Assessing leanness level with demand dynamics in a multi-stage production system

(1)

Assessing

leanness

level

with

demand

dynamics

in

a

multi-stage

production

system

Rehab

Ali

Department of Management of Technology, Nile University, Giza, Egypt, and

Ahmed

Deif

Department of Industrial Technology and Packaging,

California Polytechnic State University, San Luis Obispo, California, USA

Abstract

Purpose_– The purpose of this paper is to present a dynamic model to measure the degree of system_’s

leanness under dynamic demand conditions using a novel integrated metric.

Design/methodology/approach–The multi-stage production system model is based on a system

dynamics approach. The leanness level is measured using a new developed integrated metric that combines efficiency, WIP performance as well as service level. The analysis includes design of experiment technique at the initial analysis to examine the most significant parameters impacting the leanness score and then followed by examining different dynamic demand scenarios. Two scenarios were examined: one focussed low demand variation with various means (testing the impact of demand volumes) while the second focussed on high demand variation with constant means (testing the impact of demand variability).

Findings– Results using the data from a real case study indicated that given the model parameters,

demand rate has the highest impact on leanness score dynamics. The next phase of the analysis thus focussed on investigating the effect of demand dynamics on the leanness score. The analysis highlighted the different effects of demand variability and volumes on the leanness score and its different components leading to various demand and production management recommendations in this dynamic environment.

Research limitations/implications – The presented lean management policies and

recommendations are verified within the scope of similar systems to the considered company in terms of manufacturing settings and demand environment. Further research will be carried to extend the dynamic model to other dynamic manufacturing and service settings.

Practicalimplications– The developed metric can be used not only to assess the leanness level of the

systems which is very critical to lean practitioners but also can be used to track lean implementation progress. In addition, the presented analysis outlined various demand management as well as lean implementation policies that can improve the system leanness level and overall performance.

Originality/value– The presented research develops a novel integrated metric and adds to the few

literature on dynamic analysis of lean systems. Furthermore, the conducted analysis revealed some new aspects in understanding the relation between demand (variability and volume) and the leanness level of the systems. This will aid lean practitioners to set better demand and production management policies in today_’s dynamic environment as well as take better decisions concerning lean technology investments.

Keywords Lean manufacturing, Manufacturing management, Demand management

Papertype Research paper

Nomenclature

AOUT(t) actualoutputattimet CTi(t) cycletimeattimetforstationi

AQL acceptablequalitylevel D(t) numberofdefectsattimet

B(t) backloglevelattimet DD(t) deliverydelayattimet B0(t) initialbacklogattimet DR(t) demandrateattimet COT(t) changeovertimeattimet DSR(t) desiredshipmentrateattimet

(2)

DT delaytime(duetoquality) FGI(t) finishedgoodsinventoryattimet FOR(t) filledorderrateattimet FR(t) finishingrateattimet IT(t) inspectiontimeattimet LS(t) leannessscoreattimet MA(t) machineavailabilityattimet MOPT(t) minimumorderprocessingtime

attimet

MSR(t) maximumshipmentrateattimet

NAT netavailabletime

NOT(t) netoperatingtimeattimet OEE(t) overallequipmenteffectiveness

attimet

OSL(t) overallservicelevelattimet OWE(t) overallWIPefficiencyattimet

PDT planneddowntime

PE(t) performanceefficiencyattimet PIR(t) productioninputrateattimet

PRi(t) QC(t) QCOR(t) QCSR(t) QR(t) QS(t) SPT SR(t) TDD TGWIP(t) THCT THOUT(t) TWIP(t) UPDT(t) WIPi(t)

productionrateattimetfor stationi

qualitycontrollevelattimet qualitycontroloutputrateat timet

qualitycontrolstartrateattimet qualityrateattimet

qualitysignalattimet scheduledproductiontime shipmentrateattimet targetdeliverydelay targetWIPattimet theoreticalcycletime theoreticaloutputattimet totalWIPattimet

unplanneddowntimeattimet WIPlevelattimetforstationi

1. Introduction

Due to global competition and market dynamics, companiesall over the world are

undertremendouspressuretoreducetheircostsandincreasetheirservicelevel.There

are many tools and strategies that manufacturing organizations can implement to

achievethesegoals.Leanmanufacturingisoneofthestrategiesthatwasimplemented

bymanycompaniesintheirefforttoattaintheshortestcycletimeandreducecostsby

eliminatingdifferentwastesintheirsystems.Thesuccessofcompaniesthatadopted

leanprincipleshasincreasedtheinteresttowardleanmanufacturing.Thecompanies

and organizations that are in other areas such as the service industry and the

healthcare sectorarealsoapplying leanprinciples toreducetheircost andimprove

their systems. Although many companies are interested in lean and trying to

implementleantools,thepercentageofcompaniesthathaveasuccessfulleaninitiative

isnotveryhigh.AccordingtoChristopher(2000),leanworksbestinhighvolume,low

variety and predictable environments. However, reality reveals that today’s

manufacturing settings are faced with different dynamics and uncertainties than

these ideal conditions. An example of such uncertainty is the customer demand.

Demanddynamicsisacriticalparameterintoday’senvironmenttosustaincustomer

satisfactionandkeepcompetitiveedgeforfirms.Minorfluctuationsindemandatthe

end-userortheretaillevelcausehighvariation indemandforupstreampointinthe

supplychain.ThisisreferredtoastheBullwhipeffect.

This paper addresses two important aspects in lean systems’ implementation.

The first aspect is how to measure the leanness level from an integrated and

dynamic perspective. Second, the paper aims at studying the impact of demand

uncertainty on the leanness level dynamics. The approach starts with

proposing anew dynamic model to measure the degreeof leanness using system

dynamics (SD) approach. This is followed by a design of experiments (DOE)

technique to test the different model parameters significance on the developed

(3)

scenarios areinvestigated toprovide insightsand recommendationsfor managing

bothinternalleanparametersas wellasexternaldemandmanagementdecisionsto

secureasuccessfulleanimplementation.

2. Literature review

Researchaboutleansystems,philosophies,toolsandtechniquescanbefoundinWomack

and Jones (2003), Black (2007), Abdulmalek and Rajgopal (2007), Staats et al. (2011),

Elsayedetal.(2013)andYangetal.(2015).Thereviewinthissectionwillfocusonleanness

assessmentaswellasdynamicanalysisofmanufacturingsystemsusingSDsinceboth

topicsarerelatedtotheproposedwork.

Examplesofresearchwork-relatedleannesslevelassessmentincludeachecklistof

36indicatorsthatwasdevelopedtoassessthechangestowardleanmanufacturingby

SanchezandPerez(2001).Usingtheresultsfromasurveyformanufacturing plants

located in the Spanish region of Aragon, they analyzed which lean production

indicatorsweremoreusedtoassessthecompany’simprovementsintheirproduction

systems, and the determinants on the use of these indicators. They divided these

indicatorsintosixgroupswhichreflectthemajorcharacteristicsofleanas:elimination

of zero value activities, multifunctional teams, continuous improvement,production

and delivery JIT, supplier integration and flexible information system. Savsar and

Al-Jawini(1995)usedsimulationtoanalyzetheperformanceofjust-in-timeproduction

system. The model measured the effects of random processing times, numbers of

kanbans between stations, variabilityindemand, line length andkanban operating

policies on some system performance measures, such as throughput rate,

work-in-process inventory and station utilizations. Elnadi and Shehab (2014) presented a

conceptualmodelthatcanbeusedinmeasuringthedegreeofproduct-servicesystem

(PSS) leanness in UK manufacturing companies. The model assessed PSS leanness

based onfive leanenablers (supplier relationship, managementleanness, workforce

leanness,processexcellenceandcustomerrelationship),21criteria(supplierdelivery,

cultureofmanagement,processoptimization,etc.)andfinally73attributes.Dettyand

Yingling(2000)useddiscreteeventsimulationasatooltosupportorganizationswith

the decision to implement lean manufacturing through quantifying the benefits

achieved from applying lean principles. The current status of lean production

assessmentinChinawasalsoevaluatedbyTaj(2008).Ninekeyareasofmanufacturing

wereevaluatedinthis assessmentwhichwere:inventory;team approach;processes;

maintenance;layout/handling; suppliers;setups;quality;andschedulingandcontrol.

Theresultsweredisplayedinthescoreworksheetandaleanprofilechartwascreated

todisplaythestatusoftheplantandthegapsbetweenthecurrentsituationandtheir

specific lean targets. Almomani et al. (2014) proposed an integrated model of

lean assessment and analyticalhierarchy process (AHP)to definetheroute of lean

implementationbasedontheperspectiveprioritiesforimprovement.AHPwasusedto

settheprioritiesofimplementingleaninthedifferentperspectivesoftheenterprise,

accordingtothefollowingcriteria:leanradarscore,costofimplementation,financial

gains,timeofcompletion,technologicalandadministrativeobstacles,andthedegreeof

riskinvolved.Quality,cost,delivery,andmotivationkeyperformanceindicatorswere

used tomeasuretheimprovementsachieved viaapplying this approach. Wanet al.

(2007)useddataenvelopmentanalysistechniquestopinpointtheleannessfrontierasa

benchmarkfortheleannessscore.Theresultingleannessscoreindicated“howleanthe

systemis”or“howmuchleaneritcanbecome.”Themetricsusedintheanalysiswere

(4)

tool called the leanness assessment tool (LAT) using both quantitative (directly

measurableandobjective)andqualitative (perceptionsofindividuals)approachesto

assess lean implementation. The LAT measured leanness using eight quantitative

performance metrics: time effectiveness, quality, process, cost, human resources,

delivery, customer andinventory. The previous metrics werecaptured using fuzzy

membership function highlighting both improvement successes and needs in lean

implementation.Fuzzylogicwasfurtherusedinmanyresearchesasameasuringtool

forleanproduction.BehrouziandWong(2011)developedfuzzymembershipfunctions

toassess the lean performance inmanufacturing systems.It supported anefficient

measurement of lean performance by producing a final integrated unit-less score.

Susilawatietal.(2015)proposedaprocedureformeasuringdegreeofleanapplication

whichcombinestheadvantageofthevariousproceduresandimprovetheweaknessof

variousprocedures.Theprocedureusedfuzzynumberforscoringdegreeofapplication

ofleanpracticesfocussinginqualitativefactorslikecultureandsatisfaction.

Variousapproacheswereused todynamicallymodel andanalyzemanufacturing

systemsincludingSDintroducedbyForrester(1961).Examplesofdynamicanalysisof

manufacturingsystemsusingSDisthethree-echelonproductiondistribution system

usedasasupplychainreferencemodelforcomparingvariousmethodsofimproving

totaldynamicperformance byWikneretal. (1991).Helo(2000) showshow agilityis

built into supply chains. Three simulation models were studied: the demand

magnificationeffect insupply chain, theanalysisof capacitysurge effects, andthe

trade-off between capacity utilization and lead times. The analysis recommended

smaller order sizes, echelon synchronization and capacity analysis as methods of

improvingtheresponsivenessofasupplychain.Poles(2013)exploredtheinteraction

betweenthe physical flow, information flows and company policiesto generate the

dynamics of the re-manufacturing process. Deif and ElMaraghy (2014) presented

dynamicsystemsapproachtostudychallengesofimplementingproductionleveling

and itsassociated costs in a leancell producing at takt time. Results showed that

determiningthemostfeasiblelevelingpolicyishighlydictatedbyboththecostand

limitations of capacity scalability. In addition, delivery sequence plans of different

products/parts is needed to achieve mix leveling and lot sizes affect the feasible

productionlevelingpolicywhileimplementingleanprinciples.Zhangetal.(2012)used

system engineering concept to compare the SD models of traditional supply chain

and leagile supply chain to show the benefits of leagile supply chain. The results

showedthatshortenthelengthofsupplychain,sharetheinformation,cooperationand

productiondelaycaneffectivelyweakentheBullwhipeffect.Deif(2012)examinedthe

performance of a lean cell under uncertainty using SD. The cell performance was

compared under certain and uncertain external (demand) and internal (machine

availability) conditions. Results showed that although lean cell was expected tobe

responsive toexternal waste, however, this was not thecase under the considered

uncertainconditions.Huang etal.(2012)comparedbetweentwomodelsfor asupply

chainundertwoconditionsofsupplydisruptions,withoutbackupsupplier,andwitha

contingent supplier. The retailer’s total profits are also compared under these two

circumstancesofsupplydisruptionstoassistthedecisionmakersbetterdealingwith

thebackup purchasingstrategy.Thesupplychain studiedonlyinvolvesoneretailer

and two independent suppliers that are referred to as major supplier and backup

supplier. Georgiadis and Michaloudis (2012) investigated the impact of dynamic

disturbances in manufacturing process on the production planning and control in

(5)

different arrival patternsfor customer orders andtheexistence ofvarious real-time

events related to customer orders and machine failures. They also assess the

performance bydeterminingthebackloggedorders,WIPinventoriesandtardyjobs.

Ali andDeif(2013) usedSDtomodel amanufacturing systemandinvestigatedthe

effectofshrinkingthetargetdeliverydelay(TDD)toincreasetheresponsivenesslevel

ontheleannesslevelofthesystem.

From the aforementioned review, there are much more work related to lean

manufacturing implementation and analysis than that dedicated for assessing the

degreeofleanness ofsystems.Inaddition, fewer workaccountforsuchassessment

fromadynamicperspective. Adeeperanalysisofthereviewed papersfocussingon

leanassessmentrevealsthattheassessmentapproachesfocussedonmetricsthatare

more endogenous (internal system parameters like throughput, quality WIP, etc.)

ratherthanexogenous(externalorcustomer-relatedparameters).Thefewerresearch

whichincludedcustomer-relatedmetricstoassessleanperformancemainlyfocussed

on response time to measure customer satisfaction. It is clear that with lean

systemsdedicatedtovaluecreationandgiventhatcustomersdefinesuchvalue,more

customer-related metrics should be used in assessing lean performance. It is also

important to note that dynamic approaches are more favored in assessing lean

performance of manufacturing systems than static approaches. In the reviewed

literature,limitedworkconsideredthedynamicnatureofassessmentespeciallywhenit

comestocapturingdemanduncertainty.Also,thereviewshowsthatSDasapowerful

dynamic modelingtoolhad beendedicatedmainlytogeneralmanufacturingsystems

applicationorsupplychainwhileveryfewworkusedSDintheleanassessmentcontext.

Thisresearchcontributestothefewliteratureavailablefordynamicassessmentoflean

systems. One contribution is through integrating service level as a customer-related

measure with WIP andoverall equipment effectiveness(OEE) as internal performance

measuresinonenewmetric.Furthermore,thisworkproposesaSDmodeltocapturethe

dynamicimpactofvarioussystem’sparameters(includingforthefirsttimetheuncertainty

of demandinterms ofboth volumeandmix)on leanperformanceof amanufacturing

system.The newmodelandmetric inaddition tothepresentedanalysis offersvarious

insightsandrecommendationsforleanmanagementpolicies.

3. Dynamic manufacturing model with leanness score

The model introduced by Ali and Deif (2014) is modified as shown in

Figure 1. The model is developed using Vensim™as aninteractive SD simulation

environment.Dataforarealcasestudyinthecentralkitchenindustrywillbeusedto

demonstrate the model, measure the degree of leanness and assess the impact of

significantparametersonthenewleannessscore.Themodeledmanufacturingsystem

consistsoffourcomponents;productionsystem,backlogsystem,qualitysystemand

leannessscorecalculationsystem.Theproductionsystemincludesfivestages,threeof

them arededicatedfor assemblyand/or manufacturing,one for finishingprocesses,

andthefinal onefor finishedgoodsinventory which ishighlylinkedtothequality

inspection rate.All stagesarecontrolledbystochasticcycletime ateachstage.The

differentmodelcomponents(stages)areexplainedindetailsinthefollowingsections.

3.1 Modelcomponents

3.1.1 Production system. A multi-stage production system representing several

productionactivitiesisusedtomodelthemanufacturingsystemasshowninFigure2.

(6)

Inspection time Quality Changeo v er time Control QC output r ate QC star t r ate Quality r ate CT 1 CT 2 CT 3 production unit Def ects Time Unit WIP f or WIP f or WIP f or WIP f or finishing Quality signal stage 1 stage 2 stage 3 Production r ate 3 CT 4 Production r ate 2 Production r ate 1 input r ate Production < Theoretical Leanness AQ L output > score Ov er all WIP < Machine Finishing r ate efficiency A v ailability > T otal WIP T arget WIP Finished Deliv er y dela y Dela y time Ov er all ser vice THCT in v entor y le v el Shipment r ate goods < production unit > Bac klog Minim um order Desired Filled order r ate Maxim um Shipment r ate Demand r ate processing time shipment r ate Initial bac klog T arget deliv er y dela y OEE P erf or mance efficiency Theoretical output Machine < Production r ate < Quality r ate > A v ailability 3 > Net oper ating time Actual output < Production r ate Net A v ailab le 1 > Time Planned do wntime Unplanned < Production r ate do wntime 2 > Scheduled production time Figure1. Dynamic manufacturing model

(7)

Figure2.

Production system

where the WIP level is controlled by changing the production rate as shown in

Equation(1).TheWIPlevelateachstationiscalculatedbythedifferencebetweenthe

productionrateofthecurrentstationandtheproductionrateofthesubsequentone:

WIPið Þ ¼t INTEGðIFTHENELSEðPRi 1ð Þt–PRið Þt X0; PRi 1ð Þt –PRið Þt ;0Þ;0Þ (1)

Production rate isexpressed asthereciprocal ofcycle timeas seen inEquation(2).

Theproductionratesofeachstation(withtheexceptionoftheproductioninputrate)

areaffectedbytheavailabilityofthemachinesandtheircycletimes.Forsimplicity,itis

assumed that all stations inside the production system have the same availability.

Thisassumptionwillnotaffecttheanalysisinthiscase.Asforthecycletimestheyare

modeled using stochastic random normal distribution functions as shown in

Equation (3). Normal distribution is an acceptable reflection to the uncertainty

associatedwithtypicalmachinetimesintheconsideredcasestudy:

productionunit MAð Þt

PRið Þ ¼t (2)

CTiðtÞ 100

CTið Þ ¼t RANDOMNORMALðMin; Max; Mean; SD; SeedÞ COTð Þt (3)

Theinputproductionrateisequaltothedemandrate(noovercapacityisconsidered)

asthisreflectsthetraditionalpracticeofleansystemstryingtoproduceatdemand(see

followingequation):

PIRð Þ ¼t DRð Þt (4)

Since the manufacturing system is producing products in the same family, it is

assumedthatchangeovertimeisconstantforall stagesanditisalsomodeled using

normaldistributiontocapturemoreoftheuncertaintyintheconsideredmanufacturing

environment(seefollowingequation):

COTð Þ ¼t RANDOMNORMALðMin; Max; Mean; SD; SeedÞ (5)

3.1.2 Backlogsystem.Orderbacklogmeasuresthedelaybetweentheplacementand

deliveryoforders(seeFigure3).Suchdelayscanbecausedbyadministrativeactivities

Changeover time CT₂ CT 3 CT₁ WIP for Production WIP for

stage 1 _{Production rate 1} WIP for

stage 2 _{Production rate 2} WIP for

stage 3 _{Production rate 3} finishing CT₄ input rate

Demand rate Filled order rate

Finished goods inventory Shipment rate

(8)

suchascreditapprovalandorderprocessing,bytheneedtocustomizeorconfigurethe

producttotheneedsofparticularcustomers,andbydelaysinshippingtothecustomer

site,amongothers.Backlogiscapturedasthedifferencebetweentherequireddemand

rateandtheactualfilledorderrateasexpressedinfollowingequation:

ð ð Þ FORt

B tð Þ ¼INTEG DR t – ð Þ; B0Þ (6)

TheinitialbacklogissettoequaltheTDDofincomingorders(seefollowingequation):

B0ð Þ ¼t TDD DRð Þt (7)

Whiletheshipmentrateandfilledorderratearenumericallyequivalentandhavesame

units (product/hour), they are two dissimilar concepts as shown in Equation (8).

Theshipmentraterepresentsthephysicalproductthatleavestheorganization,while

theorderfulfillmentratecapturestheinformationflow(Sterman,2000):

FORð Þ ¼t SRð Þt (8)

TheshipmentrateasshowninEquation(9)dependsontheminimumbetweeneither

thedesiredshipmentrateorthemