunctional workers who per-
time in half. This is not an optimal solu-
form many tasks and duties, as well as operate
tion, because it adds variability to part
different processes.
sizes.
Changes in consumer desires may alter the
• The process is accelerated to reduce
product mix. Here is where the second major
machining time. There are many ways to
type of flexibility is generated. Cells make
accomplish this such as increasing depth
product families of various components and
of cut, feed rate, and cutting speed; but,
subassemblies. For example, consider a part
they can all result in complications. For
family of four different parts. All four parts
example, increasing cutting speed will
have the same sequence of processes. When
result in decreased tool life, degraded
parts differ in size (length of cut, for example),
quality, the need to make more cutting-
machining times differ because they depend on
tool changes, etc.
the length of the cut.
• The cell runs overtime at the same
However, altering machining time will not
machining time, and the difference is
disturb the cell’s production rate, because the
absorbed in kanban links. This is usually
cell’s cycle time is dictated by the time it takes
only a temporary solution, but the easiest
a worker to carry out all manual tasks at each
to implement.
process and walk around the cell, not machine
• Some operations completed on a machine
times. This is critical to cell function and
with the longest machining time are shift-
design, but moreover, it is critical to the cell
ed to another machine with time avail-
designer’s fundamental understanding of cellu-
able. This solution may not be technically
lar manufacturing. The cell design decouples
possible.
the production rate for the cell from the pro-
• The product is redesigned.
cessing rate of the machine tool. This is a pri-
• Finally, if product demand is so high that
mary concept of lean production, where the
the capacity of the cell is exceeded, then
lives of personnel in production control get sim-
the cell is cloned (replicated) to double
pler. It also means that the mix of parts in the
capacity. Proven manufacturing capacity
parts family can be changed without disturbing
is, therefore, quickly doubled, and so is its
the production rate, as long as the cycle time is
flexibility. However, this process violates
greater than machining time, and cell-setup
the concept of cellular manufacturing sys-
time is reduced.
tems by producing the same component or
If the same set of parts is made in a comput-
subassembly in two cells.
er-numerical-control (CNC) center, cycle time
for each part will depend heavily on machining When a cell is replicated, capacity is more
times, and parts of different sizes will have than doubled. Suppose a cell is making two
very different cycle times. This will result in parts, A and B, and that demand has reached
different lot or batch times and lead to sched- the point where the cell must be replicated.
uling problems. Also, the cycle time in the Now, one cell can make Part A and the other
machining center will be longer, because cell can make Part B, so cellular-manufactur-
machining times are performed serially by one ing fundamentals are maintained. Setup
spindle, rather than overlapped by many spin- between Parts A and B is eliminated, because
each cell has become dedicated. However, if decouplers. One-piece-flow movement of parts
demand should decrease, one of the cells could eliminates queues and part-storage banks
go back to making both Parts A and B. Excess between processes, reduces material handling,
capacity is available to add new parts to cells. and saves floor space. To optimize manufactur-
This leads to a condition called less-than-full- ing with cells, setup time must be reduced, so
capacity design. rapid exchange of tooling and dies is important.
The cell also adapts to changes in product Rapid exchange of tooling and dies is a highly
design for new or existing products. Cells can effective technique for setup reduction and is
be designed for design-change flexibility. Step 2 in the 10-step linked-cell strategy.
Twenty years ago, design of the manufacturing Machines also must be improved to inform
system was called plant layout. Today’s text- operators when something is going wrong, so
books call it facility design. All design work that the problem can be quickly rectified.
used to be done by industrial engineers when a Therefore, foolproof methods must be imple-
new plant was being built, or it simply devel- mented to prevent defects and machine break-
oped as the plant grew. Once the plant was laid downs. These are Steps 3 and 4 of the
out (designed), the job was considered to be lean-manufacturing strategy. Nevertheless,
complete. problems inevitably occur within cells, and
Now the manufacturing system must be buffers of inventory must protect the down-
periodically redesigned, that is, the processes stream processes and assembly areas from
reconfigured to maintain flexibility, improve delays. Inventory buffers are between the cells;
quality, reduce costs, decrease throughput that is, within kanban links (Figure 5-12).
time, and lower in-process inventory. To be flex- Parts move between cells in small lots of uni-
ible in the factory with a future, the manufac- form size using kanban links. The inventory-
turing system and assembly cells must be buffer sizes between cells equal the number of
easily redesigned. Equipment on the plant floor parts in a container, multiplied by the number
must be readily movable to restructure or relo- of containers in the link. The buffer size is con-
cate cells. Machines must be simple, program- trolled by the users of inventory, and mini-
mable, and able to be mounted on air pallets or mized by controlling the number of containers
wheels. This is another good reason to purchase in inventory links. Inventory buffers protect
or build small process-specific machines. the downstream portion of the system against
Power, water, and pneumatic connections must machine-tool breakdowns, defects, and delays
be mounted overhead on tracks for easy access. occurring in upstream cells, remote cells (ven-
Equipment should not be bolted to the floor, dors), or transport systems. The critical func-
unless absolutely necessary. tions of production and inventory control are
The product designer today knows the integrated or infused into the manufacturing
process capabilities for a family of parts, which system as cells are linked to subassembly and
are produced in a cell, so parts can be designed final-assembly lines. The system is designed to
accordingly. This is truly designing for manu- call for subassemblies and component parts as
facturing. needed (for example, when, how many, and
where parts are to go). Users of this material-
HOW AND WHY CELLS ARE LINKED information system can directly control the
At the outset, cells are linked by a pull sys- inventory level.
tem of material control, via the kanban inven- Standard operation means that a system can
tory and production-control system. Within meet customer demand with minimum labor,
cells, parts move one at a time, from machine to stock-on-hand, and work-in-process. There are
machine, controlled by kanban squares and three elements of a standard operation:
Cell Cell
1 2
Downstream
Figure 5-12. Cells are linked with controllable inventory buffers called kanban links.
• Takt time is calculated using Equation 5-6. machines within a cell. The horizontal axis is
• There is a routine or sequence of opera- time in seconds or minutes.
tions needed to make the unit. The cycle time for the cell is determined by
• There is a standard quantity of stock-on- the needs of the system for parts made by the
hand in cells. cell. Figure 5-14 shows a standard-operations
sheet for a small manned cell. Vertical columns
have been filled in with operations. The layout
STANDARD OPERATIONS ROUTINE SHEET
schematic shows the path of workers or the
The standard operations routine sheet (Figure sequence of operations. All processing times are
5-13) is used to plan the manufacture of one part less than two minutes.
in a family within a cell. The plan illustrates the Parts in a cell move from machine to
relationship between manual operations per- machine, one at a time. For material process-
formed by a worker, machining operations per- ing, machines typically are capable of complet-
formed by a machine, and time spent by the ing a machining cycle initiated by a worker.
worker walking from machine to machine. The U-shape puts the start and finish points of
Manual operations include loading and unload- a cell next to each other. Every time the opera-
ing the machine, checking quality, deburring, tor walks around the cell, a part is completed.
removing chips, marking parts, and more. As shown in Figure 5-14, machining processes
The vertical column on the left of Figure 5-13 overlap and need not be equal or balanced. The
lists operations/processes on machines or sta- diagram reveals interference between the pro-
tions. Details (drawings, photos, and instruc- cessing time (70 seconds), and the necessary
tions) on each operation are posted at the cycle time (55 seconds). Total machining time is
Figure 5-13. Example of standard operation routine sheet used to plan work for a manufacturing cell.
10 minutes, 3 seconds. A numerically controlled People who run processes are a company’s
machining center capable of performing most valuable asset. They, the users of the
machining operations could replace the cell. manufacturing system, are its internal cus-
However, the part’s cycle time will jump from tomers. These users are often the company’s
two minutes to over 10 minutes, because com- most poorly used resources. Redesigning a
bining processes on one machine prevents over- manufacturing system into cells greatly assists
in matching the skills of workers to positions in
lapping of machining times, which is called
a manufacturing organization. The functional
parallel processing. Adding additional workers and production job shops’ manufacturing sys-
can alter the cycle time for a cell. tems must be restructured because they isolate
people, restrict communications, and greatly
DESIGN FOR CUSTOMERS increase feedback time on product-quality
A cell has multi-process workers who can problems and other system concerns.
operate more than one kind of process or mul- Changes will cause some conflicts between
tiple versions of the same process. They also users of the manufacturing system and the
ultimate customer of the system’s products.
carry out inspection and machine-maintenance
The biggest dilemma is cost. However attrac-
duties, making them multifunctional. Workers tive a product or perfect a manufacturing sys-
devise ways to eliminate setup time. Note that tem, the end product must sell at a reasonable
manufacturing and assembly cells eliminate price for the consumer. Generally speaking, the
the job-shop concept of one person/one internal customer needs to dictate the design
machine; and thereby, increase worker produc- and development of a manufacturing system,
tivity and utilization. even though factors affecting the external cus-
Overview of cycle
2 units in
decoupler
Figure 5-14. Example of a completed standard operation sheet for a manned cell. The overview of the cycle time is two minutes,
stock-on-hand is 12 units (Courtesy of TRW/Koyo).
tomer (buyer of the product) are the highest OTHER SYSTEM-DESIGN TOOLS
priority. A system must be developed that Most companies design their first cell by
places top priority on the factors affecting the trial-and-error techniques. With the advent of
final customer and adapts the system to meet newer languages and programs, digital and
the highest requirements of the internal cus- high-level-graphic simulation is gaining wider
tomer (production worker). Resolving conflicts usage in the design and analysis of manufac-
between these two customers requires a man- turing systems.
agement environment where new ideas are free Another technique being extensively research-
from conventional restrictions. ed is called physical simulation. This approach
uses small robots, scaled-down models of niques work for small manufacturers and
machine tools, and mini-machines to emulate restaurants, but not in large job shops where
real-world systems. Mini-machines employ the the number of components may approach
same minicomputers and software as full-scale 10,000, and the number of processes may reach
systems. In this methodology, the development 500 or more.
of software needed to integrate hardware in the Consider the problem of a manufacturing
cell and design of cell-control logic (software) engineer trying to justify the cost of a machin-
can be developed prior to installation of a full- ing center that is computer numerically con-
scale system on the shop floor. Physical simula- trolled. The manufacturing engineer uses
tion is an ideal way to provide low-cost, experience and judgment to select as many
hands-on education for workers in manufactur- high-cost, complex parts as possible to be
ing systems. It is the best method to teach stu- machined. The engineer also performs an eco-
dents what is required to integrate hardware nomic analysis to cost-justify the new machine.
with software. Unmanned cells and flexible- The manufacturing engineer determines that a
manufacturing systems can be simulated at family of component parts (or at least a partial
reasonable costs. family) will be produced on a machining center,
which represents a type of cell (one machine).
GROUP TECHNOLOGY The same approach is valid for forming a
Group technology offers a methodology for manned cell of conventional machines.
reorganization of the functional job-shop sys-
tem, restructuring it into manufacturing cells. PRODUCTION-FLOW ANALYSIS
In a manufacturing facility, component parts of Production-flow analysis uses the informa-
similar design, geometry, or manufacturing tion available on route sheets. The idea is to
sequences are grouped into families. Machines sort through all the components and group
can then be arrayed into groups or manufac- them by matrix analysis, using product-routing
turing cells to process the family of compo- information (Figure 5-15). This method is sim-
nents. This manufacturing sequence defines ple, inexpensive, and fast, but still more ana-
the arrangement of machines and processes in lytical than tacit judgment. Production-flow
the cell. analysis is a valuable tool for system-reorgani-
Looking back in time to the first manufac- zation problems. For example, it can be used for
turing cells, there is no evidence to suggest that an up-front, or “before-the-fact” analysis, yield-
the Toyota production system designers even ing some cost and benefit information. This
knew of or used group-technology methodology. gives decision makers information on what per-
There is no mention of group technology in centage of product will be made by cellular
Toyota literature. It would appear that early methods, what will be a good first cell to under-
Toyota cells were formed using kanban. take, what other analysis methods will work
Processes were placed near each other to best, and the amount of funds that may need to
reduce movement of containers between be invested in new equipment.
processes and machines. In short, production-flow analysis can great-
Finding families of parts is one of the first ly reduce the uncertainty of making a decision
steps in converting a functional-job shop to a about reorganization of the factory floor. As
cellular-manufacturing system. There are a part of this technique, an analysis of material
number of ways to accomplish this. Judgment flow of the entire factory is performed, laying
methods using axiomatic-design principles are, the groundwork for the factory’s new linked-
of course, the easiest and least expensive, but cell layout. The use of production-flow analysis
also the least comprehensive. Eyeball tech- to identify elements of the first cell permits a
Number
Machine Tool Code Machine Tool Code
Number
Job
Job
S V H G H S V H G H
W BR EL DP M M R IN TL W BR EL DP IN TL
O M M R O
1 7
2 11
3 2 (exception)
4 5 Cell will have
5 18 three machines
6 14 HM, GR, HO for
manufacture of
7 3 six jobs
8 10
9 20
10 12
11 4
12 19
13 16
14 8
15 1 (exceptions)
16 9
17 13
18 6
19 15
20 17
company to implement that first cell without ic component elements, based on differences in
waiting until all parts in a plant have been shape, function, material, size, and manufac-
coded by comprehensive-coding systems. turing process.
No attempt to review coding/classification
CODING/CLASSIFICATION METHODS methods is made here. Coding/classification
systems exist in bountiful numbers in pub-
Many companies converting to a cellular sys-
lished material and written information from
tem have used a coding/classification method, consulting firms. Most coding/classification
which is more comprehensive and time con- systems are computer compatible, so computer
suming than production-flow analysis. The cod- sorting of codes generates classes of parts fam-
ing classification system uses design codes, ilies. The system does not find groups of
manufacturing codes, and codes that cover both machines. If a code is based on design data,
design and manufacturing. errors in forming good manufacturing families
Classification sorts items into classes of fam- will occur.
ilies based on similarities. It uses a code to Whichever coding/classification system is
accomplish this goal. Coding is the assignment selected, it should be tailored to a particular
of symbols (letters, numbers, or both) to specif- company and be as simple as possible, so that
everyone understands it. It is not necessary for components to skip some machines, and some
old part numbers to be discarded, but every machines to be duplicated. However, cellular
component will have to be coded to find part back flow is still not allowed, except under
families prior to the next step in the program. extremely strict circumstances and it is rare
This coding procedure is costly and time-con- even then. The strict and limited circumstances
suming, but most companies understand the allowing the rare back-flow cell are during low-
need to perform this analysis. volume, high-process times, and with experi-
enced cell workers.
OTHER METHODS The entire shop may not be able to convert to
Other group-technology methods, including cells immediately. Thus, the manufacturing
eyeball or tacit judgments, involve the follow- system will be a mix of job shop, flow line, and
ing tasks. cells, evolving toward a perfect linked-cell sys-
tem. Scheduling problems are created when in-
• Find a key machine, declare all parts process times for components made by cells are
going to this machine a family, and move vastly different from those made under tradi-
the machines needed to complete all parts tional job-shop conditions. However, as the vol-
in the family around the key machine. ume of parts in the functional area decreases,
Often it will be prudent to off-load opera- the total system will become more productive
tions from the key machine to other and simpler.
machines within the cell.
• Build a cell around a common set of com- PILOT CELLS
ponents like gears, splines, spindles,
The formation of parts families leads to the
rotors, hubs, shafts, etc. There are natural
cell design, but the design of cells is by no means
families of parts that have the same or
automatic. It is the critical step in reorganiza-
similar sequence of processes.
tion and must be carefully planned. Many com-
• Build a cell around a common set of
panies begin with a pilot cell so that everyone
processes. For example, the sequence of
can learn and understand how cells function.
drilling, boring, reaming, keysetting, and
Companies should proceed with the develop-
chamfering holes is commonly used in
ment of manned cells, and not wait until every
manufacturing holes on parts.
part has been coded. Many companies get start-
• Build a cell around a set of parts to elimi-
ed with coding and never form a cell.
nate setups between parts. This is the
The simplest approach is to select a logical
approach used by Harley-Davidson®.
product group and form a cell to manufacture
• Pick a product or products, then design a
it. Potential cell workers can assist in this and
linked-cell system, beginning downstream
their inclusion in the implementation team
with the final-assembly line (converting
should be mandatory. Only in this way will
final assembly to a mixed model), moving
everyone learn how cells operate and how to
upstream to subassembly, and finally, to
reduce setup time on each machine and
component parts and vendors.
process. In the vast majority of cases, machines
Part families do not all have the same mate- will not be used 100% of the time. For some
rial flow and, therefore, they require different machines, the utilization rate may not be what
designs (layouts). In some families, every part it was in the functional system. However, the
goes to every machine in exactly the same system will be highly utilized and this should
sequence. No machine is skipped, and no back be kept in mind by the accounting department.
flow is allowed. This is the purest form of a cel- The objective in manned cellular manufactur-
lular system. Other families may require some ing is to fully utilize people, enlarging and
enriching their jobs. In fact, one inherent result Cellular manufacturing offers many advan-
of a linked-cell system is that workers become tages, including:
multifunctional. That is, they learn to operate • Quality feedback between manufacturing
many machines and/or carry out many duties and assembly operations is faster.
or tasks. • Material handling is markedly reduced.
In the manned cellular system, a worker is • Setup time is reduced or even eliminated.
decoupled from the machine so that the utility • In-process monitoring, feedback, control of
of the worker is no longer tied to the machine’s the inventory and quality are very greatly
utility. This means there will be fewer workers improved.
in the cell than machines and processes. In • A smoother, faster flow of products
unmanned cells, utilization of equipment is through operations is achieved.
more important because the most flexible ele- • Cycle-time variability and line-balancing
ment in the cells, the worker, has been removed constraints are reduced.
and replaced by a robot and decouplers. • Implementation of automated manufac-
The manned cellular system provides the turing operations is easier.
worker with a natural environment for job • Process capability and reliability are
enlargement. Much greater job involvement markedly improved.
enhances job-enrichment possibilities and
clearly provides an ideal arrangement for CONVERSION CONSTRAINTS
improving quality. In the lean-production A major effort on the part of a business is
scheme for cellular manufacturing, part quali- required to undertake a conversion to linked
ty is checked between each step in the process. cells. Constraints to cellular implementation
are:
BENEFIT OF CONVERSION
• Systems changes are inherently difficult
The lean-production strategy of simplifying
to implement. Changing an entire manu-
the manufacturing system before applying facturing and production system is a huge
automation avoids many risks and facilitates job; therefore, it should be systematically
automation. Conversion to manned cellular completed. This change does require man-
systems results in significant cost savings over agement- and worker-attitude changes,
a two- to three-year period. Specifically, manu- because it empowers workers to find prob-
facturing companies report significant reduc- lems and solve them.
tions in raw materials, in-process inventories, • Companies spend freely for product inno-
setup costs, quality costs, and costs of bringing vation, but not for process innovation. It is
new designs on line. However, this reorganiza- easier to justify new hardware for an old
tion has a greater, immeasurable benefit. It manufacturing system, than to rearrange
prepares the way for automation. Progression old hardware into a new linked-cell sys-
from the functional shop to the factory with tem. However, anyone with capital can buy
manned linked cells; and ultimately, robotic the newest equipment, often creating
cells with computer control for the entire sys- another island of automation.
tem, must be accomplished in logical, economi- • Decision makers fear the unknown. They
cally justified steps, each building from the choose among several alternatives in the
previous stage. This results in computer-inte- face of uncertainty. The greater the uncer-
grated manufacturing. The key is to integrate tainty, the more likely a “do-nothing” alter-
manufacturing first, then automate and com- native will be selected, and the status quo
puterize it. maintained. While converting to cellular
manufacturing frees additional capacity to improve it. Another view strongly sug-
via setup time saved and funds not tied up gests that a worker volunteer to work at
in inventory, such conversion requires an the side of a cell designer from start to fin-
expenditure of funds for equipment modi- ish. This is extremely important when
fication, employee training (in quality, designing the first cell in a factory. Lean-
maintenance, and setup reduction), and production proponents must stack the
many more items. The assumed long-term deck to insure that the first cell imple-
payback seems a high risk in the minds of mentation is a huge success. Otherwise, it
decision makers. While in reality, cellular- may take years before a company will con-
manufacturing systems often pay back in
sider implementing cellular manufactur-
months, not years.
ing again—then it may be too late.
• Decision makers use faulty criteria.
• Lack of top management leadership: facto-
Decisions should be based on the ability of
the company to compete (attractiveness, ry leadership must have the courage to
cost, quality, reliability, and delivery time), change. Leadership driven by top-level
rather than output or cost alone. commitment is essential in changing the
• Conversion to a linked-cell system repre- manufacturing-system design.
sents a real threat to middle managers. It is not known why some people make bet-
Within the production system many func- ter leaders than others. It is known that that
tional areas that middle managers have many tangible and intangible factors make
been responsible for are shifted and inte- people good leaders; it is also known that even
grated into the new manufacturing sys-
good leaders make bad decisions. Dwight D.
tem. Also, the short-term perspective of
Eisenhower said, “It’s easy to make decisions
financially oriented middle managers, ver-
when you have all the facts.” President
sus the long-term nature of the program,
results in a resistance to change. Changing Eisenhower fully realized that many times
a system design causes a change in the facts are missing or incomplete. Therefore, good
production system that services the manu- leaders gather as much information as possi-
facturing system. ble, then make decisions. Here are some facts
• There is a lack of blue-collar involvement and comments concerning leadership qualities:
in the decision-making process of the com- • Good leaders have character and integrity
pany. Getting production workers involved — they inspire their followers to act as
in decisions is a significant change. Many they do. (Badler 2003)
managers and workers in old manufactur- • Decisions that best benefit the group
ing systems have problems adjusting to
upset some people, but good leaders make
this situation. Workers must be rewarded
these decisions anyway.
for ideas and suggestions. Bulletin boards
can be used to post notes containing sug- • Good leaders must be accessible and avail-
gestions from workers. Suggestions must able to everyone; so flat, minimum levels
be recognized and a high percentage of in the production-system design are best.
good ideas implemented. It is vital to • Good leaders are not pressured by
involve operators in the design of new accounting, and challenge experts and pro-
cells. One idea advocates that manufactur- fessionals by asking questions five times.
ing engineers allow workers to completely • Good leaders delegate, empower, and
design the interim cell, even when they understand details.
know they can do it better. Workers will • Even if things are not broken, good leaders
take ownership of the cell design and work ask how systems can be improved.
• Good leaders foster environments where demand. Cycle time is decoupled and does not
the best people are attracted, retained, depend on machining times.
and encouraged to be creative. Families of parts with similar designs, flexi-
• Good leaders pick people who are intelli- ble-workholding devices, and tool changes in
gent, have integrity, make good judg- small, programmable machines allow rapid
ments, have energy, are balanced, are changeover from one component to another.
driven, and have the ability to anticipate. Rapid changeover means quick or one-touch
• Good leaders have pizzazz, drive, expert- setup, often like flipping a light switch.
ise, charisma, and caring attitudes. Workers control quality in cells. They also per-
• Good leaders are willing to learn new form routine maintenance on the equipment
within cells.
skills, take on new responsibilities, and
Production and inventory control links
even reinvent their own positions.
define the paths materials take within the
• Good leaders are inspired to be the best.
factory, and they are integrated. As quality
• Good leaders can lead. They are lean- improves, significant inventory reduction
thinking, compelling, crisp, and clear. between cells is possible. Inventory levels can
Good leaders must have attitudes that are be directly controlled, reduced to minimum
proactive. Leaders should not be afraid to levels, and counted.
spend part of each day on the factory floor run- Most importantly, manufacturing systems
ning a process. Good leaders are in touch with must be designed to be flexible. The manufac-
manufacturing reality. They know what needs turing system design must adapt to changes in
to happen strategically down the road and, customer demand and product design. Also, it
must deliver quality products at the lowest pos-
they set things in motion for the future.
sible cost, and with the shortest possible deliv-
ery time. Using linked-cell systems composed
SUMMARY of manned manufacturing and assembly cells,
Cells make parts one at a time using a flexi- linked with a pull system of production control,
ble design. Cell capacity, or cell-cycle time, can has proven to be the method to accomplish
be altered to respond to changes in customer these objectives.
exchange of tooling and dies trains workers to The conclusion drawn from this logic is that
achieve a significant reduction in setup times if setups are long and costly, then large lots will
and many other forms of waste. justify long setups. This analysis assumes that
The objective of single-minute exchange is to setup time is fixed and cannot be changed.
reduce setup time, simplify setup procedures, Shingo and others have proven that setup time
eliminate scrap and rework, and reduce inspec- can be markedly reduced and, therefore, so can
tion times. Indirectly, as setup time is economic-order quantity.
decreased, operation frustration of the factory Another way to approach the problem is to
work force will decrease. The question becomes look at manufacturing costs. The calculation is
“Why haven’t we done this before?” The most based on the following equation in its most
probable answer is that no one ever recognized basic form:
the need. Under the archaic functional manu-
TC = FC + (VC × Q) (6-2)
facturing system, setup was viewed as a given,
something accepted. Setup time was measured where:
or estimated and converted into a cost using
TC = total cost to produce part
the economic-order-quantity calculation. In the
FC = fixed cost
past, the most economical way to handle setup
VC = variable cost per unit
was to calculate it as a computed cost.
Q = quantity to be built
Minimally, this was a suboptimal solution,
inconsistent with the needs of the manufactur- Fixed cost includes costs that are insensitive
ing system. Economic-order-quantity (EOQ) to how many items are to be manufactured in a
calculations ignore the costs of quality such as production run. Setup cost is an example of a
long delivery times, excessive material han- fixed cost. Setup cost is calculated by multiply-
dling, and many other forms of waste. ing time to complete the setup by the cost-per-
unit time. Cost-per-unit time equals labor cost
ECONOMIC SETUP-REDUCTION plus the cost of lost production time or stopping
TECHNIQUES production during a changeover.
Variable cost is a unit cost composed of
Almost without exception, every textbook on direct-labor cost per unit and material cost per
production, operations, or inventory planning unit. Figure 6-1 shows these costs plotted in
and control presents calculations for comput- classical fashion with linear relationships
ing economic order quantity. This is the quanti- assumed. The plot shows that as the total
ty that balances the cost of holding inventory quantity built increases, so does total cost.
(caused by large-component quantities), If total cost is divided by quantity, Equation
against the cost of performing setup. The equa- 6-2 becomes:
tion is usually given as:
T C /Q = FC /Q + VC (6-3)
)
xQ
VC
FC +(
TC =
)
ost (V C
b le c
ia
Total cost, Var
(TC) $
Fixed cost (FC)
0
0 Quantity Q
This simple plot has been the driving force ly with quantity. The cost per unit does
behind manufacturing/production system phi- decrease with quantity, because fixed costs are
losophies to mass produce in large lots and spread over many units; but take another look.
spread setup costs over many units. Figure 6-2 Figure 6-3 shows Figure 6-2 as an undistort-
also reflects the basic economy-of-scale idea, but ed, non-transformed relationship between cost
this plot does not extend to the right indefinite- per unit and quantity being built. When
ly. At some point, Q is large enough that the graphed on Cartesian coordinates, rather than
capacity of the process is exceeded and another log-log coordinates, the total cost per unit
process must be considered. plunges rapidly toward variable cost-per-unit
Figure 6-2 is a misrepresentation because, time. It then turns sharply and runs nearly
invariably, data is graphed on log-log paper for parallel to the variable-cost line over much of
easy viewing and graphical analysis. The data, the span of the build quantity. Cost per unit
thus transformed, gives the reader a distorted decreases while approaching variable cost per
impression that cost per unit changes gradual- unit at its minimum. The change from rapidly
Log
TC
Unit
$
FC
Unit
Q
VC
Cost
Unit
0
Log
0 Quantity Q
Figure 6-2. Typical textbook model of cost per unit versus quantity. A log-log or semi-log plot is used, which distorts the data.
changing cost per unit to slowly changing cost ed. Cost per unit is constant for all quantities
per unit occurs over a narrow range of quanti- when setup costs are eliminated. Therefore,
ties, Qa to Qb, the shaded area in Figure 6-3. manufacturers who use lean manufacturing
From Qb to Qc, the rate of change in cost per strategies can build in small lots for basically
unit is small compared to Qa to Qb. the same unit cost as a company using mass
To understand what happens to the econom- strategies and who are building in large lots.
ic picture when setup time, thus cost, is great- Eliminating setups and changeover time will
ly reduced or eliminated, examine Figure 6-4. result in a competitive edge in flexibility.
Total-cost function quickly equals variable-cost In fact, building in small lots costs far less
function. The dashed line in the plot depicts per unit. Quality improves because less inven-
setup cost as greatly reduced, but not eliminat- tory needs to be managed and stored. Layers of
Cost
Unit
TC
Unit
( )
$
Unit
VC
Qa Qb Q Qc Quantity
Figure 6-3. True picture of total cost/unit versus quantity plotted on Cartesian coordinates.
inventory are stripped to reveal problems, like unit-labor costs decrease. Figure 6-5 shows the
lowering a river to expose rocks. Throughput total cost per unit versus quantity (log-log plots
time is greatly improved. Thus, the economics for clarity) for the three different processes,
of setup elimination are persuasive. any of which could manufacture the same item.
Production volume dictates which process to
CHANGING PROCESSES select in Figure 6-5. That is, cost-per-unit data
TO MEET CHANGING VOLUMES determines the process to use for a given quan-
In the traditional mass-manufacturing strat- tity. Three competing processes are examined on
egy, an increase in volume from customer the basis of cost per unit. There is no minimum
demand dictates a shift to faster (higher) pro- cost, only processes representing the lower-unit
duction-rate processes. Suppose there is an costs at different production quantities.
item that can be produced by an engine lathe, Separating alternatives are break-even points
computer-numerical control (CNC) lathe, or or break-even quantities. These are quantities at
six-spindle automatic. These processes are which the cost per unit to build an item is equal
characterized by slow, medium, and fast pro- for either of two processes. Thus, for cost per unit
duction rates, respectively. As production rates versus quantities, this classical approach deter-
increase, setup times also increase. However, mines quantities over which certain processes
Log
TC ~
VC
Q
$
Unit
VC
0
Log
0 Quantity Q
Figure 6-4. Effect of reducing setup cost on total cost per unit.
are more economical than other processes. The could have been done long ago, but most com-
selection of the lowest-cost alternative is based panies failed to see the need.
on quantity that must be built to fill demand. If
setup cost is accepted as a given, this is a valid, Organizing to Eliminate Setup Time
but suboptimal, approach. Most setup problems are related to materi-
In lean manufacturing, a different approach als, manufacturing processes and systems,
is taken to solving the problem. Setup time is and management practices. Contrary to popu-
reduced and products are built in the smallest- lar opinion, labor is usually a minor factor.
lot sizes possible. As unit cost approaches vari- Many companies use a team approach, rather
able cost, material cost tends to dominate the than using individuals for setup-reduction
variable cost per unit. The delightful news is efforts. The recommended approach combines
that eliminating setup is not a complex, sophis- a rapid-exchange-of-tooling-and-dies team
ticated undertaking. Setup reduction requires and internal customers trained in single-
the knowledge of simple rules and the applica- minute-exchange-of-dies fundamentals (out-
tion of good operations and methods analysis, lined in Chapter 1 and addressed completely
along with a bit of common sense. Clearly, this by Shingo in his book on this subject [Shingo
Figure 6-5. Three different processes, any of which can be used to make the same item, are compared on the basis of cost per
unit versus quality. The solid line represents minimum cost per unit.
and foremen in setup reduction. The team the system’s work-in-process. Reducing lot size
should spend time establishing objectives and is fundamental to reducing work-in-process.
obtainable goals, and avoid trying to solve Reducing setup times is fundamental to reduc-
problems encountered in areas other than ing lot sizes.
those directly related to setup. The initial team The basic approach to reducing setup times
must avoid trying to solve setup problems includes several key points.
themselves, or it will run the risk of getting • It is important to believe that drastically
bogged down in one area. shortened setups are possible. Dramatic
To summarize, the role of the project team is to: reductions can be accomplished by under-
• train and involve operators, supervisors, standing that tooling change is not just a
and support personnel; matter of removing one tool or die and
• gain experience from the worst setup proj- attaching another. When setup time is
ects; reduced on one assembly line, managers
• prepare plans and set priorities; and supervisors gain direct experience
• determine installation timing; from improvements, making it easier to
• coordinate group efforts; and extend improvements laterally to other
• create and maintain the enthusiasm of assembly-line operations.
everyone involved. • The setup process is simplified so that
workers can perform it easily. Keeping
Motivation for Single-minute setup changes away from machine opera-
exchange of dies tors merely creates a class of setup
The lean-production system is a serial- experts. Dealing with simplification of the
industrial process. The just-in-time concept setup process is a principal goal of setup
means items are made when required, in quan- improvement. Workers must be directly
tities required, and as inexpensively as possi- involved in the change process.
ble. Minimizing inventories, synchronizing • Centering and locating adjustments
manufacturing-system elements, and cellular- should be eliminated. Adjustments depend
izing upstream processes to minimize work-in- on the right touch (or on luck), and are not
process are all steps in the lean manufacturing repeatable. Differences show up when dif-
methodology. ferent people make these types of adjust-
It is important to stimulate the need for ments; even the same person may take
reducing setup-change time. Lean production varying times to make the same adjust-
uses kanban. Depending upon the type, kanban ment on different occasions. For these rea-
uses indicator cards or signs that, in addition to sons, adjustments should be eliminated.
preventing overproduction and performing pro- • A setup should allow defect-free products
duction-control functions, provide information to be produced from the first piece on after
on production and transactions. Kanban also changeover. There is no logic to speeding
acts as a tool for forcing the gradual improve- up a setup operation without knowing if
ment of the manufacturing system by reducing quality products can be manufactured.
throughput time, creating fewer defects, and Total setup time is the time from the last
avoiding late deliveries. To function this way, good part of the previous setup to the first
the kanban must be located so that everyone acceptable part in the new setup. Anything
can see the sequence, amount, and timing of affecting that time frame is in the scope of a
work to be completed. setup-reduction program (Figure 6-6). The typ-
Over time, the number and size of the kan- ical industrial sequence is to set up the equip-
ban decreases. In turn, this decrease reduces ment or machine tool, run some parts, inspect
Until
finally
The existing
method of
Last Adjust 1st changeover
good
A part machine Scrap
or B part
rework
Setup Run
The improved
method
1st
Last good
A part B part
Figure 6-6. The existing method is converted to an improved method of setup by limiting unnecessary elements like adjustment.
those parts, adjust the machine, run another or process. This usually requires construction
sample, measure, adjust, and so on, until an of an intermediate workholder such as jig or
acceptable part is produced. This method gen- fixture plates to which the part workholder is
erates scrap and rework and results in nonpro- attached. Jigs and fixtures are different, but
ductive time. plates are identical. The cassette tape for a
videocassette recorder is an example of an
Functional Clamps intermediate workholder. Every cassette is
The use of functional clamps, instead of treated the same by a videocassette recorder.
screw-type fasteners, should be highly encour- Each one can be loaded and unloaded quickly
aged. Screw-type fasteners should be replaced with one handling. From the outside, every
with dovetail grooves, pins, cams, wedges, etc. tape appears to be the same, but on the inside,
When use of screws is unavoidable, they should every tape is different. Workholding devices,
be turned no more than one full turn. such as an intermediate jig, can help to quickly
achieve one-touch setups.
Intermediate Workholders A key to reducing setup time is eliminating
Workholding devices should be designed so adjustments. A significant difference exists
that they appear the same to the machine tool between setting and adjusting. The channel
selector sets the television to a channel. A produce each part within the family.
house’s thermostat sets the temperature. If set- Changeover is handled sequentially within a
tings can be manipulated, that tinkering or cell (Figure 6-9).
fine-tuning is called adjusting. Eliminate This example involves a cell with four
adjustments whenever possible. For example, it processes making different parts. In manufac-
often takes a significant amount of time to turing and assembly cells, as well as assembly
change chucks (workholders) in a lathe. Figure lines, setup work begins at the head of the cell,
6-7 shows a solution where chucks of many dif- or flow line, and moves through the cell one
ferent configurations are quickly mounted on process ahead of the first unit of that particu-
lar part or model. Setup becomes an integral
part of the cell operation. Setups in manufac-
turing cells are synchronized with those of final
assembly as lot sizes are reduced along with
setup times. As shown in Figure 6-9, when
numerous processes are involved in the manu-
facturing cell of a family of parts, sequential
setup changes are utilized. In the figure, a
changeover from Part A to Part B is described
for a cell with four processes. The worker is
moving in the same direction as part flow and
changes each machine over sequentially from
Lathe spindle Intermediate Chuck Workpiece Part A to Part B. After the worker has made
workholder (also (one four trips around the cell, the cell is making
called actuator) bolts of many
to spindle and holds different Part B exclusively. At the outset, setup times
chucks designs) may be long compared to processing times that
is, run times. Eventually setup times are
Figure 6-7. Example of intermediate workholder for lathe per- reduced to less than the time a worker spends
mitting rapid exchange of chucks. (Courtesy Sheffer Collet Co.)
at each process.
The setup operation flows through a cell.
Eventually, it should become one-touch-very
an intermediate jig that remains attached to quick, with one handling of the workholder. The
the lathe spindle. setup change should allow production of defect-
To facilitate simpler tool-and-die exchanging free products from the start. The first part pro-
procedures, the setup team should link one duced after the setup is good. Ultimately, the
movement to another. This concept is shown in ideal condition of eliminating setup between dif-
Figure 6-8. ferent parts will be achieved. However, dropping
Even though the number of setup man-hours setup time to short intervals may be difficult in
may be unchanged, setup time can be cut by lathe operations. Turning usually requires a
50% when two workers instead of one perform change in the workholder and tooling when
changes. This is particularly appropriate when parts are changed. There is no ideal setup
machines are large or change elements take a change. If setup changes are necessary, they
long time. Sequential setup changes should be need to be performed with a one-touch motion.
made when numerous processes are involved. Single-minute exchange of dies implies that
setup time is a single-digit number of minutes
Setup Flows Through Cell (nine minutes and 59 seconds, or less). When
A manufacturing cell that produces a family setup time is reduced to less than one minute,
of parts must have processes changed over to it is called one-touch exchange of dies. Non-
Old die New die 1) Insert new die while removing old die
1 (linked movement)
Linkage
touch exchange of tooling and dies is automat- higher inventory costs and other detrimental
ic. (An example is a machine tool with an auto- effects on the manufacturing organization.
matic tool or machine-pallet changer for parts.)
Rapid exchange of dies and tooling can Basic Steps for Reducing Setup Time
reduce setup time from hours to minutes. This The basic steps of a single-minute-exchange-
can be of remarkable benefit to the manufac- of-dies, setup-time-reduction program are:
turing organization. Reducing setup time 1. Determine the existing method.
increases available machinery capacity. 2. Separate the internal from the external
However, additional process capacity should elements.
not be used to overproduce as this can lead to 3. Convert the internal to external elements.
A7 A6 A8 A7 Setup A8
"B"
X X X X X
X X X X
O
Setup
A8 A5 A6 B1 A7
"B"
X X X X X X X X
X X X X X X X
Empty of "A" 4 parts in Cart of "B" 5 "A" Raw material "A" parts
Raw material "A" cart Raw materials parts "B" parts
B1 Setup B2 B1 B3 B2
"B"
O O O O O
B2 A8 B3 B4 B1
Setup
O X "B" O O
O
X X X X X X X
X X X X X X X X
Raw material "A" parts "B" parts "A" parts out to "B" parts Empty "B"
"B" parts next process cart
Figure 6-9. The flow through setup cycle showing changeover from Part A to B for a four-process cell.
4. Reduce or eliminate internal elements by pleted, when items like removed tooling and
continuously improving setup. Apply dies are returned to storage and the process
method analysis and practice doing is cleaned, etc.
setups. Eliminate adjustments. Abolish Removing and mounting workholders, dies,
the setup itself. tools, parts, etc., includes the removal of parts
5. Streamline external elements. and tools after completion of processing. This
also includes insertion of workholders, tooling,
Determine the Existing Method parts, and cutting tools for the next job.
Operational analysis using motion and time Measurements and calibrations must be
studies can be used to determine the current made prior to performing a production opera-
setup procedure. The usual objective is to tion, such as centering, dimensioning, tempera-
improve work methods, eliminate unnecessary ture, pressure, etc.
motions, and arrange the necessary motions into Trial runs and adjustments are made and
the best sequence. A setup is broken into short checked against specifications after a test piece
elements and activities that consume the most is machined. The greater the measurement and
time are noted. Problem-solving techniques can calibration accuracy in the prior step, the easi-
be applied separately to each particular activity er it will be to make the adjustments.
to achieve the lowest-possible time.
The frequency and length of test runs and
Setup procedures are usually thought of as
adjustment procedures depend on the skills of
infinitely varied, depending on the type of oper-
the setup people. The greatest difficulties in a
ation and equipment being used. Yet, when pro-
cedures are broken into elements and analyzed setup operation occur when adjusting equip-
from the single-minute-exchange-of-dies point ment. The largest proportion of time associated
of view, that setup operation includes a with trial runs is derived from adjustment
sequence of steps. The traditional setup time problems. So, manufacturers should seek to
distribution is similar to that in Table 6-1. eliminate trial runs and adjustments from
existing setups.
Table 6-1 Because existing setups can be quite long,
Steps in existing setup process videotaping two or three setups is helpful for
later review and analysis. When the worker, as
Operation Proportion of time
a team member, reviews the videotape, waste
Preparation after-process 30% in the existing setup will be revealed, even
adjustment, and checking of without carrying out an operations analysis.
raw material, dies, jigs, gages, One would expect to find the following in a typ-
cutting tools, etc.
ical metal-stamping press:
Removing and mounting new 5%
dies, etc., for next part • dies of different sizes and heights;
• different dies needing different shut
Centering, dimensioning, and 15%
setting of other conditions
heights on the press, making adjustments
necessary;
Trial runs and adjustments 50% • dies held with long, threaded bolts, often
with stripped threads and badly worn or
Preparation, after-process adjustments, damaged heads;
and checking of materials, tools, etc., ensure • missing tools, nuts, and bolts needed for
that parts and tools are where they should be changeover;
and that they function properly. Checking • operators unable to locate tools or dies to
includes the time period after setup is com- be inserted;
• at least a two-person job to take an old die external setup, the die, tools, and materials
out and/or put a new die in the machine; should be ready for insertion into the process
• in general, wasted motion and time; while it is still working on the first job. Any
• in general, nothing is standardized; modifications or repairs to tooling or dies
• 20-50% of time typically spent on adjust- should be made in advance. In internal setup,
ing (making a part, checking it, adjusting only removal and insertion operations should
settings, and repeating), and 10% on locat- occur. That is, operations must be done with the
ing and securing new tools or dies. process stopped.
A motion analysis of the process using a Lining up dies that are ready for insertion
stopwatch is the next best approach. Such an when the press stops can facilitate exchange of
analysis, however, takes a great deal of time the old for the new die. To facilitate changeover,
and skill. Another possibility is to use a work- roller conveyors can be used for staging and
sampling study. The problem with this option is changing dies, as shown in Figure 6-10.
that work samples are precise only where there A turntable cart (Figure 6-11) adds to the
is a great deal of repetition. Such a study may versatility of the exchange. Alternatively, the
not be suitable when few actions are repeated. table may hold four to six small dies in desig-
A third useful approach is to study conditions nated locations, and dies can slide in and out.
on the shop floor by interviewing workers and The entire table is on wheels so it can be rolled
having them explain what they do in a step-by- out of the way during a process run.
step manner. Even though some consultants A typical die-exchange procedure is as fol-
are advocates of an in-depth continuous-pro- lows (an actual method to be standardized and
duction analyses to improve setup, informal practiced):
observation of the setup and a discussion with 1. Detach the old die from the machine’s bol-
operators often suffices. ster plates.
