Professional Documents
Culture Documents
*l
Seeing theWhole
mapping the
by
ffi
.\.':$li
SHINGC
Pt<I.ZF
value stream
<r
theWhole Seeing
Mapping
By Dan Jones Forevvord
the Extended
and Jim Womack
Value Stream
by John
Shook
The Lean Enterprise Institute Cambridge, MA USA lean.org Version 1.1 February 2003
With gratitude to Dan Jones'scolleagues at the Lean Enterprise ResearchCcnter, Cardiff L.lniversity, in particular Nick Rich, Dave Brunt, Dave Simonsand MatthiasHolweg, who helped pioneerextended value-stream mapping. And with further gratitude to our reviewers,editors and designers(who bear no responsibilityfor the remaining faults):JoseFerro, Bruce Henderson, Dave LaHore, Graham Loewy, Dave Logozzo, Bob Morgan, Guy Parsons, Atisa Sioshansi,Peter Tassi,Jeff liimmer, Helen Zak, Maria Elena Stophe1 and Thomas Skehan of OffPiste De sign. And with specialgratitude, as always,to John Shook.
@ Copyright 20OZThe Lean Enterprise Institute, Inc. Onc Cambridge Center, Cambridgc, NIA02142 N{A tlSA Tel: 617-871 -2900 o F-ax:617 -87| -2999 . lean.org
ISBN 0-96678,+3-5-9 All rights rcserved. Design by Oi'f-PisteDesign, Inc. Printed in thc LISA May 2008
Whenever there is a product for a customer, there is a value stream. The challenge lies in seeing it.
Mike Rother & John Shook, Learning to See
When you have learned to see value streams in individual facilities, it's time to see and then to optimize entire value streams, from raw materials to customer.
tr
F
F
F F
FOREWORD
When the first item in the Lean Tool Kit, Learning to See,was launched in June of 1998, we at LEI began to hear from managersin many industriesthat "this is rhe tool we have been looking for." Readersquickly realizedthat the grear power of Learning to Seelies in focusingattention on the value stream for inclividual product families within plants. Rather than concentratingon isolatedprocesses along the value srreamor aggregated activities serving many value streams,readerscould suddenly see how to optimize the flow of each product from receiving to shipping. This insight was breathtakingfor many managers caught up in narrow techniques or looking at only one activity in a complex system. As more and more people heard aboutLearning to Seeand began to practicevalue srream mapping' we began to hear of additional needs. "How can we introduce continuous flow at the processlevel within facilities?"And, "How can we expand the scopeof value stream mapping beyond individual facilities to the extended value stream from raw materialsto the end customer?"Many readerssuspectedthat if there was vastmwda within the walls of each facility rhere was even more mwda between facilities and firms. We had been thinking about this issuelong beforeJune of 1998.Indeed, the initial ourline of Learning to Seedevoted equal attention to mapping the extended value srream.However. we knew that extended mapping is more challengingthan facility-levelmapping and we soon concludedthat we would need severalpublications.In addition, we realizedthat managers would do well to hone their skills by "learning to see" within a limited areabefore venrurins forth to "see the whole". We therefore included a diagram in Learning to Seeillustrating different levels of mapping. We've recently addressed the processlevel with Mike Rother and Rick Harris' Creating Continuous Flow.In Seeingthe-Whole we tackle rhe higher, extended levels.
ollf'
a It
arY )
Why is an extended map harder to draw? It's not becausethe fundamental concept is different. At every level of mapping, we are simply observingand writing down every step and physical transformationfor individual product families.We in information processing observethe flow of customer desiresmoving up the value stream,in the form of orders or and then observethe progressof products moving downstreamin responseto schedules, this information, from raw materialsto finished items. plant, divisional,and company we need to map across Extended mapping is harderbecause In addition,we must pay carefulattention to the variability in order and materials boundaries. flows. Finally, we need to think about untangling, simplifying, and "right sizing" complex technologies logisticsand informarion systems,large facilities,and high-scaleprocessing servingmany value streamsand operated by many firms. Conducting extended mapping requires the cooperationof many departmentsand divisions within firms and between firms. These entities rarely think about the total flow of individual productsand ofren hide information from each other while pushing in opposite directions. time to In addition, extended mapping requires that line managersdevote hard-to-spare direct observarionof each product family's value stream.Failing this, higher-levelmapping easilybecomesa staff exercise(or a consulting project) yielding only another report that's soon forgotten. However, we have had These additional dimensionsof extended mapping truly are challenges. during the prepararion in overcomingthem, including recent instances success considerable managers can meet thesechallenges of this workbook.We now are certainthat change-agent and the facility levels and we know that time alreadydevoted to learningto see at the process will prove invaluableas you expand your field of view. As with Learning to See,we hope usersof Seeingthe'Whole will tell us how to improve with the lean community. Numerous this tool and will be willing to sharetheir experiences basedon hands-onexperiencewith value stream mapping at the facility user suggestions, level, have permitted us to improve Learning to Seeseveraltimes since its first publication. We look forward ro an intense and continuing dialoguewith the lean community on Seeing ta stw@lean.org. the'Whole as well. Pleasesend your comments and suggestions
John Shook Senior Advisor, Lean Enterprise Institute Ann Arbor, MI, USA March 2002
CONTENTS
Foreword by John Shook Your Focal Plane
Introduction: Part l:
Changing Started
Getting
Part ll:
Future Future
State 1 State 2
State Future States
The ldeal
About
the Authors
Value Strearn
Facility-Level Facility-Lewel
For yearsnow we have loved to "take a walk" along the entire value stream for a given product, looking for value and waste.We've done chis for dozensof products in many industriesand followed streamsacross the world. We presentedour first example in Lean Thinking (1996)when we drew the path of a humble cola can. This simple product with only three parts (barrel,top, and "pop-top") traveled 319 days through nine facilities owned by six companiesin four countriesto progress from ore in the ground into the hands of the customer.Yet during this long march only three hours of value creating activities were performed and the great majority of the steps- storing,picking, packing, shipping, unpacking, binning, checking, reworking,and endlessmovementsof informationto managethe system's complexity - createdno value at all. Looking at the whole has alwaysseemed natural to us and doing so will always suggestways to slashcostswhile dramaticallyimproving responsiveness and qualiry. Yet most managers we have encounteredon our value streamwalks want to stand in one place and look at only one point - their machine,their department,their plant, their firm. Often, the machine, the department, the plant, and the firm are - high labor and machine utilization, low performing well on traditional measures defects,on-time shipmentsand the managers are pleasedwich their achievements.
However, when we get managersto change their focal plane from their assets and their organizationto look at the product itself and what is actually happening on its long journey, they immediately realizethat the performanceof the entire value stream is abysmallysub-optimal.Indeed, most wonder how they have worked for years in traditionally compartmentalizedoperationsand somehow failed to notice the waste everywhere.Then they wonder what they can do about the mess.
And that is the big challenge.Managersfind it easyand fun ro draw extended current state maps.And this is a critical first step becauseit raisesconsciousness. But providing a managementtool that permits the waste to be removed permanently by achievingsuccessive future stateshas been much harder.It was only when we first saw Mike Rother and John Shook drawing future sratevalue srreammaps ar the facility level and coupling these maps to an action plan for implementation that - for all extended value we begin to see how we might guide groups of managers streamsare sharedacross many departmentsand firms for entire streams. In this breakthroughguide we presenrour method. It proposesa progression through two "future states" to an "ideal state" after the current state is jointly identified and agreed.The first future state will be relatively easyand createsrhe setting for the second.The secondfuture state is considerablyharder and reaching the ideal statewill require a major commitment by all the firms touching the product. Yet we believe that the savingsin time and effort and the improvements in qualiry at every step will encouragereams to keep going once they learn how to jointly optimize the sharedstream. Eventually,with some creativethinking about processand information technologies, we believe that most value streamscan be compressedand smoothed to a point where a large fraction of the original steps and practicallyall of the throughput time are eliminated.This will be a true revolution and the value streamteam getting there first will have an overwhelming competitive advantage. More important in most cases, the team getting started first and making the quickest progressalong the path will have a continuing comperitive advantage. The key is to summon your courage,form your cross-department and cross-company team, and changeyour focal plane to focus on the product. Then learn to see rhe whole and ... get going to take out the waste!We will be urgingyou on and waiting to hear about your problemsand successes. toward similar results
Dan Jonesand Jim Womack Ross-on-Wye, Herefordshire,L.lK and Brookline, MA, USA March 2002
What is Extended
An extended vah-re streamis simply all of the actions- both value-creating and wasteful required to bring a product from raw materialsinto the arms of the customer. T'he relevantactionsto be mapped consistof two flows: (a) orderstravcling upstreamfrom the customer(clrfrom the salesdepartmentwhen forecasts substitute for confirmed orders)and (b) productscoming down the value streamfrom raw materials to cLrstomer. Tbgethertheseconstitutea closedcircuit clf demand and response. Value stredm mapping is the simple process of directly obseruing the flows of information and materials as they now occLtr,summarizing them uisually, and then enuisioning a future state with mwch better performance. Maps of the extendedvalue streamcan be drawn fcrrproductscurrently in production 'I'he or for future products being planned. only difference is that the "current state" map for a product in productionshowsconditionsas thcy cxist today while thc "current state" map for a ncw product shclws the "businessas usual" approachto makingthe productcomparedwith alternative"future states"and "ideal states" with lesswasteand greaterresponsiveness.
Se l ecti n g
a P ro d u ct
Family
The whole point of value streammapping is to disaggregate operational issues ro the level of specific products,where they can be more easily acted on by managcrs. Tir do this you need to start at the furthest point downstream(toward the customer) to be mapped and to dcfine product families at that point. Typically a product family will includea groupof productvariantspassing through similarprocessing stcpsand usingcommon equipmentjust prior to shipmentto the customer. F-orexample: o In a power tools business, a product family might be medium-sizcdelectric drills utilizing a common chassis and passing through a common assembly cell as the last manufacturing step,evcn though the finished product has many different fcaturesand customerlabels.Alternativelvthe mapping team might define the product family as the motor going into the medium drills and map back upstreamfrom that point. o In the auto industry,a product family might be a car platform (e.g.,Ford Explorer and lVlercuryN4ountaineer) produced in an assemblyplant. Alternatively it might - ls1'ssay an nllglnxgel be a major componentsuppliedto auto assemblers using a common designarchitecture and assembled in a cell, but with varying power outputs and with different attachment points for different vehicles.
PARTI: GETTINGSTARTED
. In the aerospace industry,a product family might be an entire airframe (e.g.,the Boeing 737 or Airbus A320).Alternatively,it might be a major subassembly, for example the vertical tail. The sub-assembly may have many variantsfor different buyers of the completed aircraft.F or example the tail structure might incorporatedifferent aerialsand fairings for navigationand communicationequipment.And the productswithin rhe family chosenfor mapping might differ slightly in dimensions.For example,the basicrail designmight be slightly longer for use on a srretchedairframe.However, the verticaltail clearlyconstitutesa product family because all variants follow the samemanufacturingpath and have no commonalitywith tails for other aircraft,even if they are made in other areasof the samefacilities by the samefirm utilizing partsfrom the samesuppliers. Note rhat the sameproduct family may be supplied to a number of different end customersand have cosmeticdifferencescausingthe casualobserverto overlook product commonaliry. Neverthelessfrom the standpointof the firm or facility at the downsrreamend of rhe map, the product is clearly a family. Note also from the chart below that firms along similar value streamsoften have complex relationswith eachother. Delta suppliessimilar componentsto both Summa and Zenith; Omega fabricatessimilar parts for Delta and Azimuth; and Illinois Steel suppliesmaterialsto Theta and Zeta as well as Omega. Extended mapping cuts through this clutcer to focus on jusr one streamin order to think of improvementsthar can eventually apply to all streams.
Final Assembly
Component Assembly
Part Production
Apogee
Kappa
Asia Steel
/'
Delta
\
Admiral Platform A Platform B Perigee
:.,_.
\., '-,
I
"
Monterrey
Steel
F i r m s a l o n g s i m i l a rv a l u e s t r e a m so f t e n h a v e c o m p l e x r e l a t i o n s w i t h e a c ho t h e r .
Pr o d u ct
F a mi l i e s
fro m
Summ a' s
Per spectiwe
t;;I t'"n"=
I
I
|
I
t;;-l
I
computers
Ensine
Because the product family is defined from the vantagepoint of the final srep mapped, the conceptis essentially"fractal".That is, you can define product families from many starting points and map backwardup value streamsof varying lengths.For example,whar appearsro be a product family for an armaturemanufacturer(largearmaturesfor alternators) is simply one of many component parts for an alternatorproducer (who might define a product family as largealternators). And the largealternatoris just one component among many from the standpointof the auto assembler who defines product families in terms of vehicle platforms. As you selectyour start point and move back upstream,it is best for your first map to follow the path of a single family and a single component in the product. This is becausethe first obiectiue of extended mapping is to achieue a breakthrougb in shared consciousnessof waste and to identify systematicopportwnitiesfor eliminating the waste.It is highly likely that the wastesidentified by following one component back upstreamwill occur in roughlv equal measurein every componentgoing into the finished product.The alternativeapproach of mapping the value stream of every component going into the product is time consuming and costly and we have found that it overwhelms managerswith too much data. if the collaborators in the mapping processfind ways to work together and achieve useful results- additional maps can be createdfor many or all of the componentsand parts going into a finished product. But to ger srarred,keep it simple and focus first on achieving a breakthroughin raisingyour team's consciousness! In subsequentrounds of mapping -
PARTI: GETTINGSTARTED
D e t e r mi n i n g
a Ma n a g eable
Field of Vievv
The ideal map would truly show the whole. 1'hat is, it would srartwith rhe end cusromer who usesor consumes the product.The map would then follow the product all the way up the value streamto moleculesin the ground (or in the rccyclingbin), showingall thc wasted actionsand information lossen route. However, just as trying to map all of a product'sparts back upstream is overwhelming, trying to see roo far with your current vision may be fruitless. We advisc novice mappersthat a lor can be learncd by looking one or two facilities and firms upstrcam from wherever you start.This is the minimum o f c x t e n d c dm r p p i n g . scope
Single
Facility
Field of Vievv -
Learning to See
t+1 Michisan
I steerco. I soort""t l
c;i rr',I
U-_-T
f-l
1\ \\ |,:', l-.., H
\\ \\\
n />\
46ooL 24ooR | |
t>
Coilg 5 Aaye
En -J """"^a
lClO= thour lu?time= 05%
-l
I
ffia-ffi:-#ffiHffiWffi
Eit;;;l En=iA;;;A
lcto = 10 minu|e, I lupti^.=oot" I 2Shifts | I | 27.600 5ec.avait I lcto = 1o minuLe6 | luprine=Ioo% 25hila5 | | | 127.600 'ec. avait.|
Even at this minimum scope,note that thc scaleof mapschanges dramatically between Learning to See(facility-level maps) and Seeingthe Whole. The facility boxes that are the primary units of analysis guide are the samesizeas the individual process in this breakthrough boxes("stamping", "welding", "assembly")in Learning to See.Vastexpanses of people and equipment within facilitieshave been shrunk into tiny boxes so we can see the big picture! In this guide, we will draw mapswith an intermediatefield of view, startingat the distribution (e.g.,rolls of steel). centerfor the completedproduct and proceeding upstreamto raw materials For thosc with more ambition and with full cooperationfrom upstreamfacilities and firms, it is both possible and desirable to startnear the end customerand work far back upstream toward raw matcrials.
