Learning to See

Value-Stream Mapping to Create

Value and Eliminate Muda

By Mil<e Rother and John Shook

Foreword by Jim Womack and Dan Jones

Lean Enterprise Institute

Cambridge, MA, USA
lean.org

Version 1.4
October 2009
Lean Enterprise lnstitut~-
~ •
© Copyright 1999 Lean Enterprise Institute, Inc. All rights reserved.
Lean Enterprise Institute and the leaper image are registered trademarks
of Lean E.merprise lnstituce, [nc.

Design by Off-Piste Design. Inc.

Oct◊ber2009
ISBN: 0-9667843-0-8

Lean €ntcrprisc [nsc.itutc, fnc.

One Cambridge Cemer
Cambridge, MA 02142 USA
(t) 617-871 -2900 • (f) 617-871 -2999 • lean.org
Whenever there is a product for a customer,
there is a value stream.
The challenge lies in seeing it.
-
Foreword
By Jim Womack and Dan Jones

When we latlnched Lean Thinking in the fall of 1996 we urged readers to "Just do it!" in
the spirit of 11:tiic hi Ohno and other pioneers of the Toyota system. With more than 300,000
copies in pri nt (including the Second Edition lau nched in the s pring of 2003) and with a
steady scream of e-mails, fm~cs, phone calls, letters, and personal reports from readers telIing
us of rheir achievements, we know that many of you are caking our and O hno's advice.

However, we are also aware that many readers have deviated from the step-by-step trans-
formation process we describe in Chapter 11 of Lean Thinking (and enhance in Chapter 15
of the Second Ed ition):
I. Find a change age nt (h<)" about yo u/)
2. Find a sensci (a re acher whose le arn ing curve you can borrow)
3. St:izc (or create) a crisis t<• motivate action acruss your firm

But then they have jumped to Step Five:

5. Pick somcrhing imp<1r1ant and !(et s tarte d removing waste •1u ickly, 10 su qwise
yourse lf w ith how muc h ynu c:in :1cco mpli.sh in a very s hon pcrioc.l,

Yet tbe 011erlooked Step Four is actually the most critical:

4. .t-,l ap the entire value Hrcam fr,r ,ill of your 1, 1·oducr fami lies

Un fortunatd}', we find tha t many readers have ignored ou r advice tO conduct th is critical
step before d iving imo the task of waste e limination. Instead in too many cases we fLnd
companies rush ing headlong into massive muda elimination activities- kaizen offensives or
continuous improvement blktes. T hese wel l-intentioned exercises fix one sma ll pare of the
val\1e stream for each produce and value does now more smooth ly through char course of
the s tream. But then the value now comes to a halt in the swamp of inventories and detours
a head of the next downstream step. T he net result is no cost savings reaching che bottom
line, no service and q ualicy improvements for the customer, no benefits for the supplier,
limited s ustainabili ty .as the wasteful norms of the w hole va lue stream c lose in around the
island of pure value, and frustration all arou nd.

"Typically the kaizen offensive wich its disappoin ting results becomes another abandoned
progiam, soon to be followed by a "bottleneck elimination" offensive (based on the T heory
of Constraints) or a Six Sigma initiative (aimed at rhe most visible q uality problems facing a
firm), or ... Bue rhese produce t he sa me result: Isolated viccories over muda, some of them
q uite dramatic, which fail to improve the whole.
Therefore, as the first "tool kit" project o f the L~10 Enterprise Institute, we felt an urgent
need co provide le an thinkers the most important tool they wil l need co make sustai nable
progress in the war against mucla: the value-stream map. In the pages a head M ike Rocher
and John Shook explain how to create a map for e ach of your value s trea ms and show how
this map can teach you, your managers, engineers, production associates, schedulers, suppliers,
and customers to see value, co differentiate val ue from waste, and to get rid of the waste.

Kai:ten efforts, or any lean manufacturing tec hnique, are most effective when ap1>lied
strategically within the context of building a lean value stream. T he value-stream map permits
you co identify every process in the flow, pull them out from the background clo ner of the
organi zation, and bui ld an entire value stream according to lea n principles. It is a cool you
s hould use eve!'y ri me you make cha nges withi n a value stream.

As in all of o ur tool k it pl'ojects, we have calle d on a ream with a wide variety of practical
a"nd research experience. i\1like" Rothe r studies Toyota, has worked with many manufacrurers
to introduce le an production flows, and reaches at che University of Michigan. John Shook
spent over 10 years with the Toyota Motor Corporation, much of it teaching s uppliers to
see, and is now a Senior Advisor co the L,can Enterprise Institute. ' l'hgethe r they possess a
formidable body of knowledge and experience-a pain full y constructed le arn ing curve-
which they are now s naring with you.

We hope re aders of Lean Thinking and participa nts in the activities of the Lean Enrerprise
lnstiwte will use the mapping tool immediately and widely. And we hope you will tell us
how to improve it! lkcause our own march toward perfection never ends, we need to hear
about your successes and, even more imporca nt, about the nature of your difficulties.

So again, "Just do it!" bur now at the level of the value s tream, product family by product
family-begi nn ing inside your company and then expanding beyond. T hen cell us about
your exµericnce so we can s hare your ac hievements with the emire lean network .

Jim \Nomack and Dan Jones

C:mnbridge, Massachusetts, USA
and Ross-on-Wye, Herefordshire, UK
J11n e 2003

Lean Enterprise lnstitut~

~-
Wirh gratirude to fam ily members, Jim Womack, Guy Parsons, Off-Piste Design,
and our friends at cl ient compa nies who help us fine-tune many ideas.
CONTENTS

Forevvord by Jim Womack and Dan Jones

Introduction

Part I: Getting Started

What is Value-Stream Mappi ng?
MareriaJ and Information Flows
Selecting a Product Family
The Value-Stream Manager
Using t he £\•lapping Tool

Part II: The Current-State Map

Drawing the Current-State Ma p
Your Tt1rn

Part Ill: What Makes a Value Stream Lean?

Overproduction
Characteristics of a Lean Va lue Stream

Part IV: The Future-State Map

Drawing the Future-Scare Map
Your'Thrn

Part V: Achieving the Future State

Breaking Implemenr.ation Inco Seeps
The Value-Stream Plan
Value-Stream Improvement is Management's Job

Conclusion

About the Authors

Appendix A : Mapping Icons (also inside back cover)

Appendix B: Current- State Map for TWI Industries
Appendix C : Future- State Map for TWI Industries
•
 •

Introduction

We have discovered an aiuazing chi ng. Whi le so ma ny o f us have been scratching our
heads trying to figure our why the road to lea n has been rockier than it s hould be, a viral
yec simple tool that can help us iuake real progress coward becomi ng lean has been right
u nder our noses.

One of us, Mike, had long searche d fot a means ro cie rogerJ1cr lean concepts and
techniques, which seemed more d is parace than they should be, as he worked on many
plant floor implemenration efforts. Mike noticed the mapping method while studying
Toyota's lean in, pJcmentation praccices. He realized mapping had poccmial far beyond its
usual usage, formalized the cool, and built a tra ining method around it that has proved
exrraord ina rily successfu I.

T he other of us, John, has known about the "tool'' for over 10 years, but never chought
of it as important in its own right. As Jolrn worked with Toyota, mapping was almost an
afterthought-a simple means of communication used by individuals wh() !cam their craft
through hands-on experience.

Ac Toyota, the method-called "Value-Stream Mapping" in this workbook-is know n as

"J\llacerial and Information F low Mappi ng." It isn't used as a craining method, <>r as a means
to "learn to see." It is used by Toyota Production System practitioners to depict current a nd
future, or "ideal/' ::states in rhc process of dc;vdoping imµk:mcmcarion plans to insrall lean
sys rems. At Toyoca, whi le the p luasc "valu e stream" is ra rely heard, infin ite anention is
given to esrablis hing flow, eliminating waste, a nd adding value. 'Toyota people learn about
rhree flows in manufacturing: rhe nows of material, information, and people/ process. The
Value-Scream Mapping method presented here covers che first two of these flows, and is
based o n the l\llaterial a nd Information Flow Maps used by Toyota.

Like iuany others in recent rears, we were s truggling to find ways to help manufacturers
think ol' now instead of discrete p roduction processes and co implement lean systems
inste3d of isolated proce$s-improvements. We struggled to hdp manufacturers make las ting,
systematic improvemcncs that wou ld not only remove wastes but also the sources of the
wastes so that they would never come back. Por chose who simply giv<Je the mapping cool a
try, we hi,vc been p leased to sec how exceptiona lly effective che tool has proved to be in
focusing arcenriori on flow and helping them co sec. Now we present it co you.

Mike Rother and John Shook

Ann Arbor, Michigan
May 1998
 ~◄ :srl"""'Y
orders _,~~•o«•o•
\ i c ON rieo'= ◄ :-, wee.I:.. I
IsFcm:.cc,,f 1--- -1r
t::,;,, J-,-; !,_;I,;,,.

1ococ1.o,,-,-,l,I,_
Z-Wott.k. 5ro I,>' pc • I:>''
Sd-.dslr 50:> T,rf>C "T '

-
To F.--c +
,,.......~
2
rrJ~h

.g1::. ~" l'o/?f ':L

Kr [
i=-,.~,...~
rASi!IC. Cvi SE:<..>1NG---
,os, -mi5"•PPONJ
& ~-""-'"_-_-ootcz•
2 W~I<..-, _-_
_,J_ ~I
1 &hL..: I I fx,_y
3 '21
IToh:J", ~,sos
'.:,C>e,y> ("/ ;.,iul'/;-,501,s t.'5 i)c,7'> Tol-cJ 1/T •3'lS.s 3De,y'>
G./e, : 'Z "",...... L/4 : f2{ C/o C?
:. 2.0"""''"'