2. Push table cart over to press and secure
Separate the Internal
table next to press with brake or stopper.
From the External Elements
3. Push old die onto table.
Internal, or mainline, elements refer to setup 4. Rotate table and unload new die onto bol-
actions that require the machine to be stopped. ster.
External, or off-line, elements refer to actions 5. Pull table away from press and attach new
that can be taken while a machine is running. die to machine using the same set of bolts.
The most important step in implementing sin-
gle-minute exchange of dies is distinguishing Figure 6-12 shows yet another approach to
between internal and external setup actions. rapid die exchange in a small press. This
Even though everyone usually agrees that approach uses a carousel conveyor, which is
preparation of dies and workholders, mainte- less expensive and has greater die capacity
nance, and gathering of tools should not be than the turntable cart, but is slightly less flex-
done while processes are stopped, it is amazing ible. A cart services a group of presses and
to observe how often these practices still occur. allows dies to be inserted in any order, as need-
Mastering the distinction between internal ed. A carousel conveyor is constructed around a
and external setup actions is the most direct particular press, and dies for the family of
route to achieving single-minute exchange of parts processed in this press are arranged on
dies. Internal and external elements must be the carousel in order of use. The operator, as
rigorously separated. Once the process is the first order of business, generally is respon-
stopped, the worker should never leave to han- sible for getting the right dies in the right order
dle any part of the external setup. As part of for the day’s production.
Press
Bolster Bolster
Old die set Die set Die set
Bolster
Figure 6-10. Roller conveyors used for staging and exchanging dies to reduce internal setup time.
Bolster plate
New die set
Rollers
Old die set
Off
Bolster
Rotary axis
Base of
press
Turntable cart
Brake
Figure 6-11. Die-exchange cart that holds both old and new dies on a rotary table, adding flexibility to the die-exchange process.
Top View
Punch
press X
Locator block
Carousel
roller conveyor
Mountable
die
Figure 6-12. Punch press equipped with carousel conveyor for quick exchange of dies. Dies are modified to a standard height
for quick location on the press bed (see Figure 6-14) and given a locator element.
Standardization of the height of dies going The same idea can be applied to metal-cut-
into a press and the length of bolts used to ting machines (Figure 6-16). Suppose a vertical
secure dies to the press bed greatly reduces the milling machine is scheduled to process four
internal setup time (Figures 6-13 and 6-14). jobs on four different fixtures. Each setup con-
The methods shown in Figures 6-12 and 6-15 sists of removing the old fixture, then installing
precisely locate dies every time. Adding female
and aligning the new fixture to the spindle of
locator plates to the bottom bolster plate stan-
dardizes the die set. The male locator plate is the machine. Through redesign, four fixtures
permanently mounted on the bed of the are mounted on a turntable and each is auto-
machine (press). This locator establishes the matically aligned to the spindle when rotated
X-Y position of every die set, every time a die is into position. This is an example of one-touch
placed in the press. exchange of dies.
Male die
Male die
Die A Die B
Spacer
320
Spacer added
50 30
to standardize 80 80
height 30 50
Bolt
Nut Standard height
Spacer
Die B Die C
Die A
Figure 6-14. Standardizing the height of the base plate reduces the need for different bolt lengths.
Bottom plate
X Bottom die
B Drill Handwheel
Feed
C
Before: Typically, a machine that
D C has four jobs with four different fix-
tures or jigs would need four differ-
ent setups, each consisting of chang-
ing fixtures and aligning the cutting
tool with the workpiece.
Spring stops
Spring stops
(see inset on left)
Figure 6-16. Modification of machine tool in interim cell to process family of parts with rapid changeover.
A higher-level solution might involve a turret four different parts) on one table. This is an
mill with an oversized table. Remember that example of no-touch exchange of dies.
machines in cells process families of parts.
Reducing the variety of parts coming to the Convert the Internal to External Elements
machine permits the machine to be modified so An important key to reducing setup time is
that setup times can be eliminated. Figure 6-17 converting internal setup operations to exter-
shows how setup was eliminated for four milling nal operations. Chief among elements that can
parts by permanently locating four fixtures (for be readily shifted from internal to external are:
Before
Fixture 2
Fixture 3
Fixture 1 Fixture 4
In storage
After
Oversize milling-
machine table
Digital readout or
NC drive Fixture 1 Fixture 2
Fixture 3 Fixture 4
Figure 6-17. Top view of vertical-milling machine with four fixtures for four parts in family (no setup required).
• searching time (trying to find the correct tional temperature, can be eliminated so the
die, looking for the right tools, carts, fix- production run starts sooner. Temperature sen-
tures, nuts, bolts, etc.); sors on the dies tell the operator when they are
• waiting time (waiting for cranes, carts, at the right temperature for good molding.
skids, or instructions); and External operations for preparing dies, tools,
• setting time (setting dies, tooling, fix- and materials should be routine and standard-
tures, etc.). ized. The internal die exchange also should be
If an activity can be safely carried out when standardized, documented, and posted for workers
the machine is running, then it can be shifted to view. Workers should be instructed to practice
to an external setup. An example is the pre- setups during slack times to master and improve
heating of metal molds using waste heat of the the routine method. The best setup times should
furnace before inserting the molds into a die- be posted for all to see. While a defect-free
casting machine. This means that trial shots, exchange is the primary goal, it is important not
often needed to heat dies to the right opera- to lose sight of safety during the process.
Hydraulic Systems
Hydraulic systems are small and compact;
leaks are visible; pressures are conveniently
adjustable, as with air; and they are self-lubri-
cating. But they also have limitations.
Pneumatic Clamps
Pneumatic clamps are simpler to apply than
hydraulics; supply and adjustment are conven-
ient; they are clean; and leaks are innocuous,
Figure 6-19. Spring-loaded, hydraulic, or pneumatic clamps
for holding dies or fixtures in position to further reduce inter-
other than a hissing noise. On the other hand,
nal setup time. shop-air systems are typically in the 60-120 psi
(414-827 kPa) range versus 2,000-3,000 psi
(13,790-20,684 kPa) for hydraulic systems.
1. Reduction of internal setup time—dies can Consequently, for a given clamping force, pneu-
be loaded and unloaded in less time than matic cylinders must be larger, with the size
is usually needed with manual clamping. ratio inversely proportional to the square root
Obviously, more clamping points present a of the relative pressures. For example, with air
greater advantage. A single switch or lever at 100 psi (689 kPa) and oil at 2,000 psi (13,790
can activate many clamps, either simulta- kPa), the minimum-size ratio would be 1 to
neously or automatically, in a definite 4.47 using standard-diameter cylinders, since
sequence. the size ratio is inversely proportional to the
2. Repeatability (quality)—the clamping square root of the pressure.
force is constant from operator to operator, As far as safety is concerned, using oil at
from shift to shift, and setup to setup. 3,000 psi (20,684 kPa), compared to air, at
Operators may be bright and alert in the approximately 100 psi (689 kPa), is really not
mornings, but tend to tire in the after- as dangerous as it might first appear. Air used
noons. This feature is important as the at 3,000 psi (20,684 kPa) is compressed by a
number of die exchanges increases (for factor of over 200 (3,000 psi 14.8 psi atmos-
example, as lot sizes decrease). pheric pressure). Volumetric compression is
3. Maintainability—constant hydraulic/pneu- about 7%. An accidental line rupture would
matic pressure on the clamping device produce an explosion as air reverts to atmos-
helps maintain secure gripping of the pheric pressure. By contrast, oil is relatively
device, despite vibrations and impact. incompressible. At 3,000 psi (20,684 kPa), volu-
metric compression is only 1.5%. A line rupture ways to reduce setup time and improve the
would produce an oil shower (but that would be process. This is part of increasing operators’
the worst of it). Figure 6-20 illustrates the use multifunctional abilities—the ability to do
of power clamping in the exchange of large dies many tasks beyond “just run the machine and
in a stand of presses. make parts.” Methods analysis helps operators
to eliminate unnecessary tooling movement,
reduce manual effort, eliminate extraneous
walking, etc.
Large punch presses or large molding
machines have many attachment positions on
all four sides. Setup actions for either machine
can take one worker a long time. However,
methods analysis can lead to development of
parallel operations for two workers, eliminat-
ing wasteful movement and reducing internal
setup time. Even though the total labor hours
for setup do not change, the effective operating
hours of the machine increase. Reducing setup
time from one hour to 10 minutes means that
the second worker would only be needed for 10
Common plate-less die clamping system minutes during the internal exchange. Setup
specialists perform many external setup opera-
tions and assist machine operators in setup
actions.
TRA
TRA
Press slide
Standardize Methods
and Practice Setups
Die Dies, tooling, fixtures, part design, part spec-
ification, and methods are standardized. Once
Die a standardized setup method has been
achieved, workers must document it. This
means that workers are asked to write down,
TRA
TRA
most machines or operations, especially for man- Molding machines typically require a differ-
ufacturing cells. Setting is an activity that should ent stroke for the knockout punch, depending
be viewed independently of adjustment. This can on the die size being used. The machine’s
be accomplished by adjusting the machine’s stroke is halted with a switch. To find exactly
instrumentation as necessary to re-establish ini- the right stroke position, an adjustment (move-
tial or previous setup conditions. This allows for ment of the limit switch) is necessary. A mold-
setup without any trial and error. Digital read- ing machine put into a cell environment
outs or limit switches, for example, expedite requires only five positions for the limit switch.
machine resetting without adjusting or fine-tun- Instead of the one-limit switch, five-limit
ing. Setup conditions should be determined, switches can be installed, one at each of five
recorded, and marked so they can be readily and required positions. A simple electric circuit is
accurately reproduced, time after time. Records installed to send electric current only to the
of speed, feed, depth of cut, and other process set- limit switch that needs to be activated. As a
tings should be posted along with data on tem- result, the need to adjust the limit-switch posi-
perature, pressure, and other functions. tion is eliminated (Figure 6-21). The mecha-
Step-function settings, like push buttons on an nism is left alone, and only a function switch is
automobile radio, can also eliminate adjustment. changed to accomplish the change in setting.
Knockout punch
Button
Die cavity
Limit switches to
control depth of
knockout punch
Plunger stop pin
Figure 6-21. Installing limit switches at required positions eliminates knockout stroke adjustments (Shingo 1981).
No adjustment of the limit switch is needed trates a schematic of the rotary stopper. In this
because it has not moved. way, adjustment is eliminated.
In the machining cell for pinions described in Machine-tool manufacturers usually do not
Chapter 4, four different lengths of pinions are know the applications of their products for a
made on the same lathes. Each stopper auto- particular company, so they provide machines
with continuously variable positional settings.
matically controls a cut’s length. Previously,
Machines placed in cells, however, have limited
the position of the machine’s stopper had to be
applications, so the adjustment process is con-
changed and adjusted for each different shaft. verted to steps, often with templates or digital
To eliminate this adjustment, a rotary stopper readouts to accomplish settings without adjust-
with four different stops is used. The stopper is ment (Figure 6-23). Boring machines are often
properly rotated to correspond to the shaft equipped with stops to produce the correct
length during changeover. Figure 6-22 illus- depth of cut on parts. A template can be
7"
Fixed base
Rotary
stopper 10"
20"
Shaft
15"
Figure 6-22. Rotary stopper in lathe operation provides four stopping positions, without need for adjustment.
Table 6-2 shows the result of a setup-reduc- will be felt. Setup is an interruption to the
tion program at one company where it took process. Work-in-process inventory between
four years to reduce setup time to less than cells protects downstream processes from
100 seconds for 62% of setups. Getting to this upstream problems. The goal is to develop a
level of setup-time reduction typically comes robust system of serial processes that are
in phases. The first phase requires little capi- responsive to change.
tal expenditure, and solutions can be achieved
in a relatively short time. Reductions of 20- SUMMARY
30% are typical. No analysis other than video- Here are some results that can be expected
taping is required if everyone gets involved. from setup reduction:
This is a method involving workers. The objec-
tive is to improve setup incrementally, until it • small-lot production becomes possible;
is eliminated or economically prohibited. • higher-inventory turnover;
The second phase involves operation analy- • better use of manufacturing space;
sis, minor modifications to dies, tools, fixtures, • no stock-handling operations;
machines, and procedures, and a modest expen- • no defective stock;
diture. Again, benefits of 30-50% in setup-time • no losses from product deterioration;
reduction can be achieved in a relatively short • mixed-model production;
period of time. • machine-production rate improves and
The third phase may involve methods analy- production capacity expands;
sis, design changes, and standardization of dies, • setup errors disappear, number of defec-
tools, parts, machines, operations, and proce- tive goods diminishes, and product quality
dures. Large capital expenditures may be improves;
required, and complete conversion to rapid • safer operations are possible;
setups may take years to achieve. Benefits of 10- • tool management improves;
40% in setup-time reduction may be expected. • overall total setup time is reduced;
For manufacturing cells, setups of 10 min- • setup-time reductions come at a low cost;
utes probably are achievable in the first phase, • workers no longer resent setup changes;
one minute in the second phase, and 10-20 sec- • need for special skills is eliminated as
onds in the third phase, where machine tools setup procedures become routine;
are custom built for cells. The point to remem- • flexibility improves and the system
ber is that every time lot size is decreased (size responds quickly to demand changes; and
of Container A), the need to reduce setup time • perceptual blind spots are eliminated.
used the entire company’s experience, especial- The purpose of acceptance sampling is to
ly that of the work force. Table 7-1 presents a draw a conclusion about a process by examin-
brief history of quality control in Japan. ing only a fraction of the process. Sampling
While most U.S. companies use statistical inspection is needed when it is difficult, cost-
quality control, the Japanese developed many ly, or impossible to measure an entire popula-
new methods of quality improvement and con- tion. For example, when making razor blades,
trol. These methods depend less on sampling, the expense involved in observing every blade
statistics, and probabilistic approaches and may be prohibitive. Alternatively, the required
more on self-checking and defect prevention, inspection processes may destroy a product.
what is called integrated quality control in Sampling inspection requires some decisions
make-one, check-one, and move-one-on be made. One might ask, for example, “What is
methodology. This chapter discusses the tradi- the maximum percentage of defective product
tional methods of statistical process control, that can be considered satisfactory?” That is, by
along with some of the more popular modern definition, what level of defective product must
thinking. be accepted? This defective-parts percentage is
Integration of quality control into the man- called the acceptable quality level. After deter-
ufacturing system begins with giving workers mining how many samples must be taken to
responsibility and authority to make good achieve a level of acceptance (or rejection),
products. This is key to attacking the source of sampling inspection is carried out. Samples
defects in components. Inspect to prevent a should offer true, unbiased representations of
defect from occurring, rather than finding a parent populations, but this depends on many
defect after a product has been made. This factors, such as sample size and the way it is
simple idea, in practice, however, is not always collected. Usually, it is difficult to obtain a truly
simple. It requires that a manufacturing sys- unbiased sample of a population. For example,
tem be changed to accommodate the tech- if inspectors draw parts from the top of each
niques and methods of integrated quality box of parts, operators quickly learn to put the
control. best parts on top so the best work is inspected,
and the entire lot accepted. But it is the accept-
STATISTICAL QUALITY CONTROL able quality level or the number of satisfactory
Statistical quality control began at Bell defects that creates the most problems with
Telephone Laboratories in the 1920s. Since sampling inspection. Today, in a world of high-
that time, its popularity grew across the globe quality competitiveness, defective products are
as a multitude of industries began to use it. absolutely unacceptable.
There are many different tools for statistical Control charts are used to track accuracy
quality control. The two most popular tech- via statistical means. They also track preci-
niques are acceptance sampling and control sion (or variability) via the range or standard
charts. Both methods use inductive statistics, deviation of a process by plotting selected
which means that a small amount of data (a sample statistics. When a process produces
sample) is used to draw conclusions about a products, no two are exactly alike because of
much larger, if not infinite, amount of data. variations in manufacturing processes, mate-
This large amount of data is often called the rials, and operator performance. Variability,
parent population. The decisions based on the whether large or small, is always present, and
sample cannot be stated with absolute certain- many sources can contribute to it. In 1924, W.
ty. Therefore, uncertainties are encountered, A. Shewhart of Bell Telephone Laboratories
calling for the mathematics of uncertainty, developed statistical charts for process con-
probability, and statistics. trol (Shewhart 1931).
Table 7-1
Brief history of QC in Japan
Late 1940s U.S. occupation forces showed the Japanese how to use statistical sampling tables (Military Standards).
1945–1950 Post-World War II — Japanese industry had to rebuild with only one resource, people. Quality of most
Japanese product was poor.
1950 Deming invited to Japan, introduces SQC to design engineers and manufacturing engineers.
1953 End of Korean War — industry was rocked by the termination of U.S. military contracts. Japanese
industry was reacquainted with the reality that Japan had few resources other than its people, and
that they must develop manufactured goods for export to raise capital and the standard of living.
1953 Juran invited to Japan — introduces quality control to top and middle management.
1950s–1960s Massive quality control training of Japanese executives and managers takes place.
1961 Total Quality Control, a book by A.V. Feigenbaum is published.
1962 Publication of Shop Floor Techniques began. Gemba and quality control techniques are taught to foremen.
1965 Total quality control started at Toyota and changed inspection focus to concentrate on the entire
manufacturing process. Just-in-time marries up with total quality control at Toyota. Goes to all depart-
ments. Company-wide quality control implementation is everywhere — Objective-zero defects.
1966 Quality Control to Expand a Company, a Japanese book by H. Karatsu is published
(often considered the bible of total quality control by the Japanese).
1960s–1970s Total quality control concepts and techniques developed by trial and error on the shop floor.
1970 Quality control subcontractors offer services to small-to-medium companies.
1973 OPEC oil crisis — before crisis total quality control was stagnated. This is most apparent in the reduction
of Deming Awards prior to the crisis. However, the crisis reignited the total quality control movement as
a way for industry to stay competitive.
1975 Construction quality control — nonmanufacturing industry.
1980 White-collar quality control — service industry.
1990 Taguchi methods are widely used in America and Japan; Motorola introduces Six Sigma.
2000 Defects are measured in parts per million as Six Sigma concept spreads.
Shewhart realized that it should be possible thought if the product was bad, so what? It
to determine when variations in product quality could be reworked or scrapped. Over the
are a result of random chance or major process years, an adversarial relationship developed
change (an assignable cause). He developed between the manufacturing and quality con-
control charts for this purpose. There are sever- trol departments.
al different types of control charts, but only In Figure 7-2, a decision about the parent
charts for variable data are discussed here. population based on looking at only samples,
The X-bar chart monitors the process mean. can result in two types of errors, as well as two
The range (R) or standard deviation (S) chart correct decisions:
monitors process variability. Control limits for
• Type I or error, viewing the process as
both chart are usually set at three standard
bad, when it is not producing defects; or
deviations above and below the process aver-
• Type II or error, viewing the process as
age. An example of X-bar and R charts, the
good, when it is producing defects.
most common types of control charts, are
shown in Figure 7-1. In these charts, sample The standard control chart sets its control
statistics (X-bar and R values) are plotted. limits at three standard deviations. This
The horizontal axis is time. Twenty-five sam- means that the probability of making a Type I
ples of size five are plotted. Most values fall error is remote, while the probability of mak-
within the control limits, indicating a normal ing a Type II error is usually quite large. The
prevailing condition. Shewhart showed that person who makes decisions about process
sample data will be normally distributed, quality is not the same as the person who runs
regardless of the population from which sam- the process. The decision maker determines
ples were drawn. If a point (for example, aver- error probabilities. If a sample indicates that
age of sample measurements) falls outside of a process is bad, the decision maker must take
the control limits, it is probably a result of an action; maybe even recommend stopping the
assignable cause. Another indication of an process. If no problems are found, the decision
assignable cause is a run of seven points up or maker makes an alpha error. This can make
down. A run of eight points above or below a the decision maker look bad in the eyes of
central line is another unlikely event, indicat- those who run the processes in the manufac-
ing that something about the process has turing system. On the other hand, a Type II
changed. In Figure 7-1 this occurred, since error requires no action by the decision maker
samples 18-25 fall below the centerline X-bar, and, therefore, no blame is usually assigned.
making it highly probable that the process The “do nothing” decision shifts the blame for
mean has been shifted down. defects to production workers. Of course, the
The historical function of control charts has final customers and the company’s reputation
been to control the accuracy (aim), precision suffer when defective product is permitted to
(variation), and stability (drift) of a process. leave a process.
However, people who run processes (users of In manufacturing cells, workers are given
manufacturing systems) traditionally have tools to control the quality of processes.
not used charts. Rather, people in quality con- Workers use control charts to regulate
trol departments keep the charts. Inspectors processes. While some methods of inductive
used to be sent to the factory floor to gather statistics do integrate workers into the quali-
sample data. The inspector’s job became that ty control system, such methods do not guar-
of quality enforcer or process controller. In the antee zero (or even extremely low) defect
past, production workers viewed their job rates. A common misconception about control
responsibility as meeting production-rate charts is that they indicate what goes wrong
standards, regardless of quality. Workers in a process. Control charts should be used as
Figure 7-1. Example of X-bar and R charts, X-double bar = 0.738 and R-bar = 0.169.
Quality Redefined
The decision maker
took a sample and decided It must be cheaper to do a job right the first
time. As Philip Crosby would say, quality is
Process Process free, but it is not a gift (Crosby 1979). The cost
not changed changed
of quality is the expense of doing things wrong,
Action is taken, like allowing defects. Defects must be found
but nothing is and corrected and such improvements can cost
The truth was that No action found to be wrong money.
the process had as nothing with the process;
is wrong. Type I— error, Genichi Taguchi provided another definition of
not changed.
decision maker quality: deviation from target (Taguchi 1987). By
is embarrassed. this definition, the cost of missing a process tar-
get at the optimal, which is the nominal position,
No action but
Action is taken and is the driver to make the process better. Quality
The truth was problem is found is a result of conformance to specifications or
process is making
that the process in the process;
had changed.
more defects;
decision maker requirements. This means that the standards of
Type II— error. conformance must be precisely stated. Failure to
looks good!
meet conformance standards costs money. The
fastest and surest path to low-cost operations is
Figure 7-2. When the decision maker is not the person who
runs the process, sampling can result in two types of errors in
to make a product right the first time, thereby
the decision about the quality of the products. eliminating rework and scrap. Figures on the
amount of rework done in a typical factory are
scary, and often range as high as 40%, meaning
that 40% of what a company manufactures
used the term total quality control. The idea is
requires some rework. Some refer to this as the
that, if a company takes care of quality, profit
hidden factory.
takes care of itself.
A total quality commitment for all produc- Cellular Manufacturing Systems
tion resources and levels of management is a
Integrated-quality control goes hand in hand
requirement for integrated quality control.
with the concepts of cellular manufacturing
Every person must have an understanding of
systems. Cell workers control quality in their
quality control, the methods used to obtain it,
cells. The rule is make-one, check-one, and
and the benefits. Large quality control depart- move-one-on. The concept is simple: perform a
ments are not the answer. Hiring more inspec- step in a process; check the product to make
tors to work on quality checking is not the sure that the step has been done correctly; and
answer either. Actual control must be inte- move on to the next step.
grated into the manufacturing system. Between processes in a cell, devices can be
Integrated quality control is not a series of added to assist workers in checking parts so
specialized techniques, but part of a manufac- defective products are not passed on. The
turing-based strategy that incorporates quali- checking of parts can be performed by manual
ty control at every level of an organization. or automatic operations. Automatic checking
Line personnel must be given the necessary forms the basis for autonomation. This is an
training to carry out quality control functions. important concept, though the word is often
Eventually, integrated quality control is confused with automation. Autonomation
extended to include vendors, suppliers, and refers to autonomous control of both quality
subcontractors to improve the quality of sup- and quantities. For manned and unmanned
plies and materials. cellular systems, this means that individual
processes or devices between processes are worker can check work pulled from a previous
equipped with sensors to detect: worker or an automatic inspection device
(placed between workers) can check quality
• when sufficient product has been made
characteristics. This is an example of one form
(and not overproduced);
of autonomation using decouplers.
• when something has gone wrong with a
process; and
Make Quality Easy to See
• when something is changing that can
eventually lead to a failure to meet prod- Display boards and highly visible charts are
uct specifications (also called defect pre- placed on the plant floor. Boards detail the qual-
vention). ity factors being measured, the state of recent
performance, current quality improvement proj-
Figure 7-3 shows the relationship between ects, recent award winners for quality improve-
autonomation and lean manufacturing. ment, etc. Quality and its characteristics must be
Physically, sensors and devices, such as decou- clearly and simply defined.
plers, are incorporated into machines to auto- Insist on compliance to quality stan-
matically check the critical aspects of parts at dards. Conformance to quality standards must
each stage of a process. Causes of defects are come first, ahead of output.
investigated immediately and corrective action Give workers authority. Workers must be
is implemented. able to stop a process when something goes
Within a company, quality must be the wrong. Mechanized processes might have
responsibility of manufacturing workers. Their devices to do this automatically (in-process
involvement is absolutely necessary. This inspection). More refined systems may have
means that the primary responsibility for qual- the ability to adjust or even modify a process to
ity is assigned to the people who make the correct a problem. A machine must be pro-
product. These workers must develop improved grammable. Equipping a machine or process to
habits and a desire for perfection; they must prevent defects from occurring is another form
strive for zero defects. Quality depends on of poka-yoke. For inspection of finished goods, it
everyone’s efforts, from sales, design, and pur- should be a management mandate to check
chasing workers, to those in manufacturing, 100% of the attributes.
shipping, and other departments. Changing a There should be a constant succession of
manufacturing system’s design is critical to projects for quality improvement in every work
changing workers’ attitudes. The effects of area. Continuous improvement should be a
defective products on the linked-cell system routine way of life and can be the work of qual-
cannot be ignored. Quality must be made a pri- ity circles. Quality circles are groups of workers
ority above output rates. who carry out process improvement projects,
especially those affecting product quality.
Control the Process Eliminate incoming inspection. An
The Japanese term for defect prevention is objective is to work toward elimination of the
poka-yoke. At every stage, a product must be inspection of incoming goods. This requires a
checked; thus, every worker must be an inspec- buyer to work closely with one vendor, to the
tor. Quality is controlled at the source. exclusion of others. Ultimately, a vendor
Production workers correct their own errors should be a lean manufacturer and an exten-
and there are no separate rework lines. This sion of a customer’s system.
requires one-piece flow in cells and immediate Eliminate setup time. The drive toward
feedback where problems occur on final assem- small lots requires elimination of setup time,
bly lines. This does not necessarily mean that which, in turn, makes it economical to reduce lot
workers inspect their own work. The next size. The concepts of economic order quantity
AUTONOMATION
Autonomous control of quality (zero defects) and
quantity in a process
Machine stops
Machine loads and Machine stops
when a defect is
made or decoupler unloads parts when required
finds a defective part quantity is made
(EOQ) are limiting. The optimum lot size is one. sary for a plant to improve quality and foster
This is readily achievable within cells. better work habits. Housekeeping is the respon-
Obviously, between cells, the smaller a lot, the sibility of everyone, from the plant manager to
faster the feedback on quality. In addition, the foremen and workers. Proper housekeeping
smaller lots makes problems easier to spot. is needed to improve and maintain safety in a
Keep the workplace clean. Good house- manufacturing environment and to concurrent-
keeping is fundamental and absolutely neces- ly maintain pride and improve company
morale. No one really wants to work in a dirty quality circles involve people and direct their
place. Obviously, a dirty workplace is a deter- attention to workplace problems.
rent to superior quality performance in elec-
tronics fabrication, painting, and finishing Histograms and Run Charts
areas, but also in other areas of a plant that A histogram is a graphical representation of
must be well maintained. frequency distribution using rectangles with
Organize the workplace. Organization is widths representing class intervals and areas
important in areas where there are many loose proportional to corresponding frequencies. The
tools and multiple components. The driving frequency histogram is a type of diagram where
motto here is “a place for everything, every- data is grouped into cells (or intervals), with a
thing in its place.” notation made of the frequency of observations
falling into each interval. When grouping, each
PROCESS ANALYSIS TOOLS AND observation within a cell is considered to have
the same value (midpoint of cell).
TECHNIQUES Histograms are used in many ways in quali-
Table 7-2 lists some common tools of the ty control including to:
integrated quality control methodology.
• determine process capability (control ten-
Remember, in a pull system, quality and other
dency and dispersion);
problems can be exposed through deliberate
• compare the process with specifications;
removal of work-in-process inventory from
• suggest the shape of population (normal
kanban links. Therefore, inventory is a quality
distribution versus skewed); and
control device. That is, inventory is an inde- • indicate discrepancies in data (such as
pendent control variable, rather than a depend- gaps).
ent variable. Problems are discovered and
analyzed using: Frequency values usually are shown as ver-
tical rectangles. Rectangle lengths are propor-
• control/run charts that track process tional to numerical values, while widths are
behavior; proportional to class intervals. A histogram
• process flow charts that define steps in the shows either absolute frequency (occurrence),
process; or relative frequency (percentage). Histograms
• check sheets and histograms that analyze also may be cumulative. There are cumulative
data from the processes; and relative cumulative frequencies.
• Pareto charts that identify the most There are several different types of his-
important problems; tograms, with each type possessing its own
• scatter plots that locate correlations advantage for different situations. Figure 7-4
between factors; and shows frequency versus location for 45 meas-
• cause and effect, sometimes called Ishikawa urements. One observes that the aim of the
or fishbone diagrams, which determine the process is low, but data is within tolerances.
problem causing lack of quality. Several disadvantages of the histogram include
its inability to show trends and take the time
In more recent years, Taguchi’s design of factor into consideration. If data shown in the
experiments and statistical techniques are histogram spreads over time, it is called a run
used to improve processes by reducing varia- chart or graph.
tion. Integrated quality techniques foster con- A run chart (Figure 7-5) plots quality char-
tinuous improvement in the processes. Quality acteristics as a function of time. It provides
circles are used to find (and sometimes solve) general trends and a degree of variability. Run
problems. Widely implemented in the U.S., charts reveal information that a histogram
Table 7-2
Integrated quality control tools
Category Concept
cannot, such as certain trends over time or at basic nature of a process. Inappropriate tools
certain times of the day. are sometimes used to analyze data. For exam-
Individual measurements, not samples, are ple, a control chart or histogram might hide tool
taken at regular time intervals, and points are wear if frequent tool changes and adjustments
plotted on a connected-line graph as a function are made between groups of observations. As a
of time. The graph is then used to illustrate result, use of a run diagram (with 100% inspec-
obvious trends in data. Run diagrams are nec- tion, where feasible) should precede the use of
essary, at least for a short time, to identify the control charts utilizing averages.
LSL USL
Tolerance
20
Process capability
Frequency
15
12 Nominal
10
7
5
4
2
Figure 7-4. Histogram used in quality control shows process data versus tolerances and the nominal values.
Process Flow Chart steps end with arrows showing the direction of
The first task in systems analysis should be flow of parts or information.
the creation of a flow chart. A flow chart is a Making a flow chart often leads to the dis-
pictorial representation of a process, as well as covery of omissions in a process. Flow charts
of decisions and interconnectivity between point out steps that are believed to be taking
steps and decisions. A flow chart provides place, but are not; or a sequence of operations
excellent documentation of a process at a par- that does not make sense when analyzed.
ticular time and it shows the relationships Figure 7-6 is an example of a flow chart. This
between process steps. example is complex, but it deals with the issue
Flow charts are constructed using easily rec- of what happens to component parts, sub-
ognized symbols to represent the type of pro- assemblies, etc., during manufacturing. Flow
cessing performed. A flattened oval represents charts can be used for any process within an
starting or stopping points of a process; dia- organization, from high-level-management
monds represent decisions; and rectangles rep- processes to the process of shipping an order to
resent process steps. Lines connecting process a customer.
First shift 35 40 27 30 30 34 26 31
Second shift 24 23 20 15 23 17 16 21
Third shift 15 13 28 8 20 9 5 11
Fourth shift 16 5 9 13 16 10
40
35
Length
25
2 4 10 16 22 30
5 4 5 Assembly
Fabrication of
outer casting
6
Materials
1 5
Steel sheet,
paint etc.
Fabrication
2
of lining
Packing
1 4
Components
Compressors, 7
thermostats, Parts
Stores
etc 2 fabrication
Dealers
4 Stores and
customers
2 Fabrication
Parts
of door plan
Handles,
nameplates, etc.
3
5
4
Fabrication of
1 = CERTIFICATION
door plate 2 = ACCEPTANCE SAMPLING
3 = INITIAL SAMPLE INSPECTION
4 4 = PROCESS CONTROL, 100%
5 4 5 = VISUAL INSPECTION, 100%
5
6 = FINAL INSPECTION, 100%
Foaming of 7 = QUALITY RATING
Painting
door
Figure 7-6. Process flow diagram for refrigerator manufacturing showing monitoring points.
80
72.2
70
60
50
30
23.9 23.9
20
20
16.7 15.6
13.9 13.3 13.3 12.8
11.7 10
10
6.7 6.7 4.4 3.3 2.2 2.2
0
Table of orthogonal arrays
Ch art o r grap h
F T c ha rt
Weibu ll c ha rt
P ro ce s s flow c ha rt
C au s e & e ffe ct
C o ntro l c ha rt
C hec k s he et
Simple regression
FMEA
dia gra m
H is to gra m
Othe rs
Mu lti- v aria te an alys is
Figure 7-7. Usage rate of TQC methods and techniques reported at QC conference.
diagonal, lines are drawn from the main line. effect diagram that includes the addition of
These lines organize the main factors that cards. The effect is often tracked with a control
could be the cause of the problem (Figure 7-8). chart (Figure 7-9). The possible causes of the
Branching from each factor are even more- defect or problem are written on index cards
detailed factors. Everyone taking part in mak- and inserted in slots in the charts.
ing a diagram gains new knowledge of the
process. When a diagram serves as a focus for a Scatter Diagrams
discussion, everyone knows the topic, and con- The scatter diagram is a graphical repre-
versation does not stray. The diagram is often sentation that is helpful in identifying a cor-
structured around four branches including relation that may exist between a quality or
machine tools or processes, workers, the productive-performance measure and a factor
method, and material being processed. Another that might be driving that measure. Figure
version of the diagram is called the cause-and- 7-10 shows the general structure commonly
Manpower Material
Contamination
Mishandling
Sputtered sheets
ist
er
res
mb
Pits
Ca
Pho
Not following
master production
schedule S/S trays dirty
Usage of lids
P.R. not filtered
Master production
Dirty ovens schedule correct Time
Covering photo resist
Changing of foil Transferring of photo resist
Machinery Method
Figure 7-8. Fishbone diagram showing causes that produce the effect captured by the X chart.
Causes Effects
Figure 7-9. Basic structure of a cause-and-effect diagram with the addition of cards.
used for a scatter diagram, along with some defects of each area, a tally of problems can be
typical patterns. kept for each step of a process.
The check sheet (Figure 7-11) is also an
Check Sheets excellent way to view data as it is being collect-
Check sheets are used to record data on a ed. It can be constructed using predetermined
process and determine problem areas. Their parameters based on experience with a cell or
use follows the Pareto principle in that they are system. The appropriate interval is checked as
tools to help locate defects, symptoms of data is collected. This often allows the central
defects, and causes of defects. To determine tendency and spread of data to be seen. Here
Y Y
X X
Strong, positive Strong, negative
Y Y
X X
Weak, positive Weak, negative
Performance measure
(dependent variable)
Y0 Driving factor
(independent variable)
X0
None
15
14
13
12
11
10 x
Number of observations
9 x x x
8 x x x x x
7 x x x x x x
6 x x x x x x x
5 x x x x x x x x x x x x
4 x x x x x x x x x x x x
3 x x x x x x x x x x x x x
2 x x x x x x x x x x x x x x
1 x x x x x x x x x x x x x x x
6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
central tendency is defined as arithmetic mean, check sheet, a different symbol would be used
or more commonly, the average of the data. for each worker.
Check sheets provide basically the same infor- Figure 7-12 is an example of a check or
mation as a histogram, but they are easier to observation sheet for data gathering.
build (once they are formatted).
Suppose the operation time to perform X-bar and Range Charts
caulking in a redesigned cell is being record-
Variation occurs as a natural part of any
ed. Why is the data bimodal? Looking careful-
ly at the cell design, two workers share the manufacturing system and is an initial rea-
operation. The data would suggest that the son to be concerned with quality control. In a
two workers do not caulk the same way, since perfect world, processes would make multiple
there appears to be a five- or six-second dif- parts and each part would be identical.
ference in their average times to complete the However, in the real world, differences occur
task. Clearly, further study is needed to and tracking these differences helps products
determine what is going on. For the next meet customer needs. Two aspects of this
process that are of concern are the central just the opposite in that they produce easily
tendency (or where the process is centered), detectable changes in the output of a process.
and variability (or spread of the process Efforts to eliminate assignable causes pro-
around the mean). Control charts, or more duce a good return.
specifically, X-bar (mean) and range (R) Many types of control charts have been
charts have traditionally been two ways of developed, but the X-bar and R charts are the
tracking these conditions. ones used for variable data (that is, aspects of
Variability is thought to consist of two the process that are measured). Figure 7-13
types, that which is inherent in the process shows an example of X-bar and R charts. A con-
called random or chance cause, versus that trol chart statistically determines upper- and
which exhibits observable or detachable caus- lower-control lines drawn on either side of a
es when acting on a process. These are tradi- process average. These are upper- and lower-
tionally called assignable causes. Random control limits. They help determine if a process
variation is a natural part of any process and is in control or if it has digressed to being out of
is difficult to isolate. Assignable causes are control. There are many rules to judge the state
13
UCL=12.83
Sample mean
12 X=12.00
LCL=11.16
11 1
Subgroup 0 50 100
3 1
UCL=3.060
2
Sample range
R=1.447
LCL=0.000
0
0 50 100
Sample Number
UCL = upper control limit
LCL = lower control limit
Area under
normal curve
99.7%
Calculate UCL x = x + A2R
95.5% for each sample
X ∑x
68% Chart x=
(Mean) x = ∑ x i k
n 3σx
n = sample size LCL x = x – A 2R
(Range) R = Hi – Low
or
UCL R = D 4 R
(sigma)σ = ∑(xi – x) 2
3σR
n
R
k = number of samples Chart ∑R
R=
and k
n = sample size
LCL R = D 3 R
–3σ –2σ –1σ +1σ +2σ +3σ
Sample number or time
Calculate the average range R or the
average σ and process average X after
k samples Out of control
• Run — 7 points on one side of the
+ + ....σ k
σ = σ1 σ 2
R1 + R2 + R3 +......R k or
R= central line
k k
• Trend — 6 successive increasing or
X1 + X2+X3 +......X k decreasing points
X= • Point outside the control limits
k
UCL x = x + A2R = x + 3σx
LCL = x – A2 R = x – 3σx Sample
x X Chart R Chart Estimate
Size
UCL R = D4 R
LCLR = D3 R (n) A2 D3 D4 d2 c2
2 1.88 0 3.27 1.13 0.56
3 1.02 0 2.57 1.69 0.72
R σ 4 0.73 0 2.28 2.06 0.8
Note: σ ′ = or 5 0.58 0 2.11 2.33 0.84
d2 ec22
6 0.48 0 2 2.53 0.87
σ′ R 7 0.42 0.08 1.92 2.7 0.89
3σ x =3 =3
n d2 n 8 0.37 0.14 1.86 2.85 0.9
9 0.34 0.18 1.82 2.97 0.91
so 3σ x = A2R 10 0.31 0.22 1.78 3.08 0.92
High
Low
(c) (d)
Figure 7-15. The concepts of accuracy and precision are shown in the four target outcomes: (a) The process is accurate and pre-
cise; (b) Accurate but not precise; (c) Not accurate but precise; (d) Neither accurate nor precise.
All manufacturing processes display some Some examples of assignable causes of the
level of variability, referred to as inherent capa- variation in processes include multiple
bility or inherent uniformity. For example, sup- machines for the same components, operator
pose that the task of shooting at a metal target blunders, defective materials, or progressive
is viewed as a process of putting holes in a wear on the tools during machining. Sources
piece of metal. The shooter may take 10 shots of inherent variability in the process include
at a bull’s eye. Thus, the shooter is the operator the natural variation in material properties,
of the process. Figure 7-15 gives some possible operator variability, vibrations and chatter,
outcomes. To measure process capability, that and the wear of the sliding components in
is, the ability to consistently hit a target, the the machine, perhaps resulting in poorer
target must be inspected after the shooting is machine operation. These kinds of varia-
finished. Thus, measuring process output tions, occurring naturally in processes, often
determines the process capability of a manu- display a random nature and cannot be elim-
facturing process. A product is examined to inated. In quality control terms, these are
determine whether or not the process is what referred to as chance causes. Sometimes, the
was specified in the design. causes of the variation cannot be eliminated
Process capability studies use statistical and because of cost. Almost every process has
analytical tools, similar to quality control stud- multiple causes of variability occurring
ies, except that the process-capability results simultaneously, so it is extremely difficult to
are directed at machines used in processing, separate the effects of the different sources
rather than output or products from a process. of variability during analysis.
Returning to a previous example, a process-
capability study would be directed toward MAKING PROCESS
quantifying the inherent accuracy and preci- CAPABILITY STUDIES
sion in the shooting process. The quality control The objectives of a process-capability study
program would be designed to root out the are to determine the inherent nature of the
problems that can cause defective products. process as compared to the desired specifica-
Traditionally, the objective of quality control tions. The traditional approach is to examine
studies has been to find defects in a process. A the process output under normal conditions,
progressive point of view is to inspect to pre- or what is typically called hands-off condi-
vent defects from occurring. tions. The inputs, (materials, setups, cycle
The nature of the process refers to both the times, temperature, pressure, and operator,
variability (and inherent uniformity) and aim for example), are fixed or standardized. The
of a process. Thus, in the target shooting exam- process is allowed to run without adjustment
ple, a perfect process would be capable of plac- while the output (the product or units or com-
ing 10 shots directly in the middle of the bull’s ponents) is documented carefully with respect
eye, one right on top of the other, leaving one to: (1) time, (2) source, and (3) order of pro-
hole after 10 rounds. The process would display duction (Figure 7-16). A sufficient amount of
no variability with perfect accuracy. Such a per- data must be taken to ensure confidence in
formance would be unusual in a real industrial the statistical analysis, and then calculations
process. Variability may have assignable caus- outlined in Figure 7-17 can be done. The pre-
es and may be correctable if the cause can be cision of the measurement system should
found and eliminated. Variability where no exceed process capability by at least one order
cause can be assigned and where it cannot be of magnitude.
eliminated, is inherent in the process and is Prior to data collection, these steps should be
therefore considered “nature.” taken:
x
Manufacturing Measure x with gage
process
Process plan to Parts
manufacture the part (Machine tool
+ tooling) Distribution of
measurements —
Raw materials Collect measurement of x
histogram or
Information to parent population
manufacturing
engineering
Mean or average of distribution
Part
drawing Std dev.
x
LSL USL
Nominal
LSL
Time
Figure 7-16.The process capability study compares the part as made by the process to the specifications called out by the designer.
Histogram and run charts are commonly used.