Multiple
Facilities
Field of Vievv -
/,1;afiffi'oo*"
L_
.E;;A _ .')lr'x:rl 6
I;;I
---7
r--t
\-
*-
|_____i
|
Ml
-rl
lcar,,
-/'
-.VL/t -
t;.l
6aa!.
tN-l
6da!.
+ :.
//@^i /,/ // f_--l--l
-rt ,
lft4
tIA
/ta--a
I I |
i \_'. I ,. L_ \ i
:twalt\+t-__
i.
'L
,//
<al'M,au
//! 426e1 214Hr )24 b
|
| I I
'I" -
i=r \)it
\\.\\ \\
ll*n4 l.--r---/,-pJ--.
k4. M
,r''o"
I D^ n lAphaMoroFl
;;;i
I ffi
li L
fr rrFrFrl
mr--f>
J
l--n@;-, Tsnpk-
_l rRM0 wtP2h
I l--;5h,4 T--il;;--] teE#t-l l-w-1
PARTI: GETTINGSTARTED
The Product Line Manager This individual in the most downstreamfirm needs to be much more rhan a technician concernedwith one facility. Indeed, for optimal resultsthe Product Line Manager needs
::::,rr:::,r',rp:i.
to be a business manager. This means"business"in the senseof taking responsibility for making money and growing market sharewith the product family in question.And it means "manager" in the senseof looking concretelyat the preciseactionsthat need to be taken all along the value stream to remove wasre and cost while improving quality and responsiveness. The most successful firms we have encounteredusing these techniques have Procluct Line Managerswho think about product marketing and engineeringas well as production and purchasing. With all the elementsof marketing,design,producrion,and supply chain under his or her oversight,this individual is in a unique position to judge the performance of the many functions touching the product. Indeed, as we will see in a moment, a continuing assessment of functional performancealong with preciseprescriptionsfor improvement is one of the most important benefits of product line management. However, we do not usually recommend what is sometimescalled a "product team" structure in which all of the engineering,operarions, purchasing,and marketing employeessupporting the product are put on a dedicated team. Doing this causesa large amount of organizational disruption during the transition and this structure still does not addressthe behavior of upstreampartner firms. Whai's more, it is really not necessary in most casesif the P[,M takes an energeticapproachto the job.
Perhapsche best known example of what we are talking about in the manufacturingworld today is the Chief Engineer for a car platform at Toyota (a job position also called the shusa).This individual is widely known by everyone in the company and takes responsibility for the success of the product in terms of return on investment and market share. Yet the Chief Engineer,like our proposedPLM, actually has no direct authority over marketing (which is done by a large marketing department),over engineering(which is done by the variousparts of the large engineeringdepartment),over production (which is done by the operationsdepartment),and over suppliers (who are managedby the purchasingdepartment,and the production controland logisticsdepartment.)Insteadthe Chief Engineer,working with a tiny group of assistants, is the one person who can "see the whole" and think about the necessary contributions from every functional activity and every upstream firm to createand deliver a successful product as judged by the end cusromer. The PLM in the most downstreamfirm will be even more effective if there are similar individuals in each of the upsrreamfirms so that for any product a quick evaluationcan be conducted by a small group composedof one PLM per firm. But this is not likely to be the case.Indeed, in today'sworld very few firms have true PLMs. (One of our concernsin preparingrhis breakthroughguide has been that the very managersmosr able ro benefit from it don't currently exist in many firms!) Thus to get started,someonefrom one of the functional areasin the most downstreamfirm will probably need ro take the lead and aim to achieve a breakthroughin consciousness. This individual probably will have little formal authority for overseeingrhe value stream and will therefore need to lead by example and by raising hopes about joint gains. possible We can't guaranteethat anyone anl.wherealong a value stream can succeedin raisingevery participant'sconsciousness to transformthe entire stream.We can guaranteethat anyone anywherecan raisethe important issues and make construcrive changea possibilitywhere ir was previously impossible ... if they have the courageto acr. ''.
To be successful, the mapping leader needs to be someonewho can gain the respect of upstreampartnersby conductinga rigorousand fair process. Logical candidates are production control, logistics, from purchasing, operations, or a processimprovement function like quality or processengineering.Any of these can work. However, assigninga buyer from purchasingto be a mapping leader can lead to problems if upstream participantsbelieve that the real purpose of mapping will be to uncover 'Ihus waste at suppliers,followed by demands for immediate price reductions. a purchasing function will probablyneed to assign mapping leadership to someone from its supplier development group if all participantsare to be convinced that the processis fair, balanced,and aimed at win-win-win outcomes. The value streamteam needs to include representatives of all the firms and facilities that shareownership and managementof the stream.Ideally, it would also include the relevant departmentswithin each firm - sales,operations,production control purchasing, and logistics, manufacturing engineering, informationmanagement, and product engineering.However, this can make the team too large to walk the value 'fhus streamtogether,which is often a critical learningexperience. we generally recommenda small team with a minimum of one representative per company. The team can query the functions supporting the value stream as necessary to fill in missing information.
The Wrong
and $taffs
A n u n d e r s t a n d a b lie nclination in any firm with busy linemanagers -and this surely includespracticallyall firms-is to delegatethe task of creatingvalue stream maps to outside consultantsor to internalstaff groups,typically in operationsplanning or process improvementdepartments.However,in our experience this is misguided.The findings of the consultantor staff expert are rarely credibleto the managerswho need to take action and the raisingexperienceof walking the value stream consciousness together-discovering the waste and jointly agreeingto a crossfirm action plan-simply never happens.A beautifulreport is produced by the consultantor staff team-and in our experience the beauty and precisionof the maps is generallyinversely proportionalto their usefulness-but the findings are then fildd away and soon forgotten. Remember:Only managerstaking clear responsibility can fix the mess. So the same managersought to draw the maps.
fi'|)
ilil ilil
flll
-L | t
a
\J
T a ki n g
a Wa l k
Once designated, the leaderand the team need to take a walk together along the value stream,draw the current statemap, and then ask, "Which stepscreatevalue?,Which stepsare waste?, Why is order flow so erratic?, Why is quality so erratic?, Why are deliveriesso erratic?, How can value be enhancedfor the end-customer?" Once the map is drawn so that the current state of an existing value stream is known precisely,it's time to createthe first of two "future state" maps that remove wasted stepswhile stabilizingprocesses and simplifying information flows. Future State 1 achievesthe luture state shown in Learning to Seewithin each facility touching the product. This means introducing continuous flow (as describedin Creating ContinwowsFlow) wherever possibleand instituting smooth, leveled pull between the areasof continuous flow. Future State 2 then introducessmooth, leveled pull with frequent replenishmentloops between every facility touching the product. In the process, most warehouses are eliminated, or converted to crossdock operations. An Ideal State may then co-locate at one site all of the activities required to proceedfrom raw materialsto finished goods,in the process eliminatingpractically all transportlinks and needs for information management. You may or may not find this particularsequence appropriatefor your own value streams.In particular,if you are mapping a new value stream for an entirely new product you will probably want to skip directly from the current (businessas-usual) state to an ideal state.We follow the three-step.sequence, beginning with Future State 1, in this breakthroughguide because we believethat this is likely to be the most typical approach.
PARTI: GETTINGSTARTED
Tvvo A
Final
Diagnostic
As teamsdraw their current statevalue streammaps,they are likely to make a surprising discovery. Most problems identified along the value srreamwill trace direcrly to the performanceof variousfunctions - information technology, production control, logistics, product engineering,operations,purchasing. What's more, weaknesses in functional performancediscoveredin the sample value streamwill almost certainly be present in every other value stream the firms touch. In our experience,the functions want to support the value stream for each product. But they have a hard time seeing the connection between their activities and the needs of the product. Thus an important benefit of the mapping process- in addition to a breakthrough in consciousness about the magnitude of waste and the enormousopportunities for improvement - can be to give much clearerguidance to each function about its role in supporting value streams. A real bonus can be achieuedif the improued functional performance can then be applied to all ualue stredms within the participating firms.
Diagnostic for Relations Betrrueen Firms As teams start mapping, they are likely to make yet another discovery.Today we all use languagestressing partnershipand cooperationbetween firms sharingvalue srreams. However, mapping teams in most caseswill discoveran enormousgap between these high-level principles of collaborationand the day-to-dayreality down ar the level of each value stream.If the value stream map showswidespreadconfusion and counterproductive actionsbetween firms at the value stream level, it will be obvious that "partnership" at the top isn't translatinginto competitivenessat the bottom. Fortunately,value stream mapping provides a clear and consistentlanguagefor firms to start an intelligent conversation with each other about the root causes of their shared cost, quality, reliability, responsiveness, and communicationsproblems.(Indeed, we believe a relentless,fine-grainedfocus on improving each value srream,rather than high-level agreementon principles, is what has given Toyota its edge in creating the world's leanestsupply base.)A real bonus can be achieuedif the practical lessonsof sbared ualue stream management can then be applied by each'firm to its relations with its other customers and swppliers.
The Current
State Map
With the basic principles of extended mapping in hand, it's time to accompanya value streamteam creating a map of the current state for a specific product family. This map will characterize the value stream as it is today. We've chosento focus on a high-volumeautomotivecomponentoffered with a small number of options- a windshield wiper consistingof a blade holding the actualwiping edge and the arm attachingthe blade to the vehicle. This product is similar in complexity and variery to the steeringcolumn bracket used to illustrate Learning to See. We've decided to map only an intermediate portion of a total value srream,which runs its entirety from the end user (you in your car) at the downsrreamend to raw marerials(iron ore in the earth) at the upstream end. The portion we will map srartsat Alpha Morors, rhe final assemblerof the finished vehicle, toward the customer end of the value stream.We then proceedback up the streamthrough che facilitiesof Beta Wipers and Gamma Stamping to the shipping dock at Michigan Steel, a raw materialsservicecenter.The five-member team, from the four firms sharingthis portion of the value stream,will be led by the head of supplierdevelopmentin the purchasing departmentat Alpha Motors and includesrhe product line managerand the assemblyplant managerat Beta Wipers,the value streammanagerfor this product family at Gamma Stamping, and the salesmanagerat Michigan Steel.
Windshield
W-per
Valrre
Strearn
Information flow
Tearn
Michigan
Steel
Gamma
Stampers f
Alpha
Motors
s.r". Manager
,,"n.*t*un",
Material flow
11
Windshield
Wiper
Assernbly
ancl
Fabrication
path
$
\B
.,
.--
brackets
windshield
A. B. G' D. E.
wiper
Assembly
and Fabrication
Rolled steel stamped into blade spine Four brackets attached to blade spine Wiping edge attached to blade spine and brackets assembly Blade assembly attached to arm Assembled wiper attached to automobile
Steps
Before we start mapping, let's look at an exploded view of this product showing the parts going into the wiper and its fit-point on the end product. Note that we will only map the circled areain this initial map. This is to keep the map simple and to concentrateinitially on raisingeveryone'sconsciousness of the extended value stream.
.---
arm components
13
our windshield wiper comesin two specifications - high trim and standard (HT and sr) - and in two sizes- small and large (s and L) to fit two different vehicles (A and B). The right-hand and left-hand wipers are identical on the vehicles in this example.The trim levels differ only in the paint - a matte-black finish for the standardtrim vehiclesand a glossy-blackfinish for the high trim models.The designsfor the two models differ only in the size of rhe parrs,not in their number or basic design.This means that the wipers are inrerchangeable from a final assemblystandpoint becausethey use the same fit points and require the same installationtime. The wipers clearly form a product family becauseall of the actionsoccurringupstream- component assembly, painting, and stamping - are in the same processsequencein the same firms and use the same processing equipment with a few tool and fixture changes. with the product family clearly identified, the first step for the team is to "take a walk" along the entire length of the value srreamto be mapped, recordingthe facilitiesvisited, the transportlinks, every action performed on the product' all information managementactions,and the time required. we alwayssuggesrstarting at che customer end becausethe customeris the point - indeed, the only point - of these materialflows. No product should be advancingthat the customer doesn't wanr and nothing should be happeningthat rhe cusromerdoesn'rconsiderof value! For the wiper example, the list of actionson the product is shown in the following list. Note rhat we have numbered all of the steps (73) in the left hand margin of the list and comparedthese with value creating steps (8) in the first column on the right. We have also recordedthe total elapsedtime (total product cycle time) which sums the time required to conducr all of the srepson a product (44.3 days) and compared this time with the actualvalue crearingtime (54.7minures),which is the sum of only the value creating steps.
Ph y s ica l
Total Steps
A cti o n s
R e q uir ed
to Gr eate
a W indshield
Total Time
W iper
Value Create Time
Raw Materials Supplier: Michigan Steel, Dearborn Heights, Ml 1. Load coils for twice weekly direct ship Transport Link 1 2. Directship (truck), Tonawanda,NY (500 miles) Second-TierSupplier: Gamma Stamping, Tonawanda, New York 3 . U n l o a dc o i l s 4. Receive& create ticket 5. Store coils 6. Conveycoil to Stamping Press1 7 . M o u n t o n c o i l r o l l e ra n d f e e d p r e s s 8 . S t a m p i n i t i a l( f l a t )s h a p e 9. Accumulatestamped parts during run 10.Convey parts bin to storage 1 1 .S t o r e p a r t s 12.Convey parts in bin to Stamping Press#2 13. Load parts in magazine,auto feed to press 1 4 .S t a m p f i n a l ( c u r v e d ) shape p a r t sd u r i n g r u n 1 5 .A c c u m u l a t e 16. Convey parts to storagearea 17. Store parts 18. Convey parts to paint shop 19. Rackparts on moving conveyor,clean, d i p , p a i n t& b a k e 20. Remove parts,inspect,sort & accumulatein bin 21. Convey parts to storage 22. Store parts prior to shipment 23. Load parts for twice weekly direct ship Transport Link 2 24. Directship (truck)to Harlingen, TX (1500 First-I'ierSupplier Warehouse: First-lrer
10m
8h
10m 10m 14d 10m 5m 1s 4h 10m 48h 10m 10m 10s 4h 10m 48h 10m 130m 2h 10m 48h 10m
1s
10s
52m
Har,insen. rX T:tH,ffis,
26.Formallyreceive
27. Store Parts
/rq$'
Sl*wt(.
.,;Kar-. b
ols-Slrfi
96h
/r"gqo;.f..1*$
\? V\F' . cnr$)lo'" D - ' ' n '' i !
15
Total
Steps
Total Time
Transport Link 3 29. Directship (truck)to Reynosa,Mexico ( 1 0 0m i l e sw i t h q u e u e a t b o r d e rc h e c kp o i n t ) First'lier Supplier Assembly Plant: Beta Wipers, Reynosa,Mexico 30. Formally receiveand move to storagearea 31. Store in receivingstoragearea 32. Conveyfrom storageareato first assemblystep 33. Store at first assemblystep 34. Insertfastenerclip and securewith pin 35. Accumulateparts in first assemblystep 36. Convey parts to second assemblystep 37. Store at second assemblystep 3 8 . C l a s pw i p e r b l a d ea s s e m b l y to sub assemblies 39. Accumulateparts in second assemblystep 40. Convey parts to third assemblystep 41. Store at third assemblystep 4 2 . I n s e r tw i p i n g e d g e i n b l a d ea s s e m b l y 43. Accumulateparts from third assemblystep 44. Convey parts to inspection, test & pack step 45. Store parts at inspection& test 46. Conductinspection, test & pack in protective sleeve 47. Accumulateparts at pack 48. Convey parts to shipping dock 49. Store awaiting shipment 50. Load truck for daily direct ship Transport Link 4 5 1 . S h i p b y t r u c kt o H a r l i n g e nT ,X ( 1 0 0m i l e sw i t h q u e u e a t b o r d e rc h e c kp o i n t ) First-TierSupplier Cross-Dock: Beta Wipers, Harlingen,TX 5 2 . U n l o a dt r u c k 53. Cross-Dock 54. Store awaiting full truck 55.Reload t r u c kf o r d a i l y s h i p Transport Link 5 56. Ship via multi-pick-uproute (truck)El Paso,TX { 6 0 0m i l e s )
l
6h
10m 48h 10m 8h 10s 4h 10m 8h 10s 4h 10m Bh 10s 4h 10m 8h 20s 4h 10m 12h 10m
10s
10s
10s
6h
96h
Total
Steps
Total Time
Car Company Cross-Dock: Alpha Motors, El Paso,TX 5 7 . U n l o a dt r u c k 58. Cross-Dock 59. Store awaitingfull truck 60.Reload t r u c kf o r d a i l y d i r e c ts h i p Transport Link 6 61. Directship to West Orange,NJ by truck ( 2 0 0 0m i l e s ) Car Company's State Street Assembly Plant: Alpha Motors, West Orange, NJ 62. Formallyreceive 63. Conveyto storagearea 64. Store awaiting need 65. Conveyto kitting area 66. Transfer to assemblybins 67. Store in assemblybins awaiting need 6 8 .A s s e m b l e w i p e r b l a d ei n h o l d e rt o a r m 69. Attach wiper arm with blade to vehicle 70, Line off vehicle and test 7 1 . S t o r ef i n i s h e dv e h i c l e s 72. Load train for daily direct ship Transport Link 7 7 3 . S h i pt o C l e v e l a n d Distribution C e n t e rb y t r a i n ( 5 0 0m i l e s )
96h
1m 1m
12h
Value Greating
73
44.3Days
I
54.7 Min.
Time
Distance
17
L e a rn i n g
to S e e V alue
As we write down the actions,the ability to distinguish value-creatingsrepsfrom currently necessary but wasteful steps and value-creatingtime from currently necessary but wasted time is critically important. The enormousgap between total time and value-creating time and between total actionsand value-creatingactions is the opportunity the value stream team must seize. Given the importanceof telling the difference between value and waste, it is not surprisingthat we often encounter readersand audienceswho are anxious about their ability to categorize actionscorrectly.Actually, it is very simple. Put yourself in the position of the consumerand ask if you would pay less for the producr or be lesssatisfiedwith it if a given step and irs necessary time were left out. In the caseof attachingthe wipers to the vehicle in the Alpha assemblyplant, the answeris clear.Consumersdo not expect to receive their vehicles with the wipers in the front seat,accompaniedby a polite note stating, "Some assemblyrequired". The final attachment step clearly createsvalue for the customer.So do the seven actionsof stampingthe metal arms,painting them, and sub-assembling them prior to attachment on the vehicle. By contrast,look at the many movementsof the product within each plant between processsteps,the long transportlinks between plants, the warehousingand crossdocking activities along the value stream,the numerous testing and inspection steps,and the repeatedpacking and unpacking of the product. Would you, as a consumer'be less satisfiedwith your vehicle if these currently necessary activities could somehow be left out? Of coursenor. And would you be happier if the car company could get you the model you want with the trim level you wanr quicker becausethese steps were left out? Of courseyou would. Indeed, the more these stepscausea delay in receiving exactly the product you want, the lessyou probably are willing to pay for it. Far from creatingvalue, these shipping, packing, inspecting, and warehousingactionsactually destroy it!
D r avvi n g
a U se fu l
Map
The long list of steps,categorized by waste and value, is highly provocarivebecause it helps the team realizethe enormousopporrunity for savings.. What's more, rhe ratiosof value-creating time to total time (54.7 minutes out of 6S,792or 0.08%) and of value-creating stepsto total steps(8 out of 73 or ll%) and the amount of transport distance(5300 miles) are quite typical for discretemanufacruredproducts in the world today.Our example is the norm, not the exception,and similar ratios are likely to emerge from any maps you draw.
18
State
Map -zLzL-1
I I I
ar6oEi--] 640^
|
| | | | |
I I I I I
However, for this information to be useful we need to simplify it and put it in a form managers can act on. The best way to do this is to group and summarize the data by each of the facilities and transportlinks the product encounters.Again, the place ro starr is with the customer,at the most downstreamend of the map. In this case,the customer is the Alpha Motors Distribution Center, which interactswith car dealersto get end consumers the products they want. We'll representthis organizationwith a facility icon placed on rhe right side of the map. Underneath this icon we'll draw a databox recordingthe customer requirement for size and frequency of shipmenr. Note that this facility is a cross-docking operationwhere vehiclesare sorted and sent onward as quickly as possible to several regional storageareasacrossNorth America. From there they go to auto retailersand then into the hands of the customer.Thus our intermediate-viewmap stops considerablyshort of the total value stream map thar ir may be useful to draw at some point in the future.
19
a-L-14
lVichiganSteetl
ferviceco. I
DearbornHeights,
4--14 Gamma
stampins I I
fonawanda,NY
lbetaWipers I I WarehouseI
lrr
|.t
Harlingen,TX
To get from raw materialsto the Alpha Distribution Center, the procluctflows through seven assembly, fabrication,warehousing,and cross-dock facilities.These are: o Alpha Motors' State street Assembly Plant in west orange, New Jersey o Alpha Motors' Cross-Dock, for many components from many suppliers, in El Paso,'Ibxas o Beta Wipers' Cross-Dock,for parts sent from severalplanrs ro many customers, in Harlingen, Texas o Beta Wipers' ComponentAssemblyPlant in Reynosa, Mexico o Beta Wipers' PartsWarehousein Harlingen, Texas o Gamma stamping's stamping and Painting plant in Tonawanda, New york o Michigan Steel'sServiceCenter in Dearborn Heights, Michiean
Gurrent
State
Map
Shovving
All Facilities
| ,"llr,lf""^ | Center
I I
Cleveland,OH
ZLzL-1
1160| orv
| | |
| |
I betaWipers I I Cross-Oock I
I zzoo I 213er
rc7Hr I I
I
I I I
-=
lhl +l I
Harlingen,fX El?aeo,TX
West Orange,NJ
We have createdtwo new facility icons not seen in Learning to See.One is a cross-dock icon for facilitieswhere productsare not stored but instead moved immediately from an incoming vehicle to an outbound shipping lane. The other is a warehouse icon for facilities where incoming goods are sorted and stored before shipment to their next point of use. (The iconsused in this workbook are displayedon the inside back coverand explainedin Appendix A.) You may want or need to createother icons,of course,in particularfor activitiesnot encounteredin our example.Just make sure that everyonc working on the extended map Llses the same icons.
21
I I
ll
Gamma I stampins I
I Warehciuee I
trn;Ww;1
Harlingen,TX
Tonawanda,NY
In
I I
t--RMaa6r'.-l wrPl1o h.
FG O\. I I
Dd".t a-=2oooppm
|
You will soon discoverthat you can't successfully gather and summarizerhe informarion needed for improving the value stream without drawing detailed currenr state in-facility value stream maps for products as they move through manufacruringfacilities.This is why masteryof the material in Learning to Seeis a prerequisitefor macro-mapping. We've drawn current scatefacility-level maps for the three manufacturingfacilitiesalong this value stream- Gamma Stamping, Beta Wipers Assembly,and Alpha Motors Assembly in Appendix B of this workbook, and you'll want to append yorir facility-level maps ro your current state macro map as well. Note that the data box under each facility contains data on inventories (Raw Materials,Work-In-Progress, Finished Goods),rhe amount of productive time (the number of shifts per day and the number of working days per week), the frequency of the production cycle (showing how ofren every part is made, such as
Shovving
All
Facilities
| ,"il?,lf"", I Center
I I
Cleveland,OH
/L-L-1
l%otD"v--]
| | | | | | 640^ 42651 214Hr 32oB 2135f 1o7Hr | | I I I I
lsl -l 1
Harlingen,TX
El?aso,TX
"EPE = 1 Day" meaning "every part every day"), and the defect level (in parrs per million) as reported by the customer at the next downstreamfacility (or by the customer'sinspector at the point of shipment in the caseof the Alpha Morors Assembly Plant.) We have not drawn facility-levelmaps for the Alpha and Beta cross-docks and for the Beta partswarehouse. This is partly to keep the size of this guide manageable and alsobecause we will endeavorto eliminate these facilities altogetheras we move through progressive furure states.If your value streamswill require large distribution warehoutesin any imaginable future for example for serviceparts - or cross-docks, you should also draw maps of these facilitiesas a guide to improving their performance.Exactly which facilitiesmerit in-facility maps and in what detail will alwaysbe a matter of judgmenr, so be preparedto adjustyour approachas your experienceaccumulates and you encounterdifferent siruations.
T h e Ou a l i ty
S cre e n
As we look at the data in the facility boxes,we nore a trenclworthy of further crarnination. At Alpha Motors Assembly the defect rate for wipers installed on rhe vehicle - def'ects discovered by a representative from Alpha'sDistribution Division in a final inspection just prior to shipment is 5 per million. (SinceAlpha is assembling 250,000 vehiclesper year with two wipers per vchiclc, this means that two to three wipers per ycar arc rejccted at final inspection,usually for scratches in the finish.) Yer when we look at defects emerging from Beta Wipers Assembly (asjudgcd by Alpha), we note that there are 400 def'ccrsper million and when we look at defects emcrging from (]amma Stamping (asjudged bv Beta) we note that there are 2000 defectivc parts per million. F'inally,when wc look at defccts arriving ar Gamma from N,lichigan Steel the figure soarsro 10,000pcr million. In brief, clualityis worseat every step up the value stream,a common phenomenonin practically every industrytoday.This meansthat to achieve5 clefects per million (approaching the Six Sigmalevel of 3.4 dcfectsper million), the prodr,rct is flowing through a scriesof qualitv screens in each facility,cach of which resultsin scrapand inspectioncost.The slope of this clualitygradient can surely bc reduced in firturc sraresanclit is importanr to nore carefully the current slope to aid our thinking on how to do this. We therefore recommend drawinga Quality Screen(asshown below) on the (lurrent Statemap. In this casewe have placedthe diagramin a convenientspot in the rowerright-handcorner.
O.rrality
ppm defects
Screen
2000
r500
1000
500
0 MICHIGAN TO GAMMA
GAMMA TO BETA
BETA TO ALPHA
ALPHATO ALPHAPC
24
Mapping
'fhe
the Transport
Links
next step, oncc the facilitv-level maps are drawn and the data have been summarized, in facility boxes,is to add the transportlinks between the facilities. Tb do this, you may need boat, train, and airplane icons,in addition to the truck icon frctmLearning to See. In this example,we will use the airplaneicon with a dotted line for shipmentsexpedited by air and a truck icon with the same stylc of dotted line fr-rr those expedited by truck. Thc
numbersin the regularshippingicon (a truck or a train) show the fiequency of shipments (e.g.,"1 x day" : on shipmcnt per day) while the number in the expeditingicon shorvs the number of costlyexpeditedshipmentsin the past year (e.g.,"2 x year" : rwicc a year). With these data in hand, we are ready to complete the physicalflow portion of the map Lry drawing in the normal product flows between facilities, using broad arrows.Notc that thcse arc striped,"push" arrowsbecausc productsare mclvingaheadat thc commandof a centralized informationsystemand not necessarily in accordwith the immediateneedsof the next downstream facility.l.lnder eachof the transportlinks we recordthe distancein milcs, the shippingbatch sizc,and the percentage of defectivedelivericsas reportedbv thc cusromcr. As thcse flows are drarvn,the team shoLrld notc one additionalpoint the trcnd in def'ective (the shipments: latc, early,or incorrect wrong product or in the wrong amount). As is alscltypical in most industriestoday,we note that the furthcr up the value streama facility is, the morc likely it is to make defectiveshipments. This situarionis analogous to the quality gradientand equallyworthy of improvementin future srares because ever_v defectiveshipmcnt generatcs correction costsdownstream and perturbsthe schcdule.Fclr economyof spacewc have summarized this trend in the samebox as thc qLralitv data on the Current Statemap, changingthe label to the "Quality and Delivery Screen".
O.ualitlz
PPM DEFECTS 2000
and
Delirrerlz
lScreen
% DEFECTIVE DELIVERIES 10
1500
1000
500
MICHIGAN TO GAMMA
GAMMA TO BETA
BETA TO ALPHA
ALPHA TO ALPHA PC
PARTII: THE CURRENT STATEMAP
The
"Bottom
Line"
Finally, we can draw a time-and-stepsline along the bottom of the map. Note that the first figure above each segment of this line is the total rime within each facilitv and along each transportlink, while the figure in parentheses ro the right is the value creating time. The first number below each segment of the line showsthe total actions(steps)taken on the product in each facility and transportlink, with the value creating actionsshown to the right in parentheses. Note that information needed for each facility is contained in the .,steps,, .,Time,, and summary boxes at the ends of the time-and-stepsrines on the indivicluar facility maps.
_N-
SteelCoits I
:\
.
t\
l- 2"1_
lweekl I
u---*-
r1,l
oooo.or>
stampins I I
Tonawanda,NY RM336h. wrP110h. FG4Ah.
35hifts 5Davs E?E=3Davs Defec.ts
4-/14 Gamma
I P,, l-l
U-T
-2LzL-1
ril|Drv+
^' o ' . . , r
rrrrrrr+ 1 a
3t"w4#+
Reynosa,Mexicot.... l - ' FRM f 56 h. wt? 41h.
FG12h z5ntftq
SDavs
=zoooppm
o.3d.
20.6d.(3131e.) 22(3)
o.25d.
4.6 d. (3o s.) 21(3)
Gurrent state shovving all Facilities, Transport Defects & Delivery, and Time-and-Steps Line
Links,
//
/r
| ,"flr,l&,", I Center I
I
Cleveland, OH
zz't-21--1
-1
| I I I I I
qfil
//
l xDay
| | | | | |
,rooio^v 640^
4265r 214Hr 32oB 2135r rc7Hr
2Shitrs
SDavs E?E = lDav Defects = Sppm
4.Od.