F,-.....,., rAAME -~~

&
z_week'>
e;, ~ GOs ~ r·"''I
1 /I ~
S l:lc._y', A Current-State

Ar-1,;:....::: ~ i~--~· i
Value-Stream Map
~~~

~l:s ~-~-:--,
¼' ,s-:...,
~,k' 5 t,.,,,,'

- - - n " n n n n n n n n n n n n DD ll HAQ PH g g_g_g_HQ_..J

 I
~
;:, ,...,
,'1
-A. ~.,
:I:
Ill:
I
l / I
/ III
< I-. '.c I
l
() J
..J Ji ..l
'?:,4.!il
0
X
?
\) -! ·iVI
,.._
-,,s ~- .. -~
g;it= 0
" ....-t" V,,
I X
~
I
~s ~ od,:ft
v'\
.,I,.
1
·-
?
i.::::::=::::1 2,q,sr
Week
V\ .,...,,.--.,....,.
I
I
PART I: GETTING STARTED

• What is Value-Stream Mapping?

• Material and Information Flows

• Selecting a Product Family

• The Value-Stream Manager

• Using the Mapping Tool

~ T A R T HERE

)
PART I: GETTING STARTED

• What is Value-Stream Mapping?

• Material and Information Flows

• Selecting a Product Family

• The Value-Stream Manager

• Using the Mapping Tool

What is Value- Stream Mapping?
"Va lue stream" may be a new phrase in your vocabulary. A value scream is all the actions
(boch value-added and nonvalue-added) currencly l'eq uired to bring a product through the
main flows essential co eve,·y prvduct: ( I) the p roduccion flow from raw material into the
arms of the customer, and (2) the design flow frorn concept to launch. T his workbook looks
at the production flow from customer demand back through raw material, which is the flow
we usually re late co lean manufact1Jring and precisely the area where many have struggled
co implement lean methods.

Taking a value s tream perspective means worki ng on the big picture, not just ind ividual
processes, and improving the whole, not just optimizing the pares. If you truly look at the
whole and go all the way from molecules into the arms of the customer, you will need co
follow the value scream for a product across many firms and even more facilities. But
mapping th is e ntire stream is 100 much for getting scarred!

T his workbook covers the "door-to-door" production Oow inside a plant, including shipment
co the plant's cusromer and delivery of supplied parts a nd material, where you can design a
future-state vision :i nti srarr im ple.l'ne:.nring it right nway. This is a good level at whlch to
begin your mapping and lean implementation effort.

As your lean experience and confidence grow you can expand outward, from the plant lel'el
toward the complete molecules•t0•end-user map. Note, however, that in large companies whe,1
a product's value s tream passes through more than one of your own faci lities, expanding the
mappi ng effort 10 include the flow through you r other facilities should happen very ,wickly.

SUPPLIERS CUSTOMER TO END USER

o-t
PART I; GETTING STARTED 1
 Va lue -s tream mapping is a pencil a nd pape r iool that helps you t0 see a11d undersrand the
now or material and information as a product makes its way through the valL1e s tream.
What we mean by value-strea m mapping is simple: Follow a producr's production path
from customer to supplier, and carefully draw a visual representatio n of every process in
the mate rial and information flow. T hen ask a set of key questions and draw a "fucure-
srate" map of how value should flow.

Doi ng this over and ove,· is the si mplest way-and the best way we k now-to teach
yourself and your colleagues how t() sec value and, especia lly, the sources of waste.

WHY VALUE-STREAM MAPPING IS AN ESSENTIAL TOOL

• lt helps you visualize more than just the single-process level, i.e., assembly, welding,
etc., in production. You can see the flow.

• It helps you see more than waste. Mapping helps you see the sources of waste in
your value stream.

• It provides a common language for talking about manufacturing processes.

• It makes decisions about the flow apparent, so you can discuss them. Otherwise,
many details and decisions on your shop floor just happen by default.

• It ties together lean concepts and techniques, which helps you avoid "cherry picking."

• It forms the basis of an implementation plan. By helping you design how the whole
door-to•door flow should operate-a missing piece in so many lean efforts-value•
stream maps become a blueprint for lean implementation. Imagine trying to build a
house without a blueprint!

• It shows the linkage between the information flow and the material flow. No other
tool does this.

• It is much more useful than quantitative tools and layout diagrams that produce a tally
of nonvalue-added steps, lead time, distance traveled, the amount of inventory, and so
on . Value-stream mapping is a qualitative tool by which you describe in detail how your
facility should operate in order to create flow. Numbers are good for creating a sense of
urgency or as before/after measures. Value-stream mapping Is good for describing what
you are actually going to do to affect those numbers.

2
Practice drawing value-stream maps and you wi ll lei,rn to sec your shop floor in a way char
supportS lean manu factu ring. JusLremember chat che poinr of' getting lean is not "mapping,"
which is just a techn ique. What's important is imple menting a value-adding flow. To create
th is flow )' OU need a "vision" of chc flow. Mapping helps you see and focus on flow with a
vision of an ideal, or ac lease improved, srace.

You shouldn't run our and map all your va lue streams right away. 'lo benefit from value-
stream mapping you should make use of it 0 11 the shop floor, mapping a value stream you
wi ll actually be implementing. ff you arc planning changes in a value stream, be sure to
draw a future-state map firsc. If you are d.esigni ng a new production process, first map a
futu re srnte for the value stream. Considering a new scheduli ng system? Draw rhe future
state first. Changing production managers? Use value-screa m maps co help ensure an effective
hand-off and com inued implementation prog,rcss.

Material and Information Flovvs

Within the production flow, che movemem of mare.rial through
the factory is the flow that usually comes to mind. Bur there is
another flow-of informacion-rhac ce lls each process what co
make or do next. Matccial and informacion flow are cwo sides of
the sa me coin. You mus1 map both of chem. PRODUCTION FLOW

In lean manufacturing the information flow is ueared with material

just as much importance as che materi,11 now. Toyota and ics
suppliers may use the same basic material-conversion processes
as mass producers, like sramping/wclding/assembly, b111 Toyota
plants regulate their production qu ice differently from mass
producers. T he question to ask yourself is, "How ca n we llow
information so chat one process will make only what t he nexc
process needs when it needs ic?"

To create value-adding flow you need a "vision." Mapping helps y ou

see and focus on flow w ith a vision of an ideal or improved state.

PART I: GETTING STARTED 3

 Selecting a Product Family
One point to understand clearly before scarcing is the need ro focus 0 11 one product fam ily.
Your cuscomers care about rheir specific products, not all your products. So you will not be
mapping everything that goes through the shop floor. Unless you have a small, one-produce
planr, drawing all your product flows on one map is coo complicated. Value-stream mapping
means walking and drawi ng the processing steps (material and information) for one product
fam ily from door co door in your p lane.

Identify your produce families from the cuscome,· end of rhc value s u eam . t\ family is a
group of products thar pass through s imilar processing s teps and over common equipment
in your downstream processes. In general, you should not rry co d iscern product families by
looking ar ups trea m fabrication s teps, which may serve many product families in a batch
mode. Write down cb1rly what your selected p roduct fam ily is, how many diffcre m finished
part numbers the re are in the family, how much is wa nted by the customer, and how often.

Note:

If your product mix is complicated you can create a matrix with assembly steps
and equipmeni on one axis, and your products 011 the other axis (see below).

6 7

X
X
XX
X
E X
F X XX
G X XX

4
The Value-Stream Manager
You may have: ,1lrcady noticed chat tracing the value scream for a product family will cake
you across organizarional boundaries in your company. Because companies tend to be
organii.ed by departments and functions, instead of by che llow of value-creating steps
for product families, you often find that-surprise-no one is responsible for the value-
stream perspective. (Tc's no wonder we have focused tOO heavily on process-level kaizen!)
le is astoundingly rare to visic a faci lity and fi nd one person who k nows the entire material
and information llow for a product (all processes and how each is scheduled). Yee wirhour
this, pa rts of the flow will be left ro chance-meaning that ind ividual processing areas will
operate in a way that is optimum from their perspective, not the value-stream's perspective.

To get away from the isolated islands of funccionality you need one person with lead
responsihiliry for understand ing a prod uce fam ily's value stream and improving it. We call
th is person a Value-Stream i\1lanager, and suggest that in chis capacity they report to the rop
person at your site. This way rhey wil l have the power necessa,·y to help cha nge happen.

Who is responsible for the value stream?

PROCESS 1 PROCESS 2 PROCESS 3 CUSTOMER

..
.
t

The Value-Stream Manager!

PART I: GETTING STARTED 5


JOB DESCRIPTION FOR A
VALUE-STREAM MANAGER
• Reports lean implemen-
tation progress to the
top person on site
• A line, not staff, person
with the capability to
make change happen
across functional and
departmental boundaries
• Leads the creation of the
current-state and future-
state value-stream maps
and the implementation
plan for getting from
present to future
• Monitors all aspects of
implementation
• Walks and checl1s the flow
of the value stream daily
or weekly
• Makes implementation
a top priority
• Maintains and periodically
updates the implementa-
tion plan
• Insists on being a hands-on
person driven by results
Many people gee involved in lean imp lementation, and all need
an understanding of value-stream mapping and ,he ability ,o ,cad a
futu re-state map. But the mappi ng and future-state imp le mentation
ream needs to be led by some one who can see across the boundaries
over wh ich a product's value-stream !lows a nd make change happen
there. Value-stream imp rovement- "flow kaizen" - is management
doing kaizen.
Do not make the mis,ake of splitti ng up the mapping task among
area managers a nd then hope to stitch together their individ ual
segments. Likewise, don't map you r organization. i\fap the flow
of p roducts through you r organization.
Two kinds of kaizen
s enior
mgmt. FLOWKAIZEN
!Value-Stream Improvement)
l
PROCESS KAIZEN
(Elimination of Waste)
front
lines
focus
Note:
Both flow kaizen (value-stream im provement) and process-level
kaizen \eliminat1on o f waste at the shop fl oor team level) are
necessary in your company; improvement in o ne improves the
o ther. Flow kaizen focuses on material and information flow
(wh ich require a high vantage point to seel and process kaizen
focuses o n people and p rocess flow.
 Using the Mapping Tool
Va lue -stream mapping can be a communication tool, a business
p lan ning tool, and a tool co manage your cha nge process. Va l11e-strea rn
mapping is essentially a language and, as with any new language, che
best way co learn ma pping is 10 practice ic formally at first, until you
can use it insti nctively.

Va lue -meam mapping initially follows the steps s hown ac right. 1oticc
that "fucure-srnce drawing" is highlighted, be cause your goal is 10
design and introduce a le an value scream. A current state without a
future state is not much use. The furure -srate map is most important.

product family
T he fim ste p is drawing the current state, which is done by gathering
in formation on the s hop floor. T h is provides che information you
need co develop a f'u mre State. Notice that the arrows between
current-state
current and furure scace go both ways, indicating chat development drawi ng
of che current and future st.ates arc overlapping efforts. Fu cu re-
state ideas will come u p as you a re mapping the current state.
future•ata.te
Likewise, drawing your furure s tate will ofce n point ouc important drawing
currenr-srate information you have overlooke d.

T he final step is to prepare and begin actively using an implemen- WOf'kpl an &
implementation
. cation plan rhat describes, on one page, how you plan to achieve the
future state. T hen, as your furu1e state becomes reality, a new
Initial Value-Stream
future-sca re map s hou ld be drawn. T hat's continuous improvemenr Mapping Ste ps
at che value -scream level. There m ust always be a future-state ma p.

The be auty of this burea ucracy- and PowcrPoint-frcc method is

that your mapping a nd impleme nrarion team e nds up wirh on ly
a few sheets of paper (the futu re s tate and a plan co achieve ic)
that can transform your business !

Note:
Value-stream mapping for one product family should not take
too much time. In about two days you should have a future-state
map drawn to the point where Implementation can begin. Don't
get hung up trying to make all the details on your future-state
map perfectly correct. Fine-tune your futu re-state map as
Implementation progresses.

PART I: GETTING STARTED 7

 SUMMARY - YOUR STARTING POINT

• Select a product family

• Have one person personally lead the mapping effort

• Begin at the " door-to-door" l e vel

• Consider both the material and information flows

8
PART II: THE CURRENT-STATE MAP
 Drawing the Current- State Map
Developing a fucure scate begins with an analysis of the currenc
production situation. T his section shows you how to create a "current•
state map" usi ng a simple example factory we call Acme Stamping.
Mapping begins at the level of the door-to-door flow in your pla nt,
where you draw process categories like "assembly" or "welding,"
instead of recordi ng each processing step.

We use a set of symbols, or " icons," summarized inside the back

cover of this workbook, to represent processes and flows. You can
develop additional icons of your own, but keep chem cons istent
within your company so char everyone will know how to draw and
understand the maps that you need co institute lean man ufacturing.

Once you see the overa ll flow through the plant, you can change
the level of magnification; zoom ing in to map every individual step
wlrhin :i process cacegory or zooming ouc to encompass the- value
1

stream exte rna l to your plam.

levels of mapping the value stream

for a product fam ily

I
I process level

I
single plant
I (door to door)

I multiple plants

I
across companies
I
I
I
I
 A few mapping tips :

• Always collect current-state information while walking along the

actual pathways of material and information flows yourself.

• Begin with a quick walk along the entire door-to-door value stream,
to get a sense of che flow and sequence of processes. Afte r the quick
walk through, go back and gather in formation ar each process.

• Begin at the shipping end and work upstream, instead of starting

at the re ceiving dock and walking downstream. T his way you will
begin with the processes chat are linked most directly co che custome r,
which shou ld set the pace foe other processes further upsrream.

• Bring your stopwatch and do not rely on standard times or

information that you do not personally obtain . Nu mbers in a
lile rarely reflect current rea lity. F ile data may reflect t imes when
everything was running well, for example the firsc-timc-chis-year
three-rn inuce-die-c hange, or the once-since-the-plane-opened week
when no expedi ting was necessary. Your ability co envision a future
state depends upon personally going to whe re the action is and
undemanding and cimi ng what is happening. (Possible e xceptions
to this rule are data on machine upri me, scrap/rework rates, and
changeove,· times.)

• Map the whole value stream yourself, even if several people are
involved . Understanding the whole t1ow is what value-stream mapping
.is about. If d iffe rent people map different segments, the n no one
will u nderstand the whole.

• Always draw by hand in pencil. Begin your rough sketch rig ht on

the shop floor as you conduce your current-state analysis, a nd clean
it up lacer-aga,n by hand and in pencil. Resist the cemprncion co
use a computer.

10
Always draw by hand in pencil.

Drawing by hand can be done without delay,

while you are on the floor. As you draw you will
think of further Information that you need.

Drawing by hand means that you can do it

yourself, which is key to understanding the
material and information flows.

Drawing by hand means you will focus on

understanding the flow, instead of on how to
use the computer. The point of value-stream
mapping is not the map, but und11rstanding
the flow of Information and material.

Having to manually fine-tune your drawings

will Improve your mapping ability. Keep an
eraser handy!

PART 11: THE CURRENT-STATE MAP 11

 Dra wing the Curren t -State Map
To get starced, fold out the Acme Stamping dara set inside the back cover of this
workbook and refer to it as we bu ild Acme's c urrent-State map. Also get yourself
a blank sheet of paper (l 1" x 17" ledger size paper-<:alled "A3" in E,,rope and
Japan- works well for us) and a pencil so you Clln draw along with us.

Acme's product fam ily to be mapped is a scamped-steel steering bracket. This

component holds the steering column to the body of a car and is produced in
rwo versions: one for left-hand-side drive cars, the other for right-hand-side
drive. Because there is no variabiliry in the design beyond the left-drive and
right-drive versions, the J)roducr family is very narrow in th is example.

The boundaries of Acme's first map arc the door-to-door flow of the product
through Acme's plant, including basic supplied material (coil steel) and the
shipment of completed brackets co Acme's cuscomer, the Seate Street Automotive
Assembly plant. Mapping starts with the c ustomer requirements. We'll represent
the cusromer's assembly plant with a factory icon, placed in the upper right-hand
portion of the map. Underneath this icon we'll draw a data box recording the
requi.rcmencs of Acme's assembl)' plane customer.

Note:
As noted in Lean Thinking, the critical place to begin any improverne111
effort is c lear specification of the value of a product as perceived by the
end customer. Otherwise you run the risk of improving a va lue stream
that efficiently provides the end customer with something other than
what's really wanted . Thus mapping begins with the customer requirements.

State Street Assembly operates on two shifts. This customer uses 18,400 steering
brackets per month and requires daily shipments. TypiC111ly J 2,000 "left-hand
drive" brackets and 6,400 "right-hand drive" are needed every month. State
Street Assembly requests palletized returnable tray packaging with 20 brackets
in a tray and up ro 10 trays on a pallet. The customer orders in mulciples of
trays, so the "pack size" is one tray of 20 pares. All of the bracke ts on each pallet
need to be either left-drive or right-drive style.

12
 State Street
Assembly

18,400 pcs/mo
-12,000 "L"
- 6 ,400 "R"
Tray= 20 pieces
25hi~s

First View of the Current- State Map

Showing the Customer

PART II; THE CURRENT-STATE MAP 13

 I
ASSEMBLY I The next mapping Step is to d raw rhe basic production p rCJcesses.
To indicare a process we use a process box. The general ru le or
rhumb for the door-to-door map is rhac a process bo~ indicates a
process in which die material is flowing. Because drawing one box.
for every single processing s rep wou ld make tbc map unwie ldy, we
use the process box co ind icate one area of marerial □ow, ideally a
continuous flow. T he p rocess box stops wherever processes are
disconnected and the material flow stops.

For example, an assembly process with several connected workstations,

even if there ls some WIP inventory between sracions, wou ld be
drawn as one process box. Bur if one assembly process is disconnected
from the next assembly process downstream, with inventory stagnati ng,
accumu lacing, and being moved in batches between them, the n two
process boxes would be used .

Likewise, a machini ng line of say IS sequential machining operations,

s uch as drilling, rapping, etc., that are connected by a rransfer line
between each opcrarion would be shown wirh only one process box on
rhe door-ro-door map, even if some inventory accu mulates between
machines. (Tf a dern ile.rl rm~P.S<-li,ve,I m,1p is later made for the
machining area alone, then you would draw one box for every ind i-
vid ual machining seep.) But if there ,1re distinctly separate machining
processes in the plant, with inventory between them stagnating and