Parts
σ
= ?
Step 1
X
= ? Sample statistics, sample size = n
A look at the first 70 parts X = x i
from the process n
USL (upper specification limit
desired size
15
assigned bydedesigner)
2
σ = X i
LSL (lower specification limit
2
–X
n
5 to 70 units
Sample
Frequency
pe
assigned by designer)
Step 2
X is estimated by X
k
R
R= R
;
is estimated by R/d2 or
k
R
is estimated by c
2
0.995
0.996
0.997
0.998
0.999
1.000
1.001
1.002
1.003
1.004
1.005
1.006
1.007
1.008
where =
k
Histogram of 70 measurements of a parameter.
The design specification was 1.001 ± 0.005 in.
Figure 7-17. Calculations to obtain estimates of the mean ( X′ ) and standard deviation (σ′ ) of a process.
• Design a process capability experiment. the data recorded on a data sheet. A frequency
• Use normal or hands-off process condi- distribution, or histogram (Figure 7-16), is
tions; specify machine settings for speed, developed. The histogram shows raw data and
feed, volume, pressure, material, temper- desired values, along with the upper- and
ature, operator, etc. This is the standard lower-specification limits. These also are called
process capability study approach. lower- and upper-tolerance limits. Statistical
• Use specified combinations of input data is used to estimate the mean and stan-
parameters that are believed to influence dard deviation of the distribution.
the quality characteristics being meas- The mechanics of the statistical analysis are
ured. Combinations should be run with shown in Figure 7-16. The true mean of the
an objective of selecting the best. For parent population or distribution, designated
example, speed levels may be high, nor- X ′ , is compared with the nominal value. The
mal, and low, and operators may be fast estimate of true standard deviation, designated
or slow. This is the Taguchi or factorial
, is used to determine how the process spread
approach. compares with the desired tolerance. The pur-
• Define the inspection method and pose of the analysis is to obtain estimates of X ′
means; that is, procedure and instrumen- and
values, the true process parameters, as
tation. they are not known. A sample size of five is
• Decide how many items are needed to per- used in this example, so n = 5. Fourteen groups
form statistical analysis. For a Taguchi of samples are drawn from the process, so k =
approach, this means deciding how many 14. For each sample, the sample mean, X and
replications are desired at each calibration sample range, R, are computed. For large sam-
level. ples (n > 12), the standard deviation of each
• For a standard process capability study, use sample should be computed rather than the
homogeneous input materials and try to con- range. Next, the average of sample averages,
trast these with normal, more variable input X , is computed. Sometimes the X is called the
materials. For a Taguchi approach, material grand average. This estimates the mean of the
is an input variable specified at different lev- process, X ′ . The standard deviation of the
els: normal, homogeneous, and highly vari- process, which is a measure of spread or vari-
able. If a material is not controllable, it is ability of the process, is estimated from the
considered a noise factor. average of sample ranges, the R , or average of
• Data sheets must be designed to record date, sample-standard deviation, σ , using either:
time, source, order of production, and process
parameters being used or measured while σ ′= R/d2 (7-1)
data is gathered. For a Taguchi method, this or
is usually a simplified experimental design,
called an orthogonal array. σ ′= σ / c 2 (7-2)
• Assuming that the standard process capa-
The factors d2 and c2 depend on the sample
bility study approach is being used, the
size, n, and are given in Figure 7-14. The
process is run and parts are made and
process capability is defined by ±3
or 6
.
measured. For a Taguchi experiment, the
Thus, X ′ ±3
defines the natural capability
order of part making is randomized.
limits of the process, assuming the process is
Assume that a designer specifies a part to be approximately normally distributed.
1.001 in. ± 0.005 in. (2.54 cm ± 0.01 cm). After Note that a distinction is made between a
70 units have been manufactured without any sample and population. A sample is of a specified,
process adjustment, each unit is measured and limited size and is drawn from the population.
The population is the large source of items, The condition for not producing rejects is:
which can include items a process produces
X ′ – N + 3
0.5(USL – LSL) (7-5)
under the conditions specified. The calculations
assume that the population is normal or bell- Another capability index, Cpk that combines
shaped. Figure 7-14 showed a typical normal both factors is:
curve and the areas under the curve that stan-
dard deviation defines. Other distributions are
possible, but the histogram in Figure 7-14 clear- C pk = min .
{|USL − X|′ , |LSL − X|′ }
ly suggested that a normal probability distribu- 3σ ′
tion can best describe the process. Now it
remains for the process engineer and operator to In Figure 7-18, the process capability
combine their knowledge with results from the almost equals the assigned tolerance spread,
analysis to draw conclusions about the ability of so if the process is not perfectly centered,
the process to meet specifications. defective products result. If this is the case,
Process capability studies can provide as shown in Figure 7-18, the process needs to
answers to these questions. be re-centered so the mean of the process dis-
1. Does the process have the ability to meet tribution is at or near the nominal value.
specifications? To answer this, a process capa- Most processes can be re-centered. Poor accu-
bility index, Cp, often is computed using the racy often is due to assignable causes, which
estimated
for the parent population. can be eliminated.
USL − LSL T
Cp = = (7-3) MOTOROLA’S SIX SIGMA
6σ ′ 6σ ′ To meet the challenge of both international
and domestic competition, Motorola developed
where: the six-sigma concept. The concept is shown in
Cp = process capability index Figure 7-19 in terms of four-sigma and six-
T = tolerance spread sigma capabilities. Most people do not know
USL = upper specification limit what sigma represents. Sigma represents a
LSL = lower specification limit standard and it is a measure of variability,
repeatability, or lack of precision in a process.
A value of Cp = 1.33 is considered good. Most explanations regarding six sigma use
2. Is the process well centered with respect incorrect diagrams. Figure 7-19 shows the cor-
to the desired nominal specification, the tar- rect relationship between four sigma and six
get? To answer this question, the deviation, , sigma. The distance between the upper- and
from the target is calculated: lower-specification limits does not change (or
increase) for a process to improve from four- to
estimated process mean − nominal five- or six-sigma levels. However, variability,
∆=
1/2 tolerance spread as measured by
, is decreased through process
(7-4) improvements.
X′ − N The Taguchi method incorporates the follow-
=
1 / 2(USL − LSL) ing general features:
• Quality is defined in relation to total loss
where:
to the consumer or society from less-than-
∆ = deviation from target perfect product quality. Methods include
X ′ = estimated process mean placing a monetary value on quality loss.
N = nominal Anything less than perfect is waste.
USL
Capability 6
Out of specification
(b)
Capability 6
(c)
LSL
USL
Figure 7-19. To achieve six sigma capabilities from four sigma capability, the precision of the process must be greatly improved.
Old standard R
New R
Reduced variability after
Taguchi
0 Time
V V V V
V II
V II V II
V II
VI
VI
VI
VI High
Low High Low Low High Low High
I II I II I II I II
A B C D
Figure 7-20. The use of Taguchi methods can reduce the inherent process variability as shown in the upper figure; factors A, B,
C, and D versus process variable V shown in lower figure.
• In a competitive society, continuous quali- The methods are more than just mechanical
ty improvement and cost reduction are procedures. They infuse an overriding new phi-
necessary to stay in business. losophy into manufacturing management, basi-
• Continuous quality improvement requires cally making quality the primary issue in
continuous reduction in the variability of manufacturing. The manufacturing world is
the product-performance characteristics. rapidly becoming aware that the consumer is
Taguchi methods can be an alternative the ultimate judge of quality. Continuous qual-
approach to making a process capability ity improvement toward perfect quality is the
study. The Taguchi approach uses a trun- ultimate goal. Finally, it is recognized that ulti-
cated experimental design. This is called mate quality and lowest cost of a manufactured
an orthogonal array, and determines product are determined to a large extent by
which process inputs have the greatest engineered designs of the:
effect on process variability (for example, • product,
precision), and which have the least. • manufacturing processes, and
Inputs with the greatest influence are set • manufacturing system (integration of
at levels that minimize their effect on product and process).
process variability. As shown in Figure 7- There is a new understanding of quality stat-
20, factors A, B, C, and D have an effect on ing that inherent process variability is not
process variable, V. By selecting a high fixed. Exploiting the nonlinear effects of prod-
level of A and low levels of B, C, and D, the ucts and/or process parameters on the perform-
inherent variability of the process can be ance characteristics reduces the noise level of a
reduced. Those factors that have little process.
effect on the process variable, V, are used Future manufacturing management should
to adjust or re-center the process aim. In include the following changes:
other words, Taguchi methods seek to min-
• continual training and massive implemen-
imize or dampen the effect of the causes of
tation of statistical process control;
variability and, thus, reduce total process
• use of statistical process control as an
variability. interim quality control measure until the
• Engineering product designs and their full slate of quality control techniques are
manufacturing systems determine the fully integrated into American manufac-
quality and cost of a manufactured product. turing;
• Exploiting the nonlinear, interactive • training and implementation of Taguchi
effects of a process or product parameters methods for process design and improve-
on performance characteristics reduces ment of products and processes;
the variability in product/process perform- • concurrent engineering of products and
ance characteristics. processes to reduce time needed to bring
• Statistically planned design of experiments new high-quality, low-cost designs to the
or Taguchi experiments determine the set- marketplace; and
tings for processes and other parameters • attitude adjustment—making quality a
that reduce performance variation. primary consideration and the process of
• Design and improvement of products and improvement continuous.
processes can make them robust, and thus
insensitive to uncontrollable or difficult- Teams and Quality Circles
to-control variations called noise by Popular programs are built upon the concept
Taguchi. of participatory management, such as quality
circles, self-managed work teams, and task of each shift to have cell-team meetings. These
groups. These programs have been successful meetings discuss daily events and seek solu-
in many companies, but have failed miserably tions to problems. This is real continuous
in others. The difference is often due to the process/system improvement enlisting entire
manner in which management implements the cell teams and supports members when it is
program. Programs must be integrated and necessary.
managed within the context of lean manufac-
turing-system strategies. For example, man- POKA-YOKE
agement asking an employee for a suggestion
While many people do not believe that the
that it does not or cannot use defeats a sugges-
goal of zero defects is possible to reach, many
tion system. Management must learn to trust
companies have achieved zero defects for a
employee ideas and decisions and move some
length of time or reduced their defect level to
decision making to the factory floor in tune
virtually zero using poka-yoke and source
with financial constraints.
inspection.
The quality circle movement started in
Poka-yoke is a Japanese term for defect pre-
Japan in 1962 and grew rapidly. Quality circles
vention. Poka-yoke devices and procedures
employ or embrace participatory management
are often devised for preserving the safety of
principles. A quality circle is usually a group of
employees within a department. Its organiza- operations. The idea is to develop a method,
tional structure includes members, a team mechanism, or device that prevents the defect
leader, a facilitator, frequently a manufacturing from occurring, rather than to find the defect
engineer, and a steering committee. Its main after it has occurred. Poka-yokes can be
objectives are to: attached to machines for automatic checks of
products or parts in a process. They differ
• provide workers with chances to demon- from source inspections in that they usually
strate their ideas; are attributes inspections. The poka-yoke
• raise employee morale; device prevents production of bad parts. Some
• develop worker knowledge and quality devices may automatically shut down a
control techniques; and machine if a defect is produced, preventing
• introduce problem solving methods. production of an additional defective part. The
Solutions to typical production-type prob- poka-yoke system uses 100% inspection to
lems provided by quality circles unify compa- guard against unavoidable human error.
ny-wide quality control activities and clarify Figure 7-21 illustrates a poka-yoke device.
managerial policies, while developing leader- This device ensures that the workers remem-
ship and supervisory capabilities (Table 7-2). ber to apply labels to products, thus prevent-
Quality circles implemented in U.S. compa- ing defective products.
nies meet limited success if they are not part of Poka-yoke devices work when physical detec-
lean-production strategies. It is possible for tion is needed, but sensory detection methods
quality circles to work, but they must be must be used to check many items, such as sur-
encouraged and supported by management. face finish on a bearing race or glass-plate flat-
Everyone must be taught the importance and ness. For such problems, a system of self-checks
benefits of integrated quality control. and successive checks can be used.
Under lean-production philosophies, cell Source inspection looks for errors before
teams are vital resources for continuous they become defects, and either stops the
improvement. Cell teams work together from system, makes corrections, or automatically
initial cell startups. Properly designed cells compensates or corrects for the error condi-
have roughly 15 minutes set aside at the end tion to prevent a defective item from being
Labeler
The worker had to remove a label from
a tape and place it on the product.
by upstream operators. If there is a problem that lines can keep moving. However, if prob-
with parts, defective items are immediately lem products reach the end of the work areas
passed back to the workers at previous sta- on moving assembly lines and the problems
tions. Defects are verified there, and problems still have not been rectified, then red lights are
corrected. Action is immediately taken to pre- lit. Red lights are only turned on if problems
vent any more defective parts. While this is cannot be solved quickly and lines need to be
going on, lines are shut down. stopped. When lines are stopped, engineers and
For successive checks to be successful, sev- supervisors rush to solve problems. When prob-
eral rules should be followed. All possible lems are solved, the red lights go off and every-
variables and attributes should not be meas- one goes back to work. The line start up is
ured. This would lead to errors and confu- synchronized and all workstations start at the
sion from the possible information overload same time.
in the inspection process. The time involved Every worker should be given the authority
to complete possible inspections would be to stop a production line to correct quality
excessive. Parts created by immediate problems. In systems using poka-yoke or
upstream processes should be analyzed and autonomation, devices may stop lines auto-
checked for only one or two key points. Only matically. Assembly lines or manufacturing
the most important elements or perhaps the cells should be stopped immediately and
features more prone to error are inspected. started again only when necessary corrections
Another important rule is that the immedi- have been made. Although stopping lines
ate feedback of the discovery of a defect takes time and money, it is advantageous in
should lead to immediate action. For parts the long run. Problems can be found immedi-
produced in an integrated production sys- ately, and workers have more incentive to be
tem, this is effective in preventing produc- attentive because they do not want to be
tion of more defective parts. If a cell has only
responsible for stopping lines.
one or two workers, they are not in positions
to directly check each other’s work after each
step. Decouplers can play a role here by pro- QUALITY CONTROL DEPARTMENT
viding automatic, successive checking of The basic idea of integration is to shift func-
parts’ critical features (before those parts tions that were formerly done in the staff
proceed to the next processing step). Only organization (part of the production system)
perfect parts are pulled from one process to into the manufacturing system. The quality
the next after passing the automatic-decou- control department serves as a facilitator and
pler inspection. acts to promote quality concepts throughout
the plant. In addition, its staff educates and
Line Stop trains workers in statistical and process con-
Andon boards are important components of trol techniques and provides engineering assis-
final assembly flow lines implementing lean tance on visual and automatic inspection
manufacturing methodologies. Andon boards installations. The quality control department’s
are pairs of yellow and red lights that hang most important function is training the entire
above the workers on assembly lines. The company in quality control.
boards tell everyone the status of the lines’ Other important functions of the quality
processes. The numbers on the boards reflect control department, outside of the company,
the number of stations on the lines. Flowline are to audit vendors and provide technical
workers can turn on yellow lights when assis- assistance. A vendor’s quality must be raised
tance is needed, and nearby workers who are to a high enough level that the buyer will
caught-up can assist. Yellow lights indicate not need to inspect incoming materials,
parts, or the subassemblies. A vendor must quality awards are also effective ways to show
become an extension of the buyer’s factory. the work force that a company is serious about
Ultimately, each vendor must deliver perfect quality.
materials that need no incoming inspection.
Note that this means the acceptable defect SUMMARY
level of incoming material is zero %. The The decade 2000-2010 is bringing a gradual
goal is perfection. transition from statistical process control (as a
For many years, the common thought was
primary means of quality management) to the
that better quality would just cost too much.
more extensive implementation of statistical,
For mass production systems, this was true. To
experimental-design methods. These methods
achieve the kinds of quality that Toyota,
are being used to design and to constantly
Honda, Sanyo, and many others have demon-
improve product and process quality. This is a
strated, the job shop and functional production
natural evolution essential to continual
system must be phased out and replaced by an
improvement of processes, products, and serv-
integrated quality function within a linked-cell
manufacturing system. ices. Anything less is tantamount to stagnation
Difficult, highly technical, and lengthy and potential demise.
inspections can still be carried out in the Although Taguchi methods have only
quality control department. These types of recently been introduced in the U.S., they
quality control or technical inspections have been employed for many years in
include total-performance checks, also called Japan. Taguchi methods were developed as a
end-item inspections, chemical and X-ray result of Japan’s emphasis on quality, limit-
analyses, destructive tests, or tests of long ed resources, and the urgent need to develop
duration. products rapidly. In elemental form, these
methods allowed for design and production
Making Quality Visible of products that are robust and insensitive
Spotlighting of quality programs by visual to environmental disturbances (noise fac-
display should occur throughout manufactur- tors). Further, they are relatively easy to use
ing facilities to make quality evident. These and quick to produce positive results. They
displays tell workers, managers, customers, are the next logical step.
and outside visitors about the quality factors An example of the use of Taguchi techniques
being measured, the current quality-improve- in a brewery follows. It was observed that there
ment projects, and who has won awards for was quite a bit of variation in the bottle-filling
quality. Examples of visible quality are signs machines. This situation led to bottles being
showing quality improvements, framed quali- overfilled to ensure that every bottle had the
ty awards presented to or by the company, minimum volume. This large spread (or stan-
and displays of high-precision-measuring dard deviation) led to a high percentage of
equipment. overfilled bottles. A Taguchi experiment was
Displays have several benefits. In lean man- performed on one of the bottle-filling machines.
ufacturing, customers often visit a factory to Through analysis of data, new operational set-
inspect processes. They want measurable stan- tings for the machine were derived, which
dards of quality. Highly visible indicators of reduced variability in the filling process. This
quality, such as control charts and displays, allowed the average-fill height to be lowered
should be posted in every department. and thus less beer went into each bottle. The
Everyone is informed on current quality goals savings on this one machine amounted to over
and the progress being made. Displays and $400,000 in the first year.
The whole idea of Taguchi methods is to ods are relatively easy to use and fit well into
continuously improve processes, which can the trend toward lean manufacturing. They
result in decreasing variability in processes do not require extensive training and educa-
as shown in Figure 7-22. This requires time tion in probabilities, statistics, or experimen-
and hard work on the part of everyone.
tal designs; so they can be quickly grasped
Competition speeds the rate of innovation.
Due to competition in the free market, there and employed. Every engineer involved in
is reason to expect that proper use of Taguchi process improvement should know how to run
methods will force major improvements in a Taguchi experiment to improve the process-
both products and processes. Taguchi meth- es in their factories.
Nominal Target
LSL
Time
Figure 7-22. Continuously improving processes is part of the continuous improvement effort for the system.
earlier process stages. Keep this thought in less likely to breakdown if they are not under
mind—if a manufacturing process or system is pressure. In the manufacturing arena, steady
not maintained or improved, then the process and consistent performers win long-distance
is degrading. (This is a layman’s restatement of races, not necessarily the fastest performers.
the second law of thermodynamics.) The manu- Placing a time cushion at the end of each
facturing strategy is to design methods to prop- work shift, with a 7.5-hour shift, allows the
erly train people to check processes. manufacturing cell or flow line to stop for team
meetings. The first 10 or 15 minutes of a shift
Scheduling Preventive Maintenance are dedicated to maintenance checks, machine
Scheduling preventive maintenance is a task warm-up, lubrication, tool checks, etc. There is
typically assigned to a mechanical or industri- also additional flexibility to respond to changes
al engineer. Production supervisors believe in product demand because of the wider lati-
manufacturing processes should not be shut tude it affords manufacturing and assembly
down simply for preventive maintenance, yet cells. Thus, to increase the production rate, one
higher-level preventive maintenance must be needs only to add workers to cells and increase
performed on schedule by the maintenance the process-operation rate when necessary.
department. For this reason, preventive main-
tenance should be flexible within certain lim- 4-8-4-8 Scheduling
its. Scheduled preventive maintenance should Integrated preventive maintenance covers the
occur between eight-hour shifts, in four-hour maintenance of machine tools, workholding
time blocks, or on weekends when necessary. devices, cutting tools, and training of personnel.
However, this may be inconvenient for mainte- This function is integrated into the daily regimen
nance engineers and specialists. Processes can of the factory floor. Low-level tasks are shifted
be used on an alternating basis when overhaul from the maintenance department to lean pro-
maintenance is required, and if extra processes duction workers. Daily, workers prepare and use
are available. In this case, where a replacement machine-tool checklists, much like the checklists
machine is available, a machine can be that pilots use to check an aircraft before takeoff.
removed from the cell and replaced with anoth- No machines should crash during an eight-hour
er, so the required overhaul can be carried out. shift. Workers also are responsible for most rou-
The manufacturing system’s pace in an tine machine-tool maintenance. The mainte-
entire plant is synchronized to system takt nance department still carries out major
time; and the system’s needs determine pro- machine overhauls and takes lead roles in
duction rates for specific processes or cells. repairing major breakdowns.
Machines are not operated at maximum rates. For an integrated preventive maintenance
Furthermore, an entire eight-hour shift is not program to operate effectively, the entire plant
scheduled unless it is required to meet daily is run on a 4-8-4-8 schedule. The two four-hour
production needs. Some amount of time (possi- time blocks between the two eight-hour work
bly 15 minutes) at the start and end of each shifts allow for maintenance or unavoidably
shift is allotted for routine repair and mainte- long setups. In addition, an eight-hour shift can
nance. If an entire eight-hour shift is sched- begin early or run over when needed, without
uled, this is a symptom of a serious problem disrupting the next shift. Lean manufacturing
and steps should immediately be taken to iso- systems are designed to be run at less-than-
late and correct the situation. With regular full-capacity to allow for breathing room to
scheduled maintenance, processes last longer keep everything and everybody up and operat-
and provide higher reliability. The idea is not to ing 100% of the time (see Table 8-1).
overtax processes, people, or tools. Workers are Operating machine tools, processes, and
less likely to make mistakes and machines are equipment at reduced speeds or with reduced
Table 8-1
Less than full capacity scheduling
Schedule Meet it every day. Breakdowns and maintenance do not cause stoppages elsewhere.
Marketing Output can be increased on short notice to fill a hot order; buys time to add workers if necessary.
Worker jobs The operator's number one job is making parts—direct labor—but if there is no work (schedule is met)
then the operator performs indirect labor. This is often mental; that is, thinking about improvements in
quality, productivity, and equipment. Mental work may be the worker’s most important contribution.
Automation With automation, the worker may perform 100% indirect labor, for example, quality checker or trouble
shooter. Lean production eases the transition from direct labor to indirect labor.
production rates further enhances machine cells. Companies who have adopted lean manu-
and tool life. This is a foreign concept to most facturing find the following advantages.
North American factories. The idea of less-than-
• The cell and the processes within it are
full capacity operation suggests less-than-100%
designed for the system so the cell oper-
utilization. It is best to worry about effective
ates on a make-one, check-one, and move-
personnel utilization and allow machine utiliza-
one-on basis, with processes meeting cycle
tion to meet demand. However, manufacturers
time, and cycle time is equal to the sys-
want machines to be ready to be used.
tem’s takt time.
• The cell is flexible, allowing rapid
MANUFACTURING ENGINEERING
changeover for existing products and
Manufacturing and mechanical engineers are rapid modification for new products or
responsible for designing, building, testing, and model changes.
implementing manufacturing equipment. • The cell has unique processing capabilities.
Maintainability also should be considered when • The cell has maintainability, reliability,
designing or purchasing equipment. Simple, reli- and durability.
able equipment that can be easily maintained • Equipment and methods are designed to
should be specified. In general, dedicated equip- prevent accidents (safety).
ment can be built in-house better than it can be • The cell is easy for workers to operate;
purchased. Many companies understand that it that is, it is ergonomically correct.
is not good strategy simply to imitate or copy
manufacturing technology from another compa- Equipment is designed and developed with
ny and then expect to make an exceptional prod- the needs of internal customers as a priority,
uct. Companies must perform research and even though factors affecting external cus-
development on manufacturing technologies, as tomers are the highest priority for manufactur-
well as on manufacturing systems, to produce ing engineering. Although many plants lack
effective and cost-efficient products. However, expertise to build machines from scratch, most
effective, cost-efficient, manufacturing systems have expertise to modify equipment and give it
make research and development in manufactur- unique capabilities. Equipment modification to
ing technology pay off. Once a lean production prevent recurring breakdowns requires that
system is implemented, machines can be management assign the highest priority to this
designed, built, and installed into manufacturing work. The most skilled maintenance personnel
must be given this task so breakdowns are cies between the specified and actual
eventually eliminated. speed of equipment; process defects due to
scrap and rework; and reduced yield from
TOTAL PRODUCTIVE MAINTENANCE machine start-up to stable production.
The evolution of total productive mainte- 2. Develop an autonomous maintenance pro-
nance is described as (Nakajima 1988): gram. This means workers are involved in
daily equipment maintenance. The seven
• 1950s: preventive maintenance—estab- steps of autonomous maintenance are:
lishing maintenance functions;
• 1960s: productive maintenance—recogniz- • Initial cleaning: Clean to eliminate dust
ing the importance of reliability, mainte- and dirt mainly on the body of the
nance, and economic efficiency in plant equipment; lubricate and tighten; dis-
design; cover problems and correct them.
• 1970s: total productive maintenance— • Countermeasures at the source of prob-
achieving productive maintenance effi- lems: Prevent the causes of dust, dirt,
ciency through a comprehensive system and spattering of liquids; re-engineer
based on respect for individuals and total equipment parts that are hard to clean
employee participation; and lubricate; reduce the time required
• 1980s: total productive maintenance for cleaning and lubricating.
evolves as an equipment management • Cleaning and lubrication standards:
strategy involving all hands in a plant or Establish standards to reduce the time
facility in equipment or asset utilization. spent cleaning, lubricating, and tighten-
(Without total productive maintenance, ing (specify daily and periodic tasks).
corporate survival is questionable and it is • General inspection: Follow the instruc-
a critical step in the lean strategy); and tions in the inspection manual. Quality
• 1990s: integrated predictive maintenance circle members discover and correct
is a key step in lean manufacturing’s five minor equipment defects.
pillars of total productive maintenance. • Autonomous inspection: Develop and use
autonomous inspection check sheets.
The Five Pillars • Orderliness and tidiness: Standardize
the individual workplace control cate-
The five pillars of integrated preventive
gories; thoroughly systemize mainte-
maintenance/total productive maintenance are
nance control.
(Nakajima 1988):
• Establish inspection standards for
1. Eliminate the six big losses, and thereby cleaning and lubricating; follow clean-
improve equipment effectiveness. The six ing and lubricating standards in the
big losses are: Losses caused by equipment workplace; establish standards for
failure—breakdowns of machine tools, recording data and standards for parts
material handling devices, fixtures, etc.; and tool maintenance.
losses caused by setup and adjustment 3. Fully autonomous maintenance: Develop a
failures—exchange of dies in molding company maintenance policy and goals for
machines and presses, and tooling maintenance. Record the mean time
exchanges in machine tools; speed losses, between failures, analyze the results, and
including idling and minor stoppages due design countermeasures. These steps are
to the abnormal operation of sensors, based on the five basic principles of opera-
blockage of work on chutes, etc.; losses tions management. In the Japanese litera-
caused by reduced speed due to discrepan- ture, they are known as the five Ss: seiri,
seiton, seiso, seiketsu, and shitsuke. A nance through employee participation in main-
rough translation of the five Ss means tenance activities. Employee involvement is key
organization, tidiness, purity, cleanliness, to successfully meeting customer requirements.
and discipline. Develop a scheduled main- As lean production builds high quality into the
tenance program for the maintenance production process itself, rather than achieving
department. This usually is done in coop- it later through inspections and repairs, preven-
eration with industrial engineering. tive maintenance catches trouble at the source,
4. Increase operator and maintenance per- rather than letting anything go wrong and then
sonnel skill. The operators should discuss fixing it.
problems and solutions with the mainte- The foundation of total preventive mainte-
nance people at the time preventive main- nance focuses on the partnerships between
tenance work is done on their equipment. the manufacturing personnel, the mainte-
Part of the operator’s job is to keep records nance department, manufacturing and design
on the performance of the equipment, so engineering, and other technical services to
the operators must be observant or must improve overall equipment effectiveness.
learn to be observant.
5. Develop an equipment management pro- BENCHMARKING
gram including a record of machine and
Prior to the development of a total preven-
tool use that notes how much they were
tive maintenance plan, benchmarking or
used and who used them.
recording the current status of an organiza-
tion is a good idea. This evaluation includes
ZERO DOWNTIME more than the maintenance organization. It
The goal of preventive maintenance is to should embrace all parts of the organization
eliminate machine breakdowns and associated involved in manufacturing maintenance,
problems, including tooling problems, process design, and purchasing of the assets of the
drift, and related incidents. In short, zero company. The most important benchmark for
downtime results in 100% on-demand utiliza- total preventive maintenance is the current
tion. Fighting maintenance fires, a mode that state of the equipment, that is, percentage of
most maintenance departments seem eternally uptime.
caught in, does not ensure zero downtime. Benchmarking provides the starting point
Preventive maintenance and its benefits can- from which to calculate improvement. As such,
not be realized without the efforts of the entire it also provides the means to demonstrate
organization-workers, management, mainte- progress, money savings, and that productivity
nance crews, and other support people. is affected in a way that improves the bottom
Operator involvement is critical. line. Benchmarking also shows the produc-
tion/maintenance teams that they are making
LEAN PRODUCTION progress. This keeps interest and enthusiasm
When considering the maintenance compo- levels high.
nent of the organization, the first focus should Benchmarking involves documenting as
be on a preventive maintenance program. much current equipment data as possible.
Without a good one, the equipment is never From the current status it can be illustrated to
maintained at a level sufficient to ensure that management where the equipment was (in
the organization has assets capable of produc- terms of productivity), where it is, and where
ing a world-class product or service. the trend patterns are pointing.
In the lean system, total preventive mainte- A first step in benchmarking should be an
nance’s goal is to achieve preventive mainte- analysis of the amount of additional capacity
the company could get from its machinery if Hydraulic Inspection Item OK Needs Repair
uptime improved. Such data might include Hydraulic pump Proper oil flow?
quality records for the equipment, time data, Proper pressure?
maintenance downtime histories, setup times, Excessive noise?
and procedures. Other documentation might Vibration?
include photos of the physical condition of the Proper mounting?
equipment at program start (to compare with Excessive heat?
similar photos after total preventive mainte- Intake filter Clean?
nance implementation). Free oil flow?
Directional control Easy movement?
valves Proper oil flow?
PILOT AREAS
Relief valve Proper pressure?
Selecting a pilot area is an important step in
Excessive heat?
the total preventive maintenance sequence. Lines Properly mounted?
Rather than attempting to implement total Oil leaks?
preventive maintenance plant-wide from the Loose fittings?
start, most lean manufacturing companies Damaged piping?
start with a pilot cell or target pieces of equip- Pneumatic Inspection Item OK Needs Repair
ment identified as critical to the manufactur- Compressor Proper airflow?
ing system. Proper pressure?
Excessive noise?
PREDICTIVE MAINTENANCE Vibration?
Proper lubrication?
Implementing a predictive maintenance
Excessive heat?
program, a basic function of the five pillars of Inlet filter Clean?
preventive maintenance, is the next step Free airflow?
established as a formal method for monitor- Directional control Easy movement?
ing equipment and ensuring that wear valves Proper airflow?
trends are documented. In this way, equip- Muffler Proper airflow?
ment can be overhauled, with worn compo- Proper noise reduction?
nents changed, before a failure occurs. Figure Lines Properly mounted?
8-1 shows a typical inspection form for a Air leaks?
hydraulic and pneumatic system. Such forms Loose fittings?
must be developed for belts, chains, and gen- Damaged piping?
eral equipment. Figure 8-1. A typical inspection form for hydraulic and
pneumatic systems.
COMPUTERIZED MAINTENANCE
MANAGEMENT SYSTEM
The organization’s ability to improve its tenance workers, etc. Systems commonly used
service of the equipment is followed by an for this task are called computerized mainte-
increase in the amount of data available for nance management systems, a comprehensive
performing failure and engineering analysis. relational database accessible to the entire
This highlights the need for a computerized organization.
database for tracking and trending equipment
histories, planned work, the preventive mainte- CONTINUOUS IMPROVEMENT
nance program, the maintenance of spare As an integrated preventive maintenance pro-
parts, the training and skill levels of the main- gram matures, it provides increased support to
other advanced programs. As lean manufactur- Standard work is not simply something
ing implementation is embraced, processes oper- management demands and operators comply
ate when they are scheduled and produce at the with. It is implemented to involve management
rate they were designed to deliver. Integrated and operators in the development process.
quality control is enhanced, since the equipment Anyone should be able to follow the instruc-
is stable and produces a quality product every tions of standard work. Toyota’s goal is that a
time it operates. Employee involvement pro- new operator should be able to master the stan-
grams are matured, since most problems are dard work after three days of training on a
related to the equipment. This makes the work-
process.
er’s responsibilities easier to meet and leads to
Combination tables. A standard work
more motivation to maximize their competitive
combination table displays the information at
strengths.
each work center in such a way that anyone
Standardization can follow the instructions. The standard work
combination table in Figure 8-2 shows the 12
The objective of standardization is to develop
the daily routine so continuous improvement operations in sequence and the walking time;
can occur. Clear standards are critical to con- manual time, such as unloading the part;
tinuous improvement. Without standards, checking and load times; the machine process-
internal customers have no goals by which to ing time; and the way these elements combine
judge their work. Operations and methods fall to produce the cycle time. This table is similar
back to the old ways. The result is continual to the standard operations routine sheet, but
fire fighting, not continual improvement. the standard work combination table gives con-
Standards must be set and followed. For trol to the team leader so work being conduct-
every deviation from the standard, the problem ed can be reviewed. The revision date on the
must be identified and eliminated. If possible, standard worktable indicates at a glance how
improve the standard so the same problem quickly improvement activities have taken
does not recur. As more jobs become standard- place. But, even though visual control is useful,
ized, there is less confusion. Training of new the standard work combination tables’ main
operators becomes easier, which is extremely purpose is to further improvement.
important where people rotate jobs often and Charts. An example of a standard work chart
work on multiple machines. In short, standards is shown in Figure 8-3. It has two functions. It
make jobs easier.
allocates an operator’s areas of responsibility
within a work cell or line. Second, it serves to
Standard Work
develop standard work based on cycle time.
Standard work or standard operation is a Standard work can be pursued through the
tool to achieve maximum performance with
following steps:
minimum waste. Standard work is composed of
three elements: 1. Standardize current work practices with
1. Cycle time—the time between completion the assistance of engineers, team leaders,
of the last component or product and com- and workers.
pletion of the next product. 2. Locate problem areas.
2. Work-sequence—the sequence of work 3. Solve the problem and develop new
performed by the internal customer. improved methods.
3. Stock-on-hand—the standard amount of 4. Implement the new methods.
work that is currently underway and nec- 5. If the new methods prove satisfactory,
essary to conduct smooth operations. develop new standard work.
6. Require workers to write the standard tenance crews, and other support people should
work combination tables with supervisory be trained to understand how their roles inter-
assistance. act; and thus, what they must do to support one
7. Justify the process over again. another.
The development and improvement of stan- Operators should learn how to perform rou-
dard work involves everyone, especially the line tine machine maintenance, be instructed in
operators. It is said that Toyota practices proper operating procedures, and develop an
improvement activities to reduce worker task awareness of the signs associated with early
time by as little as half a second. Standardization machine deterioration. Maintenance people
provides the foundation. should learn how to assist production people
with routine activities, readjustments, taking
Employee Involvement corrective actions, and increase maintenance
Machine failures are associated with the way skills. By involving every employee, zero
people think and act. Machine operators, main- machine trouble can be achieved. Employee
involvement ensures early detection and early associated with searching for items out of
correction of abnormal machine conditions. their proper place.
3. Seiso-sweeping-cleaning: Once the work
Five-S site is organized and arrangements are
Toyota developed the Five-S strategy to completed, tools must be kept clean and
describe in more detail what proper house- easily obtainable so there is no fumbling
keeping means. The five Ss are: or lost time. If something goes wrong, a
backup tool should be available, in proper
1. Seiri-sifting-organization: Analyze what is working condition, and stored exactly
available for the task, determine what is where it can be readily found.
required to complete it, and discard what 4. Seiketsu-spic-and-span-hygiene: The work-
is unnecessary. Anything extra is wasteful. ing environment should be as clean as
For example, having extra tools, materials, possible. Hygiene usually complements
pencils, and paper is waste and should be the other aspects of detailed housekeep-
eliminated. ing. Effective organization and work
2. Seiton-sorting-arrangement: Once the min- arrangement is reinforced by keeping the
imum requirement is determined, there entire area as clean as possible, particu-
must be “a place for everything and every- larly the floor.
thing in its place.” Assign a location for 5. Shitsuke-strict-discipline: The other four Ss
essential items. Make the workplace self- must be pursued with strict discipline. The
explanatory so everyone knows what goes rules must be followed and become part of
where. Eliminate confusion and lost time the daily routine. It seems the fifth S is the
most difficult one to follow. Working areas, will be discussed in the next chapter) to gather
tooling, and processes begin as organized, and present facts about the problem.
arranged, and clean, but over time the Team meetings usually occur during the set-
workplace becomes messy and deterio- aside time for each cell after the shift. In cases
rates. This is the second law of thermody- where the cell team needs to stay longer, they
namics in action on the factory floor. All receive paid overtime.
systems degrade with time unless main-
tained. Daily discipline greatly enhances Monthly Engineering Meetings
the Five-S tool.
Monthly engineering meetings are a Kaizen
The Five Ss help to identify problem areas activity. Usually manufacturing engineers
and waste. However, lean production depends present problems that they and the production
on everyone’s active involvement. Thus, every team are attempting to solve. The seven tools of
member of the factory must follow the Five S
quality often are used in a specific problem
principles before results are noticed and sus-
presentation format. Here, the lead engineer
tained on a daily basis.
makes the presentation to a group of engineers,
management, and production team leaders.
KAIZEN ACTIVITIES
The presentation uses as many visual aids as
Kaizen means continuous improvement. It is
possible.
the constant search for ways to improve the cur-
After the presentation, the group is asked to
rent situation, involving all factory functions.
suggest possible solutions. The engineer and the
The Team Concept group select the options to try the next month.
Through this exercise, the engineers and others
In lean production, the team is a key vehicle
through which continuous improvement is in the group learn how to become problem
achieved. Usually a production team works on solvers. This process is not intended to improve
problems in its specified area of the factory. a person’s presentation skills or writing skills.
Along with the production workers, the team The idea is to teach everyone how to use analyt-
may include engineers, maintenance support, ical tools to systematically solve problems.
and adjunct members the team may need to The Japanese demonstrate the ability to use
solve a specific problem. There are two specific Kaizen tools every day, not just for a presenta-
teams in lean production: first, there is the cell tion. They make it a habit to solve problems in
team made up of a specific cell’s workers; sec- a systematic way. As a result, a systematic,
ond, there are teams composed of members written history of the problem exists. These
from several cells or flow lines. written histories, called “story boards,” include
Anyone can make a suggestion for improve- the various implementations for improvement,
ment. Some suggestions the various teams may
and the current status of the situation. As with
implement without approval from anyone
the other elements of lean production, it takes
above it in the organization. The central point
is to give the team the responsibility and the much practice and discipline for this process to
authority to make decisions. become a daily problem-solving habit. The goal,
The team leader for each area is the moder- however, is not simply to go through the
ator. He or she has veto power over any sug- motions of lean production. The goal is a new,
gestion. If problems are of a greater scale, the more effective and challenging way of thinking
team calls on other teams or departments. The and acting that leads to a lean manufacturing
team may use the seven tools of quality (which environment.
Early Wear
Normal life
on out
Number
of failures
Breaking
in
Time
product or process. Hardware systems can fail The best improvement to machine or hard-
in many ways. The machine may fail complete- ware reliability is to decrease the number of
ly and stop running or a period of operational machines that do the same jobs. In manufac-
unavailability due to corrective maintenance turing cells, this means eliminating duplicate
may take place. A component failure in the machinery. This may require switching to a
machine or a power failure may cause the two-shift usage of a machine. This transition’s
machine to be down. The challenge in hardware advantages are threefold. First, when an error
reliability lies in the identification of the root is found, only one machine needs to be checked
causes. Difficulty arises because of the often and corrected. The same is true of process
complex relationships present in factories. improvements in that only one machine is
Before real improvements can be made to required to be replaced or upgraded. The sec-
the hardware in a manufacturing cell, the shift ond advantage is that it is easier to identify the
from crisis management and reactive mainte- source of the error and the machine associated
nance to planned maintenance and service with the fault if there is only one source for
must occur. This is essential and may seem each process. The third advantage is not a reli-
intuitive, but a recent survey of over 70 manu- ability improvement, but rather a general
facturing plants in a variety of industries improvement to the process as it simplifies the
revealed that over half of the maintenance paperwork required to identify the source of a
work performed was reactive. That is, over half fault once an error is found.
of the work done to keep machines running was Slowing the process is another way to
done after the machine stopped running. It is improve machine-tool reliability. This can be
then obvious that reliability of hardware is one accomplished without considerable financial
of the areas in desperate need of improvement. cost.
A third equipment reliability improvement human errors made while creating the code or
is through the implementation of successive designing the system components. Frequently,
checks following each process step. Inspecting this is due to some overlooked condition in the
at each station allows identification and correc- original software plan or an improperly ana-
tion of the causes of errors, before other lyzed operating environment.
processes mask those causes. When studying software reliability, it is
To improve hardware systems, operators important to track not only the system failures,
must possess detailed knowledge of equipment but also its faults. This is important because
operation and be focused on process reliability. while not every error requires the production
Lean manufacturers consistently better their process to completely halt, it may cause repeat-
competitors in hardware reliability. A lean ed effects and slow the system. To return to the
manufacturer allows the operator, someone bar code example, suppose tool parts are being
with the required level of knowledge, to be scanned. If the system scans in a 2-in. (51-mm)
directly involved in hardware design develop- drill bit as a 3-in. (76-mm) drill bit, this does
ment and improvement. This is in contrast to not bring the system to a halt. It could result in
the manufacturer who has a “specialized” or surplus inventory, or cause a machine failure if
“focused” study team making recommenda- the 3-in. (76-mm) tool crashes into the work-
tions, since this group may not be able to piece during rapid transverse. It could also
acquire such knowledge. cause a part defect, a hole 1 in. (25 mm) too
Reliability requires the operator’s involve- deep. Therefore, it is important to track faults
ment in equipment maintenance. Operators do and not failures. However, just because the sys-
not necessarily need to be able to repair every tem does not “crash” does not mean it is error-
problem, but more frequently must be able to free. Software errors can hide in code and
call upon an advanced technician when a more appear long after the system goes into service.
complex problem develops. Thus, maintenance An example of a frequently overlooked fault
should work with the operators to return the is wrong information and/or input to a system.
machine to its original condition. The goal A failed input device or incorrect input may be
should be to restore the equipment to a like- the cause. An input device failure could vary
new condition, rather than just restoration of from a smeared lens on a bar-code reader to a
operational capability. worn cable on a keyboard. An inappropriate
input could result from a mislabeled container.