4.Od.
o.5d.
2.Od. (12os.) 11(2)
Ouality
.PPM DEFECTS
defeds
o.5d.
4
o.5d.
4
and
Delivery
Screen
2000
70 OEFECTIVE DELIVERIES 10
1500
1000 s00
t ,r,',r?*'
MICHIGAN TO GAMMA GAMMA TO BETA
M a p pi n g
th e
l n fo rma ti on
Flow
The tcam has now completcd mapping the physicalflow of the product but the valuc srream map is onlv half done. This is for the simple reason that if no clrstomcrsignals a demand for products from upstream,then nothing will flow. Or at least nothing shor-rld flow! We thereforc need to go back to the upper right corner of our map and draw the flow of order and prroduction informarion going back from rhe customcr. Howevcr, as we do this we nccd to warn you that mapping the information flow is the 'l'he hardestpart of the task. productioncontrol,and operations salcs, groupswithin most companiestend to communicatepoorly and a managerwho fLrllytrnderstands rhe infrrrmation managementmethods of all thrcc groups is a rarity.Whcn you add the complexity of going across several companies and through sales, productioncontrol,and operations deparrmenrs within eachcompany, it's not surprising that very feu, line managers seem to have useful knowledge of how infclrmationis managedon a macro-scalc. Cliven this rcality, vou shor"rld start where ordersenter the sysremand follou' the ordcr flor,v fiom department to department and from information managementsystem to infbrmation managementsystem,first through the most downstreamfirm and then upstream through the supplierfirms. Bc sure to use a pencil as you skctch informationflows and kccp an eraser handy! What's more, if you can, requcstthesedata aheadof your visit because many facilitiesand IT departmcntsdo not have them readilv at hand. To actuallydraw the informationportion of the extendcd map wc will need an additional icon for production control, which we have drawn in the shapeof a compurer rerminal.The first of these is for Alpha N,Iotors SalesOrder Bank. At this point orders are aggregated and placed in inventory (shown by order queue icons along the information flows). Thcv are held trntil the weekly salesplanningmeeting that decidesthc specificarion of thc orders that should be relcasedinto the system,given orders in hand from dcalers.These ordcrs are then releasedupstream to the following firms and departments: o Alpha l\{otors HeadquartersProduction Control o Alpha N,fotors Assembly Plant ProdLrction Control o Alpha NlotorsAssembly Plant Ntlarerials Control o Beta Wipers Headcluarters ProductionControl . Bcta Wipers AssemblyPlant ProductionControl o Gamma StampingHeadquarters ProductionControl r (iamma StampingPlant ProductionControl o Ntlichigan Steel ServiceProductionControl
2A
In almostall manufacturingcompanies, the salesand prclrlucticln control departmencs actually send a seriesof forecasts, schedules, and productionreleases back upstream.For example, in the car industry a rhree-month forecasr, a one-month rolling schedule,a weekly fixed schedule,and a daily shipping releasemight be typical.For our purposes, the impolant informationis the weeklv fixed scheduleand the daily shippingrelease because rhese actually trigger production in facilities and shipmenrsbetween facilities.These arc the information flows wc will capture on this map. If we follow the weekly schcduleand write down the infrrrmation managemcnt stepsand the time involvcd, as we did earlicr with physical actionspcrformed on the producr, we note the following alongthe longestpath. Information Manage the Steps Production at Alpha Motors 1. Dealer Ordersqueue in the SalesOrder Bank 2. Tiansmit weekly ordersfrom Alpha SalesOrder Bank 3. Queue at Alpha Headquarters Production(lontrol 4. Releasevl'eeklyproduction requiremcnts to Alpha plant 5. Queue at Alpha Plant ProductionControl 6. Release of daily productionsequence Production at Beta Wipers 7. Tiansmit weekly orders fiom Alpha HQ to Beta He B. Queue at Beta HQ ProductionControl 9. Ttansmit weekly production requirementsto Bera plant 10. Queue at Bera Plant ProductionC<tntrol 11. Release of weekly productionschedule 12. Beta Plant issues daily ordcrsfrom Beta Warehouse 13. Alpha N'Iaterials control rransmirs daily requiremenrsto Beta plant 14. Beta Plant ProductionControl issues dailv shippingrelease Production at Gamma Stamping 15. Tiansmit weckly ordersfrom Bera HQ to Gamma He 16. Queue at Gamma HQ ProductionControl 17. Tiansmit weekly production requirementsto (iamma plarit 18. Queue at Gamma Plant ProductionControl 19. Release of weekly productionschedule 20. Beta Nlatcrialstransmitsrwice-weeklyrequiremcnrsto Gamma planr 21. Gamma Plant Producti'n control issucs twice-weeklyshippingrelease * All transmissions areelectronic and essentially instantaneous. l,l davs 6 davs 6 davs 10 Davs 14 davs Actions Currently Value Strearn Required to
Delays*
6 davs 6 davs
Delivery from Michigan Steel 22- Transmit weekly ordersfrom Gamma He to Michigan Steel 23. Queue at Michigan Steel 24' Gamma Materials conrrol transmirs twice-weekly requiremenm to Michigan Steel 25. Michigan Steel issuestwice-weekly shipping release Total number of steps Elapsedtime for an order from the first to the last step (along the longest path) information Actual processingtime (assuming eachMRprunsovernight) * All transmissions areelectronic and essentially instantaneous.
14 days*
The Value
of Information
Note that we have made no effort to categorize information managementsteps as .,value creating" versus "wasteful", as we did for the list of physical steps. This is becausefrom the end customer'sstandpoint none of the information processing steps creaces any value. To test this - perhapsshocking - assertion, just ask yourself whether you would be less satisfiedwith a product if it could be ordered and delivered to you with no managemenr of production and logisticsinformation. Obviously you would not be less satisfied.Indeed, you would be more satisfiedif the cost savingsfrom eliminating information acquisitionand managementcould be passedalong to you. Yet in the modern era of automatedinformation management'most managershave implicitly acceptedthe notion that information is good, more information is better, and all possibleinformation is best. In fact, information fbr control of operations is necessary waste (Type One Muda). Managersought to be minimizing the need for it rather than maximizing it's availability.In the future sraresand ideal state we will show how As the weekly order information flows across the top of the map from headquarters to headquarters, it is also flowing from each headquarters down to plant production control departmentswhere weekly schedulesfor each plant are set. For example, Alpha Motors, AssemblyPlant ProductionControl takes the schedules production from Alpha Headquarrers Control, runs them through its computerized Materials Requiremenrsplanning (MRp) system (after a delay averagingsix days),and crearesa rolling six-day ahead schedulefor the assembly plant. This scheduleis fully sequenced (e.g,a blue Model A with high trim, then a green Model B with standardtrim) and takes into account iine balancingconstraints. For example, there are limits on how many Model As or Model Bs can be run down the line in a row without overloadingsome workstationswhere work content variessignificantly between Model A and Model B. These schedulesare rhen released to the plant floor.
At the sametime information is being releasedto the floor in each plant it is also being sent upstream,from plant-level materialscontrol departments,in the form of daily shipping releases. These are the preciseamounts of each part number the upstream plant is authorizedto ship to its downstreamcustomer on the next pickup. These daily releaseamounts are basedon known order lead times and the stocksthought to be on hand at the downstreamplant. From this it is apparentthat there are two separateinformation flows coming into each plant - the weekly schedulefrom each firm's production control department and the daily releasefrom the customer.Often, these flows are nor preciselysynchronized.So a third information managementloop comes into play, which is direct communicarion betweenthe materials handling departmentin the downsrream plant and the shipping department in the upstream plant. This direct link, usually a telephone voice line, becomesthe real production control and shippingmechanismwhenevermanagers at the ends of this link overridethe shippingreleases and, in extreme situations, productionschedules. They usuallydo this basedon their direct observations of emerging shortages and their judgment about what to do in response. We have drawn these information flows between the plants with a dotted line and our information expediting icon - an old-fashionedtelephone.
A Warning on Order Data As you move upstream don't confuse the customer's official releasewith the amount each plant actuallymade. Instead gather from each facility data on what was actuallyproduced daily over an extendedperiod and compare this with customer daily requestsin the form of shipping releases so you can see the relationof one to the other and the amount of variation in both. We're alwaysamazedthat companiesawash in information about what ought to happen do a poor job of recordingand preserving what actuallyhappened. So you may need to dig a bit or even assign an observerto captureaccurateinformationon l plant-level productionand shipping performance. What you find will be invaluablefor achievingyour future states.
Gurrent
State
Map
Showing
Information
Ftovv
II "",,*
| ?roduction I
f
14 dayo Fuffalo, NY
Controt I
I lr-J I
+_
6 days
f-MEI-I
Cleveland, OH
tdr
/v44ffi
luichiqanst,utl frji. servtce co.
I I
DearbornHeights, Ml Tonawanda,NY
1 V
I lr{ | 6 daye
f--
MiF--]
Harlingen,TX
I, Dailv I I
7..''''''''' (,
oooo..>
Tonawanda,NY o RM336h. wt?110h. FG4Ah.
SShifts
2 UaVS
_/1/L.-1
1
Harlingen,TX
t .2.*.Y"7"...'
'...fr..J
o.25d.
-+
T'"-1llr+
-.''o..
,=;,,.;;, ,Trrrr+
Reynosa,Mexicot....t E TRM ] 56 h.
wt? 41h.
25hif'rs
5Davs E?E=1Dav Oefects = 4OO ppm
E?E=3Davs
Defects =2000p?m
o.3 d.
20.6 a. (3131e.)
+.od.
32 .
Plymouth,Ml
lrN I 14 daye
I lf'J
10 dayo birmingham,Ml
t--:-;----l I vaity
T
// //
f--H----l r k
32oD
z+ur I
2135T r c Zn r
I
I I
'txDay
f 1"-L
l.' 1,3"
I--r-x{':'-
'\
WeetOrange,NJ Harlingen,fX
aaaaaaaaaloaaaaaaaaaaaaaaa
RM50h wt?2h.
2xYear
FG14h. 2Shifts
4.Od.
4.O d.
o.5 d.
2.8 d, (12Oe.) 11(2)
Ouality and Delivery Screen
o.5d.
4
o.5 d.
4
MICHIGAN TO GAMMA
GAMMA TO BETA
BETA TO ALPHA
ALPHA TO ALPHA PC
Demand Amplification
For the pasr year,Alpha Motors sales order Bank has senr very stable weekly orderscalling for 960 vehicles per day, five working days per week to Alpha's Headquarters ProductionControl. And HQ ProductionConrrol has releasecl level weekly buckets of orders to Alpha's Assembly Plant Production Control and to Beta HeadquartersProduction Control. The actualbuild still variesfrom the schedule- due to pulling vehicles out of sequenceto correct defects or becauseof problems in the paint booth or due ro a lack of parts. However, by adjusting the scheduleand working overrime at the end of each shift as necessary, rhe output of Alpha Morors Assembly varies by only about SVofrom the 960 units planned for each day and all vehiclesbuilt are shipped on the daily train to the Alpha Distribution Cenrer.
Dernand
wipers/day 2800 2600 2400 2200 2000 1920 1800
Arnplification
for Arpha
Motors
amplification ot +l4Oo/"
3OVo
2OVo
10o/o
OVo -1Oo/"
1600
'20o/o
1400
-30o/o
1200
-40o/"
1000
March 15
20
25
30
ALPHA PRODUCTION
Similarly,the mix of models (A versusB) variesby only about 5% daily as does the mix of wipers (Standard Tiim with flat paint versusHigh Tiim with glossypaint.) On average, Model A accountsfor two thirds of production and Model B one third while StandardThim wipers accountfor two thirds of demand and High Tiim the remainder.Thus production and shipments are fairly stable ar rhe cusromer(rieht) end of our map. Yet, as we plot the production and order/release databack upstream,we nore that the amplitude of changesin both production and releases increases markedly from facility to facility. Minor variationsin production at Alpha Motors Assembly become much larger by the time we reach Beta Wiper's assemblyplant, as shown below.
Dernanel
wipers/day 2800 2600
Arnplification
-nclrrding
Beta
Wipers
amplification ol +l4OVo 30Vo
2400 2200
,,,
I I
, , , ,
I I
I I
2OYo 1Oo/o
0o/o '1Oo/o
I I
I I
l, 1t t
I ll
I I
?J' tl
tl
I I I
1t I
'- - 'l--,'
l, ll ll l, I t
i; {
, ,
t I I
-20o/"
-30%
-4OYo
15
30
35
Bv the time we reachGamma Stamping,the variations arc very large.Indeed, Gamrna Stamping'sreleases to Nlichigan Steel varicd by nearly 40% in the month prior to thc arriyal of the mapping team.This infbrmationfor Gamma Stampingcompletesthe Demancj Amplification Screenlbr our current state.as shown below. 'lb make this verv common phenomenonclearer, we've summarizecl the maximum perccnrage
changein dailv productionand dailv releases over rhe past month for each facility and alignecl them in a simplified Demand Amplification chart as shown at right. We've placed this chart in a box in the upper left cornerof our Current State map, as shown on the ncxr page spread.
Dernancl
wipers/day 2800 2600 2400 2200 2000 1920 1800
j
Arnplification
screen
in Grrrrent
state
amplification oI +l40o/o
ivi ti"
il-z \
t
-1Oo/o
:'r
:tt
1400
t. :
i
-40o/o
10
'Ib
extra production capacityor carry large stocksof finished goods in inventory or disappoinr downstreamcustomersa significant fraction of the time. Becausefailing to ship on tirne to meet customcrneedsis an unacceptablc alternative for suppliersin thc auto industryand bccauscextra tooling can be very expensivc,most firms in this industrv includine Beta.
36
Sirnplified
Dernand
Arnplification
Screen
% variation
Demand
Amplification
GAMMA ORDER
GAMMA PRODUCTION
BETA ORDER
BETA PRODUCTION
ALPHA ORDER
ALPHA PRODUCTION
Gamma,and N'Iichigan Stecl,carryextra inventrlrics ro protectthe customcr. T'he cost implications of demand amplification are therefcrre apparentin the amount of errra inventories in the value stream. Whv doesthis growingvariationexist?For the simple reasons that productionproblemsoccur in cvery plant (even the lcanest!), transportproblemsoccuron evcry link, feedbackon crlrrcnr conditionsand amountsof product on hand is nevcr completelyaccurate, and largeminimum productionand shipmentquantitiescausevcry small changes in the amountsnecded downstream to producemuch largerchanges in the amountsrequcstedand producedupstrcam. To take the u'orst-case example, if one wiper is discovcredto be defective at thc asser-r-rblv plant and the re-orderamount is just on thc cdge of one new;rallet (containing320 r,r ipcrs in our example) che re-orderwill jump to 2 pallets- or a total of 6,10 wipers- even though only onc additionalwiper is needcd.And this phenomenoncan be repeatedseveralmore times as the order flows back upstream,creating a wave. The reasonthis wavc grows largcr as we move upstreamis because of the numbcr of schedulingpoints (B) and the length of the delays(totaling .58days) before infcrrmation is acted upon. Each system recalculates its schedulebascdon its own (not very accurate)forccastsand on information from cusromcrs that is alreadyup to a week old. This is the familiarand dreaded"F orresrer liffect" documentedby Jay Forrestcrat MIT in the 1960s. The irregularities in the systcm are tlrther exaggerated by the misalignmentof what the official schcdulingand releasing system(in the ccntralizedcompurers)are sayingand what jobs are seeingand doing.Then, as misalignments the individuals in shippingand receiving grow,confidencein the formal systemdeclincsand more and more of thc actualsche duling and rclcasingmay be done manually despitc the largeinvestmentsin informationtcchnologv.