transferred in barches, then each gets its own process box.

Material flow is drawn from left to right on the bottom half of the
map in the order of processi ng s teps; nor according co the physica l
layout of the planr. At Acme Stamping we find six processes in che
steering bracket material flow, which occur in the following order:

• Scampi ng
• Spar-Welding Workstation I
• Spot-Welding Workscarion 2
• Asse mbly Workstation I
• Assembly Workstation 2
• Shippi ng

14
Each of Acme's spot-weld ing a nd assembly workstations are separate
processes because, as you can see in rhe overhead view o f the plant
(inside the back cover of this workbook), products do not move in a
flow from one tO the next. [nvenmry is moved in baskets and stagnates
between the workstations. On the map, each of these processes is
represented by a process box, from left t0 right on the bottom lu lf
of the page.
To help you get started here
Note: is a list of typical process data:
Unlike the Acme Stamping example, many value streams have
multiple flows that merge. Draw such flows over one another as
shown here. But don't try to draw every branch if there are too
t/ CIT {cycle time)
many. Choose the key components first, and get the others later
if you need to. t/ C/0 (changeover time)

t/ uptime (on-demand
machine uptime)

t/ EPE (production batch

sizc::.:t

t/ number of operators

t/ number of product
As you walk chis flow on the shop noor, you need co collect data that
variatfons
is important for decid ing what the future stare will be. So a data box
is drawn under each process box. After mapping several currenc and
future scares you will know instinctively whar process information t/ pack size
you need. The list at right will help you get scarred.
t/ working time
At Acme Stamping, we have the f'ollowinii; information to record (minus breaks)
in the data box under each processing step: che cycle time (time
chat elapses between one part com ing off che process co rhe nexc
pa rt com ing off, in seconds); the changeover time to switch from t/ scrap rate
producing one product rype to anorher (in th is case switching
between left-drive and right-drive brackers); the number of people
 required co operate the process (wh ich ca n be ind icated with an
operator icon as shown inside the process boxes); the available
working time per shift at tbat process (in seco11ds, minus break,
meeting, and cleanup times); and machine uptime information.

In the "stamping" data box we also show EPEx, which stands for
"every part every ___" and is a measure or production bacch
size. 17or example, if you change over ro produce a particu lar part
once every three days, then the production batch size is about
three days worth of parts. Notice that the cycle time is the rime
between pans coming off the end of the process and not the total
cycle time it takes one part to move through all process steps. Also
nocice that available work time d ivided by cycle time multiplied
by uptime percent is a measure of current process capacity, if no
changeovers are made.

Note:
Value-stream mapping uses sec,o nds as the time unit for cycle
times, takt times, and available working times. Many of you
have been taught to use decimal minutes to measure time, but
that unit is unnecessarily complicated. Value-stream mapping is
a tool everyone shou ld be able to use.

As you walk the product's materia l flow you will find places
where in ventory accumulaces. T hese points are imporcanr to
draw on che current-stare map because they cell you where the
flow is stopping. We use a "warning c.riangle" icon ro capture the
:3150pieces
2day,, location and amoum of inventory. (If the inventory accumulates
in more than one location between two processes, draw a triangle
for each location.)

16
Some Lean Measurements

Cycle Time (CIT)

n
D D D D D D □ D D Cycle Time (CIT)
How often a part or product actually
is completed by a p rocess, as timed by
observation. Also, the time it takes an
operator to go through all of their work
elements before repeating them .

VC Time

D D D
\
~ D □ □ D Value-Creating Time (VCT)
Time of those work elements that
actually transfor m the product in
a way that the customer is wil ling
to pay for.

Lead Time
I
□ □ □ □ □ □ □ □ Lead Time (LIT)
The time it t akes one piece to move all
the way through a process or a value
stream, from sta,t to finish. Envision
timing a marked part as it moves from
beg in ning to end.
or
I

~~ □ D

Usually: VCT < C/T < LIT

PART II: THE CURRENT-STATE MAP 17

 At Acme Stamping there is raw material inventory, finished-
goods i,wentory, and inventory between each ptncess. T he
observed am ou nt of inventory is recorded be low the tria ngles,
in qua ntity and/or time.

&
STAMPING S. WELD#1 S. WELD#2

Coils
200T
e, 1
&4600L
& 1100 L
5days 2400R '91 600R e, 1

CIT= 1 second CIT = 39 seconds CIT = 46 seconds

C/0 = 1 hour C/0 = 10 minutes C/0 = 10 minutes
Uptime= 85% Uptime= 100% Uptime=80%
27,600 sec. avail. 2Shi~s 2Shi~s
EPE = 2weeks 27,600 sec. avail. 27,600 sec. avail.

18
 State St reet
Assembly

18.400 pcslmo
- 12,000 "L''
- 6,400 "R"
Tray= 20 pieces
Second Vievv of the Current- State Map 2Shifts
vvith all Processes, Data Boxes, and
Inventory Triangles

&
ASSEMl3LY#1 ASSEMl3LY #2 SHIPPING

1600 L
101
&
1200 L
10 1
&2700L
Staging

850R 640R 1440R

CIT = 62 seconds CIT = 40 seconds

C/O = f1 C/O = f1
Uptime= 100% Upt ime= 100%
2Shifts 2Shifts
27,600 sec. avail. 27,600 sec. avail.

PART 11: THE CURRENT·STATE MAP 19

 After Acme's lase assembly workstation, steering brackets in trays
are raken co a scornge area (rriangle icon). They ,ire the n scaged in
Michigan the shippi ng area according to the daily sh ipping schedule and
Steel Co. delivered daily by cruck to the customer's assembly plant./\ truck
icon and a broad arrow indicate movement o f finished goods to tbe
customer. (Create rail or air freight icons if you need them.)
500-~. Coils
At the other end of the map, we'll represent the steel supplier with
another factory icon. We use che same cruck icon and broad arrow to
show movement of material from rhe supplier to Acme. The steel
supplier receives a weekly ()rder from Acme a nd ships twice a week.
We record in a data box that the supplier's pack size is a 500-foor roll
of coiled s teel. (In other words, the s upplier ~nnot de liver less than
Tues. + a full coil bur can de liver any number of fu ll coils, as requested.)
Thurs. Once at Acme, the coi I steel is taken co a storage area, as shown by
the invencory uiangle.

Note:
Do not map every purchased pan in your product family. Just
draw the flow for one or two main raw materials. Presentation of
all purchased parls is best shown on a process-level layout diagram.

5. WELD#1 5. WELD#2

&
STAMPING

& Coils
5c:laye
200T &
4600L 1100L
19,1
101 2400R 101 600R

CIT= 1 second CIT = 39 seconds CIT= 46 eeconcls

C/0 = 1hour CIO = 10 minutes C/0 = 10 minutee
Uptime=85% Uptime= 100% Uptime=80%
27,600 sec. avail. 25hi~s 25hi~s
EPE = 2weeks 27,600 sec. avail. 27,600 sec. avail.

20
 State Street
Assembly

18,400 pcslmo
- 12,000 "L"
- 6,400 "R"
fray = 20 pieces
2Shifts

Third View of the Current- State Map

Showing the Material Flow

1x
Daily

ASSEMBLY#1 ASSEMBLY#2 SHIPPING

&
1600 L
&
1200L
& 2700 L
Staging

850R 19,1 640R 1911 1440 R

CIT= 62 seconds CIT = 40 seconds

CIO= 0 CIO= 0
Uptime= 100% Uptime = 100%
2Shifts 2Shifts
27,600 sec. avail. 27,600 sec. avail.

PART 11: THE CURRl;NT-STATE MAP 21

 ..--------
information flow
But how does the Acme plane, each process within the p lanr, and the
steel coil supplier know how much to make and when? Let's add the
second aspect of our va lue -stream map; the information flow. do 1o
this we'll need,, few more icons and arrows, in particular a narrow
line co show information flows. T his line is modified with a lightning -
like wiggle when the information flows electronically (via electt·onic efectronic
data inte rchange) rather chan by paper. A small box icon or node is information flow

used to label or descri be different information-flow ,Hrows.

Information flow is druwn from right co left in the cop hal f of the
map s pace. In our Acme Stamping example, we drnw che flow of
in formation back from che State Srreer Assembly p lant co Acme's
Prod uction Control department and from there co Acme's scee l-coj l
s upplier. Notice there are separate lines for che forecasts a,,d daily
orders, as chose are diffcrenr information t1ows.

The Acme production comrol dcpanment is drawn wich a process

box, includ ing the note that Acrne usv:; ct <.;ompucerized tviacerials
Requi,·emenrs Planning system (lv!RP) to schedu le the shop floor.
Acme production control collects information from customers and
the s hop floor, consolidates and processes it, and s.c nds specific
instructions to each manufacturing process abouc what it should
produce and when. Production control also sends a daily shipping
schedu le to the shipping de partment.

Note
In your shop floor mapping efforts you may find information flows
more complicated than at Acme Stamping . For example, in many
shops supervisors couni inventory and make schedule adjustments
based on such information. (We call this "go see" scheduling and
show it with an eyeg lass icon. ) Try to incorporate these "informal "
"go see"
scheduling schedu ling processes into your map using the same information
arrows and nodes. If it looks messy, that's probably because it isl

22
 As you figure our how each process knows what t0 make for its
customer (the foll1Jwing process) and when to make ic, you can identify
a critical piece of mapping informacion: material movements chat are
pushed by the p roducer, noc pu lled by the customer. "Push" mea ns
that a process produces something regardless of the actua l needs of
the d0\\11lstream customer process and "pushes" it ahead.

Push typically re~ults from producing co a schedule that g uesses what

the next p rocess will need. Unfortunately this is nearly impossible to
do consiscemly because schedu les change and production rarely
goes exactly according co schedule. When each p rocess has its own
schedu le it is operating as a n "isolated island," disconnected from
any sore of downstream customer. Each process is able co set batch
sizes and to prod uce ac a pace that make sense from ics perspective,
inscead of the value-stream's perspective.

In chis situation, the supplying processes will rend co make parts

their customer processes don't need now, and those pares are pushed
into storage. This type of "batch and push" p rocessing makes it
almost impossible to establish the smooth flow of work from 011e
prncess ro rhe: ne:xr rhar is :l h~llrr-.:lrk o f le~n prod uction.

----►
The mapping icon for push movement of material is a striped arrow. Ar
Acme Scamping, only the shipping department is connected in any way
push
to a ·•customer." Each of the ocher processes is producing according to a
schedule, so the transfer of material from one process co the next is
occurring via push. A p ush arrow is drawn betwee n each process.

Note
Be wary of materia I movements that some-
one calls "pull," but are actually still a form
of "push." (See page 40 for a d iscussion of
supermarket pull systems.) To qualify as pull,
parts must not be produced or conveyed
when there is no kanban, and the quantity
of parts produced must be the same as
specified on the kanban. "Go see" schedule
adjustmen ts are not a 1r~1e pull.

PART ll: THE CURRENT-STATE MAP 23

 6-week PRODUCTION
.,_....---JForecast
CONTROL
Michigan
Steel Co. Weekly
Fax MRP

5OO-ft. Coils

Tues.+
Thurs. Fourth View of the Current- State Map
with Information Flows & Push Arrows

&Coils
STAMPING

200T
oar&•
46OOL
5. WELD#1
IOI&+
1100 L
5. WELD#2

5days ~0 1 240OR 101 101

6OOR

CIT= 1second CIT= 39 seconds CIT = 46 seconds

CID= 1 hour CIO = 10 minutes CIO = 10 minutes
Uptime=85% Uptime= 100% Uptime=8O%
27,600 sec. avail. 2 Shifts 2 Shifts
EPE= 2weeks 27,600 sec. avail. 27,600 sec. avail.

24
 ~ -- - -- -1, 90160130-day~ -- - -- i
_ _ _ _\~ _ Forecasts _ State Street
Assembly
Daily
Order
18,40 0 pcslmo
- 12,000 "L"
- G,400 "R"
Tray= 20 pieces
2 Shifts

Daily Ship
Schedule
1x
Daily

-=uffi♦
1600L
ASSEMBLY#1

KKF&•
1200L
ASSEMBLY#2
uu&
2700L
SHIPPING

Staging

8 50R 191 640R 191 144-0R

CIT= 62 seconds CIT = 40 seconds

CIO = RJ CIO= RJ
Uptime = 100% Uptime= 100%
2Shifts 2 Shifts
27,GOO sec. avail. 27,GOO sec. avail.

PART 11: Tl-IE CURRENT-STATE MAP 25

 In lookin g at the almost comp leted m;i p you ca n now see the basic
pattern of all value-stream maps, specifically a now of physical product
from left to right across the lower port.ion of the map a nd a now of
information about chis produce from ,·ig hr ro left across the upper
portion. You can also see h ow a value-stream map differs from the
typical visual cool used in operations analysis- the facility layout.
T he value-stream map makes the welter of evenrs s hown in the
plane layout suddenly comprehensible from the perspective of a
product's value stre am and its customer.

With the data from observation of current operations dra,vn or

recorded on the map, we can summarize the current condition of th is
value stream. Omw a timeline under the process boxes and inventory
timeline
triangles to compile the produccio11 lead rime, which is the time it
takes one pan to ma ke its way through the s hop floor, beginn ing
with arrival as raw material through co sh ipmem to the customer.

Note

The shorter your production lead time, the shorter the time between
paying for raw material and getting paid for product made from those
materials. A shorter production lead time will lead to an increase in
the number of inventory turns, a measure with which you may be
more familiar.

Lead times (in days) for each invcncory triang le are calculated as
follows: inventory quantiry divided by the daily customer require ment.
lly adding the le ad times through e ach process and through each
inventory tria ngle in the material now, we can arrive at a good estimate
of total production lead rime. At Acme Stamping this n umber is 23.6
days. (For maps with m ultiple upstream flows, use rhe longest time
path to compute coral lead t ime .)

26
Now add up only the value-addi ng times or the processing ti mes,
for each process in the value scream . Comparing value-added or
processing time to total lead time should give you q uite a shock.
At Acme Stamping the total processing cime involved in making
one piece is only 188 second s, whereas char piece tukes 23.6 days
co make its wtty through the plant.

Note:
Ar Acme Stamping the lead time through a process and the cycle
time are the same. In many cases however, the lead time for one
part to move through a process is longer than the cycle time. You
can draw both lead time through a process and the value-adding
time as follows:

lead time through

/ the process

2 hours
40seconds

~ value-adding time

GO LOOK!
Completed Current- State Map
vvith Lead-Time Bars & Data

PART II: THE CURAENT,STATE MAP 27

 6-week PRODUCTION
.,,..--- i Forecast CONTROL
Michigan
Steel Co. Weekly L - - - - - - -t""".""'.".:-:-,
Fax

500-~. Coils

Tues.+
Thurs.

&Coils
STAMPING

200T
O[W&•
4600L
5. WELD#1
~&-
1100 L
S. WELD#2

5aays 19,1 101 19,1

2400R 600R

CIT = 1 second CIT = 39 seconds CIT= 4G seconds

C/0 = 1hour C/0 = 10 minutes C/0 = 10 minutes
Uptime=85% Uptime= 100% Uptime=80%
27,600 sec. avail. 25hifts 25hifts
EPE= 2weeks 27,600 sec. avail. 27,600 sec. avail.

5days 7.6 days 1.8days

1 second 39 seconds 46 seconds

28
 . _ . - - - - - ~ 90/60/30-day
Forecasts 1------1
State Street
Assembly
Dally
Order
18,400 pcs/mo
-12,000 "L"
- 6,400 "R"
Tray= 20 pieces
2Shi~s

Daily Ship
Schedule
1x
Daily

l:D'&•
1600L
ASSEMBLY#1
IDD' I.
ASSEMBLY#2

LU •
SHIPPING

Staging
1200 L 2700L
850R ,S, 1 640R \9J1 144-0R

CIT= 62 seconds CIT= 40 seconds

C/0= fJ C/0 = fJ
Uptime= 100% Uptime = 100%
2Shi~s 2Shi~s
27.600
• sec. ava,.
'I 27600
• sec. ava,,'I
Production
Lead TI me = 23.6 days
2.7 day& 2days 4.5day&
62 seconds 40seconds Processing
Time = 188 Gee.

PART II: THE CURRENT-STATE MAP 29

 What have vve accomplished so far ?
We hope , hat now you can see the value stream and begin to
1
recogn iz.e areas of overproduction. H owever, a ~currenc-statc map"
and the effort requ ired ro create it are pure muda unless you use
your map co quickly create ,ll)d implement a "future-stare map"
chat eli minates sources of waste and increases value for the
customer. We will describe just how to do this in rhe rest of
Learning to See.

Your Turn
Before we proceed ro the creation of a "future-state map," you may
find it usefu l w ger a little more "current-state" practice. We therefore
ha.vc provided rhe following current-state factS for a more custom•
products oriente d company called ""£'WI Ind ustries." We invite you
now t0 rake another blank sheet of I L" x 17" paper and carefully draw
a currenc-state value-sueam map for TWI Industries. You can then
compare your currenr-srate map with rhe one we\,e drawn for TWl, as
shown in Appe ndix B. (However, be sure nor ro peek at Appendix Cl)

30
 Value-Stream Mapping Data Set
l
TWI Industries
l TWl Industries produces several components for tractors. This ,'aSe
concerns one product fomily-sceering arms-which are produced in
2
many configurations. TWl's customers for chis product family arc both
I original-equipme nt rracmr builders and rhc ,,fcermarker repair business.

I Bt:cause of the wide ,,ariery of product configurarions and the

J fact rhac customer configuration requ irements vary from order to
order, steering arms are a "make-to-orde r" b usiness. l L currently
J rakes a c usmmer order 27 d ays to get rhrough T W['s productio n
I processes. T his long lead time and a signi ficant order backlog have
prom pted TWI to qume a 60-day lead cimc to customers. H owever,
I TWl's c ustomers cannot accurately predict the ir size require ments
I more tha n 2 weeks out, a nd thus chey make adjustments to thei r
orders 2 weeks before sh ipment. T hese order 8djt1srments lead to
I order expedi ting o n the shop noor at T WI.
I
Although T WJ Production Control releases customer o rde rs t<)
I production roug hly in the order char chey are received , orde rs arc
batched by p roduct configuration on the shop n oor to reduce the
I
nu mber of ti me-consuming changeovers. Th is also creates a need
I for order expediting.

I The Product
I • /\ steeri ng arm is a me ral rod with a forged fitting welded to each end.
• TWI's steering arms are available in 20 d ifferent lengths, 2 d iameters,
I and with 3 d ifferent rypes of e nd fi ttings. (Each e nd of the steering
D arm ca n have a di rferenr fitting.) T his means there a rc 240 different.
steering arm parr numbers that T\.VI supplies.
D
D
D
a
a
a PART II: THE CURRENT-STATE MAP 31
 Customer Requirements
• 24,000 pieces per month.
• Acustomer order r.inges from 25 co 200 pieces, wirl1 an <1vcr~ge of 50 pie<.-e.~.
• Corrugated-box packaging with up to 5 steering arms inn box.
• Several daily sh ipments per day by truck co various c ustomers.
• Each c 1.1 scomer's configurmio11 requ irements vary greatly from
order to order.
• TW! req ui,·cs orders co arri ve 60 days before s hipping date.
• Customers often adjust their si.e mix 2 weeks before the shipping date.

Production Processes (see diagram)

• TWl's processes for the steering arm product fami ly involve
cutt.ing a metal rod followed by welding end fittings co the rod,
deflash (machine removal of excess weld menc), painting ac an
outside vendor, and suhseq uenc assembly of the end fittings. The
fcll'ged end-fitting sockets are also machined at TW I. F'inished
steering arms are s taged and shipped to customers on a daily basis.
• Switchi ng between rod lengths requires a 15-m inute changeover
ac the cutting, welding, and deflash operations.
• Switching between rod diameters cakes a 1-hour changeover at
chc cutting, welding, nnd dc:flash 01>ett1tions. T he longer chonge ·
over for <l iameters is due mostl y co an increased quality-control
inspection requirement.
• Switching between the three types of forged e nd fittings rakes a
2-hour changeover at rhe machining operation.
• Steel rods are s upplied by Michigan Steel Co. T he lead rime for
obwining rods is I 6 weeks. There are two shipments per month .
• Raw forgings for the e nd fittings are supplied hy Indiana Castings.
The lca<l time for obtain ing forgi ngs is 12 weeks. T here are two
sh ipments per mooch.

TWI Indus-tries
Steering Arm Production Steps
CUT RODS ~ WELD . _ . . WELD _ . , . REMOVE ~ PAINT -► ASSEMBLY
one end other end flash (outside vondor>
A
I
MACHIN£ FORGINGS
t

32
 Work Time
• 20 d ays in a mont h.
• 1\vo-shift operation in a ll ptoduction depamne ncs.
• Eight (8) hours every shift, with overtime if necessary.
• Two IS-m inute b reaks during each s hift.
Manual processes s rop during breaks .
Unpaid lunch.

TWI Production Control Department