Software An inappropriate output or response occurs
Software includes data and logic programs when the input to a system is correct, but the
that only control the machine tools and materi- outputs are incorrect. This reveals an error in
al handling devices, and programs for the col- the code.
lection, distribution, and analysis of data Code reliability is highly dependent on the
essential to the production process. Software time allowed between code development and its
may range from the code controlling an auto- use in the field. Time must be allowed for
mated welder to the software routines reading review and testing. Obviously, the more the
inventory bar codes. Software should enhance code can be reviewed, the more errors can be
and simplify processes and operations by doing identified and corrected.
the routine work. However, software is not flex- Software reliability depends on whether the
ible and has “bugs” in it. The bugs or faults in a user is creating the code, or simply modifying
software system arise from the errors in logic an existing code for another application.
or the code, or program, and the faults of the Suppose a program is used to scan bar codes. If
computer hardware that interfaces it to the the program has been used to scan bar codes in
environment. Those errors directly relate to a painting facility, and now the user wants to
scan bar codes in the tool area, the modified • Learn to use brains and not money.
code possesses better reliability than a new one • Eliminate waste—do not plan it forever,
developed solely for the tool area. Because the start today with a plan and be ready to
software is at least a second-generation pro- improve it.
gram, it has already undergone one complete • Refuse to delegate work that does not
round of program debugging, so it is already add value.
free of most errors that caused faults in the • Stop useless reporting.
first generation of the code. 4. Participation—can I ask my subordinates
But remember this, computers are unreli- for advice?
able but humans are even more unreliable. Any • Ultimately, people are our only asset:
system that depends on humans is unreliable. with them, much can be done. Without
them, nothing can be done. The contri-
Workers bution of others limits the effort of one
Reliability depends on the internal cus- individual.
tomers (the workers) required during the pro- • People must be educated, trained,
duction process who keep the process flowing respected, and fairly paid.
continuously. Worker reliability has been fre- • Arrange responsibilities so individuals
quently studied and theorized about. Every have an opportunity to make a mean-
manager has their own theory on how to keep ingful contribution to the organiza-
people on the job and productive. tion’s success. Without this, self-respect
Ed Adams, Production Machine Builders, suffers.
LLC, Knoxville, TN, is a design/manufacturing • Use every employee in a humane and
engineer who designs machine tools, workhold- challenging manner. Everyone is paid to
ers, decouplers, poka-yoke devices, and more think and their thoughts are valuable.
for manufacturing cells under contract from 5. Principles—is the daily routine pre-
lean manufacturing vendors. dictable? (See Table 8-2.)
The following are Ed Adam’s Five Ps for peo- • Standardize as much as possible (all
ple in manufacturing systems. activities). (See Table 8-3.)
1. Purpose—do my efforts complement the • Train involved personnel in those stan-
great task? dard practices.
• Make certain that everyone in the • Rules are made to be broken.
organization knows the purpose. • Only break rules after good participa-
• Be the best in the world at what we do. tive investigation.
• Have a design, cost, and quality that • Institutionalize continuous improvement.
permit us to pick our market.
Final rule—life is more important than work.
2. Passion—how much am I really holding
back?
• Have a sincere, deep, and unrelenting THE SECOND SHIFT
desire to pursue the purpose. The less-than-full-capacity schedule calls for
• Elevate the intensity, hours, and plan- a second eight-hour shift. Perhaps one area that
ning the amount and method. often can be improved in lean manufacturing
• Hire only those willing to do the same. systems is the use of shift work in the process.
• Be accountable for motivating and According to the Bureau of Labor Statistics,
requiring others to do the same. approximately 15.5 million Americans work
3. Pragmatism—could I explain it to my nontraditionally; that is, either rotating or fixed
grandmother? shifts.
Table 8-2
Training, rewards, involvement, motivation for internal customer
Tell all workers that part of their job is to make signs and manuals covering:
Results
Workers and foremen have primary responsibility for job descriptions and training,
rewards, involvement, and motivation. Personnel department has secondary responsibility.
The reasons to be concerned about shift work Permanent shifts allow the person the oppor-
may not be obvious at first. Workers often tunity to adjust to the nontraditional work
report that they feel isolated from management schedule. Rotating shifts does not disrupt the
and the rest of the organization on traditional schedules except during brief peri-
evening/night shifts. A major problem is that ods. The second characteristic used to classify
they experience a loss of contact with friends shift work is the speed and direction of the
and family. Another reason that shift work is rotation. The rotation between shifts is said to
demanding: workers can only accomplish cer- be in a forward direction if a worker moves
tain routine tasks during “normal” business from the day shift to the evening shift. It is
hours. Examples of such tasks are routine backwards if they move from nights to
automobile maintenance, most shopping, and evenings, evenings to days, or days to nights;
doctor’s visits. This causes most workers to typical of a three-shift cycle. The rotation speed
return to daytime schedules on their days off, refers to the amount of time the worker has to
disrupting their physical well-being. adjust to the schedule shifts. This typically
varies from a half day to half a week.
In addition, the number of routine tasks that
The third characteristic of shift work is the
must be completed does not vary depending on
work-to-rest ratios. This is important when
the amount of worker free time. This means
determining worker alertness. As the ratio
that the worker must still do “X” hours of
increases, more work, less rest, alertness of the
household chores, and “Y” hours of routine worker decreases.
tasks each week no matter how much time is The final distinguishing characteristic is
spent at work. Therefore, extra work hours add how the worker views the shift. This means, do
to fatigue and allow less time for rest. they see it as regular and/or predictable or spo-
The result of fatigue, if left uncorrected, is an radic? Workers typically adjust better to shift
increase in errors and safety violations on the rotations when they can prepare for them.
job. The worker’s ability to concentrate on tasks Some ways to improve shift work are to
is significantly reduced. Sick days increase as avoid permanent night shifts and keep consec-
the worker drains energy. In addition, home utive night shifts to a minimum. Avoid several
and family-oriented activities are missed, such days of shift work followed by four to seven day
as a child’s school play or holiday events, and “mini-vacations.” Try to plan free weekends
this can cause additional stress at home. These and keep schedules regular and predictable. In
are only typical samples of the variables in the addition, ensure good inter-shift communica-
complex reliability equation. tion by holding organizational and planning
The lean production system must be refined Final assembly should not pull products from
to bring it to maximum productivity. This refin- upstream subassembly cells, manufacturing
ing entails leveling, balancing, and synchroniz- cells, or production processes in a manner that
ing. These functions must be carried out in the causes production to fluctuate or peak.
proper sequence. Fluctuations cause production planners to set
Knowledge of the definitions of these terms production rates on upstream processes at the
is critical. They are key aspects of lean produc- maximum levels of demand spikes. This, of
tion manufacturing systems. course, results in overproduction and excess
inventory, or waste. On the other hand, it is
• Leveling smoothes production by making desirable to have maximum flexibility in final
the final assembly a mixed model, thus assembly lines so a company can produce to
creating a level demand for components. dealer orders and not to stock. This means that
• Balancing matches the quantity produced a manufacturing system must be flexible. For
by cells and subassembly cells to leveled example, suppose an order normally needs to be
demand. filled four weeks after it is received. If the actu-
• Synchronizing matches the time to pro- al manufacturing lead-time is two weeks, the
duce the component or subassembly to system should be able to change over to meet the
that required by the system. (This is also customer’s desires and easily fill the order.
called build to sequence where the suppli-
er builds components to match the Smoothing Final Assembly
sequence of the final assembly line.)
Lean manufacturing companies have yearly
• Sequencing sends subassemblies (not
production plans that forecast how many items
made in sync) to final assembly in the
they plan to produce. A yearly plan includes a
order and in the build sequence.
running two-month plan. Product types and
quantities are forecasted two months before
LEVELING the delivery month in question. A detailed plan
Leveling is the process of planning and exe- is formulated one month before manufacturing
cuting an even production schedule. In an ideal starts. The amount produced daily is the result
situation, a lean factory will produce an even of the monthly production plan. Leveling and
distribution of products every hour, each shift. balancing are important concepts incorporated
That is, every item should be manufactured the into the daily schedule.
same way, every day. Balancing is setting the Leveling or smoothing tasks is based on
rate of production to match the rate of consump- average total daily production and the aver-
tion. The principle behind leveling and balanc- aged quantity of each variety of products in
ing is simply to regulate final production output this total. For example, suppose a final assem-
and final assembly to minimize demand spikes. bly line produces 10,000 vehicles monthly (as
shown in Table 9-1). This would make it neces- Machines that reach 100% utilization must be
sary to schedule and produce 500 automobiles examined and work must be off-loaded to less-
a day and 10,000 in 20 days. The factory would utilized machines. Work must be simplified and
work two eight-hour shifts per day, separated standardized so new operators are able to
by two four-hour time blocks. Each shift would quickly achieve proficiency. Overtime can
have 480 working minutes. Suppose that 300 accommodate short-term increases. It can yield
variations (options) of these automobiles were up to a 37.5% increase in production. Process
being produced. It would be necessary to bal- improvements that net slack time can increase
ance variations in the daily schedule. output if necessary.
Continuing with the example, suppose that When demand decreases, the steps neces-
three major model types were being produced. sary to correct for this decrease are consider-
The daily average quantities of each type and ably more difficult. Temporary workers must
the sequence on the final assembly can be seen be moved or laid off in manufacturing areas.
in Table 9-1. This is called a mixed-model final Cell cycle times must be increased, meaning
assembly line. There should not be any setup each worker operates more machines. Also,
time between models in this situation. Workers cycle times on assembly lines must decrease,
on such lines need to make at least one of each again reducing the number of required work-
model every day so they do not forget how to ers. Extra workers are transferred to other
assemble each model. areas of the plant or work reduced schedules.
The assembly line receives next month’s An important goal of lean manufacturing is to
schedule from the production control depart- operate the system with a minimum number of
ment in the latter part of the current month. workers. On the other hand, it is not necessary
Each model’s daily average requirements are to operate with a minimum number of process-
calculated. Once a manufacturing cell or es. Having excess machine capacity means that
process receives a monthly forecast of average when demand increases, only temporary work-
daily usage of parts it makes, the process can ers are needed to effectively increase the pro-
adapt its operations to the new schedule. For duction rate, and subsequently, the output.
example, the load on a machine may normally Leveling the schedule is not usually easy.
be set at 90% of capacity, and each worker may Suppose (as shown in Figure 9-1) there is an
operate up to 10 processes. When demand is unbalanced monthly schedule, which is typical
increased, temporary workers must be hired of the job shop. During the first week, suppose
and each worker runs eight machines. such a system produces about 100 units; the
Table 9-1
Mixed-model final assembly sequencing to level the system
Assembly Sequence
4DS, 2DS, 4DS, MV, 4DS, 2DS, 4DS, MV, 4DS, 2DS
next week another 100; and the third week 200. what to expect, he or she will be better able to
So, the system must output 600 units in the control a production area. Production then
fourth week, creating a classic end-of-the- becomes more manageable and process flow
month crisis where everything is a rushed improvements become easier. In addition, when
order. To level such a schedule, the month can different products are produced in a mixed
be divided into two parts. Next, two-week peri- fashion every day, the finished goods inventory
ods can be divided into one-week periods speci- level is lowered. Leveled production also
fying daily quantities. Then, the actual reduces the risk of overproduction.
production profile becomes nearly equivalent to Production smoothing allows people to focus
the leveled schedule profile. Leveling a produc- on production activities without worrying about
tion schedule greatly improves the system’s sudden changes in scheduling or expediting
overall behavior. work. Combined with quick setups and manu-
facturing cells, leveling can improve line per-
Production Leveling formance and reduce parts-shortage problems.
Heijunka is the Japanese term for produc-
tion leveling and a defining goal of lean pro- Advantages and Disadvantages
duction. It is one of the central reasons for The following advantages can flow from lev-
manufacturing stability. In some factories, pro- eled production:
duction levels jump to peak capacities at times.
• Highs and lows of the production process
At other times, production levels are at low vol-
smooth out.
umes. Changes in production volumes are inef-
• There is inventory reduction.
ficient and result in waste. The more changes
• Transportation and handling costs reduce.
made in production volumes, the more waste
• Overall productivity increases in the facto-
created.
ry due to waste elimination.
Waste is created because swings in produc-
• Direct labor hours may decrease.
tion volume mean equipment is underutilized,
• Part suppliers are provided with leveled
thus wasting some of the machinery’s capacity.
loads.
This means that operators have lost the pace of
• The quality system is enhanced with suc-
their work, or the easy rhythm that lets them
cessive checks and poka-yoke devices
exert the least amount of effort over the least
implemented to prevent defects.
amount of time to complete a task. Swings in
production mean that when production The following disadvantages also may occur:
demand is at capacity, people must work over-
• Setup changeovers can be more frequent.
time, raising a product’s cost.
• Workers must be trained in multiple oper-
In a typical mass-production factory, goods
ations, routine maintenance, continuous
are produced in as large a quantity as possible
improvement, and quality control.
to reduce setups. If products cannot be shipped
• Jigs and fixtures must be redesigned to
until the end of the month, excessive inventory
accommodate multipurpose use.
must be accommodated. Double and triple han-
dling is required and there is a risk of product By keeping production volume constant,
damage or deterioration. If quality problems problems are more easily exposed and correct-
occur in the final assembly stage, parts will ed. Lay the proper foundation, outlined in steps
need to be reworked or scrapped according to one through four (discussed in Chapter 2)
the severity of the defect. before making radical changes in production
Leveling production for a given period devel- scheduling. For example, first reduce setup
ops steady material flow and, consequently, times, rearrange the work layout into lean pro-
eliminates much waste. If each operator knows duction cells to match product flow, increase
Units Units
produced a. End-of-month produced b. Two-week time
crisis pattern bucket pattern
1,000 1,000
1 2 3 4 Weeks 1 2 3 4 Weeks
Time Time
Units
Units
produced c. One-week time
produced d. Daily time bucket
bucket pattern
1,000 pattern
1,000
500
500
1 2 3 4 Weeks
1 2 3 4 Weeks
Time
Time
Figure 9-1. Leveling production means a plot of units produced versus time would be linear (dashed lines in figures).
where:
MIXED MODEL FINAL ASSEMBLY
To perform leveled production as shown in TT = takt time, minutes/part
Figure 9-1, a mixed-model assembly line must PR = production rate
be created. The following are the steps for The necessary output per month comes from
designing a mixed-model assembly line. the long-range forecast and actual orders. Once
1. Determine cycle time the cycle time is determined, production control
2. Compute a minimum number of processes. develops standard operations for each of the
3. Prepare a precedence diagram showing multifunctional workers on the final assembly
the relationships among elemental jobs. line. Standard operations are the best arrange-
4. Line balance (not the same as balancing ment of man, materials, machines, and methods
the cells’ production rates). to produce the required amount of products effi-
5. Determine the sequence schedule for ciently. Standard operations are based on the
introducing various products to the line. product demand. As the demand changes, so do
the standard operations and the takt times.
6. Determine the length of the operations of
Takt time represents external customers’
each process.
demand. It is important because it determines
equipment requirements, work balance and flow,
Determine Cycle Time
scheduling, staffing, and capital expenditures.
In a cell, the cycle time is the number of min-
utes it takes for the product to complete a pro- Preparing a Precedence Diagram
duction cycle; that is, the total time it spends in The third step in designing the mixed-model
the workstations and that cell’s processes. In assembly line is the preparation of a prece-
the final assembly, the cycle time is called the dence diagram. A precedence diagram shows
takt time. Takt time is determined by dividing the relationships among elemental jobs. This is
the operating hours per day by the necessary simply identifying which job must go before
output per day (the daily demand) and this sets another. The precedence relationship deter-
the pace for the entire manufacturing system. mines the sequence of operations on the final
MD assembly line to allow for the proper product
DD = (9-1) flow down the line.
D
Line Balancing
where:
The fourth step is line balancing. Line bal-
DD = necessary output per day ancing involves making the amount of work
MD = necessary output per month and, thus, the time it takes to perform the
D = number of operating days per month work (manual tasks at each station) as equal
as possible (in terms of time). Task times must process. This requires great flexibility in the
be less than the takt time. Generally speak- final assembly line at those stations that deal
ing, the line-balancing problem has been with different components or subassemblies on
solved using computer algorithms, which different models. Mixed-model final assemblies
attempt to distribute the tasks to minimize are beneficial because they reduce inventory,
the idle time at each station. The line balance decrease the need for line changeover, and
can change if the product demand changes. meet a variety of demand levels for multiple
When the line runs products that differ, then models.
usually the standard operations, takt time, After the calculation of monthly and daily
and line balance also change. That is the prob- schedules, determination of the daily sequence
lem with mixed-model final assembly. is critical. These schedules set the assembly
order of various models through assembly
Determining the Sequence Schedule lines. For example, a sequence may consist of
The sequence schedule for introducing mod- vehicle Type A, then B, then A, then C, etc. This
els to the mixed-model assembly line has two sequence schedule is communicated only to
main goals. These are: those stationed at the beginning of a final
assembly line and not to any of the workers in
• Leveling the load (total assembly time) at upstream processes or subassembly lines. This
each station within the line—this is still is the most fundamental aspect of the linked-
called “line balancing,” but now planners cell information system that differentiates it
and workers must deal with automobiles from other systems. The various upstream
with two doors and with four doors coming processes and subassembly lines receive only
down the same line. rough monthly forecasts. Supervisors of
• Keeping constant the rate at which each upstream processes must schedule their work
part on the line is consumed—this is forces on the basis of this rough monthly sched-
referred to as “leveling demand” for sub- ule. As a final assembly line builds a vehicle by
assemblies and components. pulling components from kanban stores near
Not all product operations have the same or lines, the withdrawal kanban for these parts is
meet the takt time in mixed-model final assem- detached and sent to specific upstream manu-
bly. Some may be shorter or longer than the facturing or assembly cells. The withdrawal
takt time. If two products that exceed takt time kanban then signals the upstream process to
are introduced to a line one after another, then produce more components in exact quantities
a delay most likely occurs and the line eventu- to be used and removed. Hence, the upstream
ally stops. For this reason, products with long processes do not need detailed production
takt times must be introduced to the line fol- schedules. Kanban informs upstream processes
lowed by short takt times. Workers assist each of downstream needs as components are pulled
other across station boundaries. Workloads are toward final assembly.
smoothed such that each station finishes an Workers on final assembly lines must know
operation at about the same takt time. only what type of vehicle they are building
When mixed-model production occurs, inven- next. This information is provided to them from
tory levels also are affected. That is, finished- the central computer via a printer or computer
goods inventory levels in the batch-production monitor. Information about vehicle types being
modes are higher than the levels of mixed-pro- produced is transmitted to those at the start of
duction modes (Figure 9-1). the assembly line via a computer terminal.
Scheduling a mixed-model, final-assembly This terminal also provides labels for each
line to achieve good line balance is a key com- vehicle. Information on labels instructs assem-
ponent of designing the entire production bly line workers to build specific vehicles.
Major subassembly lines such as engine and plan that is fixed; it then breaks that plan into
transmission lines can also use this labeling daily manufacturing orders. For example,
system. Meanwhile, other subassembly cells, deciding the quantity of vehicles to be produced
manufacturing cells, and processes use kanban in the month of March is finalized in early
to control production quantities and rates. February. Precise daily schedules are planned
Obtaining the best sequence schedule of and production is leveled to produce results
mixed-model, final-assembly production is a such as making the same amount of every
difficult task. The perfect model for sequencing product every day. This daily schedule is only
entails keeping the speed and quantity of with- communicated to final assembly workers and
drawal constant for every component. That is, information regarding specific demands for
the system needs to have variation of con- subassembly and component parts is communi-
sumed part quantities at final assembly held to cated back from the same final assembly work-
a minimum; the consumption rate of each com- ers via their pull system of inventory control,
ponent part must be maintained as constant as known as kanban. In this way the forecast plan
possible. This is obviously impossible for com- is converted to a production plan only for those
plex assemblies, but an attempt to achieve such orders received and accepted. In 1970, Toyota,
a status should be made. under the leadership of Taiichi Ohno, invented
a new kind of manufacturing system capable of
LONG-RANGE FORECASTING producing large volumes of parts in small lots
The Toyota production system is renowned to make this work. Today, in contrast, the
for inventory turns of 70 or more per year, American system is a combination of the func-
while being able to quickly respond to special tional job shop and flow lines that produce
customer orders. Toyota likes to call this an large volumes with large lots and large
instant delivery system. It has a strong mar- amounts of inventory.
ket-research program and forecasting methods Many US companies are still using job shop
to predict long-term demand for vehicles. These systems slowly adopting lean production prac-
forecasts are highly reputed for their accuracy. tices. Small-lot production requires that the job
How does Toyota do this? Does Toyota know shop is eliminated and replaced with a linked-
something about long-range forecasting that cell system where setup time between different
the rest of world does not know? To understand products or component parts is eliminated and
the answers to these questions requires that the final assembly line operates on a mixed-
understanding how the Toyota production sys- model schedule. Subassembly and component
tem works. manufacturing are done in cellular manufac-
Toyota produces a special-ordered vehicle in turing systems using multifunctional workers.
two days or less, but the production period for Workers run numerous processes and are
processing raw material to completion exceeds directly responsible for product quality and low-
this time. The body, frame, and various other level maintenance of processes. Cells are linked
parts are already processed according to a fixed to form a synchronized, integrated manufactur-
production plan, while painting and certain ing system, which has as its ultimate goal the
subassembly and final assembly processes take manufacturing of parts one at a time. Inventory
place over one or two days. levels are greatly reduced and manufacturing
Toyota performs long-range forecasts and lead times are markedly shortened. This new
market surveys of over 60,000 people twice a system can quickly and easily respond to fluc-
year and investigates other trends every two tuations in demand and allow for shorter long-
months. As a result of this forecast methodolo- range forecasting while increasing forecasting
gy, Toyota constructs a monthly production accuracy.
Cell layout
12 11 10 9 8
Output
Worker walks from
7
machine to machine
in the cell, transferring Each worker
workpieces one at a time services six
from machine to machines
machine
Input
6
1 2 3 4 5
Figure 9-2. Manned cell with two workers making two parts, A and B.
WLK
Assembly line
finished goods
storage
WLK
Final assembly Purchased
line-side parts parts storage
storage
WLK
Subassembly
Finished parts
storage WLK
Subassembly
parts storage
WLK
Parts storage on plant floor
Manufacturing Manufacturing
cells for bars cells for pinions
WLK
Suppliers
Figure 9-3. In the lean manufacturing system, manufacturing and assembly cells are linked to the final assembly area by
kanban inventory links or storage areas.
dependent on each other. The primary indicator try, or product. These strategies do not offer
that labor is out of balance is an excess or every available approach to cell formation, but
shortage of material. The amount of work in rather, they offer techniques from which some
process between cells results from unbalanced benefits of cellular manufacturing can be
cycle times. The objective is to set cycle times quickly achieved. They result in grouping sim-
as required by the schedule and then to shift ple, flexible machines to meet required cycle
tasks and operations accordingly. This results times, while eliminating inventory, floor space,
in less and less work for one worker, and even- transportation distance, and quality problems.
tually this worker can move to another area. It Cells can compete with multi-station, high-
must be remembered that systems are speed automatic equipment because they can
designed to optimize labor content, as well as be set up and debugged quickly, therefore offer-
improve labor efficiency. ing manufacturing flexibility.
Plant balancing is a dynamic, ongoing
process. Normally, a plant must be rebalanced Process Cells
whenever the production rate changes. The Process cells are useful when geometric
fear of losing line balance is the major reason information about parts is not available or
for reluctance to stop a balanced assembly line when parts appear to be geometrically dis-
once it is up and running. This is also true of an similar. This method assists in defining cells
entire factory. There are countermeasures to when an understanding of the manufacturing
this difficulty including: processes is limited.
• Visible signaling systems, like andon, Geometric data is unnecessary to create cells
allow a system to respond to temporary based on processes. Parts produced by one key
variations in part-usage rates and to process are selected and then a cell is built
changes in product mix. around that process. Current routings are
examined to find key machine tools and select
• Flexibility built into cells permits a sys-
candidate parts with common routings. Cells
tem to quickly adapt to requests for
then are developed for the part families.
increased or decreased production rates or
to changes in product mix.
Part Geometry Cells
• Less-than-full-capacity scheduling means
keeping a little slack time, perhaps 20-30 Part geometry cells are used when geometric
minutes per shift at each cell, so a system information is available, associated manufac-
can respond to variations from the planned turing processes are well known, and parts look
schedule. This adds flexibility to a system. the same or similar. In other words, geometry-
based cells group parts based on common geo-
Since a plant’s schedule changes periodically, metric characteristics. With process cells,
the operations also may have to change. current production data is used to examine
Manufacturing cells and subassembly cells can routings. However, new routings are then
usually continue working without rebalancing if developed for target part groups based on part-
the cycle times required do not vary more than feature analyses. Consideration is given to
about 10%. Cycle-time variations beyond 10% group tooling, machine loading, parts, and
usually require rebalancing. Detailed planning usage.
for production must take this into account.
Product Cells
MANUFACTURING CELL TYPES Product-focused cells are mini-factories ded-
Manufacturing cells are developed according icated to producing a product such as an
to three basic strategies: process, part geome- assembly, a subassembly, or a finished primary
part. These cells are intended to minimize mechanical transportation mechanism such as
inventory and may be cells that link to assem- a conveyor line. In these situations, the final
bly lines in support of Just-in-time manufac- assembly line would pace the line doing the
turing. A bill of materials details component door assembly. So, suppose the left front door
parts that are going into the subassembly comes off a vehicle at Station 26, goes by over-
where a common routing is developed for the head conveyor to the door subassembly line,
component parts. Many times a factory within and gets fully stuffed with components (win-
a factory results when incorporating both dows, door handles, speakers, etc.). It would
process and geometry-based cells, both of which then return to the final assembly line at
are focused on a common assembly family. Station 110, where the door would be rein-
stalled on the same car body from which it was
SYNCHRONIZATION earlier removed. Thus, the door line must be
Synchronization, refers to the process of kept in sync with the final assembly line.
timing movement of material between por- The yo-i-don method is used in body weld-
tions of the assembly line and major sub- ing. Operations for the body may be divided
assemblies. Even when material quantities into three primary processes such as under-
have been leveled and balanced, unnecessary body, side body, and top body (Figure 9-4). The
storage of in-process material can occur underbody and side-body processes in Figure
between unsynchronized operations. However, 9-4 can be divided into six processes and three
once operations are leveled, synchronization sub-processes, U1-U6 and S1-S3, respectively.
is just a matter of efficient, integrated sched- The top sides and bottom pieces come togeth-
uling. Leveling must precede synchronization er at B1.
because it helps eliminate process delays that Suppose a final assembly line is producing
make synchronization difficult. one unit per minute for the system takt time or
Subassemblies that are synchronized for final factory cycle time. Operations at each of the
assembly in the automobile industry are large subassembly, processing, and main assembly
items such as seats, panels, headliners, cockpits, areas must be completed in one minute or some
doors, and engines. These subassemblies are all
specific to certain vehicle models. They are made
in sequence with the final assembly and deliv- Process flow for body plant
ered to designated workstations on the final
assembly line at the same time as the vehicle S 1 Right
needing that subassembly. Only the best lean U side-body
1 S
manufacturers can accomplish synchronization, 2
process
U2 Assembly B1 station
since any subassembly-line failure can also stop S 3
on final
a final assembly line. Conversely, if final assem- U 3 assembly line
bly is stopped, then synchronized subassembly U U U T T T T
4 5 6 1 3 2 4
cells and lines also must stop so elements stay in
sync. Production processes that have stopped Underbody S
1 Top-body
must restart together. process S process
2
S Left
Yo-i-don Synchronization 3
side-body
Yo-i-don in Japanese means, ready, set, go. It process
is the name given to a method of synchronizing
startup of manual manufacturing processes or Figure 9-4. Synchronization of body parts in spot-welding
lines.
operations. This method is not used with a
Leveling quantities and synchronizing ing for Toyota created the linked-cell system,
processes can significantly reduce delays, thus but he never gave it a name. He simply
greatly reducing manufacturing throughput referred to it as the Toyota production sys-
time. For example, using one-piece flow to elim- tem. Toyota brought the strategy to the US
inate lot delays for two serial processes reduces and it has since been implemented in many
throughput time. The multiplier effect of elimi- forms.
nating process delays and lot delays with one- Leveling and balancing are important fea-
piece part movement in cells may improve tures in lean manufacturing. Leveling
throughput time to 98% if 10 processes are smooths the daily production schedule so the
involved. demand for subassemblies and components is
The final goal of leveling is achieving a the same everyday— same volume and mix.
minimum quantity of material as the stan- Balancing means the upstream elements pro-
dard quantity of work in process. In other duce the necessary daily quantities.
words, only the minimal part quantities nec- Similarly, when leveling and balancing are
essary to complete standard operations are applied to lean manufacturing cells, they reg-
kept on hand. This goal forces elimination of ulate and may reduce the labor content.
excessive work-in-process inventories in Synchronizing the system matches the pro-
links. Therefore, standard operations must duction rates of subassembly processes to the
consist of cycle time, standard-operating rou- rhythm of final assembly. These subsystems
tines, and minimum quantities of material. produce goods in the same sequence and at
Concurrently, the elimination of accidents, the same rate as the final assembly.
breakdowns, and defects is also a major com- A factory without production smoothing is
ponent of standard operations. often consumed by excess inventory and/or
starved for parts. Upstream processes can
SUMMARY experience inventory overflow due to bottle-
Lean production and manufacturing sys- necks, while downstream processes can with-
tems have evolved from the strategies of suc- er away waiting for stock to arrive. The
cessful factories. Lean principles are based proper application of leveling and balancing
on linked-cell manufacturing systems. will remedy these situations and provide key
Taiichi Ohno, Vice President of manufactur- building blocks for a successful factory.
Route sheet
MRP
Schedule triggers production
Final
Pull assembly
WIP = work-in-process
MFG = manufacturing
MRP = material requirements planning
• Zero inventory means there is no invento- More specifically, the single-card system
ry in the manufacturing cells. (Figure 10-3) functions according to the follow-
ing steps:
Kanban System Types
1. The withdrawal kanban from a container
There are many production- and inventory- just emptied at the downstream cell is
control systems that claim to fall under the placed on that cell’s kanban collection box.
kanban heading. Often, any type of system 2. A withdrawal kanban is taken from the
using the pulling of parts and material is called downstream withdrawal collection box for
kanban. In this chapter, however, three sys- delivery to the upstream cell as a full con-
tems are examined. The first is known as the tainer from the upstream stock point
dual-card system, which requires two types of arrives with a withdrawal kanban from a
kanban cards, a withdrawal kanban and a pro- previous cycle attached.
duction-ordering kanban. The second system, 3. Empty containers are then delivered to
single-card kanban, is similar to the dual-card the upstream cell to wait refilling accord-
system but does not use the production-order- ing to that cell’s daily production schedule.
ing card. Finally, a hybrid of push and pull sys- 4. Finally, the withdrawal kanban is attached
tems is examined. This system, known as to a full container in the upstream stock
constant work-in-process, is not a kanban sys- point to await delivery on the next cycle.
tem, but has similar functions and even uses
Although the single-card kanban system bor-
kanban cards.
rows some features from the traditional push
system, it is still based solidly in the lean revo-
Single-card System
lution philosophy. The containers, for example,
The single-card kanban system is simpler are part-specific, have a standard capacity, and
though similar in function to the dual-card sys- are found in specific numbers within the links.
tem. It is actually a combination of push-and-pull In addition, the cells are designed with quick
production-control strategies. The manufactur- setups in mind, allowing small lot sizes to be
ing portion of the system is based on a push phi- delivered in each container. This allows for the
losophy. A daily schedule is assigned to each cell, control of inventory at the downstream point of
and production follows this daily schedule rather use. However, inventory is allowed to build at
than waiting for the signal from a production- the upstream stock point due to the scheduled
ordering kanban. production there.
A withdrawal kanban (Figure 10-2) is used One advantage the single-card system pro-
to pull the parts from the upstream cell only as vides is simple implementation, since there is
needed by the downstream cell. This is the only only one kanban to learn and understand. Once
kanban needed, and it circulates between the this system is mastered, a dual-card system
upstream stock point and the downstream cell. often is developed. Thus, the single-card system
No input stock points are needed, since with- is often an intermediary step.
drawn containers are delivered directly to the The single-card system helps to relieve clut-
downstream cell, as demand requires. The con- ter and confusion around the downstream
tainers move in links according to the following input area, as the need for an input stock point
pattern: full containers from first cell to stock is eliminated. Most importantly, though, it can
point, from stock point to second cell, and final- be implemented even if a completely opera-
ly, empty containers from second cell back to tional lean system is not in place. This feature
first cell. There is accumulation of containers has made the single-card system attractive to
only at the output of upstream cells. companies struggling to move from a job-shop
The card specifies the quantity of parts that the next process can pull from the previous process.
When the card is picked up from the next process (for example, assembly line) and taken to
previous process (for example, rack bar line), it serves as a request.
BACK #:
FROM: 5150 TO: 5130
Assembly Rack Bar Line 002
CARD
PART #: SERIAL#:
P100002 0007
PART NAME:
Rack assembly
QTY: CNTR
5 S2
0020007
Figure 10-2. Kanban card in a one card kanban system,used to pull parts for use in assembly/production.
environment to a linked-cell manufacturing type and hold a precise (usually equal) quantity
system. of that part number. The two types of kanban
The single-card kanban rules are outlined in cards in a dual-card kanban system are:
Figure 10-4.
1. the production order that signals an
Dual-card System upstream cell or process to produce a cer-
tain part; and
A kanban card usually is a rectangular-
2. a withdrawal kanban that serves to link
shaped card enclosed in a vinyl packet and
two cells or processes.
attached to a container of parts. Typically, the
cards are made of plastic so they are reusable. Figures 10-5 and Figure 10-6 show examples
The containers are designed for each component of these two kanbans.
*F
*
* F ** F
* F *F
* F *F
* *
E E
F F
E E
E *F E * F E
E
E E
Rack-bar cell 1 Assembly cell
Figure 10-3. Single-card kanban links the rack-bar cells to rack-and-pinion assembly.
There is precisely one production-ordering and the users understand how it works.
kanban and one withdrawal kanban for each Therefore, the workers trust the system.
container. They identify the part number, con- The container and kanban flow patterns for
tainer capacity, the previous cell, the next cell two cells are shown in Figure 10-7. In this
or process, and other information. A withdraw- example, Cell 1 supplies parts to Cell 2 as well
al kanban specifies the type and part quantity as to other cells in the plant. Cell 1 is serviced
that a downstream process can withdraw from by Stock-point M and Cell 2 is serviced by
the upstream process. A production-ordering Stock-point N.
kanban specifies the part type and quantity Standard-size carts or containers move
that the next cell or process must produce. This material between the cells. Each container
system’s beauty is that it is simple and visual holds an identical number of parts. The carts or
Store
KANBAN RULES Preceding process
Shelf no. N Item back no.
Single card kanban
Item no. Serial # of part
1) Parts are transported in a specified number of Rack bar cell
containers and in specified quantities. Item name Rack bar
2) All parts in a container must be same. Car Type Camry Subsequent process
3) All parts in a container must be of good quality.
Box capacity Box type Issued no. Rack-pinion assembly
4) A full container should have the exact number of
parts as specified on the kanban card. a 3 of 11
5) A kanban card must accompany every full container.
Figure 10-5. An example of a withdrawl kanban for manu-
6) Containers must be stored only in designated areas. facture of rack bars.
7) Kanban cards and containers must be picked up
and delivered between departments by a mail person.
Store
8) Parts must be made or delivered only when Rack cell
requested by a kanban card. Shelf no. M Item back no.
9) As soon as the first part is removed from a full Item no. Serial # of part
container, the kanban card must be immediately
removed and placed in the return post. Item name Rack bar
10) The team leader should be immediately informed
Car Type Camry
of any discrepancies in the kanban system.
containers move in a link or on a circuit from detached from the container and placed in
Cell 1 to its stock point, then to the stock point the kanban collection box for stock-point N.
of the next process. The material then moves 2. The most recently emptied container in
into Cell 2 for processing. Empty carts return Cell 2 is transported back to stock point N
to Cell 1 for refilling. In summary, the with- where a WLK is attached.
drawal kanbans (WLKs) circulate between the 3. The WLK and the empty container return
output side of Cell 1 and the input side of Cell to Stock-point M. (Stock-point N and M
are not usually side by side; they may be
2, just as they did in the single-card kanban
in different parts of the plant or in entire-
system.
ly different buildings.) The WLK is
Production ordering kanbans (POKs) move
detached from the empty container and
between the stock point for Cell 1 and the sup-
attached to a full container of parts. The
plying work cell. For each container, there is
full container with the WLK is returned to
one POK and one WLK. Basically, the empty
Stock-point N. (This is the withdrawal of
container is the signal to the manufacturing
material from the upstream cell by the
cell to make only enough parts to refill that
downstream cell or assembly area.)
cart. Partially full containers are not allowed.
4. A POK is attached to the full container
The system works as follows:
(the one just removed from Stock-point
1. Starting at stock-point N, a full container of M). This was detached from the full con-
parts is moved into Cell 2. The WLK is tainer and placed in the collection box for
F In F
F 4 F F
3 F
F F F
F E F
E
Lean Manufacturing Systems and Cell Design
E E
2
1. Start here
E Standard container E
E F
Withdrawal kanban
F F Production kanban F F
5 Flow path material 5
In Out Flow path WLKs In Out
Flow path POKs
Kanban collection box
Cell display list or box
Flow path, empty cart
E: empty
F: full
Cell 2 uses
parts from Cell 1 Stock-on-hand
CELL 1 CELL 2
223
Chapter 10: Production and Inventory Control
Cell 1. Then, (and only then), can the con- necessary as the withdrawal of any parts
tainer be removed from the Stock-point M without a WLK undermines the kanban
and transported to Stock-point N. information system’s control element.
5. Periodically, POKs are removed from the • A kanban (either WLK or POK) should be
collection box and placed in the dispatch attached to the physical product, or its con-
box for Cell 1. These POKs become the dis- tainer, or should reside in a collection box.
patch list for Cell 1, controlling the order This key point eliminates the possibility of
parts are manufactured in the cell. These any unaccounted inventory in the system.
jobs are performed in the original order If the number of kanbans in the system and
that POKs were received at Stock-point M. the number of parts per container are
6. The parts that are produced in Cell 1 are strictly controlled, the amount of in-plant
placed in the empty containers taken from inventory can be determined and moni-
Stock-point M. The POK is attached to tored at any time. This is the backbone of
each container after it is filled. The con-
the kanban information system.
tainer then returns to Stock-point M to be
withdrawn by a downstream process when Rule 2. The upstream process should pro-
needed. duce products in quantities withdrawn by the
This sequence is repeated many times dur- downstream process or cell, according to the
ing the day. Parts are produced as needed, that information provided by the POK. This is the
is, as withdrawn from the upstream cell. Cell 1 complement of Rule 1. If parts are withdrawn
may produce parts for cells or assembly lines Just-in-time, then, by complying with Rule 2,
other than Cell 2. Other users operate in exact- the parts are produced Just-in-time. This is the
ly the same manner. The carts usually are cornerstone of cellular manufacturing using
color-coded to prevent confusion. the lean production philosophy. The main
points of this rule are:
Rule 1. The downstream cell or process • Production greater than the number of
should withdraw the needed products from the kanbans must be prohibited. This pre-
upstream cell or process according to the infor- vents a large accumulation of excess
mation provided on the WLK (the needed quan- inventory at the stock point for the
tity at the necessary time). This is the realization
upstream process (Stock-point M in Figure
of Just-in-time: parts should be withdrawn as
10-7) and eliminates the possibility of
they are needed, not before they are needed, and
unaccounted inventory in the system.
not in larger quantities than needed. The key
points for enforcing this rule are as follows. • When the upstream process produces var-
ious kinds of parts, manufacturing should
• Any withdrawal without a WLK is prohib- follow the ordinal sequence in which each
ited. This prevents a large accumulation of kind of kanban has been delivered. This
excess inventory at the stock point supply- helps ensure that each type of part is
ing the downstream process (Stock-point ready and available at the upstream
N in Figure 10-7). The number of kanbans process stock points whenever needed at
in the system is tightly controlled. (This the downstream processes or cells.
will be discussed later as part of inventory
control.) The withdrawal of any parts Rule 3. Defective products should never be
without a WLK undermines the kanban conveyed to the downstream process. If there is
information system’s control element. a defect, the line or cell should stop and imme-
• Any withdrawal greater than the number diately try to determine what corrective action
of kanbans should be prohibited. This is should be taken.
One hundred percent quality control is nec- SS = safety stock, usually 10% or less of DD × L
essary to achieve a truly effective kanban sys-
The maximum inventory level (M) is ex-
tem. Based upon Rule 1 and Rule 2, the parts
pressed as:
are produced and withdrawn in the necessary
quantities at the necessary times. If a defective M = aK = DD L SS (10-2)
part is sent to a downstream process, then that
The demand for parts is usually the daily
operation must stop since there is no extra
demand, leveled, or averaged over a daily or shift
inventory in the WLK loop to replace the defec-
amount, from the monthly demand. The lead time
tive part. In practice, the amount of inventory
in the kanban link reflects the probability of a takes into account the time needed to process a
defect occurring in the upstream cell or container of parts, including the time to:
process. Thus, downstream processes are not • change over the cell;
delayed unless the entire inventory in the link • process other items in the family; and
is used. If this happens, then each upstream • convey a container to the usage point, plus
process is stopped until the defective item is any delay times.
reworked, replaced, and the problem corrected.
The practice of integrated quality control and Delay times include lot delay and process
autonomation in a lean production system delay. Lot delay takes into account the fact that
enforces this rule. the first part produced cannot be conveyed to
the next cell or assembly line until the last
Rule 4. The number of kanbans can be grad- item in the lot is produced. Smaller lots reduce
ually reduced to improve the processes and the lot delay time. Process delay accounts for
reduce waste. This rule conveys the fact that stoppages due to machine tool failures, broken
inventory can be an independent control vari- tools, defective parts, and other manufacturing
able. The number of kanbans in the system at problems. Process delay also includes delays in
any time controls the work-in-process invento- the throughput time in the cell for processing
ry level. This number is initially the result of a time greater than the cell cycle time. For exam-
management decision. Many companies opt for ple, suppose heat treatment requires 10 min-
setting the initial inventory level in the link at utes in a cell with a one-minute cycle time. The
about half of the existing level when the pull process delay is 10 minutes.
system is implemented. The initial number of Suppose the cell is making a family of four
kanbans can be computed by: parts: A, B, C, and D. Obviously, there must be
enough carts or containers in the loop of Part A
DD × L + SS so that downstream processes do not run out of
K= (10-1)
a them while the cell is making Parts B, C, or D.
This is a form of process delay that adds to the
where:
work-in-process inventory and is a tradeoff for
K = number of kanbans or number of carts flexibility. By designing the cell to be able to
(K also equals the number of POKs or make a family of parts, a delay time for each
the number of WLKs) member of the part family is added.
DD = expected demand for parts, per day Honda, in Marysville, Ohio, where the
L = lead time, that is, processing time + Accord® is built, has one stand of large presses
delay time + lot delay and process that stamp out the sheet metal body parts for the
delay + conveyance time four-door models. This stand of presses manufac-
a = container capacity, a fixed amount, tures 24 different sheet metal body parts in runs
usually about 10% of daily demand of 300 parts. This is one-day’s supply of each part.
The presses are changed over in about 10 min- There is no change in the daily total-pro-
utes for each different part. Obviously, enough duction load. The production schedule only
parts must be stamped out to last until the next need be revised for final assembly—the
quantity of those specific parts is stamped. schedule for the upstream processes is
Honda is constantly working to reduce the time revised automatically by transferring the
required to change over these large presses from kanbans.
one part to another because this permits them to • The second case is the result of small,
further reduce the run size. short-term fluctuations in the daily-pro-
Looking back at Equations 10-1 and 10-2, if duction load, although the monthly total
lead-time, L, is relatively small and the demand load remains the same. Only the frequen-
per unit time, DD, is relatively constant, then the cy of kanban movement increases or
policy variable for safety stock can be small, decreases. The number of kanbans tends
resulting in a smaller inventory level. Therefore, to be fixed despite demand variation.
the number of kanbans can be smaller. In prac- • The third case is the result of seasonal
tice, this policy variable is expected to approach changes in demand or of increases and
zero (as will be discussed later). Eliminating decreases in monthly demand. The number
setup times reduces lead times. of kanbans in the system must be recom-
As volume increases, more cells can become puted (increased or decreased) and the pro-
duplicated and dedicated. The inventory level duction lines rearranged (the cycle time
between the cells can be further reduced. The must be recomputed and the number of
smoothing of demand is achieved by smoothing workers in the cells changed accordingly).
production in the subsequent processes, as was These rules must be followed for the kanban
discussed in Chapter 9. system to be an effective management informa-
tion system that also controls the level of work-
Rule 5. If there is no kanban card, there is in-process.
no manufacturing and no transfer of parts. The
WLK card, as a production control device, Kanban Limitations
should be attached to the carts or containers Kanban, as described in this chapter, is a rel-
unless they are in transit within the cell to atively simple manual information system.
order production. This rule reveals the visual Limitations for its use are as follows.
control nature of the kanban card. The key
manufacturing information is readily at hand. • Goods must be produced in whole discrete
The removal of the kanban card prevents a cart units. Obviously, kanban is not applicable
from being transported and used. to continuous-process industries such as
oil refineries and breweries.
Rule 6. Kanbans should be used to adapt to • Kanban should be a subsystem of an
only small fluctuations in demand (fine-tune linked-cell manufacturing system, using
production by kanban). Just-in-time philosophy. The use of a kan-
Flexibility means the system can respond to ban system without the lean production
changes in demand. There are three cases philosophy makes little sense. Rules 1 and
where kanban can be used to fine-tune produc- 2, regulating the use of kanban, require
the manufacture and withdrawal of the
tion, thus giving the system flexibility with
necessary parts in necessary quantities at
respect to changes in demand.
the necessary time.
• The first case is the result of product-mix • The prerequisites for linked-cell manufac-
changes of the final-assembly delivery turing systems, such as the design of the
dates, and of small changes in quantities. manufacturing system, standardization of
operations within cells, and smoothing of the vendor’s store worker. The driver picks
production must be implemented before up a trailer loaded with parts produced to
an effective pull system can be imple- the requirement of material-ordering kan-
mented. bans brought at 8 p.m. the night before.
• The parts included in the kanban system The driver returns to the user’s company
should be used every day (high-use parts). and delivers the parts to the correct loca-
Kanban provides that at least one full con- tion within the plant.
tainer of a given part number is available. 3. At 8 p.m., the empty trailer is returned to
There is not much inventory float if the the vendor. More material-ordering kan-
contents of the full container are used up bans are given to the vendor and the full
the same day they are produced. trailer is taken back to the plant. One
day’s supply is carried in two trailers.
Special Kanbans 4. If the trailers can be rapidly loaded and
A modified version of the WLK is used to unloaded, only one trailer is necessary.
reorder raw materials (Figure 10-8). Parts are Many material-transporting companies
withdrawn from the lot of 500, in containers of already have developed such systems.
100. When the stack of containers reaches the
material requisition kanban, this kanban is CONSTANT WORK-IN-PROCESS
used to requisition a coil of steel for the process The constant work-in-process system focuses
(Press #10 preceded by shear). When the signal on maintaining a constant level of work-in-
kanban is revealed, it is taken to the kanban process over the entire manufacturing system.
post at press #10 and placed in the queue next Work-in-process at any point within the system
to the material-requisition kanban. These two is allowed to fluctuate freely. Sometimes
kanbans combine to instruct the workers at described as a single-kanban cell encompassing
Press #10 to make 500 steel sheets in quanti- every machine in the plant, a constant work-in-
ties of 100. The kanbans are reinserted in the process system is actually a push/pull hybrid.
stack of containers as shown. Table 10-1 lists The system is not a kanban system. However,
some additional types of special kanbans. the constant work-in-process system has been a
Material-ordering kanbans are a special popular topic in recent literature. For this rea-
kind of withdrawal card. Material-ordering son, the constant work-in-process system will be
kanbans often are used to get material from examined in similar fashion to the kanban sys-
vendors. An example of how these kanbans are tems previously discussed in this chapter.
used is shown in Figure 10-9. The information The constant work-in-process system tries to
on the card is similar to what is needed on a control WIP like kanban but fails to make use
withdrawal card except that the card has of the demand-prompted pulling of material
detachable pieces that go to the user’s between cells. When a container of products
accounts-receivable department as shown in leaves final assembly for the external cus-
Figure 10-10. A two-trailer-truck system is tomer, this signals the introduction of a new
shown. container of raw material at the beginning of
Here is how the material-ordering kanban the manufacturing system. Once the material
works:
is introduced, production continues without
1. At 8 a.m., a truck delivers material-order- waiting for withdrawal from the downstream
ing kanbans and empty containers to the cells. When this product leaves final assembly,
vendor in a trailer. a new batch of raw material is introduced to
2. Upon arrival at the vendor, the truck driv- take its place in the system. Thus, a pull sys-
er hands the material-ordering kanbans to tem exists between the final-product-shipping
No. of
Lot size 500 5
container
Machine to r
use Reorde
Parts
Press storage
#10 area
Steel plates
Press
Shear Coil of
#10
steel
Raw material
Kanban post
at press #10
Figure 10-8. How signal and material kanbans combine to make parts as needed.
Vendor's
Empty Arrived storage
boxes at 8 a.m.
Empty boxes
MOKs brought at
8 a.m. this
morning
MOKs
Driver
switches MOKs
trucks brought
at 8 a.m.
MOKs
brought at 8 p.m.
previous night
MOKs
brought at 8 p.m.
last night
Parts completed
Starts to return Parts loaded at 8 a.m.
to plant at 8 a.m. this morning
at 8 a.m. this morning
Figure 10-9. Movement of material-ordering kanbans (MOKs) between plant and vendor (Monden, 1983).
cycle times. There is always some cell pre- certain processes, then a routing-control
venting the other cell from producing due methodology may be required. This is rem-
to its slower cycle. However, the individual iniscent of the traditional material-
cells in a constant work-in-process system requirements-planning system (push
are not subject to withdrawal from down- system) in the job shop.
stream. Thus, the slow cell in an unbal-
anced line does not prevent other cells INTEGRATED INVENTORY CONTROL
from functioning. The most powerful analogy presented in
• Bottlenecks near the end of the line are Japanese literature is the now-famous “rocks in
not devastating. If a cell near the final the river” (Figure 10-12) (Shingo 1981). In this
assembly in a constant work-in-process simple analogy, rocks are equated to problems
system experiences delays, this does not and the river is inventory material moving
necessarily affect every upstream cell. through the plant. The river’s level is equivalent
They are still receiving raw materials to the work-in-process inventory flowing through
from upstream, so they are able to contin- the factory, just as the river flows between its
ue production due to the pushing of mate- banks. When the river level is high, the rocks,
rial through the system. which represent hazards to safe navigation, are
covered. Table 10-2 lists problems in the work-
Limitations place and traditional solutions.
There are some drawbacks to the constant Now it may be asked, “Isn’t that good? Hasn’t
work-in-process system. Some of these include: inventory traditionally been used to circum-
• Constant work-in-process is inferior to vent the problems of poor quality, machine-tool
kanban if the final-assembly lines are bal- breakdowns, long setup times, parts shortages,
anced. Much of this chapter has been and other deficiencies in the manufacturing or
devoted to the contention that if the first production systems?” This is true, but there are
five steps of lean production have been errors in this thinking. Covering the problems
implemented (of which the fifth step is the wrong approach. Inventories are waste-
involves leveling, balancing and synchro- ful and expensive to carry. The greater the
nizing), then a kanban system will provide inventory in the system, the longer the
throughput time. Low-cost, high-quality manu-
optimal control of production and invento-
facturing never is achieved when work-in-
ry. Constant work-in-process cannot
process levels are high. In addition, the
match kanban’s performance under these
inventory, if not controlled, suddenly may drop
conditions.
because of factors outside of the company’s con-
• Performance is poor if bottlenecks occur
trol, revealing some problem in the system at
near the beginning. This is due to the fact
the most inopportune time and throwing the
that material is pushed from its introduc-
entire plant into disarray.
tion through the system to final assembly; a
bottleneck at the start of this process can
really delay the subsequent cells’ produc- INVENTORY: AN INDEPENDENT
tion. CONTROL VARIABLE
• A constant work-in-process system can Lean production philosophy alters the nature
experience routing problems. Since the of inventory completely, changing it from a
parts are pushed through the factory, a dependent variable in the classical push system
routing sequence must be set. This is not a to an independent control variable in the pull
problem if the parts travel to the same system. Taking the rocks in the river analogy a
cells. However, if some parts do not require step further, the river’s volume of flow represents
User Vendor
Receiving store
Sign (purchasing department)
Delivery
Delivery
slip
slip
Receipt Sales
slip slip
Receipt
slip
Sign
Sales
slip
Computer
OCR Filed OCR
department
Accounts Accounts
receivable payable
data data
Accounting Accounting
department department
Payment
Figure 10-10. MOK voucher movement for two truck system (Monden, 1983).
Station
Information flow
Material flow
Kanban system
capacity. The river’s flow can be compared to the FR = flow rate, ft3/min (m3/min)
materials that flow in continuous-processing D = amount of work-in-process
manufacturing systems such as chemical plants W = number of manufacturing systems
and refineries. Continuous-processing manufac- (number of final assembly lines)
turing systems represent the ideal in terms of V = distance materials travel through the
efficient manufacturing with the minimum system ÷ throughput time
work-in-process. Therefore, they represent the
vision of lean manufacturing except they are not In classical manufacturing systems, concen-
typically flexible. To make discrete parts flow like tration has been on the production rate with
water, setup time and lot sizes must be reduced little regard for the inventory level. In a linked-
and defective products and machine breakdowns cell manufacturing system, the amount of
eliminated in the system. Inventory flow rate is work-in-process is controlled. That is, invento-
an independent variable that can be described as ry levels are deliberately raised or lowered,
follows: even though lowering the inventory level
exposes problems. When this happens, the
FR = D × W × V (10-3)
inventory is temporarily restored to ease cus-
where: tomer discomfort or the company’s discomfort
is continually reduced by the removal of carts. planning (MRP), inventory levels, and produc-
Referring back to Figure 10-7, suppose the sys- tion philosophy. The pull system is compared to
tem begins with 11 carts (K) between the two the push system.
cells. Suppose each cart holds 20 parts (a). The At its initialization, kanban should be a
maximum inventory between the two cells is manual information system utilizing cards.
therefore 220 parts (11 × 20). The removal of a This helps familiarize everyone with the sys-
cart lowers the maximum inventory level to tem. The capital investment in a kanban sub-
200 parts. If no problems occur, another cart is system is small compared to the costs of
removed. The process continues until finally no changing the manufacturing system. Material-
carts can be removed without serious delays requirements planning, on the other hand, is a
occurring due to lack of available parts. computer-based system. Because the existing
Suppose this occurs until there are six carts manufacturing system and MRP are so com-
remaining. This number often depends on how plex, MRP is not manageable without comput-
close one cell is to another and on the length of er assistance. The capital investment in an
setup times between different parts. MRP system is extensive. In 2002, the cost of a
Now the number of parts in each cart can be full-blown MRP installation is estimated to be
cut in half. The new amount of parts is now 10 between $500,000 and several million dollars.
and the number of carts is restored to 11 This includes labor, software, hardware, and
(11 × 10 = 110). Thus, the inventory level is sim- training for system development.
ilar, but the frequency of lot production is Material-requirements-planning systems are
increased. The flow is smoother because small- known for taking a long time to implement. A
er lots produce smaller demand spikes in the materials manager who had just spent two
system. The setup reduction problem becomes years implementing his company’s material-
immediately apparent since cutting the lot size requirements-planning program was asked how
increases the setup frequency. So now setup for many people were employed at his company. He
that particular process is attacked to reduce or responded that there were 500 workers and 200
eliminate it. Once setup time has been reduced, others. He was then asked how many of these
another cycle of removing kanban cards can people understood the material requirements
again reduce inventory levels. This cycle is planning system. After a long pause, he estimat-
implemented across the factory and is part of ed three. How can any system function when
lean production’s continuous-process-improve- only three people understand how it is supposed
ment philosophy. to work? People will not trust a system they do
The dual-card kanban system is integrated not understand. Material is the lifeblood of the
because it is carried out entirely by the people manufacturing system; not understanding how
who run the manufacturing system. Eventually it is controlled is bad business.
the users get the inventory level as low as pos- A material-requirements planning system
sible without causing major disruptions to the uses built-in economic-order-quantity calcula-
manufacturing system. The system will require tions. Therefore, quantities vary considerably. It
upgrading and automation to go to the next is better to have fixed small quantities and vary
level. If factory management decides to go in the frequency of ordering parts. The material-
that direction, this kind of automation usually requirements-planning system was developed
is easy to justify. for planning in the job shop, not for control of the
materials moving through the plant.
Kanban Pull System Compared to A dual-card kanban subsystem is truly a pull
Material Requirements Planning system of parts ordering and control. The ordi-
In this section, various features of the pull nal production schedule is issued only to the
system are compared to material-requirements starting point on the final assembly line and
not for any other process. The transfer of parts demand of a single time period. This system
and withdrawal kanbans linking the processes has not adopted the concept of small economi-
in the system determines the production sched- cal lot sizes through eliminating setup time
ule for each preceding process. Therefore, the and streamlining the operation. In material-
parts are actually pulled through from the final planning systems, lot sizes vary considerably,
assembly line to the start of the system. so production cannot be smoothed.
The single-card kanban subsystem is a com- The ease of associating requirements for parts
bination push-and-pull parts-ordering system. with the schedule of end products is the crucial
The manufacturing aspect in a single-card kan- factor of information-system selection. Figure
ban subsystem is a push system because parts 10-13 shows the relationship between the ease
are produced according to a daily production of associating the requirements and the type of
schedule rather than for immediate needs as in information system used. This affects inventory-
the dual-card kanban subsystem. Coupled with level size, as discussed earlier.
this push system for manufacturing is a pull The major distinguishing factor between the
system for deliveries. Parts are delivered using pull system and the material-requirements-
withdrawal kanbans only as the downstream planning system is the ability of the pull system
processes need them. to accurately associate the component-part
Material-requirements planning is a push requirements with the end-product schedule.
system of parts ordering and planning; there is The dual-card kanban system completes this
no real control function. A push system is sim- accurate association because it manufactures
ply a schedule-based system in which a multi- and withdraws parts according to the system’s
period schedule of future demands for the need. Production control is truly integrated into
company’s products is prepared. The computer the manufacturing system. The material-
breaks down the schedule for manufacturing requirements-planning system does not have a
and develops a production schedule for each high degree of association or integration with
work center based upon the master schedule. error introduced into the part requirements as a
Then the parts are delivered throughout the result of changes in the end-product schedule.
system without regard to the immediate need. Although material-requirements-planning sys-
The connection between the planned schedule tems correctly calculate the part requirements
and reality may not exist. by precisely associating them with the master
Companies using pull kanban systems have schedule of end products, long lead times and
less delay or lead time between parts manufac-
large lot sizes erode the close association
ture and use, so they have only hours or min-
between the part requirements and the end-
utes worth of material in inventory. A
product schedules. The way to make material-
material-requirements-planning system car-
requirements planning into a truly effective
ries days’, weeks’, or months’ worth of material
information system is to reduce inherent error
in inventory because the parts are produced to
in its part-requirement calculation. However, if
cover the demand for a week or longer.
the setup time is reduced to make material-
A kanban information system is a logical ele-
requirements planning more effective, it is not
ment in a lean manufacturing system. The
elimination of setup time makes small lot sizes needed for material control in lean production
economical. Making lots equal in size or as systems, since pull systems are more productive.
small as possible plus redesigning the manu-
facturing system are integral parts of the pull SUPPLY-CHAIN MANAGEMENT
system. There is currently much activity in supply-
A material-requirements-planning system chain management. However, unless lean pro-
produces parts in large lot sizes to cover the duction is achieved, a company does not have a
Ease of
associating part
Easy to Hard to
requirements
associate associate
with end- product
schedule
Toyota
US Kawasaki Advanced US Many
Examples Harley-
breweries Nihon companies companies
Davidson
Zero
Minutes'
worth
Hours'
worth
Days'
worth
Weeks'
worth
Months'
worth
chain of linked suppliers. It has a network of eliminate setup and work together to eliminate
suppliers. Perhaps this strategy should be defective parts. The workers became skilled so
called supplier network management. As part they could operate different processes. Make-
of the last two steps in building a lean produc- one, check-one, and move-one-on became the
tion system, the production system consisting operational standard. But these events do not
of product design, manufacturing, sales, mar- totally explain why this unique system evolved.
keting, distribution, and purchasing must be in
This question probably lies in the nature of the
order. This requires an integrated internal
Japanese language.
manufacturing system, and the selection of the
best vendors to be the sole component or sub- As a language, Japanese is difficult to use
assembly sources. for written communication. The classical,
large mass-production system requires a
THE PAPERLESS FACTORY sophisticated information system to deal with
OF THE FUTURE the complexity of the manufacturing system
and its interfaces with the production system.
Manufacturing systems have their
dinosaurs. The production job shop, with its Since Ohno could not change the Japanese
manufacturing system kept afloat by the inge- language, he had to find a way to simplify the
nuity of people and by oceans of inventory, is manufacturing and production systems to
one of those dinosaurs. Clearly the time has eliminate the need for written communica-
come for the invention of a new manufacturing tion. Thus he replaced it with visual or auto-
system. What motivated Taiichi Ohno, then matic signals. He began to eliminate all kinds
vice-president of Toyota, to develop this sys- of unnecessary functions. In contrast, manu-
tem, and why does the system have the charac- facturers in the United States tried to com-
teristics that it does? puterize and optimize these functions.
After World War II, Japan was known as a
nation that made poor quality, low-tech, and Management by Sight
inexpensive products. Yet, the Japanese gov-
ernment wanted to develop full employment in The objective of management by sight or
their country through industrialization. To do visual control is to provide an easy method to
this, they needed to learn how to build quality exercise control of the plant and provide quick
products—products other nations would buy. feedback by simply using one’s eyes to view the
The Japanese felt that the United States knew status of operations. Whenever an abnormal
how to build these products, so they learned condition exists, the system provides a signal
about quality control from Deming and Juran. requiring that timely corrective action be
Initially US quality gurus taught quality con- taken. Management by sight calls for signals to
trol techniques to the engineers and managers. be actively changing and, therefore, providing
Then the Japanese did something quite differ- up-to-date information. Visual control causes
ent, something not done in the US factories. employees to get out into the plant on a regular
Most surprisingly, the workers were taught
basis to exercise control. Anyone at any time
quality-control techniques. Next, the workers
were given the responsibility for quality and can go to the shop floor and view the condi-
the authority to stop the processes if something tions. When successful, fewer reports must be
went wrong. Toyota and its suppliers began to sent through the organization, thereby reduc-
develop early versions of linked-cell manufac- ing paper flow. There are two kinds of manage-
turing systems for the manufacture of part ment by sight: information on displays, and
families. In these early cells, they learned to workplace organization through the shop.
• Andon boards—used by operators to sig- management to shut down machines and the
nal the occurrence of a problem—lights to production line if necessary to find and perma-
signal that assistance is needed. nently resolve problems.
• Scrap bins—red metal bins divided into
days of the week. An attached scrap tally Machines and Line Operations
sheet keeps a one-month history of the The line stop concept applies to machines
reasons for and quantities of scrap. and line operations. Process-indicator lights
• Clean stands—yellow stands designed to assist the worker in detecting abnormal occur-
hold one piece of product and indicate rences. On the line, the operator pushes a trou-
when the cleaning solution in the parts ble button to get the attention of the team
washer should be changed. After washing, leader. If the problem can be solved within
the part is put on display along with the workstation cycle time, the line will keep mov-
results of the cleanliness check. Upon ing; if not, the line will stop.
reaching a specified unacceptable cleanli- An andon light normally signals the occur-
ness level, the solution is changed before rence. A yellow light may be used when
unacceptable parts reach assembly. requesting assistance with a problem, and a
• First, middle, last-part examples—high- red light for stopping the line if the problem
quality samples of each machined part are cannot be resolved quickly. Buzzers or music
taken near the start, middle and finish of often are used in conjunction with the andon
each shift. The parts are displayed to show lights to enhance visibility. In brief, line stop is
the quality status at the point of inspection. one more technique encouraging continuous
Visual control enables factory operations to be improvement. Line stops should not be feared,
more tightly linked with improved communica- but encouraged to expose and then resolve
tion and better problem-solving routines. It is problems.
another set of techniques underscoring the driv-
ing force of continual improvement in the lean Other Control Techniques
production system. Management by sight also Many other techniques and tools help in the
reinforces another central theme of lean produc- planning and control of a lean production fac-
tion in the factory: for effective management, tory system.
employee and team involvement are critical.
Hourly Check
Line-stop Concept In addition to poka-yoke devices to spot and
Line stop is a fundamental lean production prevent defects, hourly checks of the product
factory-control technique. It refers to stopping further eliminate the possibility of passing
the production line when a problem occurs, defective work to the next downstream process.
identifying the problem, and then resolving the A buzzer sounds once per hour to trigger 100%
problem so it will not recur, thus regaining flow inspection of critical processing dimensions.
as soon as possible. The Japanese refer to the Every part is 100% visually inspected. The
concept of line stop as jidoka. It means literal- hourly check reinforces the previous efforts of
ly to make machines intelligent; that is, capa- the quality system and pinpoints problems at
ble of determining if a line should be stopped. their source.
The objective of line stop is to give operators
the authority to stop the process any time a Sample Size
problem occurs. However, it is difficult to imple- When workers are not checking parts in the
ment. It takes discipline to respond to problems lean manufacturing cells, the quality control
quickly. And it takes commitment from top personnel carry out sampling inspection on the
first, middle, and last part of each shift. These operation sheets are positioned above each
parts are taken from the line and examined in operation and provide stability and stan-
the inspection lab to ensure that specifications dardize the process so specifications can
are met. The pieces are then put on display to be met consistently. They are also used to
show that quality parts are being produced by train new operators. Leave nothing to
that shift. chance or memory. Make it easy to train
If problems are discovered, details are and retrain the operators and prevent
relayed to the production line for corrective them from making errors.
action. First, middle, and last piece inspection
also helps determine when a problem occurred. SUMMARY
If a problem was found on the last-piece inspec- Control techniques serve as some of the tools
tion at the end of the shift, but not found earli- workers can use to keep the production process
er in the shift, the first two pieces can be moving toward its goal of stability and contin-
re-analyzed to ensure that the problem only uous improvement. Production leveling and the
occurred near the shift’s end. The imperfect kanban card system keep production and
parts then can be isolated for corrective action. inventory predictable, controllable and, thus,
Various documents also help control the stable. Other techniques and aids help workers
process. monitor the operations and spot problems in
• Control charts or statistical process con- the line or defects in the products. Once high-
trol (SPC) sheets are put on display for ref- lighted, the troubles on the line or imperfec-
erence, especially by the operators. tions in the results should be tackled
Control charts and run charts (discussed immediately and at the site. Monitoring and
in Chapter 7) ensure specifications will be trouble-spotting tools include: management by
met by detailing the particular operation. sight, line stop and other techniques such as
At the same time, they give the operator a hourly checks; first, middle, and last inspec-
sense of the patterns of process develop- tion; display and control sheets, check sheets,
ment and point out the effects of machine and process documentation.
adjustments. The Toyota system was created by trial-and-
• Check sheets are another tool used in lean error processes within an environment where
production. The most common is the the language would not permit a written com-
machine checklist that aids operators in munication system (information system) that
the correct startup procedures for the could control a large, complex manufacturing
machines. Check sheets also are displayed system. So Toyota developed a manufacturing
for and provide help with maintenance system that was simple to operate and control
when the machines are down. They can be with a simple information system, one now
used in any function and enhance the known as kanban. A pull system of production
properly established procedures or rules. and inventory control, kanban uses visual
• Process sheet or standard operation cards for information transfer and control. The
sheets are the most common control tools. manufacturing system is redesigned so only
They provide the necessary information the final assembly line must be scheduled. In
for the machining or assembly of a part. fact, the long-range goal of the system is to
For example, they outline speeds and eliminate the need for kanban by directly link-
feeds, material, sequence of operations, ing the processes. This means that the output
operator safety, tooling and gaging infor- from each cell goes to only one customer. Note
mation, part number, and part name. A that this defines what the cell or process is to
part drawing also is included. Standard make. Moreover, one of the system’s operational
characteristics is the gradual elimination of its The lean manufacturing system, being man-
primary information document, kanban cards ual at the outset, is easily understood by its
and the resulting work-in-process inventory users, something that cannot be said for many
reduction. Thus, the proper path to obtain the of the computerized systems used in the job
paperless factory of the future is evident. It is shop. Employees understand how their actions
critical to eliminate the manufacturing system influence the entire system and that they can
of mass production using the job shop and make the system better.
replace it with cellular manufacturing systems Large lot sizes, long lead times, and changes
in the schedule make it difficult for an MRP sys-
linked via the kanban subsystem.
tem to accurately associate the part require-
In conclusion, the kanban system accurately
ments with the end-product schedule. The
associates the part requirements with the end- inexpensive kanban system can achieve better
product schedule. Part usage and manufactur- estimates of part requirements than computer
ing in the upstream processes determine the based material requirements planning systems.
need of the end-product assemblies through the The capital investment saved is better spent on
transfer of withdrawal and production kanban implementing lean manufacturing to improve
cards or carts. These parts are manufactured quality, lower costs, reduce inventory, and facili-
Just-in-time for the necessary products, in the tate an inventory control system. The most
necessary quantities, at the necessary time. expensive system is not necessarily the best.
renegotiated every six to 12 months.) Under may not be expensive, but they take up space
these agreements: and/or require expensive tooling.
• The lean company supplies updated fore- 100% Good Quality
casts every month (good for 12 months).
• The lean company commits to building a If a supplier is taught how to implement cel-
lular manufacturing so it can deliver the cor-
long-term quantity and to eventual
rect quantities on time, and with 100%-good
excess-material buyout.
quality, then an incoming inspection by the
• Delivery is specified by mid-month for the
buyer is unnecessary. In short, vendors seek to
next month.
become Just-in-time suppliers that carry out
Advantages of a long-term contract include lean-production methods, while striving for
the following: zero defects.
• It builds schedule stability. There is no Engineering Aid to Vendors
jerking up and down of a vendor’s sched-
Often vendors are small companies that are
ule, and there are smooth increases and
unable to afford engineering expertise in man-
decreases.
ufacturing and quality areas. Here, vendors
• There is better and more frequent commu-
and customers must work together to improve
nication between the buyer and vendor.
vendors’ manufacturing processes, productivity
• There is better visibility. The vendor sees levels, and quality standards. Customers
one year’s worth of forecasted needs as should visit vendors’ plants at least annually,
soon as the company sees it, instead of more often if there are problems. Vendors
viewing a limited lead-time view. should visit customers to see how components
• There is less paperwork. There are fewer are used in products.
(possibly none) change orders to run
through the system. This is good for the Local Sourcing
company and vendor. While it is not absolutely necessary or even
• Inventory is reduced, initially at the lean possible for suppliers to be geographically close
company and later at the vendor, as lead to customers, the closer suppliers are to cus-
time is reduced. tomers or companies, the easier it is to provide
them with daily deliveries. Every day that
Frequent Deliveries material spends being transported adds to the
In some systems, frequent deliveries are criti- level of inventory.
cal. A vendor is expected to deliver materials to
a company hourly, daily, or weekly, depending on Freight Consolidation Programs
the type of part or subassembly. Most parts can Materials from suppliers can be consolidated
be categorized according to the purchasing on one truck for transportation to customers. If
department’s “ABC” analysis. “A” parts are criti- a company has three suppliers in the same
cal, high-cost parts, and there is usually one per area who deliver daily, one truck and driver can
product. For vehicles, this would be the engine, pick up daily from each supplier and deliver to
transmission, seat sets, steering gear, dash- that customer.
board, etc. “C” parts are low cost, but numerous.
“B” parts are somewhere between these two, but Standard Packaging
critical for different reasons. For example, many Containers for parts being delivered to a
companies classify bulky parts such as packag- lean company may be standardized by contain-
ing materials and sheet metal as “B” parts. They er size and number of parts (lot size). Half-full
containers are never sent. Since the number of purchased and modified for the needs of a cell.
containers and items in the containers are eas- Many companies understand that it is not a good
ily seen, everyone knows the amount of avail- strategy to simply imitate or copy manufactur-
able inventory. It is therefore unlikely that a ing-process technology from another company,
lean company will run out of critical compo- and then to expect to make an exceptional prod-
nents, even if they are sole-source components. uct using the same technology that a competitor
Naturally, the supplier must take precautions uses. When process technology is purchased from
to ensure that it never shuts the lean company outside vendors, unique aspects are quickly lost.
down due to a lack of parts or subassemblies. The lean company must carry out research and
Take for example, at Johnson Controls Inter- development on manufacturing technologies as
national, the supplier of seats for the Toyota well as manufacturing systems to produce com-
Camry assembly plant in Lexington, Kentucky, petitive and cost-efficient products. Effective,
there is a location in the plant called “the lock- cost-efficient manufacturing is the result of
up.” This area contains a 24-hour supply of research and development in manufacturing
seats for the Camry. The seats are stored process technology.
behind a large fence. Only the plant manager
has a key to this area. Advantages of
in-house-built Equipment
THE PLANT TRIP There are unique advantages to an in-house-
A true understanding of lean manufacturing built equipment strategy:
requires a visit to the vendor’s facility to under- • flexibility—rapid tooling changeover,
stand the supply chain. So, through a plant trip rapid modification for new products, and
to see two suppliers of the same component, a less-than-full-capacity design;
comparison of two suppliers is possible at the • build exactly what is needed;
component level. This is the level where the • maintainability, reliability, and durability
lean system builds parts using lean-manufac- are built into the machine tools; and
turing cells and the mass system uses large job • ability to accommodate the needs of the
shops (see Figure 11-2). For instance, one ven- existing cell and system single cycle auto-
dor may supply Toyota Camry parts and the matics. Process delay can be specified if
other General Motors parts. needed; poka-yokes can be used; equipment
can be specified so it is easy to load, unload,
Lean Manufacturing Cells and operate (walk-away switch and fail safe
Lean manufacturing cells are different from operation); make one, check one, move on
the interim manufacturing cells described in methodology can be used; it is economical to
Chapter 5. Interim cells, designed with machine build in quality; equipment is designed to
tools originally used in a job shop, are the pre- produced single units, not batches.
dictors for true lean cells. In a true lean manu-
facturing system, manufacturing and equipment Flexibility
must be designed, built, tested, and implemented Flexibility is built in when a process and its
into manufacturing cells. This includes machine applicable tooling are adaptable to many types
tools and processes, tooling such as workholders, of products. It requires rapid changeover of jigs,
cutting tools, and material-handling devices, fixtures, and tooling for existing products and
especially decouplers. Simple, reliable equipment rapid modification for new designs. The process-
that can be easily maintained should be speci- es have excess capacity. This means that they
fied. In general, flexible, dedicated equipment can run faster if needed, but are designed for
that can be built in-house is better than if it is operating at less-than-full capacity.
Manufacturing manager
microcircultry
Mill Drill
Injection
Subassembly
molding Lathe
for seats
department
Sub-
Flow lines Subassembly assembly Grinding
lines Receiving
&
shipping
Figure 11-2. The factory layout for the mass production system is a flow shop with a job shop that segregates the process into
departments (above) and further functionalizes the departments in functional groups (below).
Build Exactly What is Needed Each process in a cell is a step in the compo-
Building only what is needed means a company nent production process, with a process time
is not paying for unused capabilities or options. that is less than the needed cycle time.
A machine may have unique capabilities Machines are typically single cycle automatics,
that the competition does not have and cannot but may have the capacity for process delay. An
get access to through equipment vendors. example of process delay would be the process
When purchasing equipment from vendors, of induction heat treatment that takes four
purchasers may be paying for capabilities the minutes in a cell operating with a cycle time of
competition can get for free (from the vendor). one minute. Induction heat treatment has a
For example, in the lean cell example described capacity for four units. Each unit gets four min-
later in this chapter, a broaching machine pro- utes of heat treating, with one unit being out-
duces gear teeth on a rack bar. The angle these put every minute.
teeth make with the bar varies for different Accidents. Equipment should be failsafe, or
types of racks. Although the broaching process designed to prevent accidents. Some machines
makes gear teeth in racks, job shop broaching have walk-away switches so they begin pro-
machines are not acceptable for the cell due to cessing after being loaded by a worker who is
their large size and long changeover times. So moving toward the next process. The start but-
for the cell, a unique machine tool for broaching ton is located on the exit side of the machine.
must be designed and built. Ergonomics. Equipment should be designed
Equipment should allow an operator to so that it is easy to operate, load, and unload.
Toyota ergonomic specialists recommend unload-
stand and walk. It should be of an appropriate
ing with the left hand and loading with the right
height that allows an operator to easily per-
hand, while walking right to left.
form tasks standing and to then move to the
Single units. Equipment should be designed
next machine in a step or two. The design of
to process single units, not batches. Small foot-
each process has a narrow footprint.
print, low-cost equipment is best. Machining or
Maintainability, Reliability, processing times should be modified so they are
less than the cycle time. Cycle time is the pro-
and Durability
duction-allotted time in which one unit should
Equipment should be easy to maintain (oil, be produced. Machine time is related to the
clean, changeover, replace worn parts, and with machine parameters selected. This approach
standardized fasteners). Many cells at lean often reduces the cutting speed, thereby increas-
production vendors are similar to each other. ing tool life and reducing downtime for tool
The sole-source company has the volume and changes. This approach also reduces equipment
expertise to get business from many other com- stoppages, lengthens the life of equipment, and
panies, making essentially the same compo- may improve quality.
nents or subassemblies for different original Self-inspection devices. Equipment should
equipment manufacturers (OEMs). The ven- have self-inspection devices such as sensors,
dors build manufacturing cells for each OEM. poka-yokes, and counters to promote autono-
Most equipment can be interchanged between mation. Autonomation is the means to control
cells in emergencies. quantity (do not overproduce) and quality (no
defects). It is not the same thing as automation.
Accommodating the Needs Often a machine is equipped to count the
of Cells and Systems number of items produced and prevent defects
It is necessary to design and build machines, from occurring.
material handling equipment (decouplers), and Movable equipment. Equipment should be
tooling for the needs of the cell and the system. movable. It should be equipped with casters or
wheels, flexible pipes, and flexible wiring. being installed (Figure 11-4). The vehicle spends
There should be no fixed conveyor lines. one minute at this station (one of 150-450 sta-
Self-cleaning. Equipment should be self- tions), while a worker installs the steering gear.
cleaning; automatically disposing of its own The steering-gear assembly is shown in Figure
chips and production waste. 11-5. A worker is able to perform every variation
Profitability. Lastly, equipment should be of the assembly of the steering gear, including
profitable at any reasonable production vol- right- and left-hand drives. The steering gears
ume. Equipment that needs millions of units to are made in another state and transported daily
be profitable should be avoided. Otherwise, to the assembly plant in boxes of 10 or 20. They
once production volume even slightly exceeds are delivered hourly to the point of use on the
the maximum capacity that the machine can assembly line. The kanban system controls this
build, it would be necessary to purchase anoth- process. The gear assembly plant makes gears at
er one. The new machine would not be prof- a daily rate that is equal to what the assembly
itable until it approached full utilization. This plant consumes. The steering gear contains a
may result in schedulers of equipment dividing rack-and-pinion subassembly, which is the criti-
volume into two machines, making neither one cal element in the steering gear.
profitable. Figure 11-6 shows a portion of a plant where
Cell equipment is designed and developed rack bars are made for American automobile-
with a priority on internal customer factors, steering gears. Steel bars are sawed into
even though factors affecting the external cus- lengths and moved in tote boxes to lathes
tomer are ultimately the highest priority of where drilling, turning, grooving, facing, and
manufacturing engineering. Although many threading are performed. Each lathe performs
factories lack the expertise to build machines the needed operations for a rack bar. The bars
from scratch, most have the expertise to modi- are then carted to the broaches, which cut teeth
fy equipment for unique capabilities. However, in the rack. Next, the bar is heat-treated, again
this is the interim cell approach, not the lean in large batches, with a quench-and-temper
cell approach. process. Then it is sent for magnetic-particle
inspection for cracks. Next, it is straightened.
Case Studies When the bar returns from the inspection
In the early 1980s, Japanese Manufacturing process, it is finish ground and ready for sub-
Techniques (Schonberger 1982) made Americans assembly into the rack-and-pinion gear. The
think about Just-in-time/total quality control throughput time may be weeks.
as a way to make low-cost, superior-quality As a system design, the job shop still exists.
goods in manufacturing systems with short Figure 11-6 is simply a modern-day version
throughput times. The system was flexible. where large volumes of goods are produced in
Then in 1991, The Machine That Changed the batches of 50-200 pieces. This modified func-
World (Womack 1991) told readers more about tional manufacturing system is a production job
a new manufacturing system that everyone shop. The design groups processes functionally,
(including the Toyota inventors of it) now refers but in this case, it has removed the walls
to as the lean production system. This knowl- between processes. Materials move from left to
edge helped manufacturers make productivity right through two acres of the plant floor. Parts
and quality gains that were significant. must leave the plant for crack detection after
Figure 11-3 contrasts the design of an assem- quench and tempering and bar straightening.
bly plant to explain the differences between lean In this design, it is impossible for workers to
and mass production. Illustrated is an automo- move a part from machine to machine (or oper-
bile-assembly plant where a vehicle is being ation), one component at a time. The manufac-
assembled. In one station, the steering gear is turing system design prevents one-piece flow.
Final Final
assembly assembly
batches
final assembly
Mixed-model
Large
Station #38
Inventory
g
rin
k Rack & pinion
ee
lin
Subassembly
K- subassembly
St
Out In Out cell
line
Steering gear
subassembly K-link
cell In
K-link
Parts storage
One piece flow in
subassembly cells
Job shop Out
Rack
cell
Mill Drill
Lathe Grind
Components in manufactured
cells "U"-shaped,
(see Figure 11-6)
one-piece flow
Figure 11-3. In the lean plant on the right, the components made in the rack cell are assembled into rack and pinion gears, which
are assembled into a steering gear, which is installed into car at station #38.
Between machines and filling the aisles are tote Contrast the American plant with the
boxes filled with racks of bars in various stages Japanese transplanted factory that is manufac-
of completion. Because it is difficult to count turing steering gears (see Figure 11-7). Again,
parts in inventory, it can be estimated that observe how a lean plant makes the rack. This
there are 97,000 racks in this area at any point cell makes racks for the Toyota Camry.
in time, enough for eight days of production. Machine tools that are not made by a tradi-
The company calls this area a manufacturing tional machine-tool builder are performing pro-
cell, but in reality, it is a large job shop with a cessing in the cell. Machines are custom-built
family of parts (rack bars). for the cell. Each machine basically performs
Pressworking
Painting
Foundry Machining
Line-off
Electronic parts
Moldings
Figure 11-4. The steering gear is installed at Station #38 after it has been received from the vendor and brought to the
assembly line.
one step on the bar. Then, the bar is moved to The lean manufacturing system has unique
the next machine, which may have devices on characteristics embodied in its manufacturing
it, such as poka-yokes or decouplers, to check and assembly cells. Multifunctional workers man
what has been done in the previous step. Or, the cells. They can perform tasks other than han-
the worker may be checking the rack bar as dling material and operating manufacturing
part of the lean production methodology. processes. Tasks include quality control and
The machines that tool builders are making inspection to prevent defects from occurring,
machine-tool maintenance, setup reduction, and
today are great examples of the super machine.
problem solving for continuous improvement.
Super machines are costly because they are Cells are usually in U-shaped or rectangular
complex with many capabilities. The user pays arrangements. The cells are designed so a work-
for these capabilities, but often never uses er can easily step across an aisle to work on
them. Machine tools in the cell are simple but machines located on the opposite side. This sepa-
precise, repeatable, reliable, and replaceable rates the machine’s work and the operator’s
with narrow footprints to reduce worker-walk- work. Changing the number of workers can vary
ing time between machines. the cell output.
Stopper
{
Rack guide
Pulley
{
Figure 11-5. The steering gear shown has two major components, the rack bar and pinion. The pinion is contained in the valve
housing.
To recap, implementation of a lean system poka-yoke devices to prevent defects from occur-
requires work that is systematic and, many ring. Cells operate on the lean-production
times, difficult. Lean manufacturing cells and a methodology of make-one, check-one, and move-
system’s design must continue to evolve and one on. Within a cell, there are an exact number
improve over time. Continuous improvement is of parts that are either in machines or in decou-
forced through the gradual removal of inventory plers between the machines. This is a lean oper-
from kanban links between cells. Most equip- ation, which points to there being no extra
ment in a cell operates untended; that is, it inde- inventory within a cell. When a part is finished
pendently completes a cycle initiated by an and exits a cell, another part starts into the cell.
operator. A lean manufacturing cell has built-in Lean production’s secret is that manufacturing
8
Crack
Detection
1 2 3 4 5 6 7 (outsources) 9 10
Saw I.H. S S
Lathe Draw furnace
Broach
Saw I.H. Grinder
Washer
S
I .H .
S S
Washer
Lathe Broach Grinder
Draw furnace
I .H .
Broach Grinder
S
I.H.
Lathe S S Grinder
Broach
S S
I.H.
Grinder
Broach Draw furnace
Lathe I.H.
S
I.H.
375 ft (114.3 m)
Raw material
Finished parts
I.H. = induction hardening; S = straightening machines
Figure 11-6. The overhead shafts are gone from this rack bar machining area but the lathes are still lined up in this modern day
job shop, a functionally designed area that produces racks for rack-and-pinion steering gears. This area is about two acres
(80,000 ft 2) and the work-in-process is estimated at about 97,000 parts. The area is manned by 20 workers.
Step 20
Step 21 Straighten manually
Crack detection
Step 19
Induction temper bar
Step 18
Induction harden teeth
Step 22 Step 17
Super finish rack teeth Wash Step 16
Broach gear
Step 23 teeth
Deburr and buff
Step 24 Step 15
Install piston seal Qualify mill for broach
Step 25
Crimp piston to bar Step 14
Centerless grind
Step 26 outside diameter
Wash
Step 13
Out to subassembly OUT Straighten manually
Step 1 Step 12
Input blanks IN Step 11 Induction harden
Wash journal
Step 10
Turn piston grooves
Step 2
Gundrill one end of rack Step 9
Tap both ends
Step 8
Step 3 Drill and chamfer pass holes
Rough drill both ends
Step 7
Step 4 Finish mill and chamfer
Finish ream, counter, face slot both ends
and chamfer both ends
Step 6
Step 5 Rough mill slot
Turn outside diameter both ends
and chamfer Transfer
gundrill end
to prevent processes or operators from making cally addresses 10 stations and Operator 2
mistakes. addresses 11 stations. Most steps involve
This cell is designed a bit differently from unloading a machine, loading another part into
most others in that the work arrives and that machine, checking the unloaded part, and
departs from the middle of the U-shaped cell. dropping it into the decoupler between the
Operations at the start (the right end of the machines. After completing tasks at a process,
cell) are the same for all the bars in the family a worker walks to the next machine, hitting a
of parts. Therefore, a transfer line is used with start switch for the machine that is being left.
automatic transfer devices, small robots, and This switch is called a walk-away switch.
mechanical arms and levers to move the part The stock-on-hand in decouplers and
from machine to machine.
machines helps maintain the smooth flow of
The machining time required for the deep-
parts through the machines. Stock-on-hand is
hole drilling is longer than the cycle time for
kept as small as possible. Sometimes decoupler
the cell. That is, machine time is greater than
cycle time. The cell produces one finished rack elements inspect or check a part, but mostly
bar per minute. In other words, cycle time they serve to transport parts from one process
equals one minute per rack. Thus, one rack is to the next. Sometimes a decoupler performs a
started through the cell every minute and the secondary operation like deburring or remove
processes of deep-hole drilling and tapping are residual magnetic fields. A steel bar can
divided into stages. On the average, machine become magnetized, causing small chips to
times for each stage are less than one minute. adhere to it and perhaps cause it to be incor-
In this area, the machines have automatic rectly located in a subsequent process.
repeat-cycle capabilities, with automatic trans- By design, one operator controls both the
fer devices moving parts to the next step. This input and output of the cell and, thus, the vol-
is possible because parts in the family are iden- ume of material going through the cell. This
tical in these areas. This is a common feature of keeps the stock-on-hand quantity constant and
lean manufacturing cells; a portion of the com- the cell working in balance with the final or
ponent is made on a small transfer line subassembly lines it is feeding. Operator 1
attached to a lean cell, with cell workers load- loads the centerless grinder, then moves across
ing and unloading parts. the aisle to unload the operation called “deburr
In total, the rack moves through 26 steps or and buff.” Operator 2 unloads the centerless
operations. Most of these operations are machin- grinder and then loads the part in the next
ing with single-cycle-automatic machines per- process, moving in a clockwise loop in the cell,
forming the work. However, there are manual
from right to left.
operations in the cell other than loading and
At the interface between the two operators,
unloading. These include Steps 13 and 20, to
either operator can perform the necessary
manually straighten the bar, which can warp
after heat-treating. Step 21 inspects the bar operations depending on when they arrive and
using inline magnetic-particle inspection and when the process in the machine is finished.
Steps 24 and 25 assemble parts onto the bar. That is, the region where the two operators
The times for these steps are variable. meet is really not fixed, but changes or shifts
The rack-bar lean manufacturing cell typi- depending upon the manner in which parts are
cally uses two operators. These workers are moving about the cell. This is called the relay
standing, walking workers who move from zone (Suzaki 1987), an added flexibility feature
machine to machine in counterclockwise loops, that requires workers who are cross-trained on
as shown in Figure 11-8. Each operator makes processes in the cell. This cell can operate with
the loop in about one minute. Operator 1 typi- one, two, or three workers.
Lean Versus Mass Production is more expensive to build, and generates more
In the mass-production system, reducing defects.
direct labor is the best approach to reducing The data in Table 11-2 presents a compari-
production cost. The mass-production plant son of rack machining in a lean manufacturing
puts a high value on reducing direct labor and cell to machining areas at a mass-production
increasing machine utilization. Instead of plant. Again, the mass-production plant has a
focusing solely on labor reduction and machine lower direct-labor rate, but requires more floor
utilization, the lean plant concentrates on a space, significantly higher inventory, longer
system design that simultaneously achieves all throughput time, and more expensive parts. In
of the goals of the lean manufacturing system.
addition, it results in more defects. So the mass
The data in Table 11-1 compares an assembly
cell at a lean plant to an assembly cell at a production plant succeeds in achieving a lower
mass-production plant. The results indicate ratio in direct labor in both machining and
that the line at the mass production plant has manual operations; but, if scrap rates and over-
a lower direct-labor rate, but that it requires time are considered, both plants produce about
more inventory, has a longer throughput time, the same amount of parts per labor hour.
While visiting both example plants, some At the lean supplier plant for steering gears,
intangible differences become evident. The work- there are about 14 cells for rack bars, with one
er attitudes are noticeably different. The push to cell for each Japanese transplant company.
reduce direct labor at the mass-production plant Each cell is dedicated to the company’s specific
causes workers to be wary of goals set by man- design, but each has similar process technology.
agement. Engineers at the mass-production So, many of the machines for the Mitsubishi®
plant seem reluctant to seek suggestions from rack cell are interchangeable with some modi-
the workers on the plant floor. They also tend to fications to the Camry cell. This interchange-
minimize the importance of internal customers’ ability keeps production moving if a machine
contributions. At the lean production plant, a tool in the Camry rack cell breaks down.
close relationship between engineering and pro- However, if there is a problem with the Camry
duction departments is developed. Workers rack bar, engineers from Toyota need only visit
appreciate having their cell’s engineer nearby the Camry rack bar cell.
and frequently provide suggestions for improve- The cell design isolates processes, which must
ments. Likewise, the engineers respect workers, be redesigned so batch flow becomes one-piece
appreciate their suggestions, and often imple- flow. Notice that crack detection, the Step-8
ment proposed solutions or improvements. process in a job shop that requires outsourcing, is
The lean-production plant’s machine tools are incorporated directly into the cell as Step 21.
custom-built and are a proprietary element of Between Steps 21 and 22 is a decoupler, which
lean manufacturing. Figure 11-9 shows the dif- transports and degausses parts before they go
ferences between lean and mass production sys- into superfinishing.
tems for the broaching process. In a lean cell, Other new processes are induction tempering,
broaching is Step 16. Notice how much larger the Step 12, and conduction hardening, Step 18,
broach is in the mass-production plant. It is which replace quench and temper processes in a
immovable and takes 45 minutes to changeover. job shop. Drawing is done in batches with long
The broach in the lean cell is a rotary broach and processing times, 1.5 hours, while induction
requires a flip of a switch to change it from left- tempering is completed in the cell with a cycle
to right-hand rack teeth angles. time of about 54 seconds. See Figure 11-10.
OUT IN
Rack machining
Step 16
Floor level Broaching
10 ft (3.0 m)
7 ft (2.1 m)
8 ft (2.4 m)
Figure 11-9. The machine tools in the lean plant are homemade for the manufacturing cells.
RULES FOR LEAN CELL DESIGN should then close so the machine can
The rules for the design of a lean manfuac- begin the processing cycle unattended.
turing cell include: The machine should be at least a single-
cycle automatic.
• Each machine, process, or operation in a • Machines should be arranged in the
cell should be ergonomically designed for a sequence of operations needed to process
standing, walking worker to approach the parts.
machine from the right and leave it from • All the processes needed to make the part
the left. should be in the cell and have machine
• Material should pass from right to left times less than the cycle time or the aver-
because most people are right-handed and age time. This requires some unique
it is easier for them to load with the right process technology.
hand and unload with the left. • The aisle in the cell should be about 4–ft
• Machines should be designed to have (1.2 m) wide, so workers can pass each
walk-away switches that the worker hits other in the cell, but also easily step across
when leaving the machine. The doors the aisle. Between each process there
when there is little or no variety in the products. assembly line anywhere in the world. This
Variety, however, is a fact of manufacturing life. means that the final assembly line for automo-
Lean-manufacturing cells take a different biles should be shorter (or leaner), and that it
approach. First, the manufacturing system will have fewer stations and longer station
must be integrated, then it must be computer- times for the installation of larger but fewer
ized and automated. The development of man- subassemblies. The vendors would not only be
ufacturing and assembly cells is the first step responsible for the Just-in-time manufacture
of integrating a manufacturing system (Black and delivery of subassemblies, but also be able
1991). Experts on computer-integrated manu- to supply labor to perform an installation on a
facturing agree that lean manufacturing must customer’s assembly line. Some vehicle assem-
occur before making an effort to computerize a bly plants in Brazil are operating in this man-
system (Ayers and Butcher 1993). While sys- ner and have become test beds for future
tem costs are difficult to obtain, early evidence designs of lean manufacturing systems.
suggests that a lean cell approach is signifi- Supercomputers will permit simulation of
cantly less expensive than the computer-inte- the assembly of an entire product, in this case
grated approach. the automobile, including simulation of each
Continuous improvement requires continuous workstation with ergonomic subroutines for
redesign through large and small problem-solv- good workplace design on the line. Similarly,
ing efforts in a manufacturing system. This is a work cells with walking workers or robots that
way of life for lean-manufacturing companies. produce components for subassembly cells can
be simulated in detail.
The enabling technology for the fourth
FOURTH INDUSTRIAL REVOLUTION
industrial revolution is the supercomputer,
The manufacture of weapons drove the first which is capable of operating high-level, 3D-
industrial revolution. The second industrial design software, including virtual reality capa-
revolution was driven by the production of bility. This powerful computer system is
equipment and weapons for World War II. It capable of designing products and processes,
should not be surprising, therefore, that the including entire manufacturing systems. Such
fourth industrial revolution has been assisted systems, though not widespread in use yet,
by technology developed to design airplanes, were in use in 2002 at Chrysler, Boeing,
specifically the Boeing 747 that carries super- Lockheed, Electric Boat, and a few other large
lasers that can shoot down enemy ballistic mis- companies.
siles from hundreds of miles away. The What will the manufacturing system design
computer system that Boeing used to design be for these factories of the future? They will
the 777 was also used to design an airborne be “e-lean,” large subassemblies manufactured
laser by combining the work of 22 design teams in sync with final assembly and delivered in
working in 11 states. Both the design and sequence for installation. The final assembly
manufacturing system were simulated in the plants will be smaller (leaner) and communi-
computer. This same technology is capable of cate with the suppliers via electronic methods
designing manufacturing systems and prod- for controlling material movement, account-
ucts simultaneously. Products may be large ing, ordering, planning, purchasing, billing,
subassemblies that are assembled on a final- and so forth.
Lean production’s success is critically depend- forming a make-one, inspect-one, and move-
ent on the worker, the factory’s most valuable one-on operation. The simulations did not use a
resource. It is imperative that workers have safe one-piece, inter-cell, parts-movement method-
and ergonomically correct workstations. There ology with single-cycle-automatic processes,
has been much written on the subject of work- physical-inventory and production-control sys-
station design and ergonomics, but little has tems, decouplers, or kanban squares (Black
been written about workstation design for man- 1991). There is little ergonomic research in
ufacturing and subassembly cells. linked-cell manufacturing and the implications
of having standing, walking workers; and,
LEAN PRODUCTION almost no research, as well, on the ergonomic
AND ERGONOMICS ramifications of this system. A systematic
empirical investigation of the effects of cell
Lean production is a productivity methodol- manufacturing on worker jobs, attitudes, and
ogy; it shortens the lead time between cus- work-related musculoskeletal disorders has
tomer orders and factory shipments. Lean not been undertaken (Jackson and Martin
production uses less of everything compared to 1996). However, in 2000, there was an investi-
the methods employed in a job shop (Womack gation that described a favorable analysis of
1991). Lean methods require less labor and three ergonomic simulations comparing linked-
manufacturing space, smaller investments in cell manufacturing to functional job shops.
tools, fewer engineering hours to develop new Individual workstations in a linked cell envi-
products. Proper lean production implementa- ronment must include typical ergonomic and
tion requires that less than half of the invento- safety considerations and functions to accom-
ry-on-hand be kept. It also means that when a modate and enhance total-cell optimization
factory adopts lean production practices, it will (Figure 12-1). Manufacturing cells require that
manufacture products with fewer defects and, the worker be mobile; that is, able to move eas-
therefore, an increase in quality will result. ily and freely from workstation to workstation.
There have been numerous studies on the Individual tools and pieces of equipment must
advantages of linked-cell systems. However, be designed or modified to meet these require-
research into human-resources issues and ments, while also maintaining worker safety.
manufacturing cells indicates that the integra- The workstation must be designed for effort-
tion of cellular-manufacturing systems with less and fast movement between cells and
good human-engineering practices is still stock-on-hand areas.
uncertain. Early computer simulations of Since employees are the top resource of a
linked-cell manufacturing operations do not manufacturing organization, manufacturing-
reflect proper cellular functions. These func- system designers are acutely aware of potential
tions include standing and walking while per- ergonomic and physiological dangers to factory
Ergonomically designed manufacturing cell with eight processes and one worker
Load/unload
Decoupler to Narrow footprint devices for processes
transport part for cell processes Decoupler to
connect part flow
P3
P1 D
P2
D D
Material flow P4
In
P8
P7
P6
Counterclockwise walk to
assist right hand load
workers. Government and private studies show musculoskeletal-disorder rates for workers
that work-related musculoskeletal disorders who perform high-risk tasks. Therefore, ignor-
are increasing. The resulting injury costs are ing moral and humanistic issues for a moment,
staggering to industry. Therefore, it is impor- it makes good business sense to design manu-
tant to design work that is safe and ergonomi- facturing and assembly cells within accepted
cally sound. There are several ergonomic ergonomic and physiological guidelines. Of
standards being proposed to eliminate human course, the optimum cell or workstation design
suffering and simultaneously reduce worker eliminates human suffering while, concurrent-
injuries and the economic burdens such ly, saving a manufacturing organization signif-
injuries impose on industry. icant workers’ compensation and related
The National Academy of Sciences has found expenses, along with maintaining productivity
a clear relationship between work-related mus- gains for which cellular systems are noted.
culoskeletal disorders and manufacturing jobs. There is no more fundamental aspect of occu-
It states that research clearly demonstrates pational ergonomics than the design of local
that specific intervention can reduce reported workstations where workers spend consider-
able amounts of time and effort doing their tems. An ergonomic response should include
jobs. Ergonomics and the laws of work, by sci- consideration of the hand, wrist, shoulder,
ence and custom, have focused primarily on the neck, elbow, and back. The kilocalorie con-
physical aspects of work such as force and sumption of these muscle areas should be mon-
energy requirements. It is well known that itored and analyzed for repetitive motion, joint
these factors are coupled with the specific deviations, and reach-and-grasp motions. This
design of workstations. The physical arrange- is especially important since these particular
ment of tasks being performed in workstations body parts are primarily at risk in manufac-
defines the parameters necessary for ergonom- turing environments.
ic analysis and for design of the work area. This
part of workstation and cell design and evalua- Musculoskeletal Disorders
tion can be thought of as applied ergonomics,
anthropometry, and safety. The two leading work-related musculoskele-
In workstation design, one must look beyond tal disorders are carpal-tunnel syndrome and
ergonomics to broader factors of engineering for back injury. Compared to other injuries, carpal-
humans. While workstation designers look at tunnel syndrome results, on average, in more
the physical aspects of work, engineering for days away from work. Typically, carpal-tunnel
humans examines the total interface between cases involve more than 25 days away from
workers and their work. In other words, design- work (Khalil 1991). Work-related problems of
ers are concerned with the interface of stimulus the back occur primarily because of manual
and the resulting response involving informa- material handling; although problems can also
tion processing and action initiation, that is, the occur from body movements without a load,
senses and physical-muscular control. Workers such as pinched nerves, ruptured discs, pulled
use their senses. Actions are taken based on the muscles, and many other conditions. Also, back
dynamics of the processes around workers, from problems can occur from lack of motion, such as
sensory-perception decisions. Good workstation sitting for long periods. Manufacturing-system
design is critical for the actions of workers to designers must take into account these facts
result in positive outcomes.
and other related physiological factors.
Work-related musculoskeletal disorders can
HEALTH ISSUES occur from manufacturing tasks related to the
Job-related, chronic health issues are on the hand and wrist, where the concerns are with
rise. According to the Occupational Safety and tendons, muscles, or nerves (Figure 12-2). There
Health Administration (OSHA), work-related are several cumulative-trauma disorders that
musculoskeletal disorders are presently the
may occur in these areas, including carpal-tun-
leading cause of lost-workday injuries and
nel syndrome and tenosynovitis. Tenosynovitis
worker-compensation costs. In 1996, for exam-
results when tendons running through the wrist
ple, work-related musculoskeletal disorders
reportedly accounted for 647,000 lost workdays abrade on the bones. If the median nerve flowing
(OSHA 2000). Each year, work-related muscu- through the wrist becomes pinched in the wrist’s
loskeletal disorders account for $15-20 billion narrow carpal-tunnel area, thumb and index fin-
in worker-compensation costs. The total costs, ger numbness and tingling result. Cumulative
direct and indirect, may be close to $60 billion. traumas of the upper extremities are physical
However it is accomplished, the designer’s ailments of the wrist, arm, and shoulder. The
focus should be on the prevention or reduction cumulative effects of repeated mechanical
of work-related musculoskeletal disorders stresses cause disorders that develop gradually
resulting from working in manufacturing sys- over time and are believed to be work-related.
considered temporary, until permanent engi- Load, magnitude, and duration factors also
neering solutions can be implemented. The pri- should be reduced. The use of suspended power
mary concern should be with engineering tools will reduce static loading of the worker’s
solutions. These solutions follow scientific- shoulder and protect it from possible injury.
analysis methods to address problem areas. Tasks are many times more flexible in man-
Solutions often include (Konz 1995): ual cells where workers can add or change
• automation, processes. This, coupled with the slow repeti-
• mechanization, tiveness characteristic of most manual-cell
• cycle reduction, operations, provides reduced probabilities for
• difficulty reduction, repetitive-motion injuries. In one cycle, the
• minimization of joint deviation, worker may perform loading and unloading
• minimization of force, particularly dura- operations on five to 10 processes, with no rep-
tion, and etition within the cycle. The fact that a worker
• job enlargement. may repeat a cycle 60 times an hour and 480
times per shift may be augmented by shifting
Job enlargement may play a particularly the worker to a different set of tasks every two
important function in the relationship between hours. Here again, joint deviation also influ-
work-related musculoskeletal disorders and ences carpal-tunnel syndrome. The worker’s
cellular-manufacturing systems. wrist must be kept in the handshake orienta-
Vibration is a risk factor because of its inter- tion to prevent harm. To implement this solu-
ference with blood flow, as well as its physical tion, workstations or tooling may need to be
damage to an affected area. Also, vibration con- redesigned. Sometimes, all that is required is a
tributes to artery constriction, which may hand-to-tool angle or a task change to modify
result in reduced blood supply. Insufficient an operator’s wrist posture, along with imple-
blood supply results in less-than-optimum menting loading reductions for that operator.
healing. Many cellular-manufacturing systems,
especially assembly cells, must provide solu- Ergonomic-assessment Measures
tions to prevent injury from assembly-tool The ergonomic benefits of an optimum man-
vibration. ufacturing system are qualifiable. Possible
Work-related musculoskeletal disorders may metrics considerations include kilocalorie
occur from manufacturing tasks where the expenditure, joint deviation, reach and grasp
hand and wrist are used. Carpal-tunnel syn- factors, heart rate, bending and reaching fac-
drome is more a result of repetition, rather
tors, and repetitive motion. Specific tools used
than loading factors. A reduction in the use of
to gather and then analyze this ergonomic and
certain joints may be required to prevent
physiological data include high-level computer-
carpal-tunnel syndrome. This is accomplished
software programs and simulation tools, rapid-
by adding mechanisms to machines to assist
upper-limb assessment (RULA), time methods,
with loading and unloading tasks. This offers a
and the National Institute of Health and
permanent solution.
Safety (NIOSH) lifting guides. These metrics
Solutions to possible awkward postures are
are a means to an end; that is, they are tools to
decreasing repetition, joint deviation, and
prevent or eliminate work-related injuries in
applied load. The best solution is to reduce the
the workplace. Therefore, manufacturers and
use and deviated angles of joints. A work task
their work force will reap two large benefits
should be designed so both hands do the work,
from using these tools:
rather than just one hand. The upper arm
should be kept in a vertical position aimed • elimination or reduction of worker injuries
downward, not horizontal or elevated positions. on the job; and
• reduction of the expensive losses for man- problems can also occur from body movements
ufacturers resulting from direct and indi- without applied force, and lack of movement,
rect work-related injury costs. such as sitting. Lifting guidelines from the
NIOSH can be used to design jobs to minimize
Rapid-upper-limb Assessment the risk of back injuries.
Rapid-upper-limb assessment can determine
if there are posture problems that can lead to Energy Expenditure
work-related musculoskeletal disorders. A Metabolic-energy expenditure or kilocalorie
RULA begins with an analyst making observa- usage is a physiological measurement for
tions of a worker’s limb and body postures for determining the amount of task intensity that
those parts of the work cycle that create the a worker can continuously carry out. By exam-
most frequent joint use or the most extreme ining energy requirements for a task, the man-
joint angles. Rapid-upper-limb assessment
ufacturing-system designer can assess the
uses two body regions for estimating the degree
worker’s capacity to perform the task, establish
of posture correctness. These areas are:
duration and frequency of rest periods, and
• upper limbs (upper arm, lower arm, and evaluate other work designs in case the pro-
wrist); and posed work is determined to be too strenuous.
• neck, trunk, and legs. The kilocalorie consumption prediction
A RULA scoring form is provided in Figure model can be used to provide an estimate of
12-4. kilocalorie consumption for various job tasks.
Kilocalorie-expenditure estimation directly
confronts the issue of how much energy is
required to carry out a worker’s task. This is a
RULA Scoring Form tool to make sure that a proposed task is with-
Task:_________________________________ in a worker’s capabilities. Thus, it evaluates the
A Upper arm physiological difficulty a task presents to a
Posture
Lower arm Score A worker. Since individual workers react differ-
Wrist Muscle Force Score C ently to the same energy expenditures, the
Wrist twist issue of kilocalorie usage, as with many
Grand
ergonomic responses, is not the same for all
B Posture
Score workers. Nonetheless, data and results provide
Neck Score B
the manufacturing engineer and ergonomics
Trunk Muscle Force Score D
expert a good indication of how much energy is
Legs expended, and hence, how an average worker
may react to cellular and job-shop designs
within a lean production system.
Analysis Methods
Figure 12-4. Rapid upper-limb assessment scoring form.
After a workstation is in place, the ergono-
mist or industrial engineer may use methods
Rapid-upper-limb assessment also can be time measurement or 3D/virtual reality tech-
used to minimize the risk of back pain develop- niques to capture the exact motions of factory
ment as a result of poor work design. Work- workers, and then simulate the results. The
related problems with the back occur primarily software reports potential problems with cur-
due to manual material handling, although rent work methods.
Oxygen uptake
The third type of metabolism is activity metab-
A
olism; it is the area that ergonomics is most
interested in. There is an increase in metabo-
lism that occurs when workers go from resting
to working states. This increase above a resting B
level represents the amount of energy needed
to perform work. State Rest Work Rest
Walking metabolism, with and without a
load, is of interest to the manufacturing-cell Time
designer because the basic principle of cell
manufacturing is to take a part to the next Figure 12-5. Graph showing oxygen uptake during a work
process and exchange parts. A new part is cycle. Oxygen debt must be repaid during light duty or rest
checked and the cell worker continues this periods (Kroemer 1997).
process of exchanging parts from the previous
process with the next process, while working
around the cell. The cell designer and ergono- heart rate, and heart output can achieve their
mist are concerned with walking metabolism required supply levels of oxygen can the work-
as it relates to carrying a load from worksta- er maintain this high work state. This work
tion to workstation in the manufacturing cell. plateau is called steady state. Thus, a well-
This area is also related to the possible cumu- trained person can attain this equilibrium
lative-trauma effects of bending and twisting between demand and supply, even faced with a
while carrying a workpiece from process to relatively high workload.
process. Human-energy requirements, also called
At the start of physical work, oxygen intake kilocalorie usage, allow the ergonomics expert
slowly follows muscle demand. As shown in to determine whether a job is light or heavy
Figure 12-5, oxygen intake rises rapidly after a from a work-output point of view. This is due
slow start and it gradually approaches the level largely to the linear relationship between heart
necessary to meet the body’s oxygen require- rate and kilocalorie usage. Light work is asso-
ments. During the first minutes of physical ciated with a small energy expenditure and by
work, there is a discrepancy between oxygen a heart rate of approximately 90 beats per
demand and available oxygen. During this minute. At this work level, the oxygen available
period, energy yield is almost entirely anaero- in the blood and the glycogen in the muscle
bic. This oxygen deficit must be repaid at some cover the energy requirements of the working
time, usually during rest after the workday. muscles. At medium work, with a heart rate of
The amount of this deficit depends on several 100 beats per minute, the oxygen required is
factors including the type of work being carried still sufficient. Lactic acid developed initially is
out and the worker performing the job. re-synthesized to glycogen during the work
If a workload exceeds roughly 50% of the process. During heavy work periods, requiring
worker’s maximum oxygen intake, then the a heart rate of about 120 beats per minute, the
worker will not be able to maintain this high oxygen required is supplied if the person is
work level. Only if the worker’s oxygen intake, physically able to do the work.
performance measures, and workers’ goals, ing posture, and the environment to design a
motivations, and morale levels. lean-production workstation. For a new worksta-
tion design, it is best to obtain this information
Workstation Design from similar existing tasks or pieces of equip-
For workstation design, existing anthropo- ment. Several methods, such as direct observa-
metric data may determine specific design tion, interviews with workers, videotaping, and
dimensions. This information is found in many other industrial-engineering methods can be
sources, such as the National Aeronautics and used. Before redesigning an existing worksta-
Space Administration (NASA) human-factors tion, a survey is often beneficial to determine the
standards (NASA-STD-3000T) and SSP 50005 effect of the equipment or the cell design on
Revision B, so a designer can readily employ employee health, comfort, ease of use, and pro-
these dimensions. Workstation-design dimen- ductivity. Survey results can be useful in rein-
sions are determined for industrial tasks such forcing the recommended modifications to an
as sitting, standing, and sit/stand positions. In existing workstation design based on ergonomics
the case of manufacturing-cell design, relevant principles and scientific data.
issues are standing and walking workers.
Sitting situations are not particularly relevant. Form a Design Team
In the past, manufacturing-workstation As in cell design, workstation design
design was often an arbitrary process, with lit- requires concurrent design with continuous
tle consideration of the anthropometric sizes of input from cell workers. Thus, a design team
workers in the design methodology. The physi- consisting of engineers and workers should be
cal dimensions in the design of a manufactur- formed at the beginning of the design process.
ing workstation are of major importance from This design methodology lays the groundwork
the viewpoint of productivity, as well as an for continuous improvement brought about by
operator’s physical and mental well being. the design team’s cooperative endeavors.
Small changes in workstation dimension can
have a huge effect on worker productivity, Determine Work Methods
health, and safety. Poor design can result in a From the start, decisions should be formu-
decrease in productivity and an increase in lated regarding task sequence, available space,
related health problems. equipment, and tools. Work methods should be
A challenge of designing manufacturing established before embarking on a new design.
workstations is the complication caused by The workstation design procedure usually
human variability in size and capability. It is begins with the collection of data by direct
difficult for designers to discover solutions that observation and with input from experienced
are optimal fits for the diverse body sizes of operators and supervisors. It is then necessary
internal customers, while still satisfying work- to identify and gather data for the appropriate
station and manufacturing-cell task demands. user population based on such factors as ethnic
An ergonomic analysis of workstation design is origin, gender, and age. Necessary anthropo-
concerned with working space, accommodation, metric dimensions of the population are then
posture, reach, clearance, body-segment inter- obtained.
ference, field of vision, worker strength, and
environment (Karwowski and Salvendy 1998). Determine Optimum Work Height
In developing a manufacturing workstation, a
Gather Information designer should take into account the worktable
It is necessary to gather relevant information height. Worktable height must be compatible
or data on task performance, equipment, work- with worker height, whether standing or sitting.
The height of the working surface should main- within the maximum working circle. The nor-
tain a definite relationship with the operator’s mal and maximum working circles both define
elbow height, depending on the type of work. The the working area in a horizontal plane at elbow
nature of the work to be performed also must be level. The frequently used area on the worksta-
taken into consideration in determining the tion should be within the normal reach circle of
proper work height of the worktable. In addition, the worker.
frequently used hand tools, controls, and bins For angle of vision and eye height, trigonom-
must be located within the maximum-reach etry can be used to calculate the measurement.
space. Standing and walking workers must have The angle of sight can be calculated from the
an adjustable work surface. horizontal distance of the work surface from
the worker’s eye position.
Workstation Simulation
Operator workstation fit should be evaluated Equipment Adjustment
with a workstation mock-up or virtual-manu- Previous design criteria assumed that work-
facturing simulation done with an appropriate place components, such as workbenches, bins,
user population. This ensures that the task equipment height, etc., were fixed and could not
demand and layout will not impose an undesir- be adjusted. However, there is an alternative
able working posture on the worker. It is desir- design criteria, and that is to provide for adjust-
able as well to check for interference of body ments in the cell or workstation design. The
groups with workstation components. The four advantage of this alternative is that a large pro-
essential dimensions for the physical design of portion of the work force can be accommodated.
manufacturing workstations are: Offsetting disadvantages include higher design
and manufacturing costs, adjustment times,
• work height, and the likelihood of an adjustment-component
• normal and maximum reaches, failing.
• lateral clearance, and While adjustment is a practical approach,
• angle of vision and eye height. the proportion of the population to be accom-
The thumb tip defines a normal reach, while modated must be determined. Typically the
the forearm moves in a circular motion on a range of adjustment includes the middle 90%,
table surface. During this motion, the upper or more, of the population.
arm is kept in a relaxed, downward position.
The maximum reach can be considered as an CONCLUSION
imaginary outer circle on the work surface that A good ergonomic approach to the design of
a worker can obtain without bending the body. manufacturing-cell workstations strives for an
If carrying out repetitive tasks, hand move- adequate balance between worker capabilities
ment should be limited within a normal work- and workstation requirements. The goal is to
ing circle. Controls as well as any other items of optimize worker productivity and the total
occasional use may be placed beyond the nor- lean-production system, while still enhancing
mal working area, but should still be located worker well being, job satisfaction, and safety.
Autonomation is the development of devices “A(0)” represents hand tools and manual
and methods in manufacturing that automati- machines where no human attribute is mecha-
cally prevent a process from producing defects nized. Such a process is without self-action
or overproducing. This should not be confused properties. It does not replace human energy or
with the word “automation.” any basic control, but may include built-in
The word automation was first used in the guides and measurements. A(0) tools include
early 1950s as a definition for automatic mate- hand tools. Hand tools and manual machines
rial handling. As automation technology pro- increase worker productivity, but they do not
gressed, the term gained wider usage. Today, replace human functions. The group includes
muscle-energized machines; that is, machines
automation refers to both services performed
that give mechanical advantages but do not
and products produced automatically. In other
replace a person’s energy or control. Some
words, human intervention is unnecessary for examples of A(0) machines are a lever, an
handling tasks. The dictionary defines automa- inclined plane, a wheel and axle, a screw, a pul-
tion as “the technique of making a process or ley, and a wedge.
system automatic.” An A(1) machine is a powered machine or
tool. The human attribute that is mechanized
AUTOMATICITY is energy. The basic machine function replaces
A yardstick for automation appears in Table muscle. Machine action and control are com-
13-1 (Amber and Amber 1962). The automation pletely dependent upon the operator. These
order is based on human attributes that are tools or machines use mechanical power (wind-
mechanized or automated into the manufactur- mill, waterwheel, steam engine, electric motor,
ing process. etc.), but the operator positions the work and
Work requires energy and information. A machine for the desired action.
The “A(2)” machine is a single-cycle automat-
human being, or the human’s substitute, pro-
ic or a self-feeding machine. The human attrib-
vides these elements. As a machine assumes
ute mechanized is dexterity. These machines
higher levels of human attributes, it advances complete an action when an operator initiates
to the next automation stage. The more the the action. The operator must set up, load, ini-
machine performs the human attributes, the tiate actions, adjust, and unload the machine.
higher the machine is in order of “automatici- A(2) machines are common in the kitchen as
ty.” Automaticity is thus defined as the self-act- well as the factory. The toaster and dishwasher
ing capability of the device. In this are A(2) level machines. A drill press that, once
classification, it is observed that ten levels are activated by a worker, completes the drilling
sufficient to describe the present machines and cycle and then automatically cuts itself off is a
those that will be invented (see Table 13-1). good factory example of an A(2) level machine.
Table 13-1
Yardstick for automation
Orders of Automation Human Attribute Replaced Examples Common Terms and Concepts
Level A(0) None; level, screw, Hand tools, manual machines
pulley, wedge
Level A(1) Energy; muscles replaced Powered machines and tools Electric motors
Whitney's milling machine
Level A(2) Dexterity; self-feeding Single-cycle automatics Unload, load, start by operator
Level A(3) Diligence; no feedback Repeats cycle; open-loop Older transfer lines; no
machine control; automatic feedback; automatic unload
screw machine and load
Level A(4) Judgment; positional feedback Closed loop; numerical control; NC machine invented in 1959.
self-measuring and self-adjusting Machining centers available in
1962 with tool changers
Level A(5) Evaluation; adaptive Computer control; model of CNC machines with sensors
analysis; feedback from process required for analysis feeding data to computers to
the process and optimization "optimize" the process using
algorithms
Level A(6) Learning; by experience Limited self-programming Expert systems and neural
networks (computer software)
provide learning by experience
Level A(7) Reasoning; exhibits intuition; Inductive reasoning Artificial intelligence; few
relates causes from effects examples in factories
Level A(8) Creativeness; performs Originality No machines or computers at this
design unaided level on factory floor
Level A(9) Dominance; super-machine; Machine is master (Hal from Only sci-fi material at this point
commands others "2001: A Space Odyssey")
(Amber and Amber, 1962, updated by Black)
An A(3) machine is an automatic machine tion. In the home, a CD player with an auto-
that repeats the cycle. The human attribute matic changer is an A(3) machine. A factory
mechanized is diligence. These machines carry example would include a manufacturing trans-
out routine instructions without human aid. fer line that automatically and sequentially
They start cycles and repeat the actions auto- processes a raw casting into a completed
matically. The machine loads, goes through a engine block, without ever having a human
sequence of operations, then unloads to the hand touch the block.
next station or machine. They are open loop, The A(4) is a self-measuring and self-adjust-
not self-correcting, and obey internal (fixed) or ing machine with feedback. The human attrib-
external (variable) programs, such as cams, ute mechanized is judgment. These machines
tapes, or cards. These machines include all measure and compare results to the desired
automatic machines and many transfer size or position and make adjustments to mini-
machines such as classical “Detroit” automa- mize any error. Although feedback control of
the workpiece surface is preferable, positional reasoning of A(8) machines. Analysis requires
control of the machine table or tool is also valu- deduction; synthesis requires induction.
able. A process may use more than one A(4) An A(8) machine has originality. The human
subsystem operating independently. Figure 13- attribute mechanized is creativeness. These
1 presents examples of A(1) through A(4) machines originate work to suit human tasks
machines found on the factory floor. Numerical and preferences. They do not copy, imitate, and
control (NC) and computer numerical control follow plans or instructions. A program for an
(CNC) machine tools are widely used in almost A(8) machine only designates the general form of
every factory. the desired action and eliminates clashes, dis-
A(5) machines have computer process con- cords, and disharmonies. The result is original.
trol or automatic cognition. The human attrib- A(9) machines command. The human attrib-
ute mechanized is evaluation. These machines ute mechanized is dominance. These machines
are cognizant of multiple factors on which the govern the actions of people, machines, and
machine or process performance is predicated, other systems. They act as a “commanding gen-
and a computer model (analysis) determining eral” or as a “dictator” to achieve results. The
the proper control action evaluates and recon- machine is no longer a servant, but the master.
ciles the proper control action. Any process or An A(9) super-machine is capable of superior
problem that can be expressed as an equation energy, dexterity, diligence, judgment, evalua-
can be computer-controlled. Such a control sys- tion, learning, reasoning, and creativeness
tem can adapt to variations in materials, (that is, the other automation levels replace the
process conditions, and the job. Limited-pur- human attributes) and would be able to domi-
pose, on-board computers are used to accom- nate humanity.
plish A(5) computer control. The computer Currently, A(4) machines are typically com-
must have a process “model” describing its puter numerically controlled and are common-
behavior in words or equations. The A(3), A(4), place on the factory floor. This machine class
or A(5) control systems are superimposed on adaptively controls processes. A(5) machines
A(2) machines to reduce the dependence on use mathematical equations and problem-solv-
operator skills. ing capabilities as part of their control system.
An A(6) machine has limited self-program- This requires the process be mathematically
ming capability. The control algorithm contains modeled and some process aspect is then opti-
an expert system or neural networks. The mized. For example, varying feed or depth of
human attribute mechanized is learning. These cut may maximize the metal-removal rate per
machines may be set up to try subroutines unit force. More typically, cutting forces are
based on the general program. By remember- either minimized or held constant during the
ing which actions were most effective in obtain- cuts, which usually reduces the variation in
ing the desired results, the machine learns by part size; that is, it improves the precision. In
experience or trial and error. newer machines, dynamic models have been
A(7) machines relate cause and effect or have implemented to control chatter and vibration
artificial intelligence (AI). The human attribute that can ruin the newly machined surface. At
mechanized is reasoning. These machines fore- the A(6) level, an expert system or neural net-
cast trends, patterns, and relationships from work is incorporated allowing experience to
incomplete facts. They exhibit “intuition” by teach the computer control system, which then
going beyond available data. The control soft- becomes the “expert.” The A(7) machine
ware has subroutines programmed with artifi- requires computer software to have artificial
cial intelligence models. Other strategies may be intelligence or the ability to reason. The com-
the basis of operation. The inductive reasoning of puter logic for levels A(5), A(6), and A(7) are
A(7) machines is not the same as the deductive still in the developmental stages. Extremely
A(1). Powered machine tools of A(2). Single cycle semi-automatic and A(3). Automatic repeat cycle machine tools
the Industrial Revolution self-feeding machines of the mass with open loop control
production era
Engine lathe
Engine lathe built by Pratt & Whitney
(circa 1865) Multi-spindle automatic screw machine,
repeat cycle machines
Transfer machine
Double-cut, double-housing planer
First powered shaper (circa 1851) A(4). Automatic self-measuring and
adjusting machine tools with feedback,
closed loop control
Figure 13-1. Machine tools of the first industrial revolution (A1), the mass production era (A2), and examples of levels of
automation (DeGarmo, 7th ed).
few machines with adaptive control are found linked processes is not easily accomplished.
on the factory floor, although they exist in other Trivial matters never considered in a job shop
human endeavors. Machines using expert sys- create serious problems in automated cells.
tems and artificial intelligence in their control When automating the lean production system
logic are found mostly in research laboratories. consider the following:
• The manufacturing system should move
AUTOMATION IN LEAN materials as quickly, precisely, and effi-
MANUFACTURING ciently as possible from raw material to
To obtain maximum benefits from automa- finished product. The time from start to
tion, the manufacturing areas must be finish is as short as possible. If this goal is
redesigned into an integrated manufacturing accomplished, whether tasks are per-
production system. The restructured system is formed by humans or automatically is sec-
further refined through continual improvement ondary. In fact, humans can easily perform
efforts. Once the system is running at peak effi- tasks designed for the robot.
ciency, it can be automated to give maximum • Preview potential problems to prepare for
return on investment for complex, sophisticated the advancement to the next higher stage
equipment. Productivity improvements in the of machine automation. Incremental, con-
individual processes have a stronger impact on tinuous improvements rather than radical
the bottom line when the system is at peak effi- changes must be made—large and small—
ciency. Within the lean manufacturing philoso- everyday.
phy, automation is not buying super-machines, • Maintain proper roles for people and
complex machines capable of producing thou- machines. Automation should not degrade
sands of parts per day, but is, instead optimizing people. In other words, machines should
the performance of the existing equipment and be used in such a manner that the opera-
improving its efficiency through preventive tors remain fully involved. Progressively
maintenance. building the level of automation requires
As the process of converting the factory to revising the interaction of the workers,
lean manufacturing is best done in stages, the machines, and materials making up the
process of automating the plant is basically an manufacturing system. Once modifica-
evolutionary and incremental process. The tions to old equipment no longer create
need to solve a problem in quality or capacity, viable returns, the system automatically
such as eliminating a bottleneck, initiates the tells the users what modifications are
process. It can begin with mechanization of needed.
simple operations like loading, unloading, • Buying a machine having more capability
inspecting, and clamping, and later move than needed is wasteful. Purchase or build
toward emulating the human attributes of flexible equipment (low-cost, small) or
sensing and correcting problems. equipment capable of being modified and
Before proceeding from simple mechaniza- maintained in-house. Flexibility is in the
tion to complex, programmed automation, con- ability to handle the parts family with lit-
sider exactly what automation is designed to tle or no setup for part changeovers. The
achieve. Automating just to avoid human pro- equipment should be easy to move (not
cessing of products is not the objective. As the bolted to the floor) and linked for auto-
ninth step in the lean manufacturing strategy, matic feeding and unloading.
autonomation is developed when it becomes • Remember that the quality, flexibility, and
clear that manual means do not increase pro- reliability of equipment, processes, and
ductivity. Establishing a factory with tightly people are important.
operations that can be automated. For these needed to replace many functions that workers
simple operations, robots can be readily perform in a manned cell. Typically, these are
employed. After a certain efficiency level is functions that neither the processes nor the
reached in performing these simple tasks, the robots can adequately perform. Clearly, the
robots can perform more intelligent tasks, such check-one and move-one-on functions are
as picking up the correct-size barstock or prime decoupler functions.
checking the surface finish. At the current When machines are placed next to each other
state of robot development, these are difficult in a cell, the processes become functionally
tasks for robots to carry out. dependent on each other. Many times the work-
In an unmanned cell, the machines are pro- ers service multiple machines. The decoupler
grammable, CNC, or other automated equip- breaks the dependency of the processes on each
ment, and there are no workers. Unmanned other in a cellular environment. Decouplers typ-
cells are typically U-shaped or circular allow- ically hold one part and have specific input and
ing the robot to access the processes. Figure 13- output points. Decouplers can perform the fol-
2 shows a typical example of an unmanned cell. lowing functions:
The robot’s full range of motion is required so it
can reach every machine. At each machine, the • production control;
robot performs critical unloading and loading • functionally decouple processes;
tasks, properly orienting the parts. This • poka-yoke inspection;
requires the robot to have a certain level of • freedom of movement for the robot;
accuracy and repeatability. This is called robot- • intracell transportation;
ic process capability and this must be known • part manipulation;
prior to designing the cell. • handling of a family of parts;
An examination of Figure 13-2 reveals that • automatic production control; and
the robot cell operates on a pull basis; that is, • branching.
when a part is removed from the output chute, Production control. The simplest type of
the process creates a demand for another part. decoupler exists in a manned assembly cell—
The empty place on the output chute is the the kanban square. Figure 13-3 shows an elec-
autonomous signal for the cell to make another tronic assembly cell with kanban squares
part. Decoupler Two holds parts already between each workstation. Kanban squares
processed through machines M1 and M2. If the provide for production control within the cell,
part family was comprised of four parts, then control the operational timing, and create flex-
D2 holds four parts. Pulling a part out of D2 ibility in the cell staffing, thus providing flexi-
trips a chain of commands back through M2, bility in the cell production rates. The
D1, and M1, and finally another part is decouplers hold only one part so an empty
removed from the input conveyor for processing decoupler is the signal to perform the upstream
on M1. operation to fill the space. As shown, two work-
ers are operating the cell with the workers
Decouplers sharing station six to achieve balance in time.