Demand
Amplification
GAMMA ORDER
GAMMA PROOUCTION
BETA ORDER
BETA PRODUCTION
ALPHA ORDER
ALPHA PRODUCTION
I f-Fuffalo. NY
I ",",* I
| ?roduction I
Controt MRp-l
I lf{ 6 daya
Cleveland, OH
I wTkry I
(anffi
rt ;c"ltt-] ,- : \--
lY2n?:"7:"1;'"*"
DearbornHeights, Ml
l-t ^,1
1 V
'fonawanda,NY
\ '.
..,.
r-]
6 days
3 ..
f2"-1---'
'...
MRp-l
I- 6.-l--
19) '\'
+.+J
x
Tonawanda,NY RM336h. wt?110h. FG40h.
SShrtbs SDays E?E=3Davs Defecte
W
I | Shipdarch =6?allets
=2oooppm
o.3 d.
20.6d.(3131s.)
22(3)
3a
Final Gurrent
State
Map
Showing
Demand
Arnplification
Plymouth,Ml
{-'N,l 14daye
-4-4-1
I
EI
|
i 4
I ,,Jl?,1,i, I center I
1 Cleveland,OH
/f
ar6ot-6oto-o" 640^
a| 9etaWivers I I Croes-Oock I
l-lJ---] --
//
//
i"f:i I | i 32ob I
|
'
21zgr I
.rr,*o^t
I-T*-L:'t.'\. \ l.i'v l.l
I -t El?aso,TX J
roznr I
Trr)
' -=*-l
l----,-+
t
|
-l
TIME
WestOrange,NJ RM 50 h. wl?2h.
FGIAh
Harlingen,TX
n-s\t-->
zl -z*
44.3 day
Year
zooo"rA
@ | = 6?alteto I
31.Oday:
2Shifte
@,"-J4
13.3 dayt
4.O d.
4.O d.
o.5 d.
2.8 d. (12Oe.) 11(2)
Ouality
' PPM DEF6CTS
o.5d.
4
o.5 d.
4
and
Delivery
Scrgon
| 10,000
defeds
2000 1500
7o DEFECTIVE DELIVERIES 10
1000
ttr?*- t
MICHIGAN TO GAMMA
of Our Map As thc team finishesrecordingthese product and information flows, it seemssensibleto conclude the Current State map at this scopeof mapping. The map does not go all the rvay downstream to the customertaking dcliveryof a car at the dealership and it doesnot go all the way upstreamto the steelmill, much lessto ore in the ground.Mapping theseaciditional stepswould doubtless provide additionalinsights,but to do so would require largeamounts of time and expenseto examine organizations whose behaviorthe team has little prospectof changing right now. Yet cven within this scope,the map coversa considerablepgrtion of a lengthy and complex value streamand uncoverssomevery provocative pcrfclrmance features.
The
Limits
Gamma
CurrentState
20.6
LEADTIME(in days) Beta + Alpha : In-plant* 4.6 2.8 31.0 + Tiansporr = 13.3
Total 44.3days
With regardto qLrality and delivery reliability,we nore that end-of-the-value srreamindicators of both measures are very good (5 ppm and lVa defecdveshipmentsto the customer)br-rt this is achieved through a sericsof screenswith significantcosrsand delays. With regardto information about customer clemand,\\,e note rhat ordcr information is acted upon up to 17 times and storedfor up to 58 daysin queues.What is morc, six inclividuals in receivingand shippingdirectly intervenein mediatingorclerflows within an expensive, tcchnicallysophisticated informationmanagement sysremrhat on its face is totallv ar,rtomated. Evcn with this intervention- and in somecase s probablybecause of it - demancl amplification, with compcnsaring inventories to protcct customers, increases steadilvto a very high level as onc looks back up the value stream.
40
Grrrrent
State
Srrrnrn
ary'
Gurrent State
Total Lead Tme
44.3 days
o.oB%
11%
Inventory Turns
5 400 B 7
Ouality Screen
(defectsat the downstream end over defectsat the upstream end)
Delivery Screen
(% defectiveshipments at the downstream over o/"defective s h i p m e n t sa t u p s t r e a me n d )
5300
of the manv Finally,we must note a suddenlyobviouspoint about the performance departmcntsand firms touching the physicalproduct on its 44-dayjourney and order information on its 5B-dayjourney: However effective the variousfunctions operations, productioncontrol, logistics,manufacturingengineering,quality, and purchasing- may be in achievingtheir own objectives,they are not at all effective in supportingthis product on inv<r.lved are common to all its path to the customer.What's more, becausethe processes productspassingthrough these departmentsand firms, it is highly unlikely that they are aspect doing a better job of supportingother product families.The functionaldiagnostic of our extended mapping process which we believe is its most important contribution to firms in the long run therefore revealsseverelymal-performingfunctions all thc wav up and down the value stream.
41
If this is an accurateportrayalof the current state- and, becausethe value streamteam has directly observedit, there is good reasonto think that it is - there are surely opportunities to speed the accuratedelivery of products to the customerwhile eliminating large amounrs of cost.To begin to do this we need to specify in rhe nexr secrionthe featuresof a lean extended value stream that can deliver these benefits.
The Power
o f S i mp l i city
"What do you consider the largest [scale]map that would be really useful?" "About 6 inches to the mile." "Only six incbes!...Weactually made a map on the scaleof a mile to the mile!" "Have you used it much?,I enquired." "It has never been spreadout... the farmersobjected [that] it would cover rhe whole country and shut out the sunlight! So now we use the country itself, as its own map, and I assureyou thar it does very well." Lewis Carroll, Syluie and Bruno Concluded, Chapter l1
As you experiment with drawing extended maps suitable for your product families,you may wonder just how much detail to include. We often find that novice reams- like want to record every conceivable{etail about the current state,as well as mapping the flow of every part in the finished product. To make room for all this detail they even createwall-sizedmaps in corporarewar rooms. But too much detail in an extended map interfereswith clear thinking about how to improve the value stream.We therefore urge teams to keep extended maps as simple as possible.The objective must be to truly "see the whole" by summarizingthe value srream on a singlesheetof paper (11" x 17" is a good size,43 in Europe) and to use this big picture to raiseconsciousness among all the value stream participants.Only then can you identify ways to quickly improve performanceall along the value stream and motivate the firms involved to optimize the whole. Lewis Carroll'smyopic mapmaker -
r-) e^
Value Stream
Fifty yearsago Thiichi Ohno at Toyota enumerated seven types of waste in value streams. You may have them memorized by now but they bear repeating becausethe types of waste are the same at the process, the facility, and the extended value stream levels of analysis: OverproductionDefectsMaking items upstreambefore anyonewants or needsthem downstream.
Errors in products,paperwork supportingproducts,or delivery performance. Products in excessof the amount needed to insure meeting Activities not adding value that could be eliminated, such as
a separateinspeccionstep replacedby a self-monitoringmachine with auto-stop,or flashremovalafter molding eliminatedwith higher mold tolerances and better mold maintenance. Unnecessarytransportation between work sites that could easilybe consolidated. Waiting waiting for machinesto cycle. lJsually production associates moving out of their work space Associates Unnecessarymotion in the workplace Moving products between facilities
to find materials,lools, work instructions,and help. level within facilities,we are always When mapping at the facility level and at the process concernedabout overproduction due to poor information flows withiz facilitiesand the desireof managers to move productsaheadto meet performancemetrics for equipment utilization. (Ohno alwaysstressed that overproductionis the worst waste.)We are also processing, looking carefully for unnecessary defects,waiting, and motion. When we more our analysisof product and information flows to the extended, macro level, overproductionis still a critical concern but now due to erratic information flows between firms and facilities.And we are now speciallyinterested in the two final forms of waste:unnecessaryinventories (due to erratic information flows as well as incapableand batch-orientedupstream processes) and unnecessarytransportation (causedby location decisionsthat seek to optimize performanceat individual points along the value stream rather than the whole value stream).Reducing these three forms of waste - largely by better managing information flows and logistic.s extended-mappingof future states. will be centrai concerns for our
What shoulda lean extendetlvalue streamlook like? First, everyone in the entire value stream should be aware of the rate of customer consumption of the product at the end of the stream. You are probably familiar with takt time, which is the amount of procluctdcmanded per unit time adjustedfor the amount of production rime available. This is a wonderfully uscful concept within every facility becauseit tells everyone rhe necessary rate of production from minute to minute to meet the neeclsof the nexr downstreamcustomer.Howevel note that takt time will vary from facility to facility along a value stream if the amount of availableproduction time differs from facilitv to facility and if downstreamsteps incorporatemore than one unit of an upsrream component'Thus takt time at the Alpha Motors Assembly Plant is 60 seconds (tcr build 960 vehiclesduring the sixteenhoursof production time available eachdav) but is 30 seconds at Beta'swiper assembly plant running the sameshift pattern (because eachvehicle needstwo wipers)and would fall to l5 seconds if the wipcr assemblyplant switched to only a single eight hour shift each day. T.hus,in most cases'there is no single takt time for the entire value stream. However, every facility along on the stream needs to be aware of the encl rate of consumption to calculatefacility-specifictakt time. Production ar every Llpsrream stageshould run on averageat the same rate, as adjusteclfor the availablcamounr of productiontime at each step and the neeclt' make multiple copiesof some productsto incorporate in productsdownstream. Any time we see a chronic pattcrn of imbalancedproduction rates in clifferentfacilities we know we don,r have a lean value stream. However, pleaseunderstandthat every facility upsrream should not conduct its activities in lock step with the current rate of the end facility in the stream.This seemsto be the implication today of many naive claims for e-commerceand the web: "If you know the rate of end consumptionright now you can schedule y'urself accordingly."In facr, what each facility should produce each m.rning is a leveled mix of what the next downstreamfacility requests for delivery this afternoonor tomorrow Knowing changesin actual consumption at the end of the stream(particularly the amplitude of the changes) is extremelyimportantfor capacityplanning but is not sufficient for controlling productioh todav.
44
What u'e can learn from comparingproductionratesupstreamwith actualconsumption downstreamis hou, faithfully the production control systemis sending true customer demand (which we call "signal") upstreamversusdistorted demand (which we call "noise"). If there is significantnoise,producing"demand amplification"unrelatedto true customerdesires(aswe see in our Current State map), stepsneed to be taken to eliminate these gyrationsin future states. A second feature of a truly lean extended value stream will be very little inventory. (2) work-inThis inventory will consistof the minimum amount of (1) raw materials, process, and (3) finished goodsrequired to support the needsof the next downstream demand,(b) the capabilityof customergiven (a) the variabilityof downstream upstream processes, and (c) the inventory reqr-rired between processing steps due to batch sizesand shipping quantities.Toyota calls the minimum amountsof invcntory neededto supportthe customers in a value streamat any given time the standard inventory. The standardis calculatedfor each categoryof inventory depending upon its function in the value stream.Toyota continr.rally seeksto reduce this amount by decreasing batch sizes, increasing shipmentfrequencies, leveling demand,and improvingcapability.
Lovv Low
Variability
and
w h o s e e kt o r e d u c ei n v e n t o r i e s We sometimes encounter leanimplementers along a value stream without botheringto calculate the standardinventory n e e d e df o r t h e c u r r e n tl e v e l so f v a r i a b i l i t y a n d c a p a b i l i t yA . n immediate "lowering of the water level" may indeed "expose the rocks" and put pressureon everyoneto go fasterto reducevariabilityand improve capability.However,a more likely consequence is chaos and outraged customerswhen the newly "lean" value stream fails to deliverthe right a m o u n t sw i t h t h e r i g h t q u a l i t ya t t h e r i g h t t i m e . A better strategyis to calculate the standardinventoryat every storage p o i n t a l o n g t h e v a l u e s t r e a mi n t h e c u r r e n ts t a t ea n d i m m e d i a t e l y eliminate g r e a t e rt h a n t h e s t a n d a r d T inventories . h e n l o w e r t h e s t a n d a r da n d r e d u c e inventories to the new standardin a future state after variabilityand capabilityissuesare addressed.
The
Uses of Inventory: Greating a Strategy We've definedthe three traditionalcategoriesof inventoryand comparedthese with severaladditionalcategoriesin common use (as shown in the next page).Note that these categories overlap."Finishedgoods" can be "safety stocks", "buffer stocks",or "shipping stocks".What's more, the same item-a pallet of windshieldwipers in Beta'sfinished goods area,for example-can be included in severalcategories-a "safety stock" and a "buffer stock" in the case of our wiper-depending on the practiceof the firm and the facility.The key point with regardto definitionsis for the members of the value stream team to agree on a consistentuse of this sometimesconfusingterminology. The key point with regard to the inventoriesthemselvesis for the team to make a strategic plan for every part in a future state,describingthe reasonsfor keeping specificamounts of materialsand goods in specificplacesas standardinventory.As they do this, many value streamteams decideto actuallyincrease the amount of inventoryin a downstreamfinished goods area near the schedulingpoint, both as a buffer stock and as a safety stock.This guards againstdemand amplification traveling upstreamand facilitates the reductionof work-in-process and raw materialsto a very low level in upstreamfacilities.By increasing inventoryat one point - seeminglya step backward- it may be possibleto reduce inventories at every other point along the value streamand for the value stream as a whole.
Many
Forms
and
The
Manyz
Forrns
of lnrrentort/
E:E
46
Tlrpes
of
lnrrentory
TRADITIONAL CATEGORI ES Defined by their positionin the valuestream Raur Materials Goodsentering a facility that havenot yet beenprocessed. Work-ln-Process processing Itemsbetween stepswithin a facility. Finished Goods Itemsa facilityhascompleted that awaitshipment.
A D D IT ION A L C ATEGORIES Defined by their purposein the valuestream Safety Stocks Go o d sh e l da t a n y p oint( in RawM ater ials, W lP, or Finished prevent Goods) to downstream customers from beingstarved process by upstream capability issues.
Buffer Stocks Goods held, usually at the downstream end of a facility or process,to protect the downstream customer from starvation in the event of an abrupt increasein point demand by a customer -a demand spike that exceedspoint production capacity. Shipping Stocks Goods in shipping lanesat the downstreamend of a facilitythat a r e b e i n g b u i l t u p f o r t h e n e x t s h i p m e n t .( T h e s ea r e g e n e r a l l y proportionalto shipping batch sizesand frequencies).
47
A third feature of an extended lean value stream is as few transport links as possible between steps in the production process. As we have noted earlier,no customer attachesvalue to moving thc product around.Indeed,customers will often be willing to pay more for a product if it can be supplied in the eract specificationthey want very quickly. Thus we need to ask about every transportlink: Is this really necessary? SLrbstituting modes, notably air fbr truck, is certainly an alternativeway to reduce thftrughputtime, but typicallyat an unacceptable cost premium. In general \\rewant to eliminatetransportratherthan speedit up. A fourth feature of a lean value stream is as little information processing as possible, with pure signal and no noise in the information flows that remain. This meanspulling information managementdown from highcr levels of the organization, in remote informationmanagementdcpartments,to the shop floor where each processing step and each facility can signal the previous srcp and facility directly about its immediatenecds.We should schedulethe cntire value streamfrom only one point, in this casethe assembly line of Alpha, and pull materials back up rhe value streamfrom this point. A fifth feature of lean value stream will be the shortest possible lead time. Indeed, this may be the most importanrof all. Thiichi Ohno ofren remarked the whole point of the Toyota Production System was simplv to reduce lead times from raw materialsto the customer.The shorter the lead timc, the morc likely it becomesthat thc entire value stream can respond to real orders rathcr than inaccurateforecasts. And the more likely it becomesthat defects,proccss variations,and every other problem will be detected before significantwasre is created. A final principle of a lean value stream at the macro level is that changes introduced to smooth flow, eliminate inventories, and eliminate excess transport and lead time, should involve the least possible or even zero cost. What's more, capital costs,when they are necessary, should be deferred until casierand quicker actionshave alreadybeen taken.
48
of this
the methods describedin Learning to See and Creating Continwous Flow. 'Ihese entail practically no capitalcostsand will creatcwhat we will call our Future Statel, as described in Part IV of this Guide. Once flow and pull have been introducedwithin eachfacility,eliminating many wastefulstepsin the process, it will be time to examine the information and transportlinks between facilities.Often it will be possiblc to smoorh the value stream and reduce the need for buffers by introducing direct feedback loops with leveling mechanismsfor information flowing from each downstream"cllstomer"to the precedingupstream"producer".We will do this in Future State 2, as described in Part V of this Guide, noting rhar a smooth pull of orderscan often be tested on an experimcntal basisfor one product family without effecting information flows for other productsgoing through thc samefacilities. With information flows smoothed and noise reduced, it will be timc to reduceshipmcnt sizeswhile increasing shipment frequencies between each facility and it's upstreamcustomer.We will also do this in FurLrreState 2. Frequent delivery in small lors will require the introductionof some tvpc of "milk run" logisticsbetween facilitiesand for the first time will raisethe issue of relations between multiple product families.This is because organizing a milk rLrnfor the parts needed for only a single product family ar rhe next downstreamfacility will often be impractical.Instead, major portions of a facility or an entire facility may need to make rhe leap from dedicated shipments arriving infrequently to sharedshipments arriving often. Finally, after Future State 1 and Fr-rture State 2 are achieved,it may make senseto begin re-sizing and relocating activitiesin order to "compress"the value stream.Doing this may make it possibleto remove large remaining blocks of time and cost and move the value streammuch clgserto perfection in an ldeal State.
49
Because value streamcompression will often require significanrinvestments by a firm at Point A that lower costsfor a firm downstreamat point B, some method will be needed to justify these investmenrs and to determine how the firms can sharethe cosrsand benefits. We'll provide some simple guiclelines in Part VI of this Guide, describing the Ideal State. A truly ideal state will be the happy circumstancein which all actionscreate value with zero defects and consumerresponseis instantaneous. No one is likely to reach this perfect realm soon,bur it is highly provocativeto ask what types of product designs,production technologies, and locarionallogic can close as much of this gap as possible.what's more, the processof developing an Ideal State can provide an invaluable North Star for steering each value stream through succeedingproduct generationsthat come closer and closerto perfection.
5()
Future State 1
Once the team completesthe Current State map and everyone agreesthat it is accurate, the key question becomes,"What should be done in what sequenceto create a better future state?" In our experience,the easiest placeto start is to createfuture stateswithin the walls of each of the facilities the product visits en route to the customer.By drawing and then achieving a future state of the type describedin Learning to Seewithin each major facility it will be possibleto achievea substantialimprovemenr in the performanceof the entire value streamand to do this within a short time. This createsconfidencein the process and give teams a sensethat much more is possible. Beginning with this step also has che critical advanrage of imposing a "price of admission" on all of the value stream participants.Drawing the current state map is fun but entails no real commitment. It's when you get to the, "What are we going to do today about the waste?"question that the hard issuesarise.Insisting that each participatingfacility and firm quickly implement actual improvements as the price of continuing with the exercise also tends to gain buy-in for the process. Yet the hurdle is nor roo onerous becauselittle capital investment is needed to achieve a future state within the individual facilities.
Level
Pull and
Flovv Within
All
Facilities
In Appendix B, we show the Future State Maps for the Alpha Morors Final AssemblyPlant, the Beta Wipers Component AssemblyPlant, and the Gamma Stamping Part FabricationPlant. At the urging of the extended value stream team, these were implemented and stabilizedover a three-month period by newly appointed value stream managersin each plant. (As noted earlier,no changeshave been attempted at the Alpha and Beta crossdocks and in the Beta warehouse. This is both to keep the exercise manageable and because we will seek to eliminate these facilities in Future State 2..) The cumulative result of these actionsat the plant level is shown in the summary boxes on the Future State 1 map.
PARTIV: FUTURESTATE 1
7ovariation 35 30 25 20 15 10 5
Demand
Amplification
GAMMA ORDER
GAMMA PRODUCTION
BETA OBDER
BETA PRODUCTION
ALPHA ORDER
ALPHA PRODUCTION
t-il-f
14 daye Buffalo, NY
+-
C-MEI-I
Cleveland,OH
@
v E 6 days / lr"#7,21,",1. ,zf Conrrol I \
fri{
-e
W | ?tant
| ?roductio
r-1
a-sr.mt'-] ------\-
i i
:.
\-_ \-".
>r..
Ionawanda.NY
ar;fi;iltlF"w,,kl l'------r-
t l l
M-Rr--]
' oor . .
DetaWipers Warehouse
| I
Harlingen,TX
RM16 h.
nlfl""'7;."'J
I I 5oom. | la zeh:,ft;--] ehipaar,ch I =7zcoits I |-jD-=2;-| I arrlr:io^'1 =8% I Derective D"r*r"= | l ta I r oooppm I f----;;;-------l
w t ?o h .
2Shilts SOays E?E=1 Day Defecls = 2OO ppm
o.3 d.
5Jd. (3131s.)
o.25d.
2.Od.
o.25d.
1.2d. (3o e.) I (3)
20 (3)
52
:
I IFJ
{ '--l
Plymouth,Ml 14 daye
10 da birminqham,Ml
//
/r
Cleveland, OH
,160 tDrv
| | |
_._
il;-l I p,v [-]
!----i5
t_
q=il
.\i.
lT---l \
//
1
| | I
|
|
32oe
21351
I
|
lxDay
lo7Hr
I betaWipers I I Cross1ock I
'Y,Ys:oi" I b:I+
-r-] f rfrrr=r}
WeetOrange,NJ RM 15 h. w t ?2 h . FG14h.
29hifte 5Davs E?E = 1 Dav Defecf,s = 5PP^
2x Year
4.Od.
4.O d.
o.5 d.
1.3d. (12O s.) 7 (2)
, ppm deteds 2000
% defedive deliveries
o.5 d.
4
o.5 d.
4
1
Ouality and Delivery Screen
r500
1000
500
MICHIGAN TO GAMMA
GAMMA TO BETA
BETA TO ALPHA
ALPHA TO ALPHA PC
Future
State
1 Ghanges
At Alpha Motors Assembly it was possibleto eliminate a kitting operation and deliver parts directly from receiving to lineside.At the same time, a simple pull system was introduced between final assembly and wiper subassembly to cur the amount of invenroryin half while smoothing the flow. At Beta Wipers the team took advantageof the approachdescribedin Creating Continuous Flow ro relocate4 formerly srand-alone tasks into one cell while reducingthe number of productionassociates needed from five to three. At the same time, the team createdleveled pull loops from the supermarketin the shipping areato the assemblycell and from the assemblycell to the supermarketin receiving,to reduce invenrories and smooth flow. Finally, at Gamma Stamping, the batch narure of che stamping and painting operationswas acceptedfor the moment. Rather than trying to introduce continuous flow by cellularizingthese operarions, the team focusedon introducing leveled pull loops berween the three operations and reducing set-up times (from one hour ro three minutes on the two stamping presses and from 30 minutes to five minutes in the painr booth). This permitted much smallerbatchesto be made,with frequent replenishmentof the downstreamsupermarketsin small amounrs. Note that the extended map itself seemshardly ro have changed.All of the facility boxes and flow arrowsare as they were. Yet the summary figures in the facility data boxes are now considerablydifferent and the data in the summary box at rhe lower right corner is different as well. Specifically,the total number of steps has been cut from 73 to 54 and total throughput time has been reduced from 44 to 24 days.All of the indicatorsof value stream performancein Future State 1, compared with the Current State,are shown on the next page. Euen more important, each firm participating in this shared ualue stream has quickly taken concrete steps to eliminate wdste and improue performance in its own operdtions.This is not an example, as we see all too often, of downstreamfirms and facilities lecturing upsrreamfirms and facilities on improving their performancewhile doing nothing about their own performance.
Frrtrrre
State
I Srrrnrn
ary'
Future State 1
Gurrent State
Total Lead Tme
44.3 daye
o.oB%
11%
15%
Inventory Turns
5 400 B 7
I 200 B 7
Ouality Screen
(defectsat the downstream end over defectsat the upstream end)
Delivery Screen
(% defectiveshipments at the downstream over o/o defective shipments a t u p s t r e a me n d )
5300
5300
At the level of the scampingplant, che component assemblyplant, and the final assembly plant these changesare often truly impressive.In the most striking instance- the Beta Wipers component assemblyplant in Reynosathe number of steps at has been cut by 60Voandthe throughputtime has been slashed by75%. However,in terms of the entire value stream,as experiencedby the customerat the end, the changein performanceis more modest:a 25% reduction in the number of steps and 46% reductioriin total throughput time, which is still much longer than the end customer is willing to wait. Thus the whole value stream is still producing to a forecastrather than to confirmed order.What's more, the performanceimprovements only assumethese magnitudeswhen every facility touching the product achievesits future state.
PARTIV: FUTURESTATE 1
55
This realizationprovides a useful insight to rhe value srreamteam abour the limits of isolated,individual action: If you want ro achieve a breakthrough- a "game changer" - that altersyour position in your industry or producesprofits far above industry averages, you'll need to optimize the entire value stream rather than stopping after improving the flow along small coursesof the streamwithin your own facility - as many managersand firms do today. Any firm unwilling or unable to implement the Future State I in its facilitiesis unlikely to be willing or able to take the next steps to achieve Furure State 2. 'Iherefore, if it becomesapparentat this point that some participantswon'r make this commitment, it will be critical to find alternativevalue stream members befcrre other participantswaste time in futile efforts.An obvious additional question for thc firms downstreamto ask is, "Do we want to keep the do-nothing upstream firms in our supply base?"
T h e D i sta n ce
S ti l l to Go
While the first five items in the summary box show a substantialimprovement between the Current State and Future State 1, the last three items - the cleliverv screen,the demand amplificacionscreen,and travel distanceshow no change. This is becausethese indicatorsare driven by relationsbetween facilities rather than activitiessolely within facilities.The next challengefor the team therefore is to tackle relationsbetween the facilities.This necessarily requires tackling operationalrelationsbetween firms.
56
Future State 2
As the value stream team achievesFuturc State 1 within each facility and begins tcr sensethat collectivemanagementof the value streamis possible, it's time to take thc next leap.This is to draw and quickly achievea Future StateZ, introducinga smoorh and leveledpull alongwith frequent shipmentsbetween eachof the facilities.
I n s ta l l i n g
L e ve l e d
Facilities
In concept,this is very simple. What we want to do is to link each point of use of the product in a downstreamfacility with the prcvious point of production or shipment in the next upstreamfacility. In this way, consumption at thc point-of-useis tluickly and exactlyreplenished by the nexr upsrreamprocess. In practice,shipping qr-rantities u'ill be considerably largcrthan minimum trrroduction quantities, even in a very lean value stream.For example,the minimum shipping quantity of wipers to the final assemblyplant in this casc is one paller with 20 tra1,s of wiper arms with each tray containing 16 wiper arms,for a total of 320 wipers. It is simply too expensiveto ship individual rrays,much lessindividualwipcrs. The minimum productionquantity,by contrast, would be one tray of 16 wipers. This is becauseset-up times and cosr ro alrernareberween -fyp" A and Type B wipers in the two trim levels are now zero at thc Beta Wiper Plant, after implementingF'utureState 1. But itwould still be too expensivefor materials handlersto wrap and movc individualwipers. Therefore, to level production to the maximum extent feasible as orderstravel back upstream,we will want to send production signalsto the work cell at Beta by travs ratherthan by palletsand to level theseorders.For example,if 20 trays(one pallet) are ordered by Alpha Motors Assembly with the order consistingof: 5 trays of Type A, High Trim (which we will call Part #1) 5 trays of Type B, High Trim (Part #2) 5 trays of Type A, Low Trim (Part #3), and 5 Trays of Type B, Low Trim (Part #4)
PARTV: FUTURESTATE 2
57
1I1I1|1I1t2t2t2t2t2t3
t3 t3 t3 t3 t4t4t4t4t4
By repeatingthis production leveling processar every link upstreamwe will continuallv smooth production rather than creatingwaves due to batchine. In practice,there are many ways to achieve this result. Some firms install pull systemson a strictly manual basisby collecting kanban cardsfrom trays and phoning or faxing these ordersback to the next upstream facility. There, kanban signalcardsare written up and sent to the finished-goodssupermarketto assemblethe next shipment. (When plants are very close together and shipments from the next upstreamfacility occur many times a day - not the casein our example - the cardscan be sent back with the truck bringing the new parts and returning the empty pallets. For many years,this was the primary method of information transferin Toyota City.) A small step up in automationwould involve the use of an electronicreader ro scanthe kanban cardsfrom emptied trays and send this information through an Electronic Data Interchange(EDI) network to the next upstream facility. There, new kanban cardscould be printed and releasedto the finished goods supermarkerto inserr in trays and place in pallets for the next shipment. (When these trays are received in pallets ar rhe downstream facility' the cardscan be scannedagain to confirm receipt and trigger supplier paymenr. They would be scannedone last time - and discardedto complete the cycle - when they are removed from the empty trays as the parts are consumed in the downstreamprocess.) The cardsremoved from trays in the upstream supermarketas product is shipped would then be placed in some type of load-leveling(heijunka) device before transmission upstream to the previous processing step. A further step in automation that has become atrractiverecently is to substitute a simple web-basedinformation transfersystem for the EDI link. The bar code scanningand the printing of new cardsat the upstreamfacility remain the samebut now the dam are senrover the web. (This configurationof information managementis shown in the diagram below.) Still a further step is to eliminate the cardsaltogetherand send electronic signalsdirectly from the downstreamprocessto the supermarketin the next upsrreamprocesswhere shipping instructionscan be displayedon screensor hand-held devices.However, we alwaysstart to get anxiouswhen information disappears into complex electronic sysrems
Electronic
I I
kanban
rrsing
a bar
code
reaeler
-t
I I I I I I
Heijunkadevice
I
t - - - r - - - - ! - - - t
I I I I
I I I I I ! I I I
---1
via Web
t@
rytttt+l
: l-------:
|Card I l?nnterl
f cr,.dl lg,cannerl
-i '
Clean,?aifii"&.Bake
A99EMBLYCELL
Gamma 1tamping
Note that the rows in the heijunka box are for the four types of parts in this product f a m i l yw h i l e t h e c o l u m n s( a c r o s s t h e t o p ) a r e f o r t h e p i t c h ( r a t e )o f w i t h d r a w a lo f t h e cards for conveyanceto the upstream paint process.
DetaWipers
whose inner workings are opaque to line managersand production associates. We advise using the simplestpossiblesystemthat can get the job done, acknowledging that some businesses inherently require more complexity in information managementthan others. The key point to note about each of these arrangements is that there is no need to send day-to-dayproduction instructionsdown from MRPs in the plant office or ar company headquarters. Nor is there a need for customersto send daily releases generaredby their schedulingcomputers.Rather than requiring elaboratecalculationsin a centralized processing system on what should be produced in each plant and at each machine - given expected operatingconditions and pre-established lead times - the new system simply, reflexively re-ordersfrom the next upstream point what hasjust been consumed by the nextdownsrream point. Note that the telephone-based expediting loop, which was often rhe real schedulingsysrem in the Current State and in Future State 1, is now gone. If small amounts of parts are re-orderedand shipped automatically, accurately, and frequently in responsero acrual use, the need for expediting is eliminated. We've drawn this new information managemenr system in our Future State 2 map.
PARTV: FUTURESTATE 2
Demand
Amplification
GAMMA ORDER
GAMMA PRODUCTION
BETA ORDER
BETA PRODUCTION
ALPHA ORDER
ALPHA PRODUCTION
RM 16
wtPo
FG 12 2 Shitrs
5 Davs
60
Wiper Value Stream Future State 2 Shovving Level Pull Betvveen Facilities
;:,^,-
rl
loiaeroanul
Dirmingham,Ml
t\
I I I
-T
I Daity I
Z1z1--l Atpha
I Distribution I Cen+er I
#r-*r
| ll
t=fl
Cleveland,OH
32oo I
bxof -%
l xDay
RM 15
wt?2
FG14 2ShifIs SDays EPE='lDav Defects = 5 ?P-
o.5 d.
and
Delivery
Screen
% DEFECTIVE DELIVERIES 10
1000
dofective deliveries
500
@....
0
MICHIGAN TO GAMMA GAMMA TO BETA BETA TO ALPHA ALPHA TO ALPHA rc
The Need
for Gontrolled
Experiments
"But," you will say,"how can you do this for information flow for only a single value streamco-mingledwith many others?The samecompurer sending signals to control this streamis alsoschedulingother streams. Surely the whole sysrem must be changedin order to changeanything and this, realistically, is a massive and costlyundertaking." Actually massivechange is nor necessary. Just as we have disconnectedour sample product family from the MRPs within severalplants in Future Stare 1, and installed simple pull loops between activities within each plant, we can disconnectindividual value streamscurrendy running between facilities under centralconrroland install simple pull loops. The key point is for the value stream ream ro take this opportunity ro try the experiment and judge the results.We confidently predict that the performance of the value stream as mapped in Future State2 will argue forconverring more and more product families to simple pull systemsso rhat the overly complex production control systemscommonly in place today are graduallyconverted to an activity where they are actually useful. This is capacityplanning on a total systembasis.
Lean
Lab
62
7^ \-t
HUU Futn
Installing Frequent Transport Loops
The logicaland necessary complementro pull systems berweenfacilitiesis increased shippingfrequenciesbetween facilities.This can be achievedby converting infrequent full-truck direct shipments between two facilities to frequent milk runs involving severalfacilities. This has an additional and substantialbenefit. The introducrion of milk runs and more frequent deliveriesmakes it possibleto eliminate the srop at the Beta Wipers warehousein Harlingen and the long excursionto the Alpha Motors crossdock in El Paso.This saveseight steps and six days of throughput time and a thousand miles of rransport.(Plus, if the parts for other value streamsusing these facilities are treated similarly,the facilicies themselvescan be eliminated with major cost savings.) We've drawn these changesin the F'uture State 2 map by substituting our icon for milk run replenishment loops for the striped push arrowsused in the Current State and Future State 1. milkrun replenishment
/c
Introducingpull loopsand milk runs on an experimental basiswill require a modest investment, but bounding the experiment can keep the amounts small until resultsare in and a decisionis made on whether whole production systemsshould undergo conversion.And often these days,other suppliers and customerswithin an industry are alreadyusing milk runs. Perhapsyour product can tag along.
PARTV; FUTURESTATE 2
6:
7o variation
Demand
Amplification
35 30 25 20 15 10 5
GAMMA ORDER GAMMA PRODUCTION BETA ORDER BETA PRODUCTION ALPHA ORDER ALPHA PRODUCTION
4-/v/l
W^^A
<-
Dearborn{eiqhts,Ml
ii
ii iv
wipe.g
RM 16 WIPO FG12
21hifr,s 5Davs E?E = 1 Dav Defects = 50 ppm
beta
I
F.M24 wtP 62 FG'12. 3 thifr,s
SOays E?E = 1 Dav Defects = 25O ppm
Reynosa,Mexico
o.3 d.
4.o d. (3131e.) 20 (3)
64
W
loraeraanu I
birmingham,Ml
L _T
I ouiry I
I Diotribution I Cenrer I Cleveland,OH
ZLa-'1 Atpha
I I I
ll I x: I I
l-J-J
(\7 "*^-1
---% Ioxoxl
1x D a y
o.5 d.
7 (2)
o'uality
PPM DEFECTS ' 2000 1500
and
Delivery
Screen
defeds
V DEFECTIVE DELIVERIE 10
1000
de{ective deliveries
500
@.....
0
MICHIGAN TO GAMMA GAMMA TO BETA BETA TO ALPHA ALPHA TO ALPHA PC
Frrtrrre
State
2 Srrrnrnary
Current State Future State 1 Future State 2
44.3 days
15.8 daye
o.o8%
11%
o.6%
21% 14
Inventory Turns
I 200 I 7
Ouality Screen
(defectsat the downstream end over defectsat the upstream end)
400 B 7
50 3
5
Delivery Screen
(% defectiveshipments at the downstream over Yodefective s h i p m e n t sa t u p s t r e a me n d )
5300
5300
4300
Totaling
the
Results
The consequence of smoothpull signals and frequent replenishment for our eight indicatorsof value stream performanceis shown in the summary boxes on the Future State 2 map and in the chart above.The most striking change from Furure State 1 to Future State 2 is the dramatic reduction in demand amplification,qualiry problems,and late shipments as the orders move back upstream.The amount of variation experiencedat Michigan Steel is now much closerto the very low level of variation at Alpha Motors Assembly.In addition the dramatic reduction in shipping complexity and lag time between the creation of a defect and its discoveryat the next downstreamprocesshas causeddefects and shipping errorsat the upper end of the value stream to convergeon the low levels at the lower end of the value stream.
66
Compressing
So far we have left every value creatingactivity in its original place,changing only informationflows and shipment frequencies while eliminating unneeded warehouses and cross-docks. Although the value stream team has cut the number of stepsfrom 73 to 39, reduced throughput time by 64%, and greatly damped demand amplification,much waste and long time lags remain. Becauseit appearsthat moscof the remaining waste and time are due to the need to move the product between many facilities and over long disrances, a logical next step is "value stream compression"to relocateand co-locate value-creatingactivities so they can be performed fasterwith lesseffort.
What
is the
logic
of
relocation?
The first principle is simply that all manufacturingsteps in the product shouldbe moved as closetogetheras possible. Ideally this would even be in the same room. A secondprinciple is that the closerthis compressedsequenceof activities is to the customer- Alpha Motors Assembly in our example - the better. The objective of lean thinking, after all, is to reduce cosrsand improve quality while getting customersexactly what they want when they want it. Remote manufacturingalwaysworks againstthis goal becauseit increases responsetime once the customers'desiresare known. The unavoidable consequence for remotely located manufacturerswho are determined to immediately serve their customersis to createinventoriesof finished units produced to (usually inaccurate)forecasts. In the currenr global security environment, where shipments acrossbordersare subject to disruptions, this is even more the case. "Do it all in one place" and "locate that place next to thg customer" are useful principles to get started.However, a critical third rule is necessary: That if proximity should entail extra manufacturingcosts(although the reversewill be more common), these cosrsmusr be weighid againstthe value of the time savinqs.
These principlesin combinationsuggest a very simple locationalgorithmfor most products: 1. If the cusromcris in a high labor-cosr counrry(e.g.,the [J.s.,Japan,Germany) and needsimmediateresponse to orders,and if the product has relativelylittle labor content, conduct all of the manufacturingsteps in close proximity and close to the customer in the high-wagecountry. 2. If the customeris in a high labor-cost countrv,is willing to wait for some shipping interval, and the product is price sensitive,manufacturethe entire product, from raw materials to finished goods,in closeproximity in a low-cosrlocale,shipping Only the final goods.In our experiencethe correct location is almost alwaysat a low-wagecountry within the region of sale.F-orexample, Mexico for the fl.S., China fbr Japan,Poland fbr Germany.Shipment of the finished product by rruck, or a short ferry ride, and across only one border can still permit responseto thc customerwithin a few days,while shipment by seafrom anothercontinent requireswecks. 3. If the customerin a high labor-cost country needsimmcdiate response but thc product has high labor content, do a careful costing exerciseto determine the correctlocationof manufacture. The best locationmight vary from a very low-wage site in anotherregion of the world, with the product even deliveredby air, ro a new technologyremoving high-costmanufacturinglabor in the high-costcounrry of sale and permitting the conduct of all manufacturingstepscloseto the customer. 4. If the customeris in a low labor-cost country and scalerequirements permit, manufacrurethe entire product - from raw marerialto finished goods- in geographicproximity in that counrry. As the wiper value streamteamslooked at the situationand ponderedtheserules it becameapparentthat the best locationfor an iclealstatein this casewould be immediatelyadjacentto the vehicle assembly plant in the high-cost country (the L.f .S.)This was becausethe amount of direct labor content in the product was actuallyvery small,indeed only thirty seconds at rhe wiper assembly plant and a vanishinglyslight amounr ar rhe stampingplant. (The number.ofwiper assembly operatorsrequired had already been reduced from five in the Current State to three in Future State 2.) The team found that a small increasein direct labor costsfrom even when traditional corporateoverheadswerc added to direct wage costs- would be more than offset by a big reduction in shipping, inventory, and generalconnectivity costs. relocation of this assembly step from Mexico ro rhe tl.S. -
68
--
l d e a l S ta te
Changes
The value stream team therefore created the Ideal State map shown on the next page.Note that wiper assembly(inclLrdingthe blade-to-armassemblystcp previouslyconductedin Alpha'sassembly plant),painting,and stampinghave now been compressed into one room in a "supplier park" clnthe site of the Alpha plant. A cheaper, Motors assembly low-speedstampingpresshas been introduced, which we call a "right-sized"tool because its capacityis prt-'portional to the requirementsof this value stream.This pressis also able to make both the ;rrimary and secondarystampingsfor all of the other parts needed for the wiper assembly (seethc schematic drawingon pages12 and 13 showingtheseparts)and in very small batchesto minimize inventories and lead times.A mini paint booth - a secondright-sizedtool - has also been designedand is located between the stampingstep and wiper assembly. Bccausethc new wiper manufacturingmodule gets an electronic signalon what plant to build next as eachvehicle leavesthe paint booth in the vehicle assembly (a 3-hour lead time) and because the time neededfrom the startof wiper assemblv until delivery to the final assembly linc is lessthan thc available lead time, wipers with high and low trim for vehicle models A and B can now be assembledto line sequence. They are then placedin line-sequenced traysof 40 wipers and conveved to the fit point on the final assembly line every twentv minrrtesby a "water spider" (a small cart pulled by a converted fork-lift). The water spider lo<-rp connectsseveral plant, bringing back similarcomponentplantsadjacentto the Alpha final assembly empty trays and needed parts to the wiper assemblyareaon each circuit.
Oc
s
PARTVI: THE IDEALSTATE
Demand
Arnplification
GAMMA ORDER
GAMMA PRODUCTION
BETA ORDER
BETA PRODUCTION
ALPHA ORDER
ALPHA PROOUCTION
Wiper
Value
Stream
ldeal
State
EaetOrange,NJ
o.4 d.
70
W
Dearborn.Ml Alpha ?roduction Alpha Control Materials Control Alpha Distribui;ion Center
%
t--J'
/
Daily
I I t
I--ELL
-l
I-cE[_-l
AlphaMotors
@Gl-ffiful
o.5 d.
1 . 1 d(.3 1 6 1 s . ) 20 (6) O.8 d. (12O s.) 1
7 (2)
Ouality and Delivery Scron
MICHIGAN TO GAMMA
GAMMA TO BETA
BETA TO ALPHA
ALPHA TO ALPHA PC
71
ldeal
State
Srrrnrn
Gurrent State
Total Lead Time Value Percentageof Time
( v a l u ec r e a t i n g time to total time)
44.3 days
23,9
daye
15.8 daye
2.&
days 1.5%
o.oB%
11%
o.16%
o.6%
21% 14
15%
27% 79 2.5
Inventory Turns
Ouality Screen
400
200
50
Delivery Screen
B 7
B 7
3 5
( % c h a n g ei n d e m a n da t d o w n s t r e a m e n d o v e r % c h a n g ei n d e m a n d a t u p s t r e a me n d )
5300
5300
4300
525
Dramatic
Changes
Throughput time from raw materialsto customerhas now been reduced by 94Vo t<tZ.t3 days, and practicallyall of the transport links, invenrories,and handoffsthe key drivers of connectivity costs- have been eliminated, from thc final assembler back through the wiper maker to the stamperand raw materials supplier.In addition,it is hard to tell where one company leavesoff and the next picks up the valuc stream bepause activities formerlv conducted by Alpha, Beta, and Gamma at locationsthousandsof miles apart arenow being conducted in continuous flow in one room located acrossthe road fiom the customer.
72
Winners
Need
to Gornpensate
Losers
As future state and ideal state maps are drawn up, it will quickly become apparent that positive change is most likely if the team can find a way for 'l'his winnersto compensate losers. is because it will commonly be the casethat a downstreamparticipant can get better value at lower cost if an upstream participant leavesout wasted steps,implements leveled pull systemswith its suppliers,introducesmore capableprocesstechnologies, and relocatesactivities.However, even when everyonecan see that the incremental savingsexceed the incremental costsof these initiatives, little is likely to happenunlessupstreamparticipants are compensatedby downstreambeneficiariesfor taking costly actionsthat <,rptimize the wholc.
If it were easily possible to compare total product cost before and after the future state improvemen[s, compensation might be
an easierissue.Howevel traditionalpurchasing and accounting systems are often incompatiblebctween value stream participantsand in any case are poorly suited for calculatingproduct costsfor each product family. Thesc systems typicallyrequire enormousamountsof data to allocate overheadsby product and they usually fail to calculatecostsin a way that all participants will acceptas valid. We proposekeeping it simple by ignoringtraditionalsystems and instead determiningthe incremental cost (in some common currencyunit) and the incremental benefit (in the same currency unit) of each proposedchange in 'I'his the value streamin future and ideal states. is surprisingly easyin manv casesand can change the focus of the value stream team from redressing (or defending)the mistakesand inequitiesof the past to discovering win-win-win alternativesft-rr the future. The problem of cross-firm compensation will not be such an issueif the product being mapped is new and the courseof the value streamis not constrainedby existing facility locationsor even existing suppliers. However, it will still be important to calculateconnectivity costsfor various configurationsof the value stream to see which one will actbally produce the best combinationof low cost and rapid customcrrcsponse.
73
Timing
Lower-speedpresses will be cheaperand more capableif used on new part designs,and a change in the raw marerial provider will be required as well. (Note that New Jerseysteel is to be substitutedfor Michigan steel in rhe Ideal State,to reduce shipping distancefor steel coils from 500 miles and eight hours to 25 miles and one hour.)Therefore, the best time to leap to the Ideal state will be with the next product generation, when new process equipment will be needed in any case. The exerciseof creating an ideal srate ro contrastwith a business-as-usual state should be conducted for every new product generation.This can lead to a very creativejoint mapping of the ideal sratefrom the very start of the next design,when the barriersto doing everything right are greatly reduced.
Final Risk to Avoid In developing the examples for seeing the'whole we have learned of another risk for the value srreamteam to avoid. This is ro turn the mapping exerciseinto a conventionalcost study for a product family by trying ro map the flow of every part going into the product. when teams do this we've found that they lose sight of the key point. This is that the types of wasre exposedand the demand amplification discoveredare also presenr in every product family passingthrough all of the participant firms. The first purpose of the exerciseis to raiseconsciousness about systemicproblems and to spur the development of systemicsolutionsrequiring better performance by the functions, not to shavea bit of cosr out of one specific product and then declarevictory.
74
Achieving
Future States
Value stream maps at the macro-levelare very useful for raising consciousness about wasteand the lack of customerresponsiveness in today'stypical current state,a situation often invisible to value stream partnerslooking only at their own operations.However, if consciousness is raisedbut no future stare is achieved the whole mapping exercisejust createsmore corporatewallpaper- pure mwda. How can you actually achieve future stateswhen many departments and firms must cooperateand no one person or firm is legally "in charge"? We have already suggestedthat progressis best made in a seriesof steps beginningwith the easiest. If a Future State 1 can be achievedthat reduces time and effort within each participatingfirm, this will give all of the value stream partnersthe courageand incentive to go further. Then, if Future State 2 can be achievedas well - addressing production control problems to stabilize demand, remove noise,cut costs,and enhance responsiveness to the customer- the momentum for improvement will be much stronger.The prospectsfor successful introduction of the Ideal State, with its requirementsfor investment and relocationof activities,then become much brighter. Running the process in the oppositedirection, beginning with a big leap to - particularlyfor entirely new an ideal state,may be possiblein some cases productsand we certainly don't want to discourage value streamteams in a position to make rhis leap. However, in the great bulk of instances, small stepswill be essentialat the srartro lay the groundwork for big leapslater.
In our experience, Furure state 1 can be achievedin about three months after completion of the Current Stare map. F uture State Z canbe in place in six months after the achievementof Future State 1. Howeveq conditions will vary and it may be more practicalfor the value stream ream ro begin implementingFuture state 2 even if Furure state 1 is not completelyin placeand stabilized. This is because many of the activitiesinvolvedare quite separate and can proceedin parallel. 'I'he timing for the ldeal State may range from "soon" (particularlyfor new products)to "much later". The team in our exampleconcludedthat
the new supplier park configurarioncan be in place in four years,at the point that the nexr generationof vehicle Models A and B with redesigned wiper sysremsis introduced. Tiying to move faster would mean that Beta and Gamma would need to continue their remote operations for their other customersand would incur substantialcostsfor duplicate rooling and underutilization of their existing facilities. Even if the precisetiming of the later srares is harclto determinenow the simple act of writing down all of the necessary steps and agreeingon specific target dates for achieving specific steps has the highly useful effect of converting vague intentions and "no year" projectsinto concrete, trackabletasks.
76
policy deployment or if you have alreadydeveloped facility-level value stream plans of the type shown in Part Y of Learning to See(and Part VI of Creating Continwous Flow). However, it will be a bit more complicated becausethis plan builds on the "YearlyValue Stream Plan" for each facility being developed at the same time, as illustrated in Learning to See. The wiper value stream team developed a simple value stream plan, as shown on the next page.
YEARLY
GOAL (measurable)
OUAR T ER LY
2oo2
F91
Leadtime=15.8 days lnventnry turne = 14 Quality ecreen = 50 Delivery screen=3 Demand amplification gcreen=5
F92
Leadtime=2.8 daye lnvenlory turns =79 Quafityscreen=2.5 DeJivery ecreen=1 Demand amplification Sareen= 1
O Start
A Completion
78
::
SIGNATURES
\ 'A L U E
STREAM
PLAN
SCHEDULE
2()()3
o o
A A
O On target
Behindtarget
CONCLUSION
At the end of this brief breakthrough guide for achievingfuture and ideal srates we must sharea secret:You'll never actually achieveyour ideal state! It rurns out that there is always more waste to removc and that value for the customer can alwaysbe further enhanccd. For example,wipers might some day be molded as a singlepiece in matchingbody colors, eliminatingthe need for the stamping,painting,and final assembly of considerable numbers of parts.If cycle times for these activitieswere at or below takt times for wipers on the final assemblyline and if changeovers from onc wiper color and specificationto the nexr were also (or at leastwithin takt time), it woLrldbe possibleto mold wipersto essentially instantaneous line secluence with total throughputtime and value crcatingtime both shrinkingto seconds. At that point, the "Ideal State" portrayed in this workbook will appearto be full of mwdal However, there'sa companionpoint that alsoseemsto be a secretto manv managers. This is that successive futurc statesgetting much closerto the ideal statecan be achieved- by real managers in real firms building current product designs- in only a short period of rime even when there is no "value streamdictator" giving orders.And even more can be accomplished with the next generationof products,before machinesand faciliticsare locked in place. The trick is to take a walk together so everyonecan see the whole. Then estimatethe "prize" availableto the group if the whole value streamcan be optimized. Then devise a mutually acceptable way to split the loot if the current state "Bank of Muda" can be robbed. It won't happen all at oncc and you'll probably never reach that huppy land of completely frictionless cooperationbut the challengeis to get started,gain some initial successes, and not look back. As firms and departments learn to see togetherit should alsobe possibleto make your maps ever more inclusive,eventuallvreachingall the way from the customer's use of the product through the life cycle back upstreamto inchoatcmatter before any processing. And wc believe it will be attractiveto map wider and wider range of goods and servicesincluding office processes, as many readershave already startedto do with the micro-mapsin Learning to See.(For example, we at LEI have alreadyheard from readersabout mapping gold mining, fish stick manufacture,postal sorting operations,insuranceclaims processing, the writing of technical manualsfor complex aerospace products,and visits to the doctor.)Bccausethere is alwaysa value stream whenever there is a product (whether it's a good, a service,or some combination), we are confident that consciousness will continue to spreadabout the pcltential of value streammapping. Wc wish you the best in your endeavors and hope to hear about your problemsand your successes.
About
Dan Jones
the Authors
Dan is co-author of rhe Machine That changed the'world and Lean Thinking. He is a SeniorAdvisor to the Lean EnterpriseInstitute (LEI), and Chairman and Founder of LEI's affiliate organization, the Lean EnrerpriseAcademy in the uK (www.lean.uk.org). He has long had an interest in mapping enrire value srreamsand took the lead in developing the examplespresentedin Chapter Z of Lean Thinking. These began with the humble can of cola that requires319 days to passthrough six different companiesand nine facilities acrossthe world, firms and facilities that collectively conduct only three hours of value-creating activities before the cola finally reachesrhe cusromer.
Jim Womack Jim is co-author of The Machine That Changed the'World and Lean Thinking and Presidentand Founder of the Lean Enterprise Institute. He fin<isit hard not to think about extended value streams including thoseinvolvinghealthcare, mobilitv food, communication, construction, defense,and logistics.
82
AP P E N D IX
A - E xte n d ed
Value
Str eam
M apping
lcons
The icons and symbolsfor current and future state mapping fall into three categories: Material FIow, Information Flow, and General Icons. Material lcons
Notes
One processbox equals an area should be of flow. All processes labeled.Also used for departments, such as ProductionControl.
O u t s i d eS o u r c e s
Data Box
tl
|
lhl ----+ |
-
.'--+
Cross-Dock
Warehouse
ln
NS_
|-lr-.---I
Lvdl-J
I
P l a n eS h i p m e n t Note frequencyof shipments. T r a i nS h i p m e n t Note frequencyof shipments.
-?
I Mon. I
| + wed.l-l
T r u c kS h i p m e n t
Inventory
C o u n ta n d t i m e s h o u l db e n o t e d .
APPENDIX
Material
lcons
Notes Material t h a t i s p r o d u c e da n d moved forward before the next processneedsit; usuallybased on a schedule.
M i l kR u n
o o
ooooooooo
ExpeditedTransport
Supermarket
A controlledinventoryof parts that is used to scheduleproduction at an upstreamprocess. P u l l o f m a t e r i a l su , s u a l l yf r o m a supermarket. Indicates a deviceto limit quantity and ensure FIFOflow of material between processes. Maximum quantity should be noted.
Withdrawal
max.20 pieces
-FIFO*
Information
lcons
Represents
Notes
+-
ManualInformation flow
Electronic Informationflow
Information
84
Information
lcons
Notes r" anban. The "one-per-containek Card or devicethat tells a process how many of what can be produced a n d g i v e sp e r m i s s i o n to do so. Card or devicethat instructsthe material handlerto get and transfer parts (i.e.from a supermarket to the c o n s u m i n gp r o c e s s ) . "a n b a n . T h e " o n e - p e r - b a t c hk S i g n a l sw h e n a r e o r d e rp o i n t i s reachedand another batch needsto be produced. U s e dw h e r e s u p p l y i n g processmust produce in batches because changeovers are required. P l a c ew h e r e k a n b a na r e c o l l e c t e d a n d h e l df o r c o n v e y a n c e .
W i t h d r a w a lK a n b a n
S i g n a lK a n b a n
t
Kanban Post
,/t\\t
I
lr
K a n b a nA r r i v i n g in Batches
Load Leveling
ControlCenter
Phone
Itr't., I
General lcons
Orders
Represents Operator
APPENDIX
plant, west
orange,
NJ
Alpha Crosb-Dock
-F-----> *
E|?aso.TX
l92OWiperslDay \ 12bOA 6406 16WiperslTray 32O Wiperel?allet 4A 26
\
Receiving
Kitting
"'?A+
2560 A zAOD
Wipers
us
160A 808 Wipere
l xDay FACILIry9UMMARY
FinalAeeembly &Test
2h. 1 60 e. 1 (1)
12h. 1
2h.
1
APPENDIX
Assem bly
Plant,
Reynosa,
Mexico
20o2
Gamma Saamping
Tonawanda,NY
Harlingen,TX
|
|
I t,ooonau"t I 1Z?allets
I
@ ZOothox
ril-lDuv I
br-'d
Receiving
A=eP
25,600 A 12,8000 ?arts 4324 224b ?arts
Assemblyl
Qvt
C I T= 1 9 e " " .
ryAsf q9r
4324 2249 Wipers
Aseembly2
E?E=1Day
B.Oh.
B.Oh.
4.3h.(1oe.) 3 (1)
aa
Alpha Motors
Harlingen,TX
h
I weetrv I
Detroit.Ml
Y
l92OWiperelDay zAO A 640b
Qr
C I T= 1 O s e c . ClO =5 min. Uptime=95% 2Shifts E?E=1Day
ryA\il
4324 2249 Wipers
lnspect, &Test
2Shifts
Q2z
CIT =20 eec.
ryAsf
640 A 3200 Wipers
B.Oh.
12,O h. 4.5h. 1
4.2h.(1oe.) 3 (1)
APPENDIX
plant,
Tonavvanda,
Ny
DearbornHeighte,Ml
Receiving
"zA\+
336 coilg
1tam pi ng2
fT]
10m.
Cleveland,OH
9etaWipere Warehouse
2 xWeekly th i p 5 ch edule
12?allets
Clean,?aint&Bake
Shipping
E?E=$Q2yg
Defects=2OOO?pm
Defective = 6%
C IT= 5 2 mi n .
CIO = 3O min. Uptime= 85% E?E=1week
4Bh.
5 (1)
APPENDIX
Assernbly
plant, west
orange,
NJ
beta Wipers
Alpha Cross-Dock
El?aso,
Wiper 1ub-Adoembly
Alpha Dist,.Center
/r
// |_LJ-_--.l
t-dl-J
FACILITY9UMMARY
l xDay
RM15 h. w t ?2 h . FG14h.
29hifts SDays
,F
E?E=lDay
Final&,eembly _F IFO*
12h. 60 s. 1
2h.
1
2 (1)
prant, Reynosa,
Mexico
Tonawanda.NY
t-
..... rNr 1T
Cell
I --
V box ,.)
Vt/'a
i----------------'
I I I l
+ J 4- -
,, &the
A9SEMDLYCELL I A9SEMFLY
ll
-1 I lo\e
= 30 sec. CIO = 5 min. =1OO% 25hifts
?tamped ?aris
g4
Alpha Motorg
Harlingen,TX Detroit,Ml
Harlingen,TX
->% toxoXl
D-
RM 16 h.
wl?oh.
FG12h.
Defecfts=4OO ppm
1 2h .
APPENDIX
Ap p e n d i x C : Ga mma S ta mping Asser nbly Plant Tonavvanda, NY Future State May 2OO2
Cleveland,OH
iw a'
DearbornHeights,Ml
Tonawanda,NY
9tam pingl
9t amping 2
96
OetaWipers Warehouse
FACILITY9UMMARY
RM4A h. w t ?6 2 h . F G1 2 h .
SShifts
thipping
Defective = 6%
APPENDIX
{-
Feedback
we've tried to make this workbook easyto use, wirh detaired instructions, simple illustrations,and clear examples.However, we know fiom yearsof experiencethat applying even the simplesrconcept in a complex organization is difficult. So we need your help. After you have tried implementing the techniquesdescribedin this workbook,pleasemail, fax, or email commentsto: Lean EnterpriseInstitute, one cambridge center, cambridge, MA 0zr42 usA trax:617-871-2999. Email: stw@lean.org o Reach us ar: lean.org
r--