• Receives cust0mer orde rs 60 days our and enters them to MRP.
• Generates one "shop order" pcr customer, w hich follows the
o,·der th rough the encire production p rocess.
• Releases s hop orders to produccion 6 weeks before shipment
to accelerate MRP's procurement of rods and forgings.
• Issues daily "priority" list ro production supervisors. Supe rvisors
sequence shop orders through the ir deparrmencs according co this list.
• Receives customer size-c hanges 2 we eks before s hipment and
advises supervisors co expedite these o rders.
• Issues daily s hippi ng schedu le co S hipping D e partment.

Process Information
1. Cutting (The saw cuts rods for many TWI p roducts)
- Manual process with I operator.
- Cycle time: 15 seconds.
- Changeover time: 15 minutes (for length) and I hour (for diame te r).
- Re liabili ty: 100%.
- Observed in vencory:
- 20 days o f u neut rods before the saw.
- S days of cue rod.

2. Welding Workstation I (ded icated to this product family)

- This operation welds the first machi ned forging to the rod.
- Automatic process, with operator load a nd un load external to
mach ine cycle .
• Cycle rime: Operator • 10 seconds, Machi ne = 30 seconds.
- Chang~over time: 15 mjnutes (for length) and I hour (for diameter).
- Re liabi li ty: 90%.
- Observed inventory: 3 days of welded a rms.

PART II: THE CURRENT-STATE MAP 33

 3. Welding Workstation II (dedicated to this product family)
- This operation welds the second machined forging co the rod.
- Aummalic process, wirh operacor load & unload extercrnl to
machine cycle.
- Cycle time: Operator= 10 seconds, Machi ne = 30 seconds.
- Changeover cime: IS min mes (for length) and J hour (for diameter).
- Reliability: 80%.
- Observed invenrory: 3 days of welded arms.

4. Deflash Workstation (dedicated to this product family)

- Automatic process, with operator load & un load external co
mach ine cycle.
- Cycle time : Operator= 10 seconds, Machine= 30 seconds.
- Changeover time: 15 minutes (for length) and I hour (for diameter).
- Re liability: 100%.
- Observed inventory: S days of deflashe<l arms.

5. Painting (steering arms are shipped to an outside vendor for painting)

- Painting lead ti me • 2 days.
- O 11c daily truck pickup of unpainted arms and drop-off of painred arms.
- Observed invenrory: 2 days at die painter, 6 days of painted anns at TWl.

6. End-fitting Assembly {dedicated to this product family)

- Manual process with s ix operacors.
- Torn! work cime per p iece: 195 seconds.
- C hangeover time: 10-minute fixture swap.
- Reliability: 100%.
- Observed finished-goods inventory in wa rehouse:
- 4 days of finished steering arms.

7. Machining of Forgings (dedicated to th is product fami ly)

- Auromacic machin ing process with one machine attendant.
- Cycle time: JO seconds.
- C hangeover dmc: 2 hours.
- Reliability: J 00%.
- Observed invencory:
- 20 days of raw forgings from the supplier.
- 4 dnys o f machined forgings.

8. Shipping Department
- Re moves parrs from finished goods ware house and stages them
for truck s hipment to c uscomer.

34
What Makes a Value Stream Lean?
The Cacch -22 of designing your futu re -state va lue stream is chat
you wil l be much more successful if you've already done ic many
times! T his is where~ sensei who has experienced the learning
curve that you need can be a big help.

I lowever, we don't all have access to a good sensci and some of

you don't wane one anyway. After all, O hno did n't have consu ltants
gu id ing him as he built Toyota's p roduction system through trial
and error after World War fl. l n fact, it is an invaluable lea rning
experience co cake a crack at a future-state value stream with your
own resources, even if you soon see prob lems wich your approach
and modify ir in the spirit of continuous improvement. And, u ntil
a day in t he fu ture when you can make your products in a complete
continuous Flow with lead time short enough to allow production
only to con.firmed order nnd zero ch~ng:eovel' times becween produces,
you will req uite a number of fucure -srate maps (no mam:r how much
help you get from a sensei), each a licrle leaner and closer to that ideal.

Bue you shou ldn't scare from scratch either. The manufacturi ng
world has now had a lot of experience with lean ma nufactu ring, so
you can begin with established principles and practices and work
to adapt them to fucure states for your own value screa ms.

Before we s how you how eo draw a fucure-state map (Part TV), lee's
summarize some of che mosc important lean principles co he lp you
get started.

PART Ill: WHAT MAKES A VALUE STREAM LEAN? 35

 Overproduction
We can see che fundamental problems with mass (or "batch-and-push")
production in the Acme Stamping current state: each process in the
value stream operates as an isolated island, producing and pushi ng
product forward according to schedules it receives from Production
Comrol instead of the actua l needs of che downstream "customer"
process. Because 1his material ou1pm is not yet needed, it must be
handled, cou nted, stored, and so on-pure muda. Oefects remain
hidden in the inventory queues until the downstream process finally
uses the parts and d iscovers the problem (which is by then extensive
and hard ro t l"dCe). As a resu lt, while the va lue-creating time for
prod ucing one product is very short, the total time that prod uct
s pends getting through the plant is very long.

To reduce that overl y long lead ti me from caw material to finished

good you need to do more than just try to e liminate obvious waste.
·roo many lean implemcncation effortS have been 1'sevcn-wasce"
scavenger hunts. While ir is good to be aware of waste, your futu re-
state designs need to eliminate che sources, or "root causes," of
waste in the value stream. Once the problems of mass production
can be seen in a way that reveals these ronr r..•n1scs. your company
can work at finding original solutions.

overproduction

-
36
1 'he mosr sign ificanr source of waste is overproduction, which
means producing more, sooner, or faster than is req uired by the next
process. Overproduction causes all kinds of waste, not just excess
inventory and money tied up in that inventory. Bacches of pans
must be stored, requiring storage space; handled, requiring peopl e All we are really trying to
and equipment; sorced; and reworked. Overproduction results in do in lean manufacturing
shortages, because processes are busy making the wrong things. 1c
is to get one process to
means that you need extra operators and equipment capacity, because
you are using some of your labor and equipment to produce part$ make only what the next
rhar are not yet needed. It also le ngthens the lead rime, which
process needs when it
impa irs )'Olli' flexibi li ty to respond co customer requirements.
needs it. We are trying to
T he constant actcntion l oyota puts on avoid ing overproduction is
link all processes - from
what most clearly distingu ishes rheir value srreams from mass
production value streams. Mass production chinking says that rhe the final consumer back

more and faster you produce, 1he cheaper it is co produce. But th is to raw material-in a
is rrue only from a direct-cost-per-item perspective as measured by
smooth flow without
traditional accounting practices, and ignores all the orher very rea l
costs associated with overproduction and rhe wastes it causes. detours that generates the

shortest lead time, highest

quality, and lowest cost.

Characteristics of a Lean Value Stream

All we are really trying to do in lean manufactu ri ,,g is to get one
process to make only what the next process needs whe n it needs it.
We are trying to li nk all processes-from the final consumer back
ro raw material -in a smooth flow without detours char ge nerates
rhc shortest lead time, hig hest quali ty, and lowesc cost.

So how can you-on your shop floor- actually get one process
to produce onl y what the next process needs when it needs ir?
Fortu nately you can follow Toyota's lead and use the gu idelines
on the following pages.
 takt time
Synchronizes pace of production to match pace of sales

a vailable w o rking time per d ay

takt tim e=
c u stomer d e mand r a t e per day

2 7,600 sec.
e x a mple:
460 piec e s
= 60 seconds

Guideline #1 : Produce to y o ur takt tim e.

"1akr d me" is how ofte n you should produce one part or produce, based on the rate of sales,
co rneet customer requiremencs. l Okt time is calculated by dividiug tlit: (.;ll$tOmcr de mand rate
per day (in units), inco your avai lable working ri me per da)' (in seconds).

Take time is used co sync hron ize the pace of produc tion with the pace of sales, panicu la rly
at che "pacemake r proce$s" (see page 43). le is a re fere nce nu mber that gives you a se nse
for the rate at which a process should be producing. Ir helps you see how yo11 are doi ng and
what you need to improve. On the future-state map, takt ti mes are noted in rhe d ata boxes.

Producing to cakt sou nd s si,nple, bur it rtq uires concentrated effort co:
• Provide fast response (within take) co problems.
- Eliminate causes of unplanne d downtime.
• E liminate changeover time in downstrea m, assembly-type processes.

Note
In some industr ies, such as distribution, custom products. and process industries,
it can take some creativity to define •units" of customer demand. One solution is
to define a " unit" as how much work can be done at you r bottleneck process in a
"takt" o f say, 10 m in utes. Then break your orders up into units of this takt in terval.

18
 Guideline #2: Develop continuous flow
wherever possible.
Continuous flow refers to producing one piece at a cime,
with each item passed immediately from one process step
co the ncxr. without scag,rntion (and many other wastes)
in between. Conti nuous flow is the muse efficienc way co
produce, and you should use a lot of creativicy in trying
tO achieve it.

The mapping icon we use co indicate continuous flow is

simply che process box. In your future-swre drawing, each
process box s hould describe an area of flow. So if you
introd u<.'C more continuous flow in your future state, chen
two or more curre nr-srate process boxes would combine
into one box on the future-state map.

Sometimes you'll want to limit the extent of a pure

conti nuous flow, because connecting processes in a
conti nuous flow also merges all the ir lead times and
downtimes. A good approach can be to begin wid1 a
combin:1rioo of continuous flow and some puH/R I FO.
isolated islands T he n extend the range of conti nuous fl.ow as process
relia bility is improved, changeover times are reduced to
near zero, and s maller, in-line equi pment is developed.

continuous flow

finished
product

PART Ill: WHAT MAKES A VAlUE STREAM lEAN? 39

 Guideline #3: Use supermarkets to control production where
continuous flow does not extend upstream.

There are often spots in the value stream where continuous flow is nor possible and batching
is necessary. There can be several reasons for this including:

• Some processes are designed to operate at very fast or slow cycle rimes and need co
change over to serve multi ple product families (e.g., scamping or injection molding.)
• Some processes, such as those at suppliers. are far away and shipping one
piece at a rime is not realistic.
• Some processes have coo ni uc h lead time or are too unreliable ro couple
directly co other processes in a contin uous flow.

Resist rhe rempration to schedule these processes via an independent scheduling function,
because a schedule is only an estimate of what the next process wi ll actually need. Instead,
control their production by linking them to their downstream customers, most often via
supennarker-based pull systems. Simply put, you usualli• need to install a pull system where
continuous flow is inte rrupted and the upstream process must still operate in a batch mode.

supermarket pull system

''produq.tfon''KANBAN ''vwithdt"aw-al''KANBAN

----0--,r--~- ----. I
supplying customer
process procoas

A product.
---D---+ B
product

SUPERMARKET

© cusTOMER PROCESS goes 10 supermarket and withdraws what it needs w h en It needs it.
@ suPPLVINB PROCESS produces to replenish what was w i thdrawn.
PURPOSE: Controls production at supplying process without trying to schedule. Controls
producti"on between flows .

Note:
A "production" kanban triggers production of parts, while a "withdrawal" kanban
is a shopping list that instructs the material handler to get and transfer parts ,

40
The purpose of placing a pull ~ystem between two processes is to
have a means of giving accurate production instruction to the upstream
process, without trying to predict downstream demand and scheduling
the upstream process. Pu ll is a method for controlling production
between flows. Gee rid of chose e lements of your /vfRP that try to
schedule t he diffcccnt iircas of your plant. Let the downstream There is another icon similar
process' withdrawals ouc of a supermarket determine what the to the supermarket icon, but
upstream process produces when and in which quantity. closed on all sides. This icon
represents "safety stocl<,"
There are several icons associated with a supermarket pull system: which is used as a hedge
against problems such as

G D • -c5SJ-- downtime, or "buffer stock,"

which is used to protect you
Withd rawal Production Kanban Arriving against sudden fluctuations
Kanban in Batches
in customer orders. Safety

~ ~ V
stock should be temporary
-used only until the root
cause of a problem is found
and eliminated. To ensure
Su pe-rmarket Withdrawal Kanban S ignal thot safety stock does not
Kanban Post Kanban
instead become a permanent
crutch, there should be strict
The supermarket icon is open on the left side, which faces the
rules for being able to use it
s upp lying process. T his is because this supermarket belongs co che
supplying process and is used co schedule chat process. On the factory and it may even be kept
floor, supermarkets should ordinarily be located near the supplying under lock and key. Typically
process to help that p,·ocess maintain a visual sense of customer usage this means getting permission
and requiremencs. T he "cuscomer" process material handler then from a high-level manager,
comes tO the supplier's supermarket and withdraws what is needed. who will want to see a root-
These withdrawals trigger the movement of p reprinted kanban cause problem analysis and
(typically cards) from the s upermarket co the supplie r process, where countermeasure plan before
they are used as the on ly production instruction for chat process. giving permission!

Before you decide to use any supermarket pull sys tems, be sure
that you have introduced continuous now across as many process
seeps as is possible. You don't wane supermarkets of inventory,
and the extra material handling requirements, between processes
un less you muse.
 Note:
Pu ll systems are a nice way to control produciion between processes that cannot be tied
together in a continuous flow, but sometimes it is not practical to keep an Inventory of
all possible part variations In a pull-system supermarket. Examples include custom parts
(where each part produced is unique), parts that have a short shelf life, and costly parts
that are used infrequently.

• In some of these cases you can use a FIFO {"first in, first out") lane between two
decoupled processes to substitute for a supermarket and maintain a flow between
them . Think of a FIFO lane like a chute that can hold only a certain amount of inventory,
-FIFO- with the supplying process at the chute entrance and ihe customer process at the exit.
If the FIFO lane gets full. the supplying p rocess must stop producing until the customer
has used up some of the inventory.
For example, you ship to an outside plating process one time per day. The plater can only
handle 50 pieces per day, so you set up a FIFO lane sized to hold at most 50 pieces of
plating work. Whenever the lane is full the upstream process stops producing parts to
be p lated. In this manner. the FIFO lane prevents the supplying process from overpro-
ducing, even though the supplying process is not linked to the plater via continuous
flow or a supermarket. When a FIFO lane is full no additional kanban are released to the
upstream process. (Note that some people refer to the FIFO approach as "CONWIP.'1
• Sometimes you can install a •sequenced pull " between two processes, instead of a
complete supermarket that has all components represented in it. Sequenced pull means
sequenced- that the supplying process proci 11cP.s" prP.rlP.tArminP.d quantity (often one subassembly)
pull ball directly to the customer process' order. This works if lead time in the supplying process
is short enough for productio n to order. and if the customer process follows strict
·ordering• rules. Sequenced pull is sometl mes called the • golf ball system· because
colored balls or disks (that roll nicely down a chute to the supplying process) are
sometimes used to provide production instruction.

an example of a ••FIFO Lane"

.
* ,' ------1k•nban)------------------------------------------ ----,
' ''
' '
s~bf\ - - - - i:u1..1..'? :
'•
t \
max. 50 pie<OC5
. - - ----, '
Upstream Downstream
Process - FIFO Lane - 1--_P_
ro_c_es_s_ --1

Supermarket

42
Guide lin e #4 : Try t o send the c u st o m er sc h edule to only o n e
production process.
By using s upcrrnackct p ull systems, you will typically need to schcd ul.e onl y one point
in you r door-to-door value stream. Th is JlOint is ca lled the pacemaker process, because
how you control production at this process sets the pace for all the upstream processes.
For example, fluctuations in production volu me at the pacemaker process affect capacity
req ui rements in upstream processes. Your selection of this schedu ling point also determines
what elements of your value stream become part of the lead r.imc from customer order to
finished goods.

Nore thar material transfers from the pacemaker process downstrea m to finished goods
need co occur as a now (no s upermarkets or pul ls downstream of the pacemaker process).
Because of this, the pacemaker process is freque ntly the most downstream continuous-flow
p rocess in the door-t0-door value stream . On the future-state rnap the pacemaker is the
production process that is controlled by the outside customer's orders.

selecting the ..pacemaker process"

,. . . . , ,. . . .., ,. . . . , Ii'
t ' t ' t '
p.-oc•oa LJrooe•• u•oaes• LJ
........
4
1 ~ ·
c
pull 2 ~ ·
c
pull 3 ~ ·
c
pull
customer

6UPERL ARKET ~

Ii'*
C_•_•o_••_•_•_,-FIF0-1"-•• •1-flFO-._,._•o_••_•_•__,
2 4

customer

►
t
SUPERMARKET

*Note:
With custom products and job shops, the scheduling point often needs to be further
upstream like this.

PART Ill: WHAT MAKES A VALUE STREAM LEAN? 43

 Guideline #6: Distribute the production of different
products evenly over time at the pacemaker
process. (Level the production mix.)
Mosr. assembly de pamnems probably find ic easie r to schedule long
runs of one product type and avoid changeovers, but chis creates
serious p roble ms for the rest of th e value scream.

G rouping the same products and p roduci,ig them all ac once makes
ic di fficult to serve custome rs who wa nt some thing d ifferent from the
batch being produced now. This sequ ires you to have more fin ished-
good s in ventory-in the hope that you will have on-h.i.nd what a
customer wa nes-or more le ad time to fulfill an order.

Batching i,i assembly also means that fabrkated components will be

consumed in batches, which swells che in-process inventories needed
in upstream s upe rmarkets throughout the e ntire va lue stream. And, rd
because variation in a final assemb ly schedule is amplified as you c::;
move upstream, those in-1>rocess in ventories will te nd co grow larger
the fu rther upstream you go.

Leveli ng the prod uct m ix means distributing the production of

<J ifferenc produces evenly over a time period . For example, ins tead
of assembli ng a ll the "Type A" p roducts in che morning and all the
"Type B" in the afternoon, leveling me ans a lternating repeated ly
between s maller batc hes of "A" and "B."

The more you level the p,·oduct mix at rhe pacemaker process, che
more ab le you will be to respond co different customer requ irements
with a short lead rime while holding little finished -goods lnvencory.
Th is also a llows your upstream s uperrnarkets robe s maller. But be
warned that leveling the mix require s caki ng some pains in assembly,
such as more changeovers and trying to keep all component variations
at the line ac all times (to e liminate changeover time). Your reward is
the elimination of large amounts of waste in the value stream.

The icon for leveling is this symbol, which

is inserted into an information flow arrow, loxoxf load
leveling

44
Guideline #6: Create an "initial pull" by releasing and
withdrawing small, consistent increments of work at
the pacemaker process. (Level the production volume.)
Too many compan ies release large batches of work to their shop floor
processes, wh ich causes severa l problems:

• T here is no sense o f takt time (no "takt image") and no "pull" co

which the value stream can respond.
• The volume of work performed typ ica ll y occurs 11nevenly over time,
with peaks and valleys t hat cause extra burden on machines, people,
and s upermarkets.
• The situation becorn~ difficult co monitor: "Are we behind or ahead/"
• With a large amount of work released to t he shop floor, each process
in the valu e stream can sh uffie orders. T his increases lead time and
the need to expedite.
• Responding co changes in customer requirements becomes very
complicated, which ca n often be seen in very complex information
flows in current-state drawings.

Establish ing a consistent, or level, production pace creates a predictable

production flow, which by its nature advises you of problems arid enables
you to take quick corrective action. A good place to scan is co regu larly
release only a small, consistenc amounc of produccion instruction (usuall y
between 5- 60 minutes worth) at the pacemaker proce-ss, and simultane-
ously ta ke away a,, equal amount of finishe d goods. We call this practice
a "paced withdrawal."

We call the consistent increment of work che pitch, and ofte n calculate
the pitch increment based on packout cont.1 ine,· quantity (the number of
parts one finished-goods cont.1iner holds), or a multiple or fraction of that
quantity. For example: If your takt time = 30 seconds, and your pack
size~ 20 pieces, then your pitch • 10 min uces (JO sec. x 20 pcs. = 10
minutes). In other words every JO minutes:
a) Give the pacemaker process instruction co produce one pack quantity;
b)Takc awa)' one finished pitch quantity.

So in this case pitch means multiplying your rnkt cime upward to a

fin ished-goods transfer quantity at the pacemaker process. This then
becomes the basic unic of your production schedule for a product family.

PART Ill: WHAT MAKES A VALUE STREAM LEAN ? 45

 What is your management time frame?
• What increment of w ork are you reieasing?
• How often do you know your performance to customer demand?

1 Week

1 Day L--- - - -------': : i i l -

1 Shift

1Hour [:::=J
1 Pitch 0

O ne way ro chink about pitch is as your "management time frame." How ofte n do you
k now your performance co customer dema nd ? If you release a week of work co the noor
at one time, then the answer is probably "once a week." ft is im possible to prod uce co takt
time in this situation. T he re is no "cakt image"! However, if you are scheduling and checking
p roduction every pitch, then you can rapid ly respond to problems and maintain cake time.
J ust as we don't wane co transfer materia l in large batches, we don 't wam to tni nsfe r
production instruction (information) in big batches e ither.

load-leveling box
kanban are responded to from left to right at pitch increment

8 8 ,o 8 20 ~ S"'o 8 00 9 00 8 ,o
one row per 1----+....;'-+-.;:;...~....;'---+-..:;..---,i-;'-+....::;.-+-"--+-=----I

-
product -•
type A

B
==f
typo

C ==i
c:;
t
one column per pitch
here pitch = 10 m in.
kanban
t 9
~

46
 ' !'here ,ire many ways co practice paced withdrawal of small, consistent q uantities of work.
A tool used at some companies to help level borh the mix and volume of production is a load•
k;veling (or hcijunka) box. A load-leveling box has a column of kanban slots for each pitch
interval, and a row of kanban slots for each product type. In rhis system kanban ind icate not
only the q uantity to be produced, but a lso how long it takes co produce chat quantity (based
on takt time). Kanban are pla<--e<l (loaded) into the leveling box in the desired mix sequence
by produce type (see leveling box illustration). The material handler then withdraws those
kanban and brings them to the pacemaker process - one at a time, at the pitch increment.

T he icon for leveling the production pace is the same symbol as for leveling the mix (see
G uideli ne #5 above), because a prerequisite for lean manufacniring is that both the mix and
volume of production be leveled.

an example of ••paced withdrawar•

Customer
Requirement
Drop kanban Pick up next
s _ _ _ _ _ _ _ _ _ __,_ 1\ kanban
at p roces~

''
''
(repeat cycle every pitchl ''
PACEMAKER ''
PROCESS ''
''
''
"'>'" \\
@
--J\q--\
Pick up o ne finished Move finished parts to
p itch qua ntity supermarket or shipping

PART Ill: WHAT MAKES A VALUE STREAM LEAN? 47

 Guide line #7 : D e velop the a bility to make "ev ery part e very d a y "
(then every shift, the n ever y hour or p a lle t or pitc h) In fabrication
pro cesses upstream of the pacema ker process .
By shortening changeover times and runn ing smaller batches in your upstream fabrication
processes, those processes will be able co respond t (> cha nging downstream needs more
qu ick ly. In rum they will requi re even less inventory to be held in their supermarkets. This
applies to both d iscrete-pares man ufacturing and process induwies.

ln general, we note either rhe batch s izes or "EPEx" in the data boxes. EPEx srands for
"every part every ... " afrer which you write a time such as week, day, sh ift, hour, p itch, or
rakt. This describes how frequently a process changes over ro produce a ll pare variations.
An initial goal at many planrs is to make at least "every pan every day"' for high-running
part numbers.

Note:
One method for determining initial batch sizes at fabrication processes is' to base them
on how much time you have left in i he day to make changeovers.

For example, if you have 16 hours available per day and it takes 14.5 hours to run the
daily requirement, then there are 1.5 hours available for ch~nQP.OVP.rs. (A typical target
is approximately 10% of available time to be used for changeovers.) In this case iF the
current changeover time is 15 minutes, then you can make 6 changeovers per day.
To run smaller batches more frequently you will need to reduce the changeover time
and/or improve uptime.

available time -------►~----~

pe r day

48
:THE FU:TURE-S:TA:TE MAI'

• Drawing the Future~State Map

• Your Turn

l I
 current-state

( dravving

future-state
drawing

worlc plan

The Future- State Map

The purpose of value-stream mapping is co highlighc sources of waste and e li minate them
by implementation of a fucurc-scatc value stream that can become a reality wichin a short
period of cime. T he goal is to bu ild a chain of production where the individ ual processes
arc li nked lO thci, c usLum~1(s) t:illu~L Uy c-uttllnuuu:s Ouw ur puJI, amJ t:ac..:11 p1vt:t-:-i:.> gee~ m;
close as possible co produc ing only what its cuscomer(s) need when they need it.

Assuming you are working at an existing facility with an existing product and process,
some of the waste in a value scream wi ll be the result of the product's design, the processing
machi nery already bought, and the remote location of some acciviries. T hese features of the
current state probably can't be changed im mediately. Unless you arc involved in a new
product incroduccion, the first ite ration of your future-scare map should take produce designs,
process tech nologies, and plane locations as given and seek co remove as quickly as possible
all sources of waste nor caused by these featu res. (With the exception of minor purchases
th i11k, "What can we do with what we have?") Subseq uent iternrions can address your
product desig n, technology, and location issues.

We have found that the most useful aid for helping people draw furure-stace maps is the
following list of questions. As you develop your future-state concepts, answer the questions in
roughly the following order. Based on your answers co these questions, mark your furure-
scare ideas directly on your current-state map in red pencil. Once you have worked ouc your
fu rure-state thoughts in this way, you can draw a futu re-state map.

PART IV: THE FUTURE•STATE MAP 49

 KEY QUESTIONS FOR THE FUTURE STATE

....
~ ---~
.,A.
11 What is the takt time. based on the avaflable working time of

21
your downstream processes that are closest to the customer?

wm you build to a Uoished-aoods supermarket from which

the customer puns. or directly to shfRPln97 (The answer to this
question depends on several factors such as customer buying
patterns. the reliability of your processes, and the characteristics
of your product. Building directly to shipping will require either
a reliable. short-lead-time, order-to-delivery stream, or more
safety stock. Fortunately, your order-to-delivery lead time
involves only those processes from the pacemaker process
downstream to delivery.)

3) Where can you use continuous flow processing?

41 Where wm you need to use supermarket pun systems to control

production of upstream processes 7

51 At what single point in the production chain (the #pacemaker

process"! will you schedule production? {Remember that all
material transfers downstream of the pacemaker process need
to occur as a flow.)

61 How will you level the production mix at the pacemaker process?

71 What Increment of work wm you consistentJv r e ~ and take

away at the pacemaker process?

8) What process improvements will be necessary for the value

stream to flow as your future•state design specifies? {This is the
place to note any equipment and procedural improvements that
will be necessary, such as reducing changeover time or improving
machine uptime. We use the kaizen lightning burst
icon to indicate these points in the process.) ....,J-_,.j.,~-::"7
cnanieover

,o
 Drawing the Future-State Map
When we look ai,:ain at the current-state map for Acme's steering
colu mn bracke t, what problems do we note? Perhaps the most
striking arc the latge amounts of inventory, the unconnected
p rocesses (each rroducing co its own schedule) pushi ng their Note:
o utput forward, and rhe lo ng lead time in comparison co the short Try to cycle your pace-
processing time. What can be done about th is? We'll let the key maker process as close
questions be our gu ide. as possible to takt time.
A significant gap between
Question #1 : What is Acme's takt time for the
takt time and cycle time
chosen product family?
indicates the existence
T he takt time calc ulation sca rts with the avaih,blc working time for of production problems
one shift ,n Acme's assembly area, which is 28,800 second s (8 hours). that cause unplanned
F rom this you subu acr any non-working time, which is rwo 10-minute downtime. When you
breaks per shift. Tbc customer demand of 460 units per shift is t hen compensate for production
di vided into the available working time ro give a tak t of 60 seconds. problems by cycling much
faster than takt, the incen-
tive to eliminate those
Available Working Time: 28,800 • 1200 = 27,600 seconds per shift
problems evaporates. If
you cycle faster than takt
Available Working Time 27,600 sec. + 460 units per shift there should be a plan for
Customer Demand closing the gap.

Acme Steering Bracket Assembly Takt nme = 60 seconds

What this takt time number means is that co meet customer de mand
within its available work ti me, Acme needs ro produce a steering
bracket every 60 seconds in its assembly process. This number
includes no time for eq uipment downtime, changeovers between
left-drive a nd ritsl1t-drivc brackets, or for producing scrap. Acme may
decide to cycle assemb ly faster thi,n take, if it cannor im med iately
e limi nate downtime p roblems for example, but the takt t ime is a
refe rence nu mber defined by tile. customer and ca nnot be c hanged
by Acme Stamping.
 Question #2: Should Acme build steering brackets
to a finished-goods supermarket or directly to
shipping?
At Acme, steering brackets arc small (easy co store) parts that have
only cwo varieties. The c ustomer's demand rises and falls somewhat
unpredictably, and Acme is uncertain about the reliability of future-
State changes to be made. So Acme has opted co begin with a finished-
goods supermarket and co move closer co "prod uce co shipping" in
the future.

Acme can use the c ustomer's 30-day forecast co determine the

amount of procJ 11ction capacity needed in the period immediate ly
ahead. (Lean p lanes periodically adjust che n umber of operators in
assembly- iind redistribute the work e lements-co match output to
changes in dema nd.) Acme will dete rmine actual production by means
of kanban coming back upstream to the weld/assembly cell from the
finished-goods supermarket.

Ber.:111se rhe c\1s-rome, buy.t.. in mu lri1,le:s nf ?.0-ht~r.k~r rrny~. rhis is

the simple choice for "kanban sire." T hat is, each tray of 20 left-
drive OJ right-d rive bri,ckecs in the fin ished-goods s uperma rket has
on it one production kanban. As the shipping department wirhdraws
trays from the supermarket to stage chem for deli very, the kanban
from chose trays are senr back co assembly. Each of those kanban
essentially says, "The customer has just consumed 20 left-d rive
(or right-drive) brackets; please make another 20."

Note:
For custom products you may not be able to create a supermarket
of finished goods. (See diagram at bottom of page 43.1

52
Example: Building to a Supermarket
The supermarket schedules assembly (Acme's choice)

Customer
Requirement

r-- ·~ -,
I l__)J-J I

t
ASSEMBLY SHIPPING
G ....

Example: Building Directly to Shipping

Production Control schedules assembly

Customer
Requirement

p
I ASSEMBLY It------►~ SHIPPING
....

PART IV: THE FUTURE-STATE MAP 53

Question #3: Where can Acme introduce continuous flow?
The "operntor-balance chart" below summarizes chc current total cycle times for each process.
The scamping operation C)'cles very quickly (J second per part) and changes over to serve
severaJ product lines. So incorporati ng it into a continuous flow, which would mean slowinp,
its cycle 10 near rnkt rime and dedicaring it co the s teering brac ket product family, is not
practical. Thac would result in a vastly underutili1.ed press and rhe need to buy another
expensive srnmping press for Acme's other produce lines! lt rnakes more sense to run Acm<:'s
stamping press as a batch operation and control its production with a supermarket- based
pull system.

Examin ing the rwo assembly workstations, we notice that their cycle times are not coo far
apart and near the takr time as well. T hese worksracions are also already ded icated co che
sceering brackec p,·oduct fami ly, so continuous flow in assembly certai nl y is a possibility.
The same is true for che two welding workstations, where work could also pass directly
from one welding s tep to the next in a continuous flow.

What prevencs Acme from using continuous flow all tbe way frorn ,,,elding through assembly,
a condition with no inventory (or a maximum of one piece ~t aucomared processes) between
steps? In focr nothing. The lean approach is to place these four processes imrnediarely
adjacent to each ocher (typically in a cellular arrangement), have the operators carry or
pass pares from one process stcr> to the next, and distribute the work e leme nts so rJ1ac
each operator's work concenc is just below cake time.

Acme stamping current cycle times

62s.
takttime
60
soconds

1s.
~
E
.,IV
~

~ ~ ~
D ividing the total weld ing and assembly work con rent by the, takt time(] 87 seconds
divided by 60) reveals that 3.12 operar.ors would be needed co nm welding and asse mbly
in a continuous flow at take. Four operncors would be quite underutilized, but a redistrib-
ution of work e lementS will not be sufficient ro elim inare che need for a fourth operator.

Our next option is to eliminate waste rhrough process kai1.en to bring the work content
u nder the take rime ceiling. A kaizcn target might be to reduce each operator's work
content to 55 seconds or less (or :5: 165 seconds total work content.) If that fails, use of
some overtime may be necessary. With eirhec approach, the fourth operawr a nd the
material handle r who currently moves parts between the isolated processes can be
reassigned to other activities thac actually creace value.

To allow produccion to cake tim e and m ix leveli ng, a pacemaker process should ideally
~
incur little or no changeover ti me and change over very frequen tly. So the left-drive to
right-drive weld ing-fixture changeover times will need to be reduced from rhe current
~
·10 minutes to a few seconds. Focused acrenrion on improving che reliability of the
second spot welder ( perhaps through an improved maintenance approach) w ill also
be needed .

Acme stamping
weld/assembly cell cycle times after process kaizen

takt time ◄-➔------------

60
soconda

~ ~

PAI\T IV: 1•HE FUTURE-STATE MAP 55

 Notice that on this future-state map the four welding and assembly process boxes have
been combined into one process box t0 ind icate the continuous flow. A small schematic
sketch of a cell inside the process box also indicates the cellular manufacturing idea.
d
9

5fAMF'ING
'
WELD+A55Y.

~ 0
~
Takt = 60 sec.
C/T=55sec.
C/0 = RJ
~
Uptime = 100%
25hi~s

58
 State Street
Assembly

18,400 pea/mo
- 12,000 "L''
- 6,400 "R"
fray= 20 pieces
2Shifts
First View of the Future-State Map Showing
Takt Time, Weld/Assem bly Cell, and t h e
Finished-Goods Supermarket

1x
Dally

L SHIPPING

R G Staging

PART IV: THE FUTURE•STATE MAP 67

 Question #4: Where will A c m e need to use supermarket pull
systems?
Acme has decided to produce steering brackets to a finished-goods supermal'ket
(sec Question #2). Two addkional supermarkets-one for srn m ped parts and one for
coils- are necessary to comple te Acme's in-plant value stream for steering brackets.

Stamped Parts
Ideally we might introduce a tiny s camping machine dedicated ro steering brackets-
what we call a "right-sized tool"-and incorporate this mini-press int◊ the weld ing and
assembly continuous flow. Unfortunately, this is nor possible in the immediate future
because machinery of this type does not yet exist. So we need to set up a supermarket
and use withdrawals from that supermarket (pull) to control sramping's production of
left-d rive and right-drive parts.

Pu ll system design begins with c ustomer reql•ircmcnts, and scamp ing's customer here
is the weld/assembly cell. T he cell currently requires approximately 600 I . I I and 320 RH
stamped pares per day. Contai ners for the s!~mpcd parts should be sized to allow close-
to-the-fingertips placemem in the cell (for exa mple, plastic bins that fit into gravity•
feed racks near the operators), not primarily for the convenience of the stamping or
material handling depamnenrs! Small containers allow Acme co keep both LH and RH
stamped pans in the cell at a ll times. This furcher reduces LH-co-R H changeover rime
at the pacemaker process, where very frequent c hangeovers (leveling the mL~) is a key
lean objective.

Each container in the cell-for example , a b in that holds 60 scamped parts, or about
one hour of current steering bracket assembly-will have a withdrawa l kanban with it.
When a cell operator begi ns taking parts our of another h in, its with drawal kanban is
g iven to the material hand ler so chat he/she knows ro go ro the scamping supermarket
and "withdraw" another bin of those parts.

Wlithdrawal kanban trigger the movement of parts. Production kanba11 trigger the
product.ion of parts. Acme can attach a production kanban to each bin of 60 stamped parts
in the supe rmarket. Every rime the material handler removes a bin from the supermarket
a kanban will be sent back co rhe stamping press. T his instructs stamping to produce 60
pares, place them in a bin, and move it to a s pecified location (the "market address") in
che stamping superma,·ket.

68
Now scamping no longer receives a schedule from production control. With mapping icons
the 0ow looks like th is:

,----~·------, ,---- ~
I ~
~ -----, ,, I
• : • I
Ev6ry 60 --.... Every 60 _ __,;I:___ _..,
pieces L piocos WELD + ASSY.
STAMPING
---- - --♦ ---1
R
----- - - ♦

HcMever, you may have already M ticed a ptoblem wirh rhis pull system. Wirh a cycle ti me
of one second per p iece and a cha ngeover time of one hour, stam ping would take one hour
to set up co run only 60 seconds (60 pieces) of production. Unti l c hangeover time on the
srnmping press is grearly reduced , replenishing what is withd rawn from rhe scampi ng
supermarket on a b in-for-bin basis is clearly nm practical.

Due co changeover time, srampi11g needs to produce batches larger than 60 pieces becween
changeovers. With the initial goa l of "every part every day," srnmp ing's target ba tch size
for steering brackc r.s wou ld be approxima te ly 600 LH an d 320 RH pieces (which will still
require c hange()vcr-time reduction). Stamping will keep l .5 days of parts in its supermarket,
one-half day extra co allow for replenish ment delay and some stampi11g problems.