In unmanned cells, the robot replaces the Do not confuse decouplers with buffers. Buffers
worker. It is not an easy task to replace human only store parts. In robotic cells, the decoupler
flexibility. To deal with this problem, a new ele- performs various functions that a worker in a
ment, called a decoupler, can be introduced into manned cell performs routinely, imparting flex-
automated cell designs to increase the system’s ibility back into the system.
capability and flexibility. Functionally decouple processes. Decoup-
Looking at the various functions a decoupler lers separate or decouple processes so they are
can carry out, it is observed that decouplers are not dependent on each other. This relaxes the
Output
Input
CNC
grinder
M3
D2
Robot
CNC
turning Robot
center controller
C
CN ing
l
m l ter
i
M1 cen
l
Too nger
D1 cha
M2
l
Cel ber
Decouplers hold m
m troller
e
one unit of
con
each part in family
Figure 13-2. Robotic manufacturing cell using conventional CNC machine tools with decouplers.
Station 2 Station 7
Test power cord,
Install printed attach label
circuit board
Kanban
square
Station 1
Station 8
Place disk drives Package, add count at
into base video display terminal
150 1
75 2
30 30
50 3
37.5 4
10 10
30 5
20 20
Figure 13-3. A manned assembly cell with kanban squares (decouplers) with two or three workers.
need for precise line balancing when the work in rect, repeatable registration of the part for the
the cell is redistributed. Also, decouplers can robot’s gripper. The transportation process can
overcome the problem of variations in manual be automated. Figure 13-5 shows a decoupler
process time or machine time. chute transporting and reorienting parts.
When workers have increased variation in Part manipulation. In an unmanned robot-
their processing times about the mean, pro- ic cell, decouplers handle a parts family and
cessing time output from the cell decreases. can locate a part for the robot gripper. Often
That is, variability in processing time decreas- parts must be manipulated or reoriented for
es the output but increasing the decoupler insertion into the next process. Figure 13-6
capacity improves output. shows a simple decoupler designed for part
Poka-yoke inspection. A decoupler reorientation. The robot places a part in slot
checks critical part dimensions. The feedback “A.” Before rotation, surface One is at the top.
signal from sensors is sent to the process con- After rotation, surface Two comes to the top so
troller so the process can be modified or cor- a new orientation is obtained. The wheel holds
rected before any defects are made. Thus only four slots, one slot for each part in the family of
good parts are pulled to the next station. This four parts.
is an example of 100% inspection. Figure 13-4 Handling a parts family. In a cell, han-
shows a decoupler for inspection. This decou- dling a parts family requires flexible decou-
pler is designed for handling a parts family plers along with flexible workholders. A
that the cell is producing. decoupler designed with a family of chutes (as
Suppose the family has three basic sizes of shown in Figure 13-4) allows parts to pass
parts—small (S), medium (M), and large (L). each other in the cell or to skip a process. The
Just as the machine tools must have flexible simple rotary device shown in Figure 13-6 can
fixtures to handle the different part sizes and handle a family of parts while providing part
the robot’s gripper must adapt to different part reorientation.
sizes, the cell must have flexible decouplers to Automatic production control. Decouplers
handle a family of parts. This flexible decoupler act as automatic production control devices
chute is designed for a family of three parts (S, between two processes when automatically shut-
M, and L). ting off an upstream process that cycles faster
Freedom of movement for the robot. The than the downstream machine. An example of a
robot’s freedom of movement is the most impor- decoupler performing this function is shown in
tant function of the decoupler as far as cell Figure 13-7. The decoupler consists of an inspec-
automation is concerned. In staffed cells, the tion station, a gravity chute, and a limit switch.
addition of the decoupler allows the human The limit switch automatically stops the opera-
worker to move upstream, which is opposite to tion of process “A” when three parts are in the
the part-flow direction. That is, the decoupler chute. One part has been inspected, the second
allows the worker to move in any direction part is being inspected, and the third part is
within the cell. The same is true for the robot. ready for inspection. Process “A” makes no addi-
Remember, most robots have only one arm and tional parts until they are needed, that is, until
cannot walk, so compared to a human, robots the controller for process “A” receives a signal
are severely handicapped. from the decoupler that the previous part has
Intracell transportation. Decouplers in passed inspection. In Figure 13-7, machine “A” is
the form of gravity slides or chutes can trans- not being adjusted based on the output inspec-
port parts from process to process within the tion. Instead a light is turned on that attracts the
cell. This eliminates the need for precisely worker’s awareness of the problem.
locating the part on the decoupler’s input side. Branching. When two or more processes
Precise location on the output side ensures cor- are being fed by one machine or one machine is
MCU
Machine A
Makes 3 sizes
of parts S, M, L
Inp
ut c
hut
e
Quality feedback
Input to Machine A
Inspection station
D
o
L w
S M
ute n
S
u t ch
utp
Machine B
O
Inspection
M station
Floor
L
A M
Part Inspection
T
Verification
A Reorienting
Spring M part for robot
Output
A side
Gravity chute
Inspection Inspection Inspection
device device device
Sensor
Figure 13-5. This decoupler transports the part and also reorients and inspects the part with feedback to the previous process,
detecting the part's presence or absence as part of the control system for the cell.
90 0
Side 1
A D
D B D B C 1
C C B
Load Unload
Reorientation in either direction Reoriented part
Figure 13-6. Decoupler reorients by the rotation of the part and provides exact registration (location) of the part for a robot doing
the unloading.
being fed by two or more processes, this is The lack of adequate robot process capability
called branching. When branching occurs in can lead to robot installation failure. It is
automated cells, a decoupler is needed to con- embarrassing to remove a robot from an assem-
trol the branching function. In general, this sit- bly line or manufacturing cell because the
uation should be avoided since it clearly robot was not able to accomplish desired tasks.
presents a violation of the lean production phi- Manufacturers and users need more effective
losophy. The lean production reasoning is clear. techniques for measuring robot process capa-
The branching methodology induces variability bility so they can evaluate a robot’s ability to
into the system. Sometimes this is unavoidable perform required tasks and select a robot meet-
but other times preventable. ing the company’s needs. Robot tasks or
processes requiring process capability meas-
Robot Process Capability urements include:
Machine-tool process capability is defined
as the natural capability of the process. That • assembly,
is, the machine’s ability to consistently per- • insertion,
form a job with a certain degree of accuracy, • spot welding,
precision (repeatability), reproducibility, and • inspection, and
stability. In studies, which traditionally were • loading and unloading parts.
performed on metal-cutting machine tools, In unmanned cells, equipment is placed any-
the parts made on the machine could be where within the robot’s reach, but the robot’s
examined to determine the machine tool’s ability to perform the task within its reach is
process capability. However, in many robot highly variable.
tasks (for example, material handling or Robot process capability is a function of:
assembly) there is no product or output that
can be directly examined or measured to • move speed,
determine the robot’s process capability. • move position,
Therefore, the problem of how to measure • arm orientation, and
robotic process capability is important. • the weight being transported.
Chute
Limit switch
Inspection station
Machine A 3
2 Machine B
1
Feedback on
quantity
Ready
for inspection
Part ready
Feedback on quantity for processing
by Machine B
Figure 13-7. Automatic process-control function of a decoupler, combined with quality control.
Table 13-2
Summary of measuring techniques
Method Resolution Accuracy Repeatability Advantages Disadvantages
Contact Sensing
Dial indicator 0.0025 mm n/a n/a a. low cost a. for point to point only
b. easy to setup for b. measuring range limited
one dimensional by dial construction
measurement c. difficult to set up for
accuracy test
Extensible ball bar 5 x 10
6 m 32 x 10
6 m 20 x 10
6 m a. simple in operation a. limited to one dimensional
b. both positioning and measurement
patch accuracy can b. measuring speed is limited
be measured
Latin square n/a n/a ±0.15 mm a. experimental setup is a. slow speed
defined by statistical b. for point to point test only
procedure c. velocity must be very slow
when the probe contacts the
tooling balls
LVDT sensor 0.0025 mm n/a n/a a. both position and a. slow speed
orientation can be b. for point to point only
measured c. measuring range is limited
(2 mm)
Noncontact
Sensing
Acoustic-based a. uses lightweight sensor a. environmental effects are
system; 0.2 mm n/a n/a b. can improve accuracy likely
acoustic-range by using ultrasonic
sensor sensor
Optical-based a. high speed (4,700 Hz) a. lighting environment
system; multiple markers should be controlled
active-video b. can be used for both to avoid reflection
system: PTP and continuous path b. LED position may
SELSPOT and 12 bits 1 mm Point: 0.005% c. pre-surveyed calibration cause inaccuracy
WATSMART (1:4,096) Path: 0.01% frames are available c. high temperature due to
extensive use may cause
innaccuracy
Theodolite 13 x 10
6 m n/a n/a a. simple algorithm and a. method very slow (about
operational procedure one point/min)
b. system capability is b. manual operation
good enough for may cause eye fatigue
industrial robots for a long operation
c. for point to point only
Proximeter 25 x 10
6 m n/a n/a a. can measure both a. linear measure range
position and orientation is limited (2 mm)
b. for point-to-point only
c. difficult to set up for
accuracy test
and computer link the assembly. The cells also in more than a conventional data-processing
are linked to the production system. Questions sense. Instead of merely performing repeti-
to consider include: tive calculations on data, the system can
understand the inherent relationships in the
• Will the unmanned manufacturing cells
data. For example, an engineering-design
be able to produce 100% perfect quality?
change related to a product would automati-
• Will the processes have computerized
cally be propagated throughout the databases
adaptive control?
affected by the change. New process plans
• Will the system have a short lead time and
and tooling would then be automatically
low unit cost?
brought into the manufacturing system.
• Will machines run perfectly and never
Ultimately, automatic changes in flexible fix-
break down?
tures needed to accommodate process change
• Will setups be eliminated?
would be made as well.
• Will all the material be transferred one
The factory with a future substitutes
at a time, from process to process, auto-
machines for human workers. The implications
matically?
for future employment in factory operations
Likely, no factory can operate in a perfect are clear. As automation is implemented, there
manner, but perfection can be a goal. will be a shift from direct to indirect labor jobs.
Continuous improvement of the products and Human workers will not participate directly in
processes are within reach, and the better the manufacturing and assembly processes, but
manufacturing system prior to automating and will be required to manage and maintain the
computerizing, the better it will be afterward. processes. This will require more highly edu-
The kanban method used in cellular manu- cated people. Included in that education must
facturing is still valid in a computerized manu- be an understanding of technology and manu-
facturing environment, except that the facturing systems. Education is the keystone to
computer monitors the progress of materials survival for a manufacturing nation.
and bar codes are substituted for kanban cards. The new era in manufacturing-systems
Automated guided vehicles (AGV) may replace design is here. New manufacturing giants have
kanban carts because they can be computer emerged. Honda and Toyota are examples of
controlled and are flexible. New methods for such giants. The development of lean and inte-
transferring parts between devices will be grated-manufacturing systems chiefly fuels
devised along with flexible modular fixtures these giants. In Alabama, Mercedes Benz,
and decouplers to handle the diverse product Honda, and Hyundai have constructed a plant
mixture made in the factory. However, trouble that embraces the tenets of lean production.
lights will still flash in the factory with a future Volkswagen has such plants in Brazil. Honda
when the monitoring computers beep to notify has exceptionally lean manufacturing plants in
maintenance personnel that something has Ohio and Canada. In total there are now 16 auto
gone awry. assembly plants on U.S. soil representing 10 for-
Information systems in the factory with a eign car companies. These are factories with
future will be capable of data interpretation futures.
Factories are turning to high-powered 3D ics feedback for the industrial engineer can-
graphic-simulation software and low-cost not be overstated.
powerful desktop computers for manufactur- Simulation efforts have increased signifi-
ing-system design and ergonomic analysis. cantly as the personal computer has evolved.
There are several commercial software pack- The personal computer’s emergence as a
ages available. These software programs gen- viable graphics workstation has had a sig-
erate as well as analyze human engineering, nificant impact on the application scope of
anthropometry, and physiological functions technology. There are increasing numbers of
of a simulated human model. With these soft- companies participating in high-level simu-
ware packages, engineers and ergonomists lation. Simulation of the system, work cell,
collect data and analyze potential problem and human tasks can be economically gener-
areas during design or after system or cell ated and run alongside other computer
implementation. applications.
The power of using 3D simulation and the An engineer with graphic simulation capa-
virtual environment for solving manufacturing bilities has an opportunity to contribute to
and ergonomic problems can be extraordinary. manufacturability or assembly analysis, and
process, layout, factory throughput, and
It gives engineers the ability to see the interac-
ergonomics analysis while in the cell design
tions of moving people, machines, and parts
stage (see Figure 14-2). Computer technology
across the intended factory floor (see Figure 14-
allows all engineering groups to participate in
1). The factory floor can be brought to life from
the product life cycle.
the original CAD data, and the user can model Traditionally, industrial engineers han-
the accurate physical characteristics of entities dled workstation design, layout, productivi-
and human motion. ty, quality, and human-factor analysis.
Simulation in a virtual manufacturing Workstation design often focuses on a single
environment allows multiple disciplines cycle of operation to address layout, process,
within an organization to contribute to better ergonomics, and safety issues. Discrete
solutions. It is a safe, low-cost method to ver- event simulation models, which are 3D-
ify product design and manufacturing related graphics-based, are used for many manufac-
factors for human engineering, process, tool, turing activities including final assembly,
material flow, and safety. Computer simula- subassembly, manufacturing cells, various
tion is a powerful tool for communication. The flow lines, material handling, or entire facto-
power of visually communicating design or ry-production operations. Metrics such as
process plans to management, providing visu- throughput, production rates, process uti-
al instruction to workers, process evaluation lization, ergonomics analysis, and standards
for the manufacturing engineer, or ergonom- can be accurately determined.
Figure 14-1. High-level 3D simulation of a manufacturing cell. Figure 14-2. Simulated worker fitting large fabricated panel
into position prior to welding.
HISTORY
Computer simulation began around 1960 placement within the cell had a high impact on
when NASA pioneered computer use for satel- cell performance. Accordingly, the process lay-
lite design. By 1970, larger manufacturing out function was found to be critical to manu-
operations, especially in aerospace, used com- facturing cell performance; and is, therefore,
puters and proprietary software that they best completed with computer simulation’s aid.
developed or contracted from engineering soft- In the early 1990s, several studies indicated
ware companies. Other companies followed the that the purposes of simulation of manufactur-
lead of aircraft manufacturers and soon most ing systems are to better understand the
medium- and larger-sized companies had com- dynamics of the proposed system. This ensures
puters for processing business data and numer- that the designed system satisfies the con-
ic control (NC) machine-tool programming. straints outlined in the problem definition, and
In 1973, Richard Muther was one of the first that it protects workers’ health. In 1988 it was
to apply a systematic methodology to the man- reported in a study that simulation models
ufacturing systems planning process. Muther could increase comprehension and improve
considered the planning process a loop that had insight into the performance of a manufactur-
to be executed twice; once for the general over- ing system (Kamrani et al. 1998). The process
all layout and a second time to detail the layout of constructing the simulation model forces the
plans. system’s designer to question to foster this
Early manufacturing cell designers were pri- insight. Analysis of the numerical results of the
marily concerned with cell layout. Two major simulation runs could identify true perform-
studies of manufacturing cell implementation ance indicators for the system. Table 14-1 illus-
in 46 U.S. factories were conducted in 1989 trates manufacturing engineering areas
(Wemmerlov 1997). The researcher reported particularly suited for evaluation by simula-
that the formation of manufacturing cells was tion models. A systematic simulation methodol-
not especially problematic, but the equipment ogy was developed in the study, which outlined
Though they may vary in anthropometry, size, was well-established engineering practice. A sim-
shape, and capability, they are intended to mir- ulated manufacturing environment, complete
ror human characteristics. Virtual humans can with human models interacting with existing
be programmed to carry out the tasks required layout data, was extremely important to avoid
by their human counterparts. They act within the time and expense of costly physical mockups
simulated environments, manufacturing com- of processes and tasks.
ponents, assembling, and maintaining process- Cellular manufacturing systems are quite
ing equipment. The computer models also can complex and their successful implementation
be used to train real workers on how to perform requires development of a sound, replicable
future assignments. design methodology for cells and systems
The growth in the interest and development (Massay et al. 1995). Therefore, animation-
of digital humans can be traced to at least two enhanced simulation plays an important role in
major factors. First, the advances in computer the logical design of manufacturing cells and
graphics technology have delivered the speed other manufacturing and production systems.
and performance necessary to efficiently dupli- Real-time, 3D animation technology is used to
cate and visualize human motion. In fact, abil- simulate, analyze, and provide optimization
ity to run human simulations on middle- to capabilities for the manufacturing system
high-end personal computers has put this tech- designer. Today, simulation is an even more
nology into the mainstream of engineering valuable tool due to globalization of world mar-
design tools. Second, engineers are becoming kets, where industrial firms, in an ever-increas-
increasingly aware that ergonomically correct ing competitive environment, are required to
designs are critical and must be accounted for reduce their production costs, produce superior
in every phase of the product life cycle. From quality, and shorter delivery times. This must
concept to production, the application of be accomplished in a flexible manner. Computer
ergonomics principles helps ensure that the design and analysis employing high-level
product can be safely made. graphic software adds to that flexibility.
High-level simulation has achieved formal Manufacturing design simulation helps engi-
recognition by many major manufacturers. neers identify design errors, make better deci-
Within these companies, the design engineer, sions, and more quickly evaluate the effects of
manufacturing engineer, and industrial engi- those decisions (McLean 1998). By improving
neer need a set of human-factor tools to sup- process specifications and simulation capabili-
plement the business process, just as they need ties, a much greater percentage of products are
design tools, dynamic analysis tools, and cost- produced correctly the first time with better
ing tools. Whereas most early human-modeling use of production resources. The best resources
packages were aimed primarily at ergonomics for the job are selected more often and nonpro-
professionals, today they are used to solve the ductive work consumes less time. Also, the
complex human/product-related issues of many overall time to perform the engineering func-
different engineering functions. tion is reduced if fewer changes to plans and
Several major industries have embraced programs are required once a product goes into
human modeling and simulation. Among the production. These improvements result in less
first to apply the technology were the aerospace, scrap and rework. The integration of software
automotive, and shipbuilding industries. These packages and common databases ensures that
industries were driven by long product develop- less time is wasted re-entering the same data
ment cycles, high-cost startups, and intense into multiple engineering tools.
manual labor demands. These are also industries The utilization of mathematics models appears
where the use of computer-aided-design (CAD) to be more optimistic in predicting manufactur-
technology to develop digital mockup of products ing cell performance (Mital 1995). While it seems
Graphical 3D-simulation modeling facili- derived from the motions created within the
tates engineering’s ability to design both tool- worker simulation. These software packages
ing and equipment, and validates human include techniques for determining safe pos-
capabilities such as reach, clearance, and tures, lifting, energy-expenditure evaluation,
vision. Simulation allows human motions or and physiological analysis.
labor elements to be prototyped for a range of Ergonomic and physiological functions
operator sizes so many of the ergonomic included in many of the 3D and virtual reality
aspects of a task or job can be evaluated prior simulation packages are listed in Table 14-4.
to purchasing tooling or equipment. Some soft- These are most frequently used for ergonomic
ware packages have the ability to rapidly pro- studies that relate to manufacturing-system
totype human motion using entire libraries of design. There are several types of simulation
whole-body and hand postures, and point-and- software programs available and they range
click routines to generate walking, lifting, and from the relatively simple to complex. Some
carrying. The use of such software, with use mathematics models while more sophisti-
ergonomic functions, allows engineers to quick- cated models use 3D and virtual reality simu-
ly and effectively evaluate manual assembly lation coupled with sophisticated analytical
tasks to improve cycle times and eliminate or subroutines.
reduce injuries.
While using simulation software, engineers Metrics
and ergonomists may proactively address
The ergonomic benefits of an optimum man-
human/manufacturing system interface issues.
ufacturing system, whether it is a job shop
A wide anthropometrical range of simulated
humans is used to carry out tasks while main- workstation or a manufacturing cell, can be
taining a proposed product or work cell design. observed by several metrics to monitor and
Four immediate benefits are found: track the health and well-being of the work
force. Some of the ergonomic and physiology
1. While in the design stage, designers may metrics are: joint deviation, nerve conductivity
virtually eliminate the time and costs of rate, venous pooling, energy expenditure, reach
expensive tooling rework or changes in and grasp, heart rate, bend and reach, and
design. repetitive motion. The tools used to gather
2. Simulation eliminates costly and time con-
suming physical mockups.
Table 14-4
3. Manufacturing engineers reduce time-to-
Ergonomic and physiological functions needed
market by visualizing and validating
in simulation programs
processes digitally before the product
design is frozen, previous to committing 1 Visualize the feasibility of certain tasks
resources, and before purchasing or modi- 2 Reach and grasp
fying equipment and tooling. After the 3 Bend and reach
simulation is validated, engineers may use 4 Eye windows to view what the model sees
the product and process models for train-
5 Kcal prediction model for energy expenditure
ing, maintenance, and documentation.
4. Ergonomics, anthropometry, physiology, 6 Motion time measurement
and safety issues can be analyzed and 7 Rapid upper-limb assessment posture analysis
addressed, while the system is still in the 8 National Institute of Safety and Health lifting
design stage. guidelines
Most leading software programs predeter- 9 Anthropometry switching for human models
mine time standards using the cycle times (Deneb 1998)
and/or analyze this data include rapid upper- 2. anthropometry and the structure of the
limb assessment (RULA), motion time meas- human model;
urement, and National Institute of Safety and 3. model manipulation, reach, and visual
Health lifting guides, high-level ergonomic sim- analysis functions; and
ulation software, and other means. These met- 4. other ergonomics evaluative functions.
rics are tools to identify and prevent or
From a manufacturing-engineering point of
eliminate work-related musculoskeletal disor-
view, the programs must include machines,
ders from the workplace. Doing so, manufac-
equipment, and the tools necessary for a real
turers and their work force reap two significant
manufacturing cell to work. The various pro-
benefits: (1) elimination or reduction of worker
grams currently available have varying system
injuries on the job and (2) reduction of the
requirements, operating characteristics, appli-
incredibly expensive financial losses that bur-
cability, and ergonomics evaluation functions
den manufacturers resulting from direct and
for ergonomic simulation and analysis.
indirect costs of work-related musculoskeletal
disorders. Delmia Software
Software The primary tool used in the following case
study was the 3D ergonomic analysis simula-
Since the worker is the most important aspect
tion software produced by Delmia Corporation
of workstation and manufacturing cell design,
based in Auburn Hills, Michigan. The software,
more leading-edge technology, such as computer
Delmia Envision ERGO, has these capabilities:
modeling, has been developed to assist in human
drawing in two and three dimensions; design
engineering and manufacturing. Computerized
working models of processes and tooling; and
ergonomic and physiological simulation pro-
design, simulate, and analyze manufacturing
grams for workstation and manufacturing-cell
systems. It is an interactive, 3D simulation-
design provide a convenient interface for the
based tool for ergonomic analysis that uses
user to generate and manipulate 3D images of
anthropometrically correct digital human mod-
human and workstation characteristics graphi-
els (Deneb 1998). The software also supports
cally. Through the use of high-level ergonomics
virtual reality devices where designers can
software, the designer can construct many
immerse themselves and virtually manipulate
anthropometric combinations to represent
a product or walk through a manufacturing
human workers. The programs give the user
workstation or system.
complete control over the development of the
The human simulation tool can be used for
digital model. They provide comprehensive
workplace design and analysis, allowing the
analysis to evaluate human/machine interaction
manufacturing-systems designer to prototype
through programmed commands. The user is not
human motion within a work area using propri-
required to be a computer expert to use these
etary graphical motion programs. This allows
programs. However, most require either a rea-
the manufacturing engineer to design and/or set
sonable learning curve or professional instruc-
up the motion sequences for the simulated work-
tion to be used productively.
ers of the manufacturing cell. With the graphical
The cell and workstation designer must ana-
motion programs, a motion sequence consists of
lyze and choose carefully to determine if a
an ordered collection of worker postures gener-
given program fulfills the designer’s require-
ated by the designer. Computer software manip-
ments. Four broad criteria should provide the
ulation of the worker’s limbs generates
basis for comparison.
simulated worker postures. The software pro-
1. usability in terms of hardware and soft- gram uses a combination of forward and inverse
ware; kinematics. If the worker posture exceeds the
reach of the worker’s arm, the inverse kinemat- ulate the simulated worker’s body joints. Drop-
ics provide a solution by automatically bending and-pick screens code program tasks such as
the worker’s torso. Capabilities are provided for walking, stooping, squatting, bending, and
developing time standards and studying the grabbing.
ergonomics of a job, including parameters relat- The human-motion-programming interface
ed to lifting, energy expenditure, and posture is based on the graphic programming method.
analysis using percentile-based, fully articulat- A main objective of software designers was not
ed 3D human simulated models. to burden the cell designer, engineer, or ergono-
The simulated human worker is central to mist by having to write extensive computer
the case study, just as real workers are the code; but to allow simplified methods to govern
essential resource in a manufacturing facility. the actions and motions of the product, system,
Anthropometrically, the simulated worker uti- or worker. Commanding worker action and
lized was a 50th-percentile female, and is used work elements was done by selecting drop-and-
for both the manufacturing cell and each of the pick commands then manually manipulating
five functional system workstations. the worker via mouse movements. Thus, the
The design of the manufacturing cell or designer was able to exactly program the work-
workstation, jigs and fixtures, and workstation er-motion sequences and form them into a
support equipment could only be done after series of postures. A motion sequence is an
examining the inherent human factor interface ordered collection of postures where the user
in the design process. Design and industrial manipulates the model worker’s limbs using
engineers traditionally relied upon expensive task-based and graphic programming. A pos-
and time-consuming mock-ups to evaluate ture contains information regarding the joint
their designs and workplaces. To avoid these values, attachments, and analysis. There are no
expenses, it was important to evaluate avail- limits to the number of postures in a motion
able design alternatives early in the initial sequence or the number of sequences attached
design stage. Three-dimensional simulation of to the worker. The interface allows quick selec-
human workers where the software carries out tion for moving back or forward through pos-
human motion, reach ability, anthropometry, tures for quick visual verification and editing of
biomechanics, cycle times, and ergonomic motion sequences (Deneb 1998).
analysis was extremely important. This Kilocalorie prediction. The Delmia ERGO
methodology was used to eliminate or reduce software provides a kilocalorie prediction tool
human suffering and the many upfront costs that is used to obtain kilocalorie consumption
associated with product and new system devel- estimates for various manual industrial and
opment. The utilization of such simulation soft- assembly tasks that simulated workers carry
ware can result in significant reduction of the out. This tool is primarily designed to ensure
design effort and related costs in both time and that work designed by the manufacturing engi-
financial resources. neer is within the worker’s energy capabilities.
Before the designer can analyze the product, Metabolic-energy expenditure is a physiologi-
manufacturing system, and ergonomic func- cal measurement used to accurately estimate
tions, they must draw or import into the soft- the task intensity that a worker can continu-
ware programs a product, workstation, or ously execute. Thus, by examining the energy
system. Next, for ergonomic functions, the requirements for a task, the manufacturing
designer must teach the digital worker how to engineer or ergonomist can assess the capabil-
carry out the designed tasks. The teaching ity of the worker to perform a given task. Then
process, for Delmia software, is done through a they can use this data to assist in the estab-
dedicated human-motion-programming inter- lishment of duration and frequency of rest peri-
face. This interface provides utilities to manip- ods and evaluate alternative work methods in
case the task is too intense. The kilocalorie pre- nals the action level. However, in use, the color-
diction model provides strong support for man- code scoring appeared to be conservative.
ual handling activities that involve walking, Delmia engineers confirmed that this was the
carrying, lifting, pushing, or pulling. For this case, so a combination of 3D simulations that
study, it was arbitrarily determined to use a generated postures carried out the RULA
work classification endurance limit of four kilo- analysis for this study. Then the postures were
calories per minute as the maximum limit for viewed at various points of view and manual
female workers, who are engaged in light RULA assessments were made based on simu-
industrial tasks. lated joint angles. Analysis was then carried
The way the simulation software works is out manually on the worst-case selected pos-
relatively simple. It uses designer inputs and ture for a given task. This was done by decom-
then breaks the motion sequences into sub- posing the individual process jobs into work
tasks that the energy expenditure prediction elements. Then, by posture flipping the simula-
model understands. Once the user supplies this tion, the worst-case posture for each job ele-
input, the system computes the geometric dis- ment was found. This relatively simple
tances and energy requirements from the sim- procedure froze the simulated worker in the
ulation data. An energy-expenditure log from worst-case posture and the postures were then
the prediction model is written to a file. The file manually scored.
contains the data in report form; there, it sums With the manual/simulation analysis tech-
the total energy expenditure and time, and nique, the simulation records the position of
averages the rate for the motion sequences in the upper limbs and head, then the trunk and
the analysis. legs using two separate number codes derived
Rapid upper-limb assessment (RULA). from the worker’s postures. The procedure
The Delmia simulation software has a posture requires that segments of the human body be
analysis system to investigate the workers rated on a simple scale that produces a series
exposure to risk associated with work-related of numbers, which, in turn, were matched
upper limb disorders brought about by stress- against a grid. The value of the numbers and
ing postures. The RULA tool can be used to their position on the grid determines the sever-
reduce or eliminate workplace hazards due to ity of the posture. The engineer or ergonomist
workstation design and layout. RULA-based can then take action based on this data to
analysis examines the following risk factors: reduce or eliminate the problem.
number of movements, static muscle work, NIOSH lifting tool. This analysis function
force or load, working posture (body-part of the software is based upon the NIOSH lifting
angles away from neutral), and time worked equation. It analyzes the two-handed-lifting
without a break. These factors are combined to activities of workers. The lifting equation is
provide a final score for a worker’s particular aimed at reducing lower back injury and relat-
posture. For working postures, RULA focuses ed musculoskeletal disorders. The output can
on the use of arms and wrists, head position, alert manufacturing cell and workstation
and upper-body posture. RULA also takes into designers to the need for assist devices for
account the worker’s legs. workers to perform their jobs without back
The designer provides input to the system injury risk.
regarding muscle use for arms and body load- The NIOSH lifting equation gives a measure
ing. The system combines this information with of human lifting capacity. The output of the
the posture category, based on a comfortable equation is a weight limit suitable for men and
joint range of motion, to give an action level for at least 67% of women. The NIOSH limit is 51
the posture. A color change of the simulated lb (23 kg) under ideal conditions and if all lift-
worker’s body part stressed by the posture sig- ing risk factors are 1.0. The Delmia software
calculates the risk factors associated with the ditional functional job shop. In both cases, the
lifting task by computing the input variables simulation software generated and collected
from the postures. Based on the computed risk data on the various ergonomic stressors and
factors, the system outputs recommended reported those back for further analysis.
weight limits for the specific task. The weight
limit takes into consideration the risk factors The 30-Second Rule
at the lift’s beginning and ending. The Delmia Cumulative trauma disorders of the upper
system also allows the design engineer to carry extremities are physical ailments of the wrist,
out “what if” scenarios. arm, and shoulder caused by the cumulative
Cycle time. The Delmia simulation soft- effect of repeated mechanical stresses. They
ware comes with an accurate performance develop gradually over time and are believed to
analysis subroutine for rapid evaluation of be work related. Typically, a task is considered
work measurement. This module determines repetitive if the basic cycle time is less than 30
time standards for tasks to support productive seconds—hence, the 30-Second Rule (Konz
task planning. The standard for this simulation 1995). The 30-Second Rule is primarily con-
tool is methods time measurement.
cerned with repetitive motion of the hand,
wrist, shoulder, and back. Repetition occurs
Case Study: Ergonomics Simulation
when exertions, motions, recovery, vibration, or
In the following investigation, which utilized cold temperature exposure are repeated by the
the simulation-driven cycle-time calculation worker in the course of carrying out a task. Of
feature of the Delmia software, the manufac- particular concern are repetitive motions that
turing cell and the five job-shop workstations involve force and result in extreme deviations
were analyzed for potential repetitive-motion for the body part carrying out the task. It has
injuries based on the 30-Second Rule. The soft- been reported that highly repetitive work has a
ware was used heavily for cycle-time calcula- 2.8 times higher risk of injury than the same
tions and the kilocalorie energy-expenditure
work with longer cycle duration (Silverstein
function. In addition, rapid upper-limb assess-
1986). Also, work that involved extreme devia-
ment and repetitive motion analysis via the 30
tions coupled with high force had a 30 times
Second Rule were performed.
higher risk of worker injury.
The research for this case study used Fourth
Industrial Revolution technology along with
Fourth Industrial Revolution
traditional industrial and manufacturing engi-
neering methods for data collection and analy- Some believe that the state-of-the-art tools
sis to answer the question as to whether and analysis now being used in the manufac-
manufacturing system design has an impact turing design process are catapulting manufac-
upon worker ergonomics. turing into the Fourth Industrial Revolution.
The research plan included simulating two Teams of design, industrial, manufacturing,
competing manufacturing systems, lean pro- quality assurance engineers, and ergonomists
duction and the job shop, in which the modeled will design and test a product and the process-
human carries out various computer-generated es to produce it. These teams will use powerful
tasks. These identical manufacturing tasks desktop computers that run high-level simula-
allow the modeled worker to be exposed to var- tion, design and analysis software packages;
ious physiological stressors while carrying out and, everyone will be working off the same
simulated cell or job-shop manual tasks. The database. In the Fourth Industrial Revolution,
manufacturing processes are simulated using a virtual product will be designed and tested;
two manufacturing system designs: (1) an the manufacturing system will be built and
interim lean manufacturing cell; and (2) a tra- workstations designed; subsystems built and
tested, such as product quality assurance; and oz. (0.4 kg). The cast cylinder blanks are
ergonomic and safety factors ensured; all in a shipped from the die caster to Greer for final
3D/virtual environment. This completed and machining.
tested system will be finished before the first The John Deere cell and functional job shop
chip is cut, the first piece of metal bent, the were modeled using Delmia software. The John
first die made, or the first prototype is actually Deere manufacturing cell was simulated and
produced. Design and analysis facets will be verified first. Then, the job-shop processes were
completed via computer simulation. When the redesigned using the exact manufacturing
first part is made, it will be to specification cell’s processes as simulated for the modeled
with no scrap, rework, or starting over. cell. The processes are exactly the same for
both the cell and job shop. Only the methods
The Cylinder Manufacturing Cell
employed for process usage were different.
The manufacturing cell used for the 3D sim-
ulation model in this study is a viable and Decomposition of the Job Shop
mature interim machining cell. The cell is a
vital part of the John Deere Corporation facili- Typically, lean manufacturing cells are
ty located in Greer, South Carolina. Figure 14- designed using the former job shop’s processes.
3 shows a schematic of the cell. The cell The process was reversed for this research
produces cylinders for small gasoline engines. where each cell process was decomposed or
The engines power string trimmers, leaf blow- redesigned into a typical job-shop workstation.
ers, and chain saws. The factory is located in a As is the normal practice, job-shop workers
2 2
240,000 ft (22,297 m ) building and operates stay with the process the entire shift, tending
with roughly 160 people, including two manu- the process by loading and unloading parts,
facturing engineers. The cylinder cell and its and waiting for the machining cycle to end to
processes form the core of the study. Both the initiate another round of servicing. The simu-
cell and job shop simulations were designed lated job-shop workstations were designed
from the Deere cylinder cell; thus the compar- with typical functionality for that manufactur-
isons and analysis have this cell as a common ing system.
origin. At each job-shop workstation, the worker has
The Greer plant produces about one million two boxes of parts, input and output; one is for
cylinders a year from two manufacturing cells. raw material or semi-finished parts and the
These cells replaced two former transfer lines; other for completed parts. The normal job-shop
there is still one other cylinder transfer line in processing cycle for any one of the five process-
operation. The two manufacturing cells have es redesigned into the functional system would
cycle times that are greater than the transfer
include unloading a just-finished part, then
line. However, tolerance problems, machine
loading a new part into the process. Some
breakdowns, and an array of other technical
problems continually plagued the transfer line. workstations may include other manual tasks
Therefore, the quantity of acceptable cylinders such as cleaning the machine tool’s chuck or
produced by the transfer line is below either of workholding device with an air gun and dip-
the two manufacturing cells. ping the finished part into a bucket of coolant
The completed cylinders are assembled into solution to wash off chips. After starting the
small gasoline engines for consumer products machine tool, the worker may give the just-fin-
designed and built for the commercial market. ished part a visual inspection and then wait for
The cylinders are also produced for aftermar- the machining cycle to end. Once the process
ket spare parts. The cylinder is die cast from has completed the machining cycle, the job-
aluminum alloy and weighs approximately 14 shop worker/machine cycle starts over.
M2 M3 M4
M1
Parts Flow
Input Output G1
M5
stations work in series. Therefore, there is zero and its effects on the workers in a work cell or
idle time for the cell worker. Table 14-5 does not other system.
indicate the total idle time difference for the Kilocalorie expenditure estimation directly
workers operating the five machine tools in the confronts the issue of how much energy is
job shop. However, the idle time component of required to carry out a worker’s task. Thus, it is
the total job-shop cycle time was calculated at a gage of how difficult, physiologically; a task is
91.66 seconds. So, the five workers’ total idle to a worker. Since individual workers are going
time accounted for 41.4% of the total job-shop to react differently to the same energy expen-
cycle time. diture, the issue of kilocalorie usage, as with
many ergonomic responses, will not be the
Ergonomic Analysis same for all workers. Nonetheless, the data and
The primary purpose of this study was to results presented will give the manufacturing
concentrate on the ergonomic ramifications of engineer and ergonomist a good indication as
lean manufacturing cells versus the functional to how much energy and hence how the aver-
design. The use of high-level 3D simulation age worker may react to the cellular and job-
focus and ensuing analysis provided the data. shop designs.
Kilocalorie expenditure. The amount of Table 14-6 illustrates the kilocalorie energy-
energy used by the workers working in each expenditure data generated from the worker
manufacturing system type was measured and operating the five processes in a cellular man-
the results compared. According to Delmia ufacturing methodology, and the five workers
Corporation, the kilocalorie analysis model operating each of the five workstations in a
assumes that the work can be broken into sim- functional job-shop design.
pler tasks. Once the work has been divided, the Both sets of simulated workers serviced the
average kilocalorie rate for the whole job can be identically same simulated processes with the
estimated by summing the energy requirements only difference being the manufacturing-sys-
for those individual tasks. Also, the energy tem design. To manufacture one part, the cell
required to maintain the posture is added in and worker consumed 1.1674 kilocalories, as com-
then averaged over time for the completed work. pared to the job shop, where five workers
A kilocalorie prediction model was pro- expended a total of 7.2705 kilocalories or
grammed in this Delmia simulation software 1.4541 average kilocalories per worker. These
package used for this study. The software was calculations yielded a 19.7% decrease in kilo-
extremely useful for calculating the amount of calorie expenditure for one worker going from
energy expended by these two manufacturing the job shop to the cellular system. If the total
systems during simulation. Also, on an on- kilocalorie expenditure of the cellular system
going basis, the energy expended could be col- versus the job shop is compared, then the five
lected by time periods and charted to track job-shop workers use 622.8% more kilocalories
progress of continuous-improvement projects than the one cell worker.
Table 14-5
Cellular manufacturing system versus job shop cycle times
M1 (sec) M2 (sec) M3 (sec) M4 (sec) M5 (sec) Totals (sec)
Cell 18.05 2.64 2.68 4.38 11.99 39.75
Job Shop 38.89 47.92 30.50 52.50 51.30 221.11
Difference 20.84 45.28 27.82 48.12 39.31 181.36
% Difference 53.6% 94.5% 91.2% 91.6% 76.6% 82.0%
Table I4-6
Cellular manufacturing system versus job shop worker’s kilocalorie expenditure comparison
M1 M2 M3 M4 M5 Total Kilocalorie/ worker
The simulated expenditure rate for the cell Rapid upper-limb assessment (RULA).
worker was 2.292 average kilocalories per This study was concerned with harmful worker
minute, while the job-shop workers expended postures associated with the following: chronic
2.479. This data indicates that the job-shop musculoskeletal injuries to the hand, wrist,
worker was utilizing energy at an 8.2% faster shoulder, neck, elbow, and back; and repetitive-
rate than the manufacturing-cell worker. motion injures related to overuse and excessive
Another interesting comparison was the force. The ergonomic features of the Delmia soft-
number of kilocalories required to produce one ware and other analytical means assisted with
part under the two different manufacturing the evaluation of those body parts that were
systems. For instance, the simulation model subject to work-related musculoskeletal
indicated that the job-shop workers (a total of injuries. The RULA tool was also employed to
five) must expend 12.4033 kilocalories per fin- evaluate postures relating to the ergonomic dif-
ished part. The manufacturing cell worker (a ferences between the lean shop and the job shop
total of one) used 1.1674 kilocalories per fin- manufacturing system designs.
ished part. This difference reflects 10.6 times The focus of the RULA analysis was to deter-
mine whether there was any advantage from an
per part more kilocalorie usage. In other words,
ergonomics and physiological standpoint to
the five job-shop workers must use 1062.5%
choose a cellular manufacturing system over the
more energy to produce one component when
functional job-shop-manufacturing system or
compared to the one cell worker.
vice versa. Although both systems could produce
The simulation proved that the lean cell sys-
the exact same product, the methods employed
tem consumed fewer kilocalories in every work- were completely different. The data was gener-
station comparison. Therefore, in this study, ated by manually completing the RULA assess-
the criterion for determining the better manu- ment worksheets for each task in both systems.
facturing system was the selection of the sys- Both manufacturing designs were subdivided
tem that was less stressful to the workers, that into five major processing tasks and further
is, the system that used fewer kilocalories. The division resulted in a subset of elements for each
cellular manufacturing system was shown to major task and for each system. Naturally, each
expend significantly fewer kilocalories per process task and the resulting elements were
worker to achieve the same output. Therefore, different; this is inherent for different manufac-
it is readily apparent that a lean production turing system methodologies. Once the two sys-
cellular system design can lower the risk of tem tasks were decomposed into elements, then
human suffering while possibly saving the the process of applying RULA took place on each
manufacturing organization a significant systems’ elements. Then it was only a matter of
amount of workers’ compensation, medical, and compiling the information into tabular form for
related expenses. analysis.
Summarizing the data in Table 14-7, it is Table 14-8 gives the average RULA scores for
apparent that the cellular manufacturing sys- the individual body parts by manufacturing-sys-
tem is much less stressful on the worker than tem design. The final RULA score averages both
the same work performed under the job shop the job shop and the cell workstations. The dif-
system. The scoring is based on the RULA ference scores of the job shop over the cell design
action levels, which range from one, indicated average from a low of -6.0%, indicating that the
by a final score of one or two, to an increase in job shop faired more favorably than the cell for
severity level of four, indicated by a final score the neck column, to 25.6% in favor of the cell
of seven or more. In almost every case, the cel- over the job shop. All but the -6.0% score showed
lular manufacturing system scored fewer that the cell was superior as far as RULA
points and the points were less severe than the assessments were concerned.
The final score for the job shop is 3.348;
job shop. Thus, Table 14-7 indicates that the
clearly the average RULA final score for the
cell should be much less harmful from an
functional design is in the unacceptable action
ergonomic postural standpoint.
level. The cellular design final score is 2.664,
A comparison of the total number of RULA
and still below the unacceptable 3.0 action
task elements of the total job shop versus the level. The difference between the two scores
cellular system is significant. In this case, there represents a 25.7% increase of the job shop
were a total of 56 RULA points. The job shop over the cell. Since lower is better, the higher
scored 33 points as compared to 23 for the cell. job-shop average score represents a substantial
This indicated that for the same tasks per- increase and is viewed as having a greater
formed in the cell and the job shop, the job shop potential for ergonomic problems associated
required 43% more assessment than was with poor work-related postures.
required by the cell to capture the potential Conclusion. Although the Delmia software
harmful postures. Further, the average scores incorporates a RULA function, the RULA func-
for the job shop were more severe than the cor- tion had limited utilization for this study. Mainly
responding scores from the manufacturing cell. because it only highlighted the simulated work-
Table 14-7
Summary of RULA assessment element data
Area Number of Elements Total RULA Average Standard Deviation Range Score
Table 14-8
RULA element comparison
Upper arm Lower arm Wrist Twisted Wrist Neck Trunk Leg Final Score
Cell 1–5 2.375 2.070 1.804 1.000 2.198 2.236 1.000 2.664
er’s stressed body part with different colors dur- ogy and personnel to be widely used in sever-
ing the simulation run to indicate the Action al locations simultaneously.
Level of the postural stress. This reason dictated
the manual RULA assessment along with the The Virtual Factory
use of the Delmia 3D simulation to determine The idea of a virtual factory is not new. But
exact body-part angles, etc. Nonetheless, the in the past, the development and application of
graphics generated by the simulations of the sys- large usable simulations of factory operations
tem designs were essential, and played an have been almost universally unsuccessful.
extremely important role in the analysis process. This is in the face of the fact that local simula-
Therefore, there is assurance that the utilization tions, for example, those simulations at the
of an ergonomic computer simulation with pow- process or workstation level focusing upon
erful ergonomic and physiological functions dur- manufacturing and ergonomical situations,
ing the design state for a manufacturing system, have been successful. The past models have not
workcell, or workstation would ensure lower pos- had the detail to properly capture factory oper-
tural stress in the worker. ations. These early models were too poor to pro-
vide solutions that benefited factory demands.
Virtual Reality Even small events may produce large fluctua-
Virtual reality (VR) is a special kind of sim- tions in factory operations, and adequate mod-
ulation that depends upon a human/computer els must be capable of reflecting these subtle
interface technology. It allows three-dimension- influences. Also the early simulations were too
al, multi-sensory interaction between the user slow to provide usable solutions. Often graphic
and the computer. With VR, instead of looking representations of the process were inaccurate,
at the flat computer screen, the user interacts which lead to wrong solutions. The user inter-
with a 3D computer-generated environment, faces were complicated and/or incomprehensi-
based on the way manufacturing or design ble and were unusable to manufacturing and
areas look in the real world. human factors engineers. A requirement for
VR is a real-world-grounded, business-ori- manufacturing models is that detail levels be
ented technology of particular interest to consistent since small factor variables at local
many areas of product design, manufactur- levels may have a significant impact at the
ing, safety, and ergonomics. For example, VR highest factory level.
allows companies to bring employees from Powerful simulation programs to capture the
the manufacturing side into the design entire manufacturing environment while being
process much earlier. This includes those detailed enough to simulate and analyze the
involved from conceptualization to post-pro- individual details of human and machine inter-
duction maintenance. VR is used for training face are required. Simulation software running
purposes, allowing a company’s best technol- on low-cost, yet powerful, computers, using
manufacturing and ergonomically correct soft- In the automotive industry there is a highly
ware with a short learning curve, is the key. defined, repetitive cycle of activity that occurs
The simulation, whether two-dimensional or with each new car program. At the General
three-dimensional, must provide fine-detail Motors’ North American Operations, engineer-
modeling of work cell or factory conditions. ing activities are associated with the product
Software must be powerful to emulate compli- life cycle, from concept to production. For the
cated factory operations. Computers driving most part, many individuals who are currently
this state-of-the-art software must be powerful performing this type of work are either design
and fast to produce solutions to real-world or manufacturing engineers who must validate
problems in real time. There are such VR sim- a product and process design plan several
ulation systems being developed and used in years before production begins.
various manufacturing operations, primarily in Many of General Motor’s divisions now
incorporate worker simulations as part of a
the aerospace and automotive industries.
strategy to evaluate manufacturing and robot-
ic processes. In addition, engineering groups
INDUSTRY SUCCESSES contracted to develop processes and tooling for
In the aerospace area, many organizations specific manufacturing cells are required to
including Boeing, Lockheed, as well as NASA, deliver robotic and worker simulations prior
are using human simulation to dramatically to process implementation. The design of a
decrease the time and expense associated manufacturing cell; the interaction of workers
with the design, redesign, prototyping, pro- with parts, equipment, and tooling; and the
duction, and maintenance phases. In the com- cycle time of operations are all important
petitive aerospace industry, one design or issues addressed through the use of 3D
production mistake can be costly. Every com- human simulation.
petitive edge must be used and virtual proto- Shipbuilding is another area where virtual
typing capabilities provide a huge advantage manufacturing has been utilized successfully.
by eliminating the need to build multimillion- Within the shipbuilding industry, human simu-
dollar mockups. To eliminate building a proto- lation is a key component to what the Navy
type, the company must have considerable refers to as simulation-based design. Historically,
confidence in the accuracy of the digital proto- shipbuilding, as an industry, has one of the
type. Human simulation provides that assur- longest product life cycles, with submarines in
particular taking as long as 14 years to go from
ance by graphically demonstrating worker
design to launch. Traditional prototyping meth-
interactions with products and processes, and
ods may include several iterations of producing
delivering analysis on the worker’s productiv-
wooden mockups of submarine sections. These
ity and physical capabilities. physical prototypes are expensive, time-consum-
Lockheed has had great success with the ing efforts. Eliminating or reducing the number
application of virtual manufacturing in the of prototypes will save development time and
aerospace and defense industries. Human sim- millions of dollars.
ulation activities are ongoing in many of its General Dynamics’ Electric Boat Corporation
companies. For example, an analysis of manual uses a 3D virtual environment to build digital
labor tasks on an F-16 fighter’s vertical fin prototypes of nuclear attack submarines.
involves a simulation of maneuvers required of Working with CAD packages, Electric Boat is
astronauts aboard a space shuttle. The objec- able to simulate the kinematics, dynamic,
tives are wide ranging, from pure visualization mechanical, and ergonomics aspects of the sub-
of human activity to more advanced productiv- marine and its components. The simulation cre-
ity and ergonomics analysis. ates an environment in which to evaluate a wide
range of parameters, including ergonomics, and However, in spite of the benefits of computer-
then optimizes the preliminary design based on ized human modeling, it has been used to prove
the results. Virtually tested activities, such as few industrial workstation designs.
torpedo loading and engine maintenance, are The use of 3D high-level simulation contin-
prominent examples of where human simulation ues to gain additional following. Human inter-
contributes to better design. face to a product or process is critical to the
Electric Boat Corporation was one of the first success or failure of that process. The ability to
to demonstrate the use of motion capture and predict human performance enables correct
immersion techniques to enable engineers and engineering decisions to be made earlier and at
their naval customers to “walk through” a digi- less cost.
tal submarine; this includes visual feedback. For industrial and manufacturing engineers
Motion-capture techniques, coupled with virtu- with an outlook toward proactive participation
al reality and human simulation, are expected in the product life cycle, virtual manufacturing
to provide solutions to real-world training and may be the perfect partner. Computer simula-
design problems at Electric Boat. tion can assist the engineer in seeing the envi-
ronments in which real humans will later be
SUMMARY placed. It can help engineers and designers
Successful organizations continue to build experience, in a virtual workstation or manu-
upon their digital enterprise, drawing upon facturing cell, the physical characteristics of
product, manufacturing, plant, and operations parts to be handled and the effort required to
data to simulate the entire product life cycle. execute production processes. Simulation will
The benefits received from the CAD revolution ensure that the manufacturing cell or worksta-
have enabled companies to shorten the time tion will be successful from the productivity,
required to bring new products to market. quality, ergonomic, and safety points of view.
After World War II, in Japan, production cords to stop the assembly line. Line-stop
volumes were minuscule compared with auto- problems are then permanently resolved.
motive output in the U.S. Those small pro- Trust in the company encourages employees
duction volumes did not allow Japanese to acquire whatever skills are necessary to
automakers the luxury of using specialized perform the work that needs to get done. By
equipment for each model. Nor did they allow mastering a wide range of skills, they equip
for stocking huge inventories of parts. themselves to help and fill in for each other as
Automakers in Japan thus needed to develop necessary. Employees also reinforce their job
flexible methods for adapting the same security by acquiring diverse skills. The versa-
machines to different vehicle models. And they tility of multi-skilled employees gives man-
needed to find ways to ensure reliable supplies agement the flexibility to redeploy people in
of needed parts and materials without main- response to changes in demand. However, to be
taining large inventories. able to deploy people, the design of the manu-
Another factor involved the deterioration of facturing system must be changed. The three
labor-management relations. Severe recession basic parts of the manufacturing system are
buffeted Japan’s rebuilding effort in the early the final assembly line, the subassembly lines,
postwar years. Adverse economic conditions had and the component-part manufacturers. Each
catalyzed an ugly cycle of strikes and lockouts. of these areas must be redesigned to change
Toyota experienced the same labor strife that the functional job shops and flow-line systems
was endemic in Japanese industry in the early into the Toyota lean manufacturing system.
1950s. Ultimately, labor and management at So, TPS has inherited Ford’s practice of
Toyota came to an understanding. Labor would breaking down work into simple steps and dis-
cooperate with drastic restructuring, including tributing those steps among employees along
job cuts, which were necessary for the company’s the line. This is called the “division of labor”.
survival. Management would guarantee lifetime But employees in the Toyota system are in
employment for about 30% of the employees. charge of their own jobs. Through their teams,
Together, they would refrain from increasing they manage their own work sites. They identify
employment casually, even in good times. opportunities for making improvements and
Instead, they would hire temporary workers and take the initiative in implementing those
try to find ways to accommodate upturns in the improvements under the umbrella of coop-
demand using the minimum number of eration with management. And they share fully
employees. in the fruits of their own labor via a profit-
Mutual trust between labor and man- sharing system.
agement at Toyota developed over the years. Yet another element of Ford’s system that
Management rewarded employees for produc- remains critical in the Toyota production
tivity gains with improved compensation and system is the concept of one-piece flow. On
working conditions. Employees took the ini- Ford’s final-assembly-line conveyor, vehicles
tiative in activities for raising productivity were manufactured one at a time. TPS extends
and otherwise enhancing the company’s com- that methodology to the rest of the manufac-
petitiveness. turing system, so it is like a vast machine that
Trust has enabled management to delegate starts with customer orders and culminates in
authority to the workplace. The people who run deliveries of finished products (Figure 15-2).
the TPS are the ones on the line, those who can
see problems when and where they occur and TPS HISTORY
who can act immediately to resolve them. Four men figured prominently in creating
Workers are empowered and provided with pull TPS and bringing the revolution to the U.S. The
first was Sakichi Toyoda, the inventor of the small production volumes of the Japanese
automatic looms, who later founded the Toyota market.
group. In 1902, he invented a loom that would Kiichiro’s solution was to provide the dif-
stop automatically if any of the threads ferent processes in the assembly sequence with
snapped. His invention opened the way for only the kinds and quantities of items needed
automated loom works where a single operator and only when they were needed. Production
could handle dozens of looms. and transport activities took place simulta-
neously and synchronously throughout the pro-
ROOTS OF AUTONOMATION duction sequence, inside and between all the
Sakichi Toyoda’s invention reduced defects processes. Kiichiro thus laid the groundwork
and raised yields, since a loom would not go on for Just-in-time (JIT) production and gets
producing imperfect fabric and using up credit for coining the term JIT.
materials and production time after a problem The man who did the most to structure TPS
occurred. The principle of designing equipment as an integrated framework was Taiichi Ohno.
to stop automatically and calling attention to In the late 1940s, Ohno was in charge of a
problems immediately is crucial to TPS. It is machining shop. He worked his way up to
evident on every production line at Toyota and become an executive vice president at Toyota.
at other companies that use the system. He experimented with various methods of
When Toyota set up an automobile manufac- setting up the equipment to produce needed
turing operation in the 1930s, Toyoda’s son, items in a timely manner. But he got a whole
Kiichiro, lead the new venture. Kiichiro new perspective on JIT production when he
traveled to the United States to study Ford’s visited the U.S. in 1956.
system of operation. He returned with a strong Ohno went to the U.S. to visit automobile
grasp of Ford’s flow-line system and an even plants, but like many Japanese who came to
stronger determination to adapt that system to visit, he made two side trips. One was to an
Subassembly
WIP
Empty cart
Withdrawal WLK
Kanban Full cart Raw
Mfg Cells Material
Figure 15-2. The lean production system provides for arranging all the processes from the raw materials to finished products in
a single, smooth flow. Employees and managers at companies that employ the system learn to identify and eliminate any waste
that occurs in the flow.
American supermarket. Japan did not have (SMED) system. The key to SMED was simplifi-
many self-service stores at that point and Ohno cation of the setup process so workers could
was impressed. He marveled at the way cus- perform setups quickly and routinely as part of
tomers chose exactly what they wanted and in their daily routines. This was critical in the
the quantities that they wanted. Ohno admired press forming area of the plant where sheet-
the way supermarkets supplied merchandise in metal body parts were produced. Also, it was
a simple, productive, and timely manner. critical in manufacturing cells where families of
In later years, Ohno often described his pro- components were being made on the same set of
duction system in terms of the American super- processes. The machines were arranged in U-
market. Each Toyota production line arrayed shapes according to the sequence of operations
its diverse output for the next line to choose needed to make the family of components. The
from, similar to merchandise on supermarket processes were extensively modified so that a
shelves. Each line became the internal changeover from one component in the family to
customer for the preceding line; each line the next was simple and quick.
became a supermarket for the following line.
The downstream line would come and choose
GETTING READY FOR LEAN
the items it needed and only those items were
chosen. The feeder line would produce only the Companies around the world have demon-
replacement items for the ones that the fol- strated the value of lean production in making
lowing line had selected. This format, then, was automobiles and many other products. Some
a pull system, driven by the needs of the fol- have even used the concepts of the lean pro-
lowing lines. It was in direct contrast with con- duction system successfully in service sectors.
ventional push systems, which were driven by As applied to lean production, the basic
the output of the feeder lines. concepts of the original TPS are unchanging.
Ohno developed a number of tools for con- But all companies implement those lean
trolling his production system in an integrated concepts differently. This has been true for all
way. The best known of those tools is the manufacturing-system implementations, but
kanban system, which provides for conveying the TPS is particularly adaptable for many dif-
information in and between processes by the ferent products in many different countries.
utilization of instruction cards. Some common threads are apparent in the
Ohno’s other American side trip was to ride experiences of many companies that have
the cable cars in San Francisco. Here, he implemented the lean system successfully. To
observed riders pulling an overhead cord when benefit from the TPS, companies must satisfy
they wanted to get off the trolley car. Pulling three basic conditions:
the cord sounded a bell and the cable car
operator stopped the car and the customer • Top management must make a strong and
would disembark. Ohno later implemented a visible commitment to the system, par-
system of pull cords on the final assembly line ticipate directly in implementing the system,
for the operators to stop the line whenever they and instruct middle-level managers to do
found a problem. likewise.
Another contributor to the development of the • All employees must participate in the
TPS was a consultant, Shigeo Shingo. He system (100% participation). The workers
developed a methodology for setup-time must be committed, not just involved. This
reduction and wrote one of the earliest books requires enlightened management lead-
detailing the TPS. In the 1980s, Dr. Shingo ership. In the manufacturing of a ham and
visited many companies in the U.S. demon- egg breakfast, the hen is involved but the
strating his single-minute-exchange-of-dies pig is committed.
• Companies must put in place a solid The second prerequisite for adopting the
framework for cultivating capable leaders lean production system is participation by all
and providing internal customers with employees. This requires a strong commitment
necessary practical skills to do the job. from those in top management, in addition to
The commitment by top management is those in the rest of the organizational
especially crucial. It is a beacon for everyone in framework. Full participation is essential
the company, a clear and compelling mandate because the lean production system works by
for change. establishing a smooth, continuous flow of
Companies often introduce new schemes for material through the entire production
raising productivity and quality. But they sequence. For example, streamlining work and
rarely stick with demanding regimens unless eliminating inventories at one work site means
required to do so. Even inefficient companies little if work just piles up at the next stage of
can get by when business conditions are good. the production sequence. Lean production gen-
And all but the most nonproductive companies erates the greatest benefits for companies
can muddle through the modest economic when it is implemented throughout the entire
downturns. manufacturing system, including all of the
It usually takes a company-threatening companies’ suppliers. Companies strengthen
crisis, a severe market slump, for example, or a
each other by developing a smooth flow of
technological breakthrough by a competitor, to
material that extends all the way from sup-
put mortal fear into management and their
employees. It usually takes a crisis to put the pliers of raw materials to assemblers of
fear of going out of business into everyone. finished products in the final assembly line.
Only in crisis do people fully awaken to the
need for fundamental change. Vendors
Once top executives see the gallows and rec- Suppliers who participate in the lean pro-
ognize that corporate survival depends on fun- duction system enjoy the same benefits that
damental change, they will lead that change. the main company does from the system.
They must venture onto the factory floor. They Lean production can reduce inventories at
must let employees know what kind of change part suppliers just as readily and effectively
they envision, how they will be involved in as it does at main assembly plants. Product
change, and why the change is important to the quality improves as well. This is because the
long-term survival and success of the company. lean production system includes measures for
And they must put middle management on exposing defects whenever and wherever they
notice that their tenure will hinge on active occur.
cooperation in promoting these changes.
Suppliers who become lean producers also
Middle management has the most to gain and
report improvements in their employee-man-
the most to lose.
Companies need to develop programs for the agement relations. This is mainly because the
training of everyone, from managers to system provides for expanding their employee
employees, to operate and maintain a new lean roles in designing their own workplace and
production system. The natural teachers are managing their work areas. It brings employees
the people who have participated in lean man- and management together in the joint pursuit
ufacturing with other companies. They should of improving the quality of the product and
impart their skills to the other managers and working conditions, thus reducing the unit cost
employees through classroom instruction and of the product and throughput time, and
hands-on sessions. increasing the overall flexibility of the factory.
The Internal Customer cycle for any task that is not absolutely nec-
Responsibility and authority are motiva- essary to generate value.
tional. Nothing is more demoralizing over the The lean production system enforces a creative
long term than spending time in an unpro- attitude in the workplace. Employees do not
ductive manner. Experience has proven that coast. Lean production demands continuous
the more authority employees have to manage vigilance. Continuous improvements demand
their own work, the more inclined they are to unflagging efforts to find better ways of doing
pursue improvements in that work. Employees things. The lean manufacturing system structures
who can translate their own ideas into visible a workplace environment that nurtures employee
improvements in production flow and product initiative. The overall result is a stimulating
quality take pride in their work, jobs, and com- workplace where employees can take charge of
panies. their own destinies.
In lean manufacturing, teams of employees
undertake tremendous responsibility for JIT PRODUCTION
designing standardized work procedures for Customers want the best possible products
their own jobs and strive continuously to find at the lowest possible prices and they want
ways to improve those procedures. Team them as soon as possible. The Toyota pro-
members use kanban to manage workflow and duction system provides for fulfilling cus-
obtain parts and materials. They each work to tomers’ demands productively and promptly by
master every job in their cell or on their line so linking all production activity to marketplace
that any member of the team can help or fill in sales.
for any other member of the team. This means In the lean manufacturing system, all the
that the factory must be redesigned so that the processes in the manufacturing sequence are
processes and operations can be operated by arranged to produce in a single, smooth line of
just enough people to meet demand. For flow. Basically, each process produces items for
example, take a look at Wendy’s, a fast-food the next downstream process to withdraw and
restaurant that can be considered a manufac- use as needed. Each process withdraws items
turing cell for hamburgers. At lunchtime, from some preceding process and produces
Wendy’s has numerous workers in the cell to items to replace ones that the following process
meet the daily demand. By mid-afternoon, one has withdrawn. Each process only makes addi-
person can handle the demand. Wendy’s burger tional items to replace items that the following
cell is designed for flexibility in handling process has withdrawn. All the processes are
changes in customer demand and product connected by kanban links. A kanban link is a
design, or making the burger to the customer’s set of containers, cartons, baskets, boxes, etc.,
specifications. which hold a specific number of parts. The
For people who have worked in regimented system’s work-in-process (WIP) inventory is
job/flow shop manufacturing systems, the lean held in the kanban links.
production system and the broad-ranging JIT manufacturing, another term for lean
responsibilities it assigns to employees can production’s cellular manufacturing, means
come as a shock. Many conventional systems making only what is needed, when it is needed,
are rife with rigid job designations. The lean and in the amount that is needed. It eliminates
production system cultivates employee flexi- many types of waste and the need to maintain
bility by having employees acquire multiple large inventories, which reduces storage costs.
skills. Conventional manufacturing systems JIT eliminates waste that occurs when changes
buffer schedules for many tasks with so-called in specifications or shifts render stocks of old
“reserve” time. Lean leaves no time in the work items worthless. It also eliminates waste that
occurs when defects go undetected in the man- plants and suppliers to provide a rough basis
ufacture of large batches. for production planning. Dealers give Toyota
Cellular manufacturing, though simple in information every 10 days about actual orders
principle, requires dedication and hard work to they have received. Based on that infor-
implement properly. Once managers and mation, plants and suppliers draft detailed
employees have mastered the basic concepts, 10-day production plans. Dealers also provide
they can learn how to devise various tools and Toyota with daily information about customer
techniques for putting those concepts into orders and requests. Production plans can be
practice. Notably, they learn to: adjusted up to three days before production
• distribute production of different kinds of begins to accommodate requests for changes
items evenly throughout the day and in some specification, such as body color.
week, allocating work evenly and thereby Based on the daily production plan, an exact
using resources optimally; sequence of items to be produced at the plant
• link each process using kanban to the pre- is determined. Different specifications are dis-
ceding and following process (pull system); tributed evenly over the day.
• make items one at a time whenever At a Toyota vehicle plant, a variety of body
possible, emulating one-piece flow; types can be seen moving along the same
• establish a time frame for linking the pace assembly line at the same time. The production
of work in every cell to the pace of final of different body types is staggered evenly over
assembly (system takt time); and the course of the day to make efficient use of
• where batch processing is necessary, such the basic elements of production processes,
as stamping sheet-metal components, including machine tools, tooling, equipment,
reduce batch size to as small as possible by and material-handling devices.
reducing setup time between batches. Suppose Toyota spends all morning pro-
ducing one item, all afternoon producing
In summary, lean production requires a level another, and all evening on a third item. This
demand for subassembly and component parts. might seem productive, since it would enable
This level demand is accomplished by making Toyota to run larger batches without changing
the same mix of products in small quantities paints, tools, and dies. However, it would
every day. The output from feeder processes is distance the pattern of production from the
balanced then to match the daily demand. This pattern of sales in the marketplace. Even
process seems easy, but is often difficult to worse, it would impose a disproportionate
accomplish. burden on one team at times in the upstream
processes. Some teams might be idle, while
LEVEL PRODUCTION others would be busy, and this would be an
Lean production can help companies unproductive utilization of resources.
achieve spectacular gains in productivity and Production should be scheduled evenly in the
quality. It is impossible to attain, however, final assembly processes. Leveling and
unless companies distribute work evenly by scheduling final assembly to evenly distribute
leveling production. various models of the cars and their options
At Toyota, leveling can be looked at (based throughout the day is also called smoothing of
on customer sales) by examining how the production in Toyota literature. Leveling
planning department gets information each enables suppliers to distribute components
month about the number and kinds of vehicles evenly in their manufacturing processes. It
it expects to sell in the following month. This enables workers to function with a minimum of
information is then passed on to vehicle manpower and equipment.
An operator removes the kanban The operator deposits the kanban Team leaders gather the contents ...and takes them to a sorting
from a new box of items when he that he or she should be removed of the kanban mailboxes at pre- room. There, an automatic sorter
or she uses the first item from the from boxes of parts in a kanban scribed times—several times a places the kanbans in separate
box. mailbox nearby. day—and place them in collection boxes for the different suppliers.
boxes. The kanban postman picks
the kanbans up from the collection
boxes...
The drivers that bring parts from Back at the suppliers’ plants, the A withdrawal kanban goes onto Then, the supplier delivers the new
the suppliers stop in at the sorting drivers deposit the kanban in a new box of parts in place of box of parts to the plant indicated
room after unloading their trucks collection boxes for subsequent the production instruction kanban. on its kanban.
and pick up kanban to take back sorting. The latter goes back into the plant
to their plants. as a production order for a like
quantity of the same part.
➡
Here is a withdrawal kanban that Toyota used with an outside supplier.
It indicates the name of the supplier, the receiving area at the plant, the
location in the plant that will use the item, the part number, the part
name, and the quantity. This kanban also has a bar-code label to permit
automatic invoicing.
Figure 15-3. Kanban Flow: Withdrawal kanban (courtesy Toyota Motor Company).
productivity is maximal. And the employees redesigning the factory floor so that work
are completely in charge. proceeds directly from one cell to the next
without any detours to storage. It means
LEAN MANUFACTURING devising logistics so that work moves smoothly
In the lean production system, work is and on schedule from raw materials plants
arranged in a single, smooth flow. That means through machining cells, to subassembly cells, to
arranging work inside each cell to flow smoothly assembly lines, and on to distributors, dealers,
from one step or operation to the next. It means and customers, like water flow through a pipe.
Boxes of finished parts at this Toyota supplier go into a rack to await pickup and delivery. Note the arrows. They indi-
cate which boxes to pick up first when more than one box of the same kind of part are awaiting delivery. This first-in,
first-out system avoids the collection of dust and other problems that can occur when items sit idle. Shipment orders, of
course, are in the form of withdrawal kanban. Those kanbans go onto the boxes in place of the production instruction
kanbans. The production instruction kanbans go back into the manufacturing line as production orders.
Kanban do not need to be printed cards. They can be triangular metal plates and color-coded washers, like the ones above.
They even can be colored balls, like those used in some processes at Toyota.
Figure 15-4. Kanban Flow: Production instruction kanban (courtesy Toyota Motor Company).
Ideally, items should proceed one at a time on whole batches in the following processes.
through the entire production sequence. This is One-piece flow allows for completing 100 parts
the fastest way to translate raw material into a in 102 minutes, just enough time to produce all
finished product because it minimizes the 100 parts, one at a time, in turn, through the
amount of material in process at all stages of three processes.
production. One-piece flow is impractical, In Figure 15-5, Operators A, B, and C are
however, for work where dies, molds, or other performing successive steps in an assembly
tools are changed to produce different products, sequence. Work begins for each operator when
as in forging, castings, stampings, as well as receiving a box of 100 semi-finished units from
moldings. Other batch-type processes, such as
a preceding process. Workers perform addi-
plating and chemical milling, must be con-
tional work on each of the assemblies and then
tinued until technology catches up and single-
piece production is feasible. A good first step is take the 100 assemblies to the next operator.
to use the smallest possible batches. Since This approach entails inefficiencies of the fol-
changing dies or molds is time-consuming, lowing types:
manufacturers traditionally will favor large • Each operator has 100 items at a time,
batches in batch-processing machines and also which is resulting in extremely long lead
in upstream stages of production sequences. times.
Inventories can be reduced and flexibility • It is difficult to balance the distribution of
increased by using smaller batches or lots. work.
Suppose 1,000 units per day each of Parts A
• The large amount of items per operator
and B are needed. Can these be produced in
means extra handling on the workbench.
four batches of 500 units, rather than two
batches of 1,000 units? To do this without com- • If Operator B discovers an improperly
promising productivity, ways must be found to assembled workpiece, the operator cannot
shorten the changeover or setup time for the determine exactly when or how this
dies or molds. That is, the increased number of problem occurred. Operator B knows only
production runs must not greatly increase the that it has happened at some point during
total time the machines are idle for setup. the processing of 100 items.
Historically, advances in decreasing change- • When production shifts to a different kind
over times have been an important dynamic of of assembly, operators must remove all
TPS. Those advances are easy to achieve once parts for the previous assembly from the
internal customers in the workplace recognize shelves to avoid mixing them with parts
the importance of decreasing changeover time for a new assembly.
to reduce batch sizes. The trick is to analyze
changeover work systematically using the Assuming the time for Jobs A, B, and C are
single-minute-exchange-of-dies system. equal, a company can avoid problems in the
preceding example by placing workers next to
Example each other in an assembly cell and having each
Here is a look at the difference in throughput operator handle parts one at a time. Workers
time between one-piece-flow processing and perform their work on an assembly and hand it
conventional batch processing for 100 pieces. over to the next operator before reaching for
This example is of three consecutive processes, the next workpiece. This is one-piece flow in
each of which requires one second to complete assembly operations. This change, which
one item. Batch processing entails a production requires no increase in the number of operators
lead time of 300 minutes because parts sit idle for the same span of work, yields the following
at upstream processes while work is under way improvements:
operators does not accommodate changes in second operator would handle the drill and
takt times. Instead, when the takt time for a tapping machine. Turning all four processes
production line becomes shorter and more back over to one operator would accommodate
demanding, the flow of work is streamlined and a subsequent decline in demand.
additional workers are added as necessary.
When takt time becomes longer, fewer workers INTEGRATED QUALITY
are assigned to a line. Assigning more people to The principle of stopping work immediately
a line means that each operator handles a when problems occur and preventing the pro-
narrower range of work. Assigning fewer duction of defective items is basic to the Toyota
people means that each operator handles a production system. In Japanese, this principle
broader range of work. Flexibility in allocating
is called jidoka. We call it autonomation.
work is possible because the cell is designed for
When under the Toyota production system,
standing, walking workers. People master a
equipment is designed to detect abnormal con-
much broader range of skills, and multiprocess
ditions and to stop automatically whenever
handling (instead of multi-machine) is
they occur. Operators in the assembly cell can
employed.
stop the production flow whenever they note
Traditionally, manufacturers assigned their
anything suspicious. Mechanical and human
operators to several machines of the same type.
autonomation prevent defective items from
That is, a lathe operator might process five
progressing into subsequent stages of pro-
items on lathes and then pass them on to a
duction, prevent the waste that would result
milling machine operator, who processes them
from producing a series of defective items, and
on milling machines. The items then might
prevent overproduction (see Figure 15-6).
move on to drilling and tapping in the same
manner. Even if the lathe operator runs Another advantage of autonomation is that
multiple lathes, multi-machine handling it exposes the causes of problems by stopping
entails long lead times and excessive handling. the equipment exactly as it is when a problem
The item that the lathe operator processes on first occurred. Autonomation also calls
the first lathe, for example, sits idle until the attention to the problem immediately with a
operator has finished processing the other signal light or some other kind of indicator. The
items on the other lathes. Overproduction most fundamental effect of autonomation is the
occurs frequently, and feedback on quality way it changes the nature of cell management:
problems is difficult to obtain. it eliminates the need for an operator to watch
In manufacturing cells, an operator handles over each of the machines continuously, since
different kinds of machines to keep work machines stop automatically when abnor-
moving in a continuous flow. The operator malities occur. It therefore opens the door to
might use a lathe, a milling machine, a drill major gains in productivity.
press, and then a tapping machine on each Autonomation is thus partly a humanistic
item in a sequence. approach to configuring the human-machine
Multi-skilled operators and multiprocess interface. It liberates the operators from the
handling thus enable the accommodation of tyranny of the machine and leaves them free to
changes in takt time while maintaining short concentrate on tasks that enable them to
lead times with one-at-a-time production. In exercise skill and judgment.
the example, an increase in demand could be
accommodated by assigning two operators, Line Stop
instead of one, to the four processes. The first On the final assembly line, the fixed position is
operator might narrow his or her scope of work clearly marked for work in each station. Oper-
to the lathe and milling machine, and the ators in the process assume responsibility for
Chuck
Buzzer
Work
Stop button
guaranteeing the quality of all work up to that Operators would be hesitant to pull the line-
line. They make sure no defective items progress stop cord if pulling it stopped the line imme-
beyond that line and into the next process. When diately; they might be subconsciously inclined
an operator signals a problem by pulling on the to let minor imperfections pass rather than
line-stop cord, the line keeps moving until it take responsibility for stopping the line.
reaches the fixed position. That gives the team Employees are more inclined to call attention
leader and operator a chance to resolve the to possible problems when they know that sum-
problem before the line stops (see Figure 15-7). moning help will not stop the line at that
A B
C D
Operators in the Toyota production system can stop the production line whenever they spot anything suspicious. Along the conveyor on this
assembly line, an operator (A) has noted a part that does not fit correctly. He pulls on the line-stop cord (B). That lights the andon lamp and
summons the operator’s team leader to have a look (C). The line will continue moving until it reaches the next “fixed position”— the position
where each process on the line has completed one work cycle.
When the team leader arrives, the operator explains the problem. The team leader discovers a fitting that has slipped out of place and is able
to resolve the problem (D) before the line reaches the fixed position. If the leader required more time to resolve the problem, the line would stop
at the fixed position.
Figure 15-7. The fixed position is clearly marked for work in each station with a yellow line. Stopping work movement at the line
assures all tasks on the line will be completed when stoppage occurs (courtesy Toyota Motor Company).
instant. This methodology results in more occurs. Thus the errors and quality problems
rigorous quality control. that happen all too easily when interruptions
Another reason for keeping the line moving occur in the midst of work are avoided.
until it reaches the fixed position is that
stopping the line in the middle of a work cycle INTERNAL CUSTOMER SATISFACTION
would be disruptive to all the other stations. External customer satisfaction is a reflection
Stopping the line at the fixed position ensures of employee (the internal customer) satis-
that all processes along the line will have com- faction. In that sense, the Toyota production
pleted their work cycles when the stoppage system has been successful in earning
customer satisfaction because it provides 1. takt time, which is set by the pace of mar-
employees with fulfilling work. ketplace sales;
Measures for enhancing employees’ sense of 2. working sequence for the manufacturing
job fulfillment are an important emphasis in cell, which is the series of steps that are
the new ways Toyota is implementing its pro- determined to be the best way to perform
duction system. For example, Toyota knows a task; and
that work is more fulfilling when employees 3. standard stock-on-hand, which is the
can take part in completing an entire product minimum number of workpieces needed to
or assembly. Employees take pride in their be on hand for a process to maintain a
work when they can see their efforts take smooth flow of work.
shape in a functional unit.
Standardized work results in detailed, step-
Assigning a single small team to assemble a
by-step guidelines for every job in the lean pro-
whole vehicle would be grossly inefficient. But duction system. Team leaders determine the
work can be arranged so teams can handle all most efficient work sequence. They make con-
the steps of assembling complete systems. For tinuing improvements in the work sequence with
instance, Toyota traditionally has installed their team members. Continuous improvement
electrical wiring a small portion at a time at thus begets new patterns, or new cell designs of
different stages in the vehicle assembly standardized work.
sequence. This also has been the case in the The operator in the series of photos in Figure
installation of exhaust systems and other 15-8 is abiding by a carefully designed working
important systems. Recently, Toyota began con- sequence. The sequence appears in a stan-
centrating the installation steps for such dardized work chart posted at the workplace.
systems at single work sites. This heightens The chart shows the sequence of 12 operations
the sense of teamwork among the 20-or-so for the right half of the manufacturing cell. It
workers at each work site. It reinforces their shows the location of the quality checkpoints,
sense of accomplishment by enabling them to stock-on-hand, and cycle time compared to the
complete a crucial and integrated part of each takt time.
vehicle. Because standardized work involves fol-
The ultimate responsibility for raising lowing procedures consistently, any inherent
quality and productivity lies with the operators problems in the working sequence surface
in the workplace. Success in implementing the repeatedly and, therefore, conspicuously. This
TPS hinges on earning their trust, securing allows cell team leaders and members of the
their active participation, and providing them team to identify and fix problems easily.
with sufficient skills and understanding to Similarly, monthly changes in production
fulfill their role fully. volumes require changes in the standardized
work. Standardized work allows team leaders
STANDARDIZED WORK and members to devise new standardized work
AND CELL DESIGN procedures to accommodate monthly changes
Standardizing work is a way to maintain in production volumes.
productivity, quality, and safety at high levels.
It provides a consistent framework for per- Kaizen
forming work at the designated takt times Kaizen furnishes the dynamism for continuous
and for illuminating opportunities for making improvement and encourages individuals to take
improvements in work procedures. Three part in designing and managing their own jobs.
elements are used for structuring stan- Kaizen improvements in standardized work help
dardized work: maximize productivity at every work site.
Figure 15-8. Standardized work chart for a lean manufacturing cell (courtesy Toyota Motor Company).
Case Study
Table 15-1
In 1993, Grand Haven began working with Luggage arm assembly
the Toyota Supplier Support Center. Its core Results of implementing the Toyota
competencies are centered on stamping, production system (first two years)
welding, and assembly. This company makes Productivity
floor-mounted transmission-control systems, Output/person 0.84%
hood- and deck-lid hinges, and clutch, brake, Overtime hours/week –100%
and accelerator pedals. Personnel needs on line – 45%
Like most companies that adopt TPS, Inventory (on production line)
Grand Haven began in one segment of its Work-in-process – 93%
operations. It set up a continuous flow line in Raw material – 92%
a cell for assembling luggage arms. The Finished goods – 68%
improvements in productivity and quality Floor space – 60%
were immediate and dramatic, and have con- Quality
tinued (Table 15-1). Customer complaints/month –100%
industrial and systems engineering manufac- Center. Before NASA, Hunter taught manufac-
turing with a minor in ergonomics. turing engineering courses at the University of
Hunter has been interested in manufacturing Memphis. Earlier, he worked as a civilian at the
systems design since 1980 and has designed and Air Force and Naval Aviation depots serving as
implemented many lean production systems. an engineering project manager. His work in
His recent research has included the ergonomic lean production includes the design and imple-
ramifications of manufacturing system design mentation of many commercial manufacturing
using computer simulation. This research cells, as well as the design of an 18 cell system
attracted the attention of NASA where he was for manufacturing and assembly for the Navy.
honored as a NASA Fellow. He carried out While working for the Department of Defense,
primary human engineering research on the Hunter won numerous outstanding service
ergonomics effects of micro-gravity on awards for his contributions to national defense.
astronauts for NASA’s Marshall Space Flight He is married and his wife is a nurse.
There are many people who made significant The authors would like to thank the engineers
contributions to this book. Co-author, Dr. Steve and managers at Delmia Corporation for their
Hunter is one of the world’s experts in cellular continued support of manufacturing system
manufacturing system design and implemen- design and ergonomics research. They would like
tation. He is now at Mississippi State to especially thank Delmia’s president, Bob
University continuing his research into manu- Brown, and David Papp, Southeastern sales
facturing system’s design while assisting the manager, for their continued support through the
Mississippi furniture industry with the design use of Delmia’s remarkable ENVISION ERGO®
and implementation of lean manufacturing high-level 3D/virtual-reality software.
systems. The axiomatic design material was extracted
Author J T. Black thanks his wife, Carol, who from the unpublished work of Nam Suh, a pro-
did all of the typing and helped with editing fessor at MIT whose work on axiomatic design
prior to submission. She also made many sug- has long been admired. This text also reflects
gestions that greatly improved the readability the work of David Cochran, a former student of
of the final manuscript. Thanks also to Dr. Dan Black, who is now on the faculty at MIT and
Sipper, Rich Wilson, Angeline Honnell, and Dr. leading the axiomatic systems design work
Brian Paul who provided many suggestions for there.
improvement. Two people inspired this book. Tom Gelb is
The authors give thanks to students from all an engineer at Harley-Davidson, who described
courses at Auburn University and earlier at the with great passion the conversion of Harley-
University of Alabama in Huntsville. They con- Davidson to lean manufacturing. The second
tributed greatly to the book with their great influence on the book was Dick
questions and discussions, both during and Schonberger. Drs. Schonberger and Black pre-
after class. Some of the students, Dave sented many Just-in-time seminars in the
Hanning, Yen Shi “Hopper” Tsai, Doris Lizotte, early 1980s and they are considered pioneers in
Jill Williams, Alice Carter, Jack R. “Rick” Wade, this field.
Carlos Engle, Dr. Lewis Payton, and Kavit The authors would also like to thank the
Antani, provided written contributions. Jason SME editorial staff Cheryl Zupan, Bob King,
Wang and Allen Chen did the figures for the Chris McGorey, Kathye Quirk, Jon Newberg,
first version. Dr. Steve Hunter and Ranjit Frances Kania, and Rosemary Csizmadia for
David developed many of the new figures for their work on the book.
this book, along with extensive commentary
and editorial.
accidents 248
acclimatization 270
accuracy 285
Amber and Amber 1
American Armory System 3
andon 212
assembly cells 83 94 257
automatic production control 282
automaticity 273
automation 4 184 273
autonomation 31 41 273 310
calibration 286
case studies 249 298 325
cells 25 83 98 112
207 210 258 262
281 299
cellular manufacturing system 250
cleanliness 197
CNC machine tools with decouplers 280
coding/classification methods 111
decouplers 93 279
Delmia 295
design
axioms 79
for customers 108
for flexibility 62
process hierarchy 76
dual-card kanban system 219 234
ergonomics (Cont.)
rapid-upper-limb assessment (RULA) 267
scoring form 267
exchange of dies 30
kanban (Cont.)
integrated manufacturing
production system 217
limitations 225 230
material 227 229
production ordering 221
push production control system 216
rules 221
signal 227
single-card system 218
special types 226 228
stock on hand 217
systems 218
withdrawal 221
work-in-process inventory 217 233
kilocalorie 269 296 301
linked-cell systems 2 8 59 84
106 213
long-range forecasting 205
long-term contracts 244
production (Cont.)
systems 16 18 48 58
profitability 249
pull system 310 315
push production control system 216
quality 30 101
safety 262
scoring form 267
second industrial revolution 4
second shift 195
self-inspection devices 248
sensing 287
sequence schedule 199 203 205
setup reduction 28 117 128
simulation 289
3-D 289
evaluation areas 291
history 290
single
card kanban system 218
minute exchange of dies (SMED) 29 117 125 143
311
sourcing 244
single (Cont.)
units 248
smoothing
final assembly 199
production 314
software 194 198 295
special kanban 226
stability 285
standard operations 107 212 256
standard packaging 245
standard work 88 159 323
station design 197 271
steady state 269
steps to lean production 25
1 25
2 28
3 30
4 33
5 34
6 35
7 36
8 40
9 41
10 42
stock-on-hand 94 188 217
studies 192
subassembly flow lines 10
suggestion system 192
suppliers 40
supply-chain management 236
support routines 198
synchronizing 199 211
system design tools 109
throughput time 20
total preventive maintenance 206
total productive maintenance 185 197
Toyota production system (TPS) 205 307
UVW
U-shaped cells 25
vendors 243 312
venous pooling 265
virtual factory 304
walkaway switch 26 255
waste 201
withdrawal kanban (WLK) 221 315
work
flow 319
force 198 265
height 271
in-process 36 94 226 253
methods 271
sequence 188
simulation 272
station design 197 271
XYZ
yo-i-don 31 211
zero downtime 186