So Acme will use a signal kanban to schedule stamping. In this case the kanban (of1en a metal
triangle) for left- and right-drive pares is brought from che s upermarke r to the scamping press
whe never the number of bins remaini ng in the s upermarket d rops to a trigger ("minimu m")
poi nt. When a triangle kanban arrives at 1he stamping press' scheduling board, ir initiates a
changeover and production of a p redetermined batch size of a specific part. Scamping sti ll
does not receive a schedu le from producrion conrrol.

PART IV: THE FUTURE-STATE MAP 59

 D rawn with icons, t he flow now looks like this:

~atch

:------v-------i
•
I

Every 6 00 LH
or 320 RH
plece,s
I
I
...i---.
.
1----~-----,

L
I

Evory 60
pieces
I

:
WELD+ ASSY.
STAMPING
-------♦ ---c------- ♦
R

t 5day•

The stamped-pares s upermarket, withdrawal and signal ka nban, and kanban flows (dotted
lines) ate drawn on the future-state map.

Coils
To build a pla nt-level lean value stream the future-state map musL also show a third s uper-
marke t at the receiving dock, which holds coils of steel. Even rhoul(h Acme's steel s upplier
is ,,or ready to receive ka nban a nd produce according co chem, Acme can s till attach an
internal withdrawal kanban to every coil and send chose kanban to its own production
control de partment whenever another coil is used. Production control c,, n the n order coils
based on actual usage, ins te ad of based on MRP's best g uess of what furore usage will be.
(MRP may sti ll be used to provide a capacity-planning forecast for the coi l supplier, bur
day-co-day orders should be based on pu ll.)

Once prOd\lCtion conrrol has made the day's orde r for coils, the co,.,.espond ing kanban can
be placed in kanban slots at the receiving dock. T hese indicate the day rhat coils should
a rri ve. If there are kanban still left in yesterday's receiving sloe, then something is wrong
at t he s upplie r.

Currently the steel supplier is shipping coils weekly. By lin.ing up other c ustomers along
a "milk run" deli very, it rnay be possible ro get the necessary amount of s teel on a daily
basis, even if the sreel s upplie r does noth ing to reduce its minimum ba tch size for sli1ti n!!,
coils. Simply moving to daily delivery elim inates 80% o f the inventory ar Acme, while
provid ing smoorh, steady demand for the steel su pplier.

60
 Our Progress So Far
We have now proposed a ce ll of the type many firms have imp lemented in the past few
years, the introduction of pull to control stamping production a nd coil deli very, "every
part every day" in stamping, and institutirtg milk runs for delivery from the raw material
supplier to Acme. By constructing a "before-and -afte r" table (see be low) for the current
state and the futW'e srace so far, we can sec that a large amou nr of waste can be te,noved
through these actions.

These arc big steps forward. However, if the rest of the infom1ation flow ac Acme is not
fundame ntally changed, it will be very difficult co ope rate a lean value stream. So we need
t◊ go back to the customer and rethink the flow of information abourcustome t· desires as it
is sent back co Acme and used the re.

Acme Stamping Lead Time Improvement

Stamped Weld/Assy finished Production Total

Coils
Parts WIP Goods Lead Time Inventory
Turns

Before 5Days 7.6Days 6 .5Days 4 .5 Days 23.6Days 10

So Far 2Days 1.5 Days RJ 4.5Days 8Days :30

PART JV: THE FUTURE,STATE MAP 61

 6-week PRODUCTION
..------I forecast CONTROL
Michigan
Daily
Steel Co. Order

Daily
(Milk Run)

!7atch

:--v-------. .~
-·~-.
•
STAMPING
: I
I
I
I
I

•
-------,
I
I

WELD+ASSY.

Takt = 60 sec.
(at the press) 1.5 days CIT= 55sec.
CIO = f1
Uptime= 100%
2Shi~s

62
 State Street
Assembly

18,400 pcs/mo
- 12,000 "L"
-6,400 "R"
Tray= 20 pieces
2 Shi~s
Second View of the Future-State Map
Showing Stamping and Raw Material
Supermarkets

1x
Daily

-------~
•

SHIPPING
L

R G Staging

PART IV: THE FUTURE-STATE MAP 63


How can we flow information
so that one process will make
only what the next process
needs when it needs it?
Years ago, Toyota discovered
a very different method of
managing the schedule:
stop trying to guess what
the customer will want.
Instead, shorten lead times
within production and install
supermarl<ets with small
inventories of each product
between processes that can-
not be coupled to one another.
These will permit upstream
processes to simply replace in
the supermarket what down-
stream processes have just
withdrawn. Then, instead of
ending customer information
o a centralized MAP system,
hich then sends instructions
o each production activity,
evel the customer orders and
end them only to one place -
ither directly to the pacemaker
rocess, where the requested
roducts will be produced in
ime for shipment, or to a
lnished-goods supermarket,
here the requested products
ill be withdrawn and staged
r shipment.
Currently, the customer is send ing by fax a 90-day forecast, revised
once a month and frozen at 30 days. In addition, it is sending a daily
release during the night by EDI (Eiecttonic Data Interchange- in
plain language, by phone line) to Acme's scheduling computer for
the next day's shipping requircmem. F inally, there are occasional
revisions in shipping requirements on a n emergency basis, These
are sent by phone from the c uscome r\ materia l handling clepamnent
to Acme's shipping deparrment d uring the day as the assembly plane
d iscovers that needed parts are not on-hand for whatever reason.
What happens to the information sent from the cuscomcr once it
reaches Acme? In the current case, the weekly schedule is fed over
the weekend into the computeri~d MRP, which chen sends instruc-
tions by Mond:,y morning 10 each department- sta mping, welding
I & II, and assembly I & 11 -abouc what to make t he coming week.
Then. as additional in formation is received each n ight nnd as each
department reports back periodically co the M RP on what it actually
d id that day (because production does nor go as scheduled), the 9
~
dai ly prod uction schedu les are continually adjusted to bring what
Acme is making into sync wich what the customer wants.
If this sounds complicated ic is because crying to run operations off
of MRP systems doesn't work well. T here is still a frequent need
for humans to override rhe system to avoid shortages at various
d
stages of production. The occasional call from the customer for d
emergency changes in orders requ ires human intcrvenrion as well
~
and upsets the entire production schedule, requiring recalcularion
and retransmission to the processing areas.
Question #5: What single point in the production chain
(the pacemaker process) should Acme schedu le ?
Because alI process steps downstream of the pacemaker process need
to occur in a flow, in the Acme example the scheduling point is cle:irly
the welding/assembly cell. We cannot sched ule any further upstream
(at rhe stamping process) beca use we are plannjng to introduce a pull
system berween stamping and weld/assembly. Th is single scheduling
point will regulate Acme's entire sccering-brackeLvalue scream.
 Question #6: How should Acme leve l the production mix at the
pacemaker process?
When the dail y delivery is made to the assembly pl<1 nt, 30 trays of left-drive brackets (600
p ieces) and 16 trays of righ t-drive brackets (320 pieces) are typically staged and loaded onto
the trnck at one time. If we are not careful, the 46 production kanban removed from these
crays before loading will be sent back to the weld/Hssem bly cell in a batch, as shown on the
future-state map so far. If ch is happens the weld/assembly cell wi ll probably batch-produce
these pares. Tbat is, che cell will produce all 30 rrays of left-drive brackers, and then change
over to make the 16 trays of rig ht-drive brackecs, which would look like th is:

1st Shift 2nd Shift

LLLLLLLLLLLLLLLLLLLLLLLLLLLLLLRRRRRRRRRRRRRRRR

F rom the cell's perspective chis seems co make sense because it minim izes the num ber of
requi red weld-fixture changes. However, from a value-stream perspective batching is the
wrong way co go. Barch-producing bracke ts in assembly will increase the impact of problems.
lc11gd1cu tlu; k;ad ti 111t:•, Hllll 111c11JL that dte :;rnmµ~U-pan:s :;uµc:a1m1rkt:.t has tu I.Jc.: 1c::a<ly tu
meet sudden demand s urges. "Being ready" means keeping more stamped parts inventory
in the supermarket, which aga in increases lead rime, obscures stamping's quali ty p roblems
and, in general, causes all chose wastes associated with overproduction.

Instead, if the weld/assembly-cell levels the mix of brackets it produces evenly over the
shift, then the stamping press (with " shortened setup time) will have plenty of time tO
react to the cell's pulls for left-drive and right-drive pares. lt will have time to replenish
what was ta ken away without the nee d for so much inventory in the stamping supermarket.

With leveling, which requires much more freq uent changeovers, the cell's production mi x
of trays o f brackets would look like this:

1st Shift 2nd Shift

RLLRLLRLLRLLRLLRLLRLLRLLRLLRLLRLLRLLRLLRLLRLLR

I
I PART IV: THE FUTURE-STATE MAP 65
 -
-
-·-
Note:
Enabling such frequent changeovers in the cell will probably require keeping all fixtures and
component varieties ready, near the operator's fingertips. However, when all components
--
are kept on the line you may need some fail-safe devices to prevent the wrong ones from
being assembled. --
Here the benefit of a value-stream perspective should become evident to you. By raking a few
pains to level the production mix ac the pace maker process, which may seem unnatural at
t.hat location, the e ntire value strea m will show improved lead time, quality, and cost. A$ you
--
can imagine, chese bene fics are amplified greacly in value streams chat are longer and more
complex than otrr Acme examp le. --
How can we e nsure thac kanban coming back to the weld/assembly cell, which are rhe
production instructions, come back in a seque nce th at levels the mix of products over the
sh ift? At Acme there are cwo places where rhe batch of kanban can be intercepted and chis
-
--
leveling can take place. (We'll assume that Acme has decided to use a load-leveling box to
help maintain a level production mix, paced withdrawal, and genuine pu ll.)

Option A. Production comrol can place withdrawal (" move") kanban corresponding co che
customer order in a load -leveling box near the shipping dock in a mixed, left-drive/right-drive
--
sequence. A materia l handler then r,ulls these kanhan out of the leveling box one-by-one ~t
-
the pitch incre me nt (20 minutes in chis case), and moves trays of b rackets from che fi nished
goods s upermarket t◊ the staging area one-by-one accord ing tO the withdrawal kanb:1n.
--
PRODUCTION
COl'ITROL
--
Batch of Kanban Option A --
One Kanban
at a lime
--
-
-
--
66
--
As each tray is pulled from the supermarket, the production kanban on those trays are brought
back to the cell in rime incremc□ts and a left-drive/righc-drive paccern that mirrors exactly the
mix and pitch incre ment char production control had set up. (This leveling opt ion is the one
shown in Acme's completed future-state map.)

Option B. Production control can send today's customer orckr to the material handler, who
pulls all the corresponding trays out of the finished-goods market at once and stages them
for shipment. Pulling the trays produces a batch of production kanban, which are placed in
a load-leveling box near the cell, in a mixed, left-drive/right-drive seq uence. T he weld/
assembly material handler then pulls production kanban out of che leveling device one at
a rime at the p itch increment and as a result assembly pruclucC$ in a left-drive/right-d ri ve
mixed pattern.

The drawback of Option 13 compared to Option A is chac in B an e ntire batc h of finished

goods is moved to shipping at once. L,ean manufacturing srrives to avoid or minimize hatching
as much as possible, always getting closer and close, co contin uous flow. Also, if Acme is
someday able to shrink its finished-goods s upermarket ro less than one day of invenro,-y,
withdrawing a one-day quantity all ar once will not be possible. However, Oprion A does
re quire someone co re peatedly move one cray ar a time (at the p itch increment) not only
from the weld/assembly cell to the fi nished-goods supermarket buc a lso from finished
goods co the s hipping dock.

Batch of Kanban

PRODucnoN
CONTROL
One Kanb~n
at a Time
I
I Dally Order I
Option B
~-IOXOX !-- ~ o I

\ I
I
20
§ 20
A

•
I
I I

WELD+ASSY.
SHll'PING

PART IV; T!iE FUTURE-STATE MAP 67

 Question #7: What consistent increment of work should Acme
release and take away at the pacemaker process?
How will Acme provide cakt image co the weld/assembly cell, and how frequen tly will it
check production there? Returning all 46 kanban (two sh ifts worth) co the cell ac once would
p rovide n<> take image to the cell. Batching the volume of work instruction like this musr be
avoided. A natural increment of welding/assembly work in Acme's case is the 60-secor\d takt
time x 20 pieces per cray - 20 minutes. T his is the steering-bracket pitch, which corresponds
to one kanban for one tray of 20 steering bracketS.

Does chis mean that every 20 m i,1 u1es someone walks over ro the weld/assembly cell and
asks, "How are things going?" Not exactly. What chis pitch means is t hat Ac me will practice
paced release of work instruction, one kanban at a time, and paced withdrawa l of finished
goods at itS weld/assembly cell.

Each column in Acme's s teering-bracket load-leveling box represencs a 20-minute pitch

increment. T he cwo rows are designated for left-drive and right-drive kanban. Every 20
minutes, a material handler brings the nexc kanban (the nexc increment of work) co the weld/
assembly cell and moves the j ust-fin ishe d tray of bracketS to the finished-goods area. If a tray
is nor finished at the 20-m inute pitch increment, then Acme knows there is a production
pmblern (for example, a problem with the spot-weld ing equipment) that needs atte ntion.

Acme load-leveling box for steering brackets

weld/assembly cell gets kanban from left to right at pitch increment

shift 1 ., .,.. .,.. 8 B"' s•• 9•• ....

&hift 2 4 42• 440 6 5 .. 5 .. s•• &"'
one row per

-
product
type Loft
Drive

Right
Drive

t
one column
t
kanban
per pitch increment
pitch = 20 min.

GS
 Question #8: What process improvements will be necessary for
Acme's value stream to flow as the future •state design describes?
Ach ieving the materi~I and information flows we e nvision for Acme Stamping requ ires the
following process improvements:

• Reduction in changeover time and batch sizes at the stamping p ress, to allow faster
response to downstream usage. T he goals are "every part every day" and 1ben "every
part every shift."

• 8 limina1ion of the long time ( 10 minutes) required to change between left•drive and
right-drive fixtures in weld ing, to make possible conti nuous flow and mixed production
from welding through assembly.

• Improvement in on-demand uptime of che second spot-weld machine, as i1 will now be

tied to other processes in a continuous flow.

• Elimination of waste in the weld/assembly cell, to reduce cotal work content down to
165 seconds or less (which allows use of 3 operators at the cu rrent demand level).

We mark these items on our future-state m~p witb the kaizcn lightning burst icon.

We shou ld also figure out how rn user.he existing stamping tcch noloi,,y- dcsignecl to produce
stampings in much higher volume than the c ustomer for this product desires- in a less
wastefu l way. The secret here is to have the stamping press, which also stamps parts for
other product families in the plant, make smaller batches of the two pares our value stream
needs and make them more freque ntly. T his will require addirional reduction of the
changeover t ime.

In fact, the methods for reducing setup times on a stamping press are well known :1nd a
reduction in time ro less tha n 10 minutes can be achieved quickly. With that, we can imagine
the press making on ly about 300 left-drive stampings and 160 right-drive stampings (the
per-shift p roduction need); then producing parts for other value streams; then making
more lefts and righrs on the next shift.

8PE will now equal eve,·y part every shift! This way the amount of inventory stored between
the stamping p rocess and the weld/assembly cell would be reduced by about 85%.
I
We can now draw Acme's complete future-state value-scream map,
with information flows, material flows, a nd kaizen needs specified. GO LOOK!

I
I PART IV: THE FUTURE-STATE M AP 69
 PRODUCTION
6-week
CONTROL
Forecast
Michigan
Steel Co. Daily
Order

Daily Order

Daily
(Milk Run)

.~.
1,atch

.--------
I
I
I

•:--v------i -·~--
STAMPING
:
I
I
I

WELD+ A55Y.

o,angeover

EPE =15hih Takt = 60 sec.

(at the press) 1.5days CIT= 55 sec.
C/0 < 10 min.
C/0 = /if
Uptime= 100%
25hihs

1.5 days 1.5days

1 sec I 165sec

70
 ~ - - - - --11 90/60/'30-day
_ _ _ __\~ . Forecasts
I
- -- - - - - - -1
r_

6tate 6treet
Assembly
Daily
Order
18,400 pcs/mo
- 12,000 "L"
- 6,400 "R"
Tray= 20 pieces
2Shi~s

oxox 1x
Daily

~
~
r, - - - - - ~

' ------ ~ --; I

I
I I
I I
I I
SHIPPING
.•.•- L~----•.,___S_t-ag-i-ng--~
----------• R

2days
Acme Stamping
Steering Bracket
Future State

Production
Lead lime = 5 days
2days
Proce,;;sing
lime = 166 ,;;ec.

PART IV: THE FUTURE•STATE MAP 71

 The beauty of asking question #8 !:1st is that your process-improve-
ment efforts become subordinate to the overall value-stream design,
as opposed to vague, stand-alone improvement activities.'leams can
now be dispatched co work on these process improvements with a
clea,· unde rstanding of why they arc making chese improvements.

However, be s ure tO kick off these improvement projects by creating

a "pull" for che improvements. That is, instead of "pushing" a team
t(> reduce setup time on the stamping press, begin instead by stating

chat in 30 days che batch sizes on che stamping press will be ceduced
co 300 & 160 pieces. This creates a sense of urgency about ma king
che process improvement. Likewise, don't simply send a team co
eliminate the weld-fixture changeover time a11d wait for chem co be
finished. Begin by stating chat in 14 days 1he welding and assembl y
seeps will be placed into a continuous flow orientation.

Summing Up
When we compare the s ummary scaciscics for Acme's current state
and its future state, the resulcs arc quite striki ng. In particular, due
to leveli ng production in ics weld/assembly cell and developing the
ability to scamp every part every shift, Acme can further reduce the
amount of coils and scamped pares held in supcrmarke1s. Of course,
chis put.~ great pressure on maintaining equipment reliability and
pcediccabili1y of production to takt.

Wich che shonened production lead time through its shop Ooor,
the pacemaker process operating consistently co cake time, a nd fast
response co problems, Acme can comfonably reduce che amou nt of
finished goods it holds to cwo days. (If Acme's customer were co
level ics schedu le, this finished-goods inventory cou ld be reduced
even funher.)

Compared wi1h 1he interim irnp1ovcrncnts shown in 1he table,

leveling production at Acme has further reduced production lead
time by another 3.5 days and nearly doubled inventory t urns.

72
 Acme Stamping Lead-Time Improvement

Stamped Weld/Assy Finished Production Total

Coils Parts Goods Lead Time Inventory
WIP
Turns

Before 5Days 7.6 Days 6.5 Days 4.5Days 2:3.6 Days 10

Continuous
Flow&Pull 2Days 1.5 Days RJ 4 .5Days 8Days :30

With 1.5 Day,;; 1 Day RJ 2D.ays 5Days 48

Leveling

PART IV: THE FUTURE-STATE MAP 73

 Your Turn
Before you head off co create the fuwre state for your own value
streams-which we hope :ind trust you will do very shortly- you
may need some practice. We invite you to take another clean sheet
of paper and draw a fu mre•State value-stream map for the ''TWI
Industries" example on page 31 you used 10 prnctice creating a
c urrent-state value stream. You can compare your future-state map
with our map by turning to Appendix C.

Remembe r that what is "right" in terms of a value-stream map is one

that permits you to quickly remove root causes of waste. T herefore,
please take our map and mapping tec hniques as a provocative
suggestion and revise the basic ideas to s ui t your particular needs.

74
. ..

• Breaking Implementation Into Steps

• The Value-Stream Plan

A c hieving the Future State
Value-stream mapping is only a cool. Unless you achieve the furut'e scare chat you have
drawn- and achieve parts of it within a short period of time--your value-stream maps are
nearly worthless.

This final section of Learning to See discusses developing and using a yearly value-stream
p lan and ends with some management guide lines for developing lean value screams.

THE PLAN FOR ACHIEVING YOUR FUTURE-STATE VALUE

STREAM CAN BE A COMPACT DOCUMENT THAT INCLUDES
THE FOLLOWING ITEMS:

1) Future-state map

2) Any detailed process-level maps or layouts that are necessary

3) A yearly value-stream plan

PART V: ACHIEVING THE FUTURE STATE 75

 Breaking Implementation into Steps
A value-stream map looks at the entire flow through your faciliry, as
opposed tO only ind ividual processing areas, and in most cases it will
nor be possible ro implement your e ntire future-state concept at
once. T he re is coo muc h to do! So it is the value-stream manager's
responsibili ty to break implementation into steps.

Perhaps the most important point about your furure-srate implemen-

racion plan is not to think of it as implementing a series of techniques,
hut to envis ion it as a process of b uilding a series of connected Oows
for a family of products. To help you do this. try to think about
((value~strcam loops.,.

D ivide your future-state value-scream map int() segments or loops,

as described be low and shown at right:

The pacemaker loop: The pacemaker loop e ncompasses the flow of

material and information between your customer and your pacemaker
prQCess. This is the;: most downstream loop in your facility, and how
you manage this loop impacts all th.c upstream processes in rhat
va lue stream.

Additional loops: Upstream of the pacemaker loop there are materia l-

now and information-flow loops between pulls. T hat is, each pull-
system supermarke r in your value stream usually corresponds with
the end of another loop.

You can c ircle these loops on you r future -stare map, 10 he lp you see
rhe flow segmencs that make up your val ue stream. These loops are
an excellent way co break your future -state implementation effort
in ro managea,ble p ieces.

In the Acme Stamping future-state map there are rhree loops-

pacemaker, stamping, a nd supplier-shown on page 79. With these
three loops in mind, Acme's steeri ng-bracket va lue-stream manager
can break implementation into steps by loop, which might look
like page 78.

16
 L.
\)
E

[
Cl 1
'-'
\!\
:,
CJ
~ -- _J
Q.
0

.. 1_!\
0
..I
G)
~
(ll
E
G) 3:
Cl
(J
(ll u::
/, 0.

\
z
0 _,
//
Ill
Q.
1
'
0
~~ ... 0
Q.
0 3:
ul- ..I C
::,z 0 O IL
co G) ..I
~u a.E E
0.

-L (ll
(ll G)
x~
w (I)
I
C G)
~ .2 \ 1 \
~
/ Q. Q.
0 0
0 0
..I
..
..I :]:
Cl
cr:
·-
G)

Q.
I
Q.
:J
(I) j

J/
\ 1 \
',,

PART V: ACHIEVING THE FUTURE STATE 77

 Acme Future-State Implementation Steps

Loop 1: Pacemaker loop

Objectives:
• develop conrinuous flow from weld rhrough assembly (cell)
• kaizen work elemenrs to reduce coral cycle cime ro 165 seconds or less
• eliminacc weld-fixture changeover time
• improve uptime on welder :/2 co 100%
• develop pull system wirh finished-goods supermarker (eliminate sched ules)
• develop material-handler rouces between the s upennarketS and che cell

Goals:
• only 2 days of finished-goods invemory in supermarkec
• no in ven tory between workstations
• operace the cell wich 3 peop le (ac c utrcnc demand race)

Loop 2: Stamping loop

Objectives:
• establish p ull system wich scamped-pares supermarket (elimi nate
scamping sched ule)
• reduce scamping batch .s i,.es to 300 (LI I) •nd 160 (RH)
• reduce scamping changeover cime 10 less chan 10 minutes

Goals:
• only I day of stamped-bracket inventory in s upermarket
• bacch sizes of 300 & I 60 pieces between changeovers

Loop 3: Coil-supplier loop

Objectives:
• develop pull system with steel-coil s upermarket
• incroducc daily coil delivery

Goal:
• only LS days of coi l in vcncory in supermarket

78
 ..◄-------".,, ""'''-"••<
◄_j Dal~y ~
-.:-
I 6 •wc~k • PROD/ Ci!ON
co l<:OL ◄
◄
3.--
ro.,ii;7 '
9
: %~~-::.·~,
~----------
_J _ 5't,a~.e-etrr..ct
'osemS:,y
I
~
,f I '-._
. I
•
o,i:," I-· - -
\, , I
s r

l
\\ I L , .. ' D..til'v Orasr
'\\ upp ,er oop ,•'
p "-

. \\. @J
I
F. :°~0
::,11i• h . -·!-°a
I
(~-Vlk~

-
.,.'
l- .- ~
v,~h
\ -~ '-"-- ~

~
loxoxl
Pacemaker Loop
'~
B'.,7-7'>.. \
\\ J_ ,------, 20 ;3
' Cc..,.:.&
\
/ I
, /.;

\ \ 11 ,-~·,?----~ _f~~:?i - ,------~'Q ------l~·- ~-- 1/

•
//
<. 7
V
I
1
I

:
~ ..:.L#f
\/ :'\ ' I I
I
I I (/
I'
I /,
WG.l.J" /./6'::' - ' I

~
&fdP!-'JNG
il/
':l ----♦ l
,;:
t:,:
;;;
& l~ ing LL ,_
~l
------- ------ ♦
R
5~glr~

$ EPEX = 15hifT. ...,i«;.,..OO!'!eo.

z
C> C/0 < !V m.n CIO = 1' t)',f.t•
....
:c
y1z_!'o'U'
~~n:e, c.i +OOA.
m
251!irt-.;.
~
C
:,,
m Prt-,h.,c+vltJi' s d.-ip1
-=,;,e Acme's Value-
--::=-~~
·~-
(/) .2.
1.5~:t~ 2d!Z._'$.
~ ~.;:;,·'Srl~ Stream Loops
;;l '✓.:iluc•
Ad.;!6d -= l·SG ~
~tr-

.,...
 The Value-S t ream Plan
You r future-srace map s hows where you wam w go. Now you need w create one more
sheec: a yead y valt1e-stream plan. This plan shows:
• Exactly what you plan co do by when, seep-by-seep.
• Measurable goals.
• Clear checkpoints wirh real dead lines and named reviewer(s).

The first question chat usuarly arises in planning implementation is, "In what order should we
implement?" or "Where do we starr?" We suggest you answer these questions by considering
che loops in your future-state value scream.

1b pick a starting point you can look for loops:

• Where che pcocess is well-u nderstood by your people.
• Where the likelihood of success is high (co build momenrnm).
• W he re you can predict big bang for die buck (but beware, th is sometimes leads
you to areas that have many major problems co be solved, which can lead 10
conflicts wich the previous criteria).

If you have highlighted the loops on your future-state map, you can number chem in order
of your implementation plan. (Use a pencil, because plans do change!) One effective strategy
is co begin iinplcincntation in your do"vnstrcam "paccmak cru loop and ci:ansition up~~ream
as necessary. The pacemaker loop, being closest ro the final customer, aces as che inte rnal
"customer" and controls demand in upstream loops. As the flow in the pacemaker becomes
lean and consistent it will reveal upstream problems that need acrencion. However, the
"movi ng upstream" strateg)' docs noc preclude s imultaneously implementing your future -
state objectives in more rhan one loop of the value sucam. For example, we have frequently
opted to begin working on batch-size reduction and pull in an upstream fabrication loop
while we are still fine-tuning continuous flow a nd introducing leveling in the downstrea m
pac<:makcr loop.

Within a value-sueam loop, a sequence that your value-stream improvements may generally
follow is one chat mirrors the key questions for furure-scare design listed on page j Q,

Specifically, improvements to a loop often follow thi.s pattern:

I. Develop a continuous flow that operates based on take rime
2. Establish a puU system co control productio,1
3. introduce leveling
4. Practice kaizen co continually elim inate waste, reduce batch sizes, s hrink
supermarkets, and extend che cange of concinuottS flow

80
 Natumlly you'll find that this sequence will va ry from case to case and the distinction
between the steps may b lu r to the degree that they are rea lly happening at the same time.
Even so, it is beneficial for you co have t his general sequemial model in your mi nd as you
proceed, recognizing that th.csc implementation objectives are b uilding on each othe r.

Why the implementation sequence listed ahove? First, continuous flow g ives you the
biggest bang for the buck in terms of elimi nating wasce a nd shortening lead time. It's
also the simplest area to begin wo,·king on. (Don't set up a pu ll system if you can create
continuous flow.}

Continuous now with minimum waste means e li minating overprod uction, which rhen means
rhat you musr (and can) standardize your work e lemencs so that production is consistent and
predictable to your take time. You'll then need pull as a means of giving production instruction
to the now (and to k ick off the entire production sequence from the customer in che case of
rhe pacemaker loop). Finally, you'll need leveling to ach ieve a lean !low anytime you have
multiple productS, simply because lack of leveli ng will mean that you are scill batching your
builds of different products. Even if you produce only one product, you sti ll need co level
rbe production volume.

T har leaves the last key question, "What process improvements will be necessary for
the value scream to flow as your future-state design s pecifics?" Successfully developing
continuous now, pull, and levehng will demand varying levels of preparatory work. Por
example, before you can reach a high degree of levelizarion you will have co gain the
ability to execute qu ick changeovers. Or, before you can expect your assembly cells w
operate effectively at take rime, you wi ll need a high level of first-time-through ca1)ability
and machine reliab ili ty. O r your order-e ntry process needs to be changed.

He re is the source of another potential Catch-22: Which comes first, rhese preparatory
process improvements or introduci ng continuous flow, pu ll , and leveli ng improvements?
Cen ain ly, they both need to move hand-in-ha od to some degree. However, we have found
that when in doubt, move forward on making flow improvements and Ice those flow improve-
mentS drive the implementation of s upporting process improvements. Otherwise, you ca n
work forever simply scriving- and waiting-to attain a h igh level of process capability.
Pitfall w avoid: Don't get stuck in process improvement!

Once you have a sense for the basic orde r in which you want co implement the elements of
)'0llr futu re-scare vision, rhe value-stream manager needs co write th em down as the yearly
value-stream p lan. The format for th e value-stream plan is shown wich the Acme Stamping
example on the next page. The plan will be fam iliar w you if you have experience with
policy dep loyment, or it may look like a variation of a Gantt chart.

PART V: ACHIEVING THE FUTURE STATE 81

 DATE:
JAN 2, 2003
.
FACILITY MANAGER: Barb Smith
YEARLY
V. S. MANAGER: Paul Doe

Product Family v.s. Va lue-Stream GOAL

2003
Business Objective LOOP Objective (measurable)

1 2 3 4

Improve • continuous flow from zero wip

profitability in weld·aseeml:>ly
steering brackets 1 *kaizen to 165 sec. < 1655 cit -
pace-
maker *eliminate weld clo < 30s clo -
~
'
• uptime welder #2 100%
- .
' finished goods pull
•materia l handler
routes
2days FG +
pull schedule
--
.

•stamping pull 1 day inventory

+ pull schedule

2
stamping

• stamping l:>atchsize
changeover 300/160 pcs
clo < 10 min

3
supplier
• pull for coils wit h daily delivery &
dally delivery ~ 1.5 days of coils
at press

82
 Signatures
PLANT UNION ENGINEERING MAINTENANCE
MANAGER
VALUE-STREAM PLAN

REVIEW
PERSON IN RELATED SCHEDULE
MONTHLY SCHEDULE CHARGE INDIVIDUALS
& D EPTS
R EV IEWER DATE

5 6 7 8 9 10 11 12

.
•

PRODUCT FAMILY: Steering Brackets

PART V; ACHIEVING THE FUTURE STATE 83

 As you mighr imagine, the key to making your yearly value-scream p)ao useful is incorporating
it into your normal business process, partic ularly in the budgeting process. No money
approved without a value-strea m plan! This will make ir easier for both sides-requester
and approver- once everyone gees used to using the tool (value-stream mapping in its
"communication tool'' role).

You can also use the yc,irly value-stream plan to evaluate manufocturing performance
quarterly or monthly as a key method of performance ceview: "Bring me your futu re-state
value-scream map along with an honest progress evaluation every monrh." We have inc.luded
an example value-scream review form on the next page. Prior co a review, the valuc-sm:am
manager should honestly evaluate each implemenrn,ion objective as on target (0), slighcly
be hind (.0.), or unsuccessful (X ).

T he key to effective progress review is co "manage the exceptions." That is, during the
review do nor focus first on what has been accomplished. Focus inste.~d on the X items and,
if e nough meeting time remains, the A ite ms. For each of these behind-schedule items the
manager should ask the value-stream manager, "What do you need in order to get this item
on uack?" Then the manager can pro,,ide cargeted support as required.

The idea behind >J value-stream review is something we call "plan-based trial-and-error,"
which refe rs to blending two usually opposite thought processes: "planning" and "trial-and- 9
error." "7'hal-a nd-error" indicates acceptance that all will not go according ro plan and that,
9
in face, we can learn useful things from failuxes. Bue "pla n-based" indicates that the typical
la issez-faire, "let's try chis next" scyle of trial-and-e rror is not acceptable. 'l'here must be an
c
effort 10 strive coward pla n accomplish me nc, even as we recognir,c th.ac the value-stream plan
will c hange and evolve annually. Deviations from the p lan are questioned rigorously and
accepted only after trial has shown the plan co be less than optimal. T his provides the
discipline necessary co achieve improvement.

You may or may nor wane co incorporate the quarterly value-stream review into your personnel
appraisa l process. The plus of incorporating it i1lcludes the facts char it is ( I) a good cool for
evaluating performance, and (2) nothing gives teeth 10 a process more than linking it to
appraisals- and cherefore paychecks. T he negative here, though, is that a key co mi1king
the continual current-state/future-state revision cycle wol'I, is ensuring honest, aggressi,,e
goal-setting and evalua tion. When cied co personnel performance appraisals, future-state
plans rend to become less aggressive and progress evaluations more generous.

A final suggestion is to conduct you r va lue-stream reviews on the shop floor wh ile walking
the flow.

84
 DATE:
SIGNATURES
•ACtUTY MANAGE R:
VALUE-STREAM REVIEW
V .$ M.ANAG-lil't!

PLANT~LEVEL v.s. OBJECTIVE & PROGRESS EVALUATION REMAINING PROBLEMS POINTS AND IDEAS FOR COMING
OBJECTIVE LOOP MEASURABLE CONDITIONS YEAR'S OBJECTIVES
G·OAL

)1
~
,,:
l!;
:,:
iii
<
z
Cl
=t
m
..,,
~
C
ill
(/)

~
0: SUCCESS L:::::,.: LIMITED SUCCESS x.. = UNSUCCESSFUL I
PRODUCT FAMILY:

..
.,
 Value- Stre am Improvement is
Management's Responsibility
As noted earlier in chis workbook, value-stream improvemenc is
How can we
make this work? p rimarily a management responsibili ty. Managemenc has co under-
stand chat its role is to sec the overall flow, develop a vision of a n
improved lean flow for the furure, and lead ics im plemencarion. You
can't delegate it. You can ask rhe from lines ro work on eliminating
waste, but only management has che perspective co see the cotal flow
as it curs across depamn e ncal and functional boundaries. F' rom our
combined expe,·ienoe with many companies in a range of industries
over the past 15 years we can st.ate that the following are needed :

• Constant efforts co e li minate overproduction. If you eliminate

overproduction, you wi ll have great flow.

• A firm conviction that lean p rinc iples can be adapted co work in

your setting, coupled with a willingness to try, fail, and le arn.

What you don't often hear a bout is the process of tria I-and -error chat
' H1iid 1i Olmo ,,vcnc th rough as he :sought co elimi nate overproduccioo
ac che you ng Toyota Mocor Corporation . Many e rrors s imply go with
che territory when implementing change in long-estab lished mass
production practices. If you do it right, each approach will be nearer
co the carger and will add to your understanding. Such reiteration is
a normal part of a ny lean implementation effort a nd s uccess will be
achieved by those who have the determination ro personally work
through the obsrncles.

• Ma nagement needs to dedicate time and to really learn chis scuff

for chemsclves-learn ic to chc point rhar they can actually reach ic.
And then rhey need co actually teach it, not primarily in rhe classroom
(alchough there is a place for that), bur in their daily imeracrions
with their scarf.

86
At whatever level, from CEO to plant floor s upervisor, che words
and deeds of managers must be pushing t he creation of a lean value
scream. le simply won't work if ic's relegated to a few minutes ar the
weekly staff meeting. It's got to be part and parcel of every day's
acrivities. Practice the mapping concept presented here to the poi nt
that it becomes an instinctive means of communicacion.

• You'll need a way co get people co follow your lead, wichouc always
waiting for you to lead them. Begin by focusing you r organization on
a relatively small numbe r of specific ta rgets (e.g., manage by value-
stream maps). You may recognize chis process as policy deployment.

Eventually, you should evolve to policy management, which is a

much more dynamic process where lower levels of the organization
take part in formulating policy as well as carrying it out. As your lean
organization marures, you' ll find thac policy begins co emerge from
interaction between levels of the organization, rather than simply
e manating from above to be d eployed below.

• Close-ro-rhe-opcration support, nor "sel f-directed work reams."

Close-to-cile-operacion support means (I) all "ind irecc" operations
are considered "support" for direct operations and (2) the work of
s upport operacions (production control, supervision, team leaders,
material handling, maincenance, problem response) must be tied to
the cakt and pitch times of the direct value-adding operations.

If we are asking operators to work at takt cime, we need co be able

to manage within the same takt time framework. For example, ask
yourself, "Can my s upport organization (maintenance, etc.) respond
co problems encountered by production operators with in take time?''
If the answer is "no" (and it almost always is), chen your organization
isn't ready for production to operate at cake time.

PART V: ACHIEVING THE FUTURE STATE 87

We are continually amazed • C hanging the organizationol focus rrom departments to prod uct team,.
at the responses we get when
we walk through firms, pick • A "value-scream manager" whose job is ro lead the people operating
a product at random, and ask d1e process, not just in manu facturing but in all business fu nctions,
the simple question: "Who is and to take responsibility for the cost, quality, and delivery of the
responsible for the cost, quality, product in the p resent State while mappi ng and lead ing implemen-
and on-time delivery of this tation of the future state.
product from start to finish?"
• Lean manufacturing speciaJists who can hel1, value-s tream managers
The usual answer is: see waste and introd uce the appropriate practices needed to remove
"Well, Material Handling is its sou rce;;.
responsible for moving parts
between the production steps; In the beginning most value-stream managers and their team members
the Stamping Department will benefit from a hie of technical assistance in sharpening their vision,
manager is responsible for imroducing and relin ing continuous tlO\v, instirnting quick changeove,·
meeting his schedule; the of eq ui pment, installing a pu ll system, leveling the schedule, and so
Welding Department manager on. H owever, the lean specialists must be the coaches rather than
is responsible for meeting her
the acttoal implementers, with che clear objective of transferring all
schedule; the Shipping
of cheir le an experience to the value-srream manager and others as
Department is responsible for
quickly as possible.
getting the product shipped on
time; Production Control is
Make s ure your lean promotion grou p is actua lly on the plane floor,
responsible for setting every
leading rhe changes, e m brac ing a "hands-on" approach co problem
department's schedule; the
solving, while pay ing atte ntion to the accual needs of the organization
Quality Assurance Department
a nd c ustomer. Also make s ure that your lea n promotion group helps
head is responsible for ensuring
all business fu nctions, not just manufacturi ng.
that defects are below the
maximum acceptable level
and ... "
Sen ior Man agement
In short, no one is responsible.

/ I \
Value-Stream Va lue-Strea m Value -Stream Le a n
Ma na ge r Manager Ma nage r Specialists

Lean specialists support t he implementers and report to senior management.

 Note:
The on ly way we have found to actually learn lean methods is to apply the techniques
ourselves hands-on w ith a bit of coaching. We are certain that this is true for the great
majority of managers. The tragedy is that so many managers want to retain a consultant
expert to solve their immediate problem for them without need for their active involvement.
Of course they discover rhal they are never able to solve their own problems themselves
and often fall into a spiral of consultant dependency. Just say no!

• A new sec of performance measures for produce teams focused on reducin11, lead time,
space, efforr, defects, and missed deliveries, rather than the traditional financial metrics
of asset utilization and burden absorption.

Measures should provide meaningfu l information for managing a lean operation, and musr
not be co11nterprod uctive co lean goals. Unfortunately, our traditional measurement systems
do us little good when it comes to trying to manage a lean value stream. After all-remem-
bering that our concern should be improved flow-what impact on now ca,, we expect from
measures that emphasize such things as equ ipment utilization or labor and overhead?
Answer: negative impact!

Unfortunately, few of us are in the position of being able to discard our traditional financial
rnear.ures {in which-the perfect example of a counte rproductive mcas u,·c-invcntory is an
asset!). However, what we can do is determine that: those measures rhar provide us wirh
little co no help in running our plantS do nor interfere with how we manage our plants.

Lean measures for manufacturing performance shou ld adhere co the followi ng 1>rinciplcs:

Principle 1: Measures should encourage desired behavior by the front li nes.

Principle 2: Measures should provide information for senior managers to make decisions.
Principle 3: Principle J cakes precedence O\ler Pri nciple 2,

That being said, here is a fine lean-manufacturing measure for any plant: "ls our take time-
based production target achieved every day at the pacemaker p rocess?"

I
I PART V: ACHIEVING THE FUTURE STATE 89
.
 CONCLUSION
Clearly, there is no end to rhe "future-becomes-present" cycle. T his should be the hean
of day-co-day management in any organization with a product to sell, whether ic be a good,
a service, or some combinarjon constituting a solution to the cuscome r's problem. As we've
discovered again and again. when you remove sources ofwasce during a cycle you discover
more wasce lurking in che nexc cycle chat c~n be eliminated. The job of lean managers and
thei,· teams is keep this virtuous circle going.

This workbook has focused almost entirely on technical aspects o f introducing a lean value
stream. This is a good place co begin because to be comperirive t he value scream needs to
flow in a way that serves rhe customer with the overall shortest lead rime, lowest cost, highest
quality, and most dependable delivery. It shouldn't be sub-optimized to serve the desires of
individual processes, depa rtments, functions, or people.
I
However, making the tech nica l c ha nges will also "pu ll" the need for c ha nge in the people
I
side of the va lue stream. Adversarial labor/manageme nt relations, for example, will hamper
I im ple mentation efforts and traditional job classifications will nor mesh with a truly lean
ope ration. Current ways o f measu ring pe rformance- led by standard cost accounti ng-
I
will encourage a regression co mass-production methods. And so on.
I
Lean value st,·eoms must be developed with respecc for people. 13ut. ,·e~pect for people shou ln
I nc)t be confused with "respect for old habits. " Developing lean value screams can be hard
I work, often with one step back for every two forward . Developing a lean value stream exposes
sources o f wasce, which means that people in all busi ness fu nctions may have co c hange
I habits. We believe that everyone-management and e mployees- has a role to play in lean
I impleme ntation, and that everyone should feel a benefit from it. These benefits can come i,1
many forms: increased competitiveness of the company, a be tte r working e nvironment, greater
I trust between management a nd e mployees, a nd- not least-a sense o f accom plishment in
I serving your customer.

I Whenever there is a pmduc t for a custome r, there is a value stream. The challenge lies
in seeing ic. Va lue-scream ma ps can be drawn in the sa me way for practically any business
I
activity and expanded up- a nd downstream from your company to spw from "molecules
Cl to c ust0mer." We can't provide an example of a value-stream map for every activity here.
l.l uc we hope Learning to See will spur your thinking abom your own value streams, and
I

, .....,'!?
help you introduce lean value screams that fi t you r industry.
I
TAKE THE ,....
ii
rl
I
II I 91
 About the Authors

Mike Rother
Mike began his career in the manufacturing division ofThysscn AG and
has spent 10 years learning to app ly lean practices through consulti ng at
several different companies-both large and small. Mike also teaches at
the University of Michigan, Depamnenr of Industrial and Ope,·acions
Engineering, and srudies 'Toyota. He finds there is always another level
of lean co practice and understa nd .

John Shook
John has been leaming about lean since 19$3 when he joined 'lbyora to
he lp them transfer its production, cnginccrin~ and management systems
from Japan t0 its oversees affiliates and s upplie rs. Si nce 1994, he has split
his time between directing the University of Michigan, Japan Technology
IV!anagemcnt Program, serving as senior advisor for the Lean Enterprise
Jnst.iruce, and working wlch 0omp-.-nies and individuals to undernrand and
implement lean man ufacturing. And he is ever scudying and learning
about lean .

92.
App e ndix A : V a lue- Stre am Mapping I cons
T he icons and symbols for current• and future-state mapping fall into three categories:
Material Flo"~ Information Flow; and General [cons.

Mate r ial Icons Repr ese nts Notes

IA5~5CY I Manufacturing Process One process box equals an area

of flow. All processes should be
labeled. Also used for departments,
such as Production Control.

XY2
Corporation Outside Sources Used to show customers, suppl iers,
and outside manufacturing processes.

CIO = 30min Data Box Used to record information concerning

35hi~s a manufacturing process, department,
customer, etc.

Inventory Count and time shou ld be noted .

300plcce•
1Day

Truck Shipment Note frequency of shipments.

••••••►
Movement of production Material that is produced and moved
material by PUSH forward before the next process
needs it; usually based on a schedule .

._____> Movement of finished

goods to the customer

Supermarket A controlled inventory o f parts that

is used to schedule production at an
upstream process.

APPENDIX A:. VALUE-STREAM MAPPING ICONS 93

 Material Icons Represents Notes

G Withdrawal Pull of materials, usually from

a supermarket.

Transfer of controlled Indicates a device to limit quantity

max. 20 pieces quantities of material and ensure FIFO flow of material
between processes in between processes. Maximum
-F IFO- a "First-In, First-Out" quantity should be noted.
sequence.

Information Icons Represents Notes

◄ Manual Information flow For example: production schedule

or shippi ng schedule.

◄ s Electronic Information flow For example: via electron ic data

interchange.

Weekly
S<:hedulo Information Describes an information flow.
~
~
,-- 8--- Production Kanban The "one-per-container" kanban .

• ~
I {dotted line indicates Card or device that tel Is a process
kanban path) how m any of what can be produced
and gives perm ission to do so.

,--~ --- Withdrawal l<anban Card or device that instructs the

•
I
material handler to get and transfer
parts (i.e., from a supermarket to the
consuming process).

:--v---- Signal l<anban The "one-per-batch" kanban.

• Signa ls when a reorder point is

reached and another batch needs to
be produced. Used w here supplying
process must produce in batches
because changeovers are required .

94
 l

Information Icons Represents Notes

Sequenced-Pull Ball Gives instruction to immediately

produce a predetermined type and
quantity, typically one unit. A pul l
systern for subassembly processes
without using a su permarket.

Kanban Post Place where kanban are collected

and held for conveyance.

Kanban Arriving
in Batches

Load Leveling Tool to intercept batches of kanban

loxoxl and level the volume and mix of
them over a period of time.

''Go See" Production Adjusting schedules based on

Scheduling checking inventory levels.

General Icons Represents Notes

''Kaizen Lightning Burst" High lights improvement needs at

specific processes that are critical
to achieving the value-stream
vision. Can be used to plan kaizen
workshops.

Buffer or Safety Stock "Buffer" or "safety stock"

must be noted.

Operator Represents a person viewed

from above.

APPENDIX A: VALUE-STREAM MAPPING ICONS 95

 Appendix B: TWI I ndustries Current State

PRODUCTION
Orders CONTROL
Rod + Forging
Suppliers

Daily Priorities

2xper
Month
Shop
Orders Q To Weld-Area
A Supervisor

CUT WELD#1 WELD#2

Rods 5days 19, 1

20days
CIT = 15 second
CIT = Oper 10 sec CIT= Oper 10 sec
C/0 = 15·60 min Mach30sec Mach30sec
Uptime = 100% C/0 = 15 min-1 hr C/0 = 15 min -1 hr
Uptime=90% Uptime=80%
MACHINING 1 Operator 1 Operator

IJCJ
Forgings
20days 4days
Ctr = 30 second
C/0 = 2 hours
20days Uptime=100% 5days 3days 3 days
30 seconds 30 second s
30sec

96
 Orders
\ 60-daysout
OE + Aftermarket
Customers
r------11 Slze adjustment J \
_ 2 weeks out !

Daily
Daily Ship
Schedule

er. DEFLASH

m+
PAINT

OUTSIDE 5UPP1.IER

&2doyo
~
ASSEMBLY

~
SHIPPING

Staging
<Q, 1 5days 6days <Q, 6 4 days

CIT = Oper 10 sec Total CIT=

Mach BO sec 195 sec
CIO = 15 min - 1 hr CIO = 10 min
Uptime= 100% Uptime = 100%
1 Operat or 6 Operators

5days 2days 6days 4 days Li= 48daY$

BO seconds 195 sec. VA = 315~

APPENDIX 8 : TWI INDUSTRIES CURRENT STATE 97

 Appendix C: TWI Industries Future State

PRODUCTION
Orders CONTROL
Rod + Forging
Suppliers

\\
Weekly
loxoxl
Pitch = :30 min.

Cut
Rods
Rode G '-__c_u_r__~

C/0 = 5min.
1day G ------,
CIT= 15 sec. WELD + DEFLASH -------
m _. ----:C,
,+- v----i::r-~-----:fl .______
Forgings G L-----~
MACHINING L - - - - -'
C/T =:39 sec.
C/0 5min
-
- .

~
Uptime=100%
Maohine<i
Forgings 1 Operator
Total CIT= 90 sec.
r:;
7days C/0 = 30 min. 1day It:=
Ctr= 30sec.
~
7days 1day
30sec 90sec
.____ _ lr-- ~

98
 Orders
2 weeks out
OE+ Aftermarket
Customers

Daily

-----------
I max 1200 pcs. PAINT ·--------·
empty daily
ASSEMBLY
.,. SHIPPING

1- FIFO -
L__________
OUTSIDE SUPPLIER

& 1day -FIFO-

---------
~99999~ ~- Staging

TAKT = 45 sec.
5 Operators
C/0 = lo

1day 1 1day I 1 day same dav

I I I~-----~
195sec. I

APPENDIX B: TWI INDUSTRIES FUTURE STATE 99

 Comments to the TWI Industries Future- State Map

TWl's shop n,1or is sw,1mpcd with orders thai were released too
soon and are s huffied and reshufficd co optimize changeovers and
meet the most urgent customer requirements. lnsce,id of releasing
so many orders co the shop Ouor, if T \:VI were m send only 30 minuccs
of work (one pitch) at a time co cl1e the first weldi ng 01)eration and
now it in Fll:0 order (first in. first out) from the re through s hipping,
the lead cime for producing a customer order could be reduced co
under three days (with che middle of those days being ac che outside
painL supf}lier). T his requ ires reduction of changeover ti me in the
welding and dcf1ash operations to five minutes or less so that different
configuration steering arms cou ld be produced nearer co the sequence
in which the cusro,n e r orde rs chem.

Because customer configuration rcquiremcntS for tractor steering

arms vary so widely from orde r to order-nearly to the point of
being custom produces-and th e lead time co replenish an order is
sci ll q uite long, TW I decided chat it would be impractical to hold
fi nis hed sreering arms in a supermarket ac chc downstream end of
ics value st.ream. So T\.VJ ,yiJI need to s.c hedule further 1.tps:uenm. in
th is case at the first weldin~ operation where prod uce variation arises,
and thus have co employ a FIFO flow downstrea m of rh:,c point (see
the FIFO lane discuss.ion on page 42). 13y on ly releasing 30 minutes
worth of work in 30-minute increments at this scbcduling point (the
"pacemaker proccss"- sec page 43) and following FIFO procedures,
TWl can avoid overprod uction and "push" in this long FIFO flow.

TWJ can develop continuous flow through the weld a nd dcflash

operations, which could then be run by one operator who loads and
transfers parts from one au tomatic machine co the next. TWl will need
10 cycle rhe weld/dcflash now faster than the take time of 45 seconds
-at approximately 39 seconds -to leave time for 12 changeovers per
s hift. Because assembly involves no changeover, it c,m cycle ne arer
co takr rime, which allows assembly to ru n wirh five operators.

100
 TWl's 30-minure pitch in this case is based on the SO-piece average
order size and the fact that the weld/dc flash now requires a 5-minu te
c hangeover between orders. With cusmmer dcmn1\d of600 pieces
per shi ft and a cycle time of 39 seconds, there is one hour left over
per shift for the 12 changeovers between pitches. "fh establish the
pitch increment, production conuol wi ll comhine s mall orders and
break down large orde rs in SO-piece increments. Production control
wi ll also inmx luce leveling of the producrion mix, so that the cut-rod
a nd machined-forging s upern1.i rkets upstre,1m can be sma ller. So
TW I will nor produce orde rs in exactly the sequence as they are
recerved, but very close ro it.

With the changes noted above. TWl's customers can now place their
orders on ly two weeks out. Produciion of cut rods and mach ined
forgings, of which there are fewer varieties than finished steering
a m1s, can be conrrolled by superm,1rke1 p ull systems. Likewise,
uncut rods and raw forgi ngs can be ordered b;ised on withdrawals
out of raw-material supermarkets. This e lim in,it.es the i1ecd fot
production control to release orders early to trigger MRP-based raw
material orders.
I
I
I
I
I
I
I
I

•
•
I

APPENDIX 6: TWI INDUSTRIES FUTURE STATE 101

 Lean Enterprise Institute

Continue Your Learning

The Lean Enterprise l nstiture (LE I) has a wide range of learning resources, all with the
practical knowledge you need to sustain a lean transformation:

Learning Materials
Our plain-language books, workbooks, leadership guides, and training materials reflect che
essence of lean thinking- doing. T hey draw on years o r research and real-world experie nce,,
from lean tra nsformations in manufacturing and service organizations to provide tools chat
you can puc to work immed iately.

Education
Facult)' members with extensive implementation cxpe,i.c nce teach you actual applications
with the case studies, work sheets, formulas, and me d1odologies you need for implcmemacion.
Select from courses chat address technical copies, cuJcu re change, coaching, senior manage-
ment's roles, a nd muc h more.

Events
Every t-.'1arch che Lean Transformation Summit explores the latest lean concepts and case
studies, presented by executives and implementers. O ther evencs focus on an issue or
industry, such as starting a lean transformation or implementing lean in healthcare.
Check lean.o rg for derails and co gee first nocicc o f these li rn iced -atcendance events.

lean.org
;\ qu ick and secure sign-up delivers these Online learni ng resources:
• Jim Womack's thoug ht-leading e -letcer delivered monthly to your inbox.
• Use of t he Connection Cen ter co network or benchmark with fellow Lean Thinkers.
• Entr\' co a range o f Forums where you can ask questions or help ochers.
• Access co the Lean Road l\•1ap for c ustomizing and tl'acking a personal learning path.
• Use of the Lean Notebook for saving a11d s hari ng importa nt arricles.
• Firs t notice about LEI events, webinars, and new learning materials.

About the Lean Enterprise Institute

T he Lean Enterprise Institute, Inc. was founded in 1997 by management expert James P.
Womack, Ph.D. , as a nonprofic research, education, publishing, and conferencing company.
As part of its mission co advance lean think ing arou nd the wol'ld , LE I supports the Le an
Global Netwtlrk (leanglobal.org), the Lean Edue;nion Academ ic Net.work (teachinglcan.org),
and r.he Healthcare VaJue Leade rs Network™ (healthcarevaluc leadcrs.org).

102
 MATERIAL FLOW ICONS

I I
MS™~y XYZ
Corporation
C/T=45,;ec.
CIO= 30min
35hifts
2%5crap
&
:300 pieces
1 Day
Man ufacturing Outside
Process Sou rces Data Box Inventory

Truck
••••••► PUSH Finished Goods
-FIFO-
First-In-First-Out
Shipment Arrow 10 Custome r Seqvence Flow

G ENERAL ICO N S
G
- Supermarket Withdrawal

Kaizen
-
Buffer or
Lightning Burst Safety Stock Operator

INFORMATION FLOW ICONS Weekly

Schedule loxoxl
Schedule Load leveling
◄ ◄
s
Manual Electronic
Information Flow Information Flow

V
,--~ --- ,-- 0 --- Signal Kanba11 Kanban Post
' 1

't Withdrawal 't Production

Kanban Kanban
&I
Sequenced-Pull Ball " Go See "
Kanban Arriving Sched11ling
in Batches
Acme Stamping Factory Floor Layout

PRESS DEPT. A & SHIPPING

w w & Q I ~ AieldJ

WELDING DEeT.
~
w »~~s~A A AQ &C-) , 0
9 "'111111
eldinge AAAAAAAA F".
101she d II
w w lnspecllon
Area9
Welded Parts Storage
AAAAAAAA
Goods
Storage 111!1
11111
··1
Bl ill
I llll

Ai□*~i□i 101 l~il

!JIil iii
Steel Coll Stamped Parts • '■
Storage Storage II !JIil Ill 111!1 Ill
111111 ill Ill
I
•••••••••••
~
ASS EMBLY DEPT. (II N ~at] 1!111 111!1 ■ Ill IJI
••••••••••• l~TI
•••••••••••
•••••••••••
•••••• 1Di 1D11D1 I ~I 111111
Ill 111!1 ■ 1!111
Ill Ill

9 = Work in Progress (WIP) in Baskets

Ill = Finished Goods in Trays
Acme Stamping Data Set
Acme Scamping Comp~my pl'Oduces seven1I ,~rnpooem.s fo1·
vehicle assembly ph1ms. T his case concerns one product family,
;1 steel inscrumcnr•panel hrackcr suhasscmhty in two rypes:
ouc cadl fur left-hand ~\nd right-hand drive versions of rhc
s.arne automobi le model. These components are ~cm co rhe
State Street Veh icle Assembly Plant (the cuswrner).
Production Processes
• Acme's prote..•tscs for this producr family involve stamping a
mct.11 part followed by welding a nd subsequent 11.sscmhly.
The components ure then staged & sh ipped to the vehicle
nssembly pl:H1t on :1 dn ily busis,
Process Information
All proc..-esse.'i- oc<.:ur iJl the following order and each piece
goes throug h all processes.
1. Stamping ( l'he press mak<.-s l'""~s for m,iny Acme prudul'IS)
• Aurmnated 200.'fon press wich coil (au(om:1cie) rn1ttcrial feed
• Cycle time: I second (60 pieces per minurc)
-Changeover time: l hour (good piece m good piece)
• Mnch ine wliability: 85%
• Ohserved invenrory:
- .'l da)'S of coils h~fot'e suunping
• 4,600 pieces of' l'y pc "LH" finished siumrings
-2,400 pieces of Type " RH·• finished srnmpings

• Swicc hing hctween 'l i•pe ''l,H" (lcrt-h:tncl drive) ,and 2. Spot-Weld Workstation I (dedicated to this product family)
'] ypc "RH" (right-hand drive) hrackc r.s requires n I-hour - i\fanual process with one oper:1wr
changeover in stamping and .i !{)-minute fixture c.hnngc in - Cycle time: 39 seconds
,hi; welding processes. • C:h,ingeover rime: 10 minutes (fixn11·e change)
• Rcliohility: 100%
• Swel coils nre su1>plied by M ichignn Steel Co. Steel .. Observed inventory:
clcli\•eries co Acme arrive on 'Ihesd;1ys :tnd ·rhursd~)ys. • 1,1 0() pic<•es or'Typc "LH"
• 600 pieces of' lype "R H"
Customer Requirements
• 18,-H>O pieces per month 3. Spot-Weld Workstation II (dedicated to this product family)
• 12,000 per month of Type " LH" - Manual pr()ccss wlrh one open1tot
. 6,400 per month or · fype " RJ I" • Cycle cimc: 4-6 sccond.s
• Customer plant operate..~ on two shifts - Changeover time: 10 minutes (fixcu re chcmge)
• Pc,llcciied rccurnahlc cmy packaging wirh .20 brackcrs in a • Reliabili ty: 80%
tray and up co 10 trnys on a pallet. The cuswmcr orders in - Ohset'v.cd invcntory:
rnulriples of trnys. • 1.600 pieces of 7ype "LI-I''
• One d:iil y sh ipment to the :1ssembly plant by truck. • 850 pieces of ' ly11c "R H"

Work.lime 4. Assembly Workstation I (dedicated to this product familyl

• 20 days in !l rnonth - Ivhanw\1 process with one operator
• 'H,'<h'ih ift operation in all production dcpamncnrs - C\lclC time: 62 seconds
• l!ighc (8) hours ever)' .shif½ with ovcrcime if nec..·cs.'i-ary - ch11ngeovcr time: none
• Two IO~minuw breaks during e:1<.:h shifr. - Reli:1hi li1y: 100%
Ivfanual processes st◊p during IJr<:aks. • Observed inventorv:
Unpaid lunch. • 1,200 pieces ,;fTypc " I ,H"
· 640 pieces of Type " RI I"
Acme Production Control Department
• lkccivcs Srntc Strcer's 90/60/30-d,ty forecasts :11ul enters 5. Assembly Workstation II (dedicated to this product familyl
them m MRP. - M:.rn ual process with one operator
• Issues Acme 6-week f'otc'CaSt 10 Michigan Steel Cu. via M RP. . Cvcle time: 40 seconds
• Secure.s coil steel by weekly Fi\Xed order release to • Ci1,mgcovcr time: none
Michigan Steel Co. • Rc liahilirv: IOO%
• Receives dai1v lirm orde r from Sc:ue Sneer. - Observed· finishc<l-good.s inventory in ,v;irchouse:
• ( ;enernces rvli{P-h;lsecl weekly dcparcff1cnr.al requirements • 2. 700 pieces or Type "LH"
h:.1sed upon customer order, \ VIP invcnmry levels:. F/G . 1.440 pieces or ' lype "R H''
invcmory levels. :m<l antici p,ucd scrap :.md <low mi me,
• l.ssues weekly bui ld schedules to Stampi11g, \Vdding, ~ulll 6. Shipping Department
Asseltlbly processes. • Removes pn.-cs from fioishcd-~oods warehouse and stages
• Issues d:1ily shipping schedule tO Shippinl( DcpM11nem. them fvr rrnck shipmcm c() customer.

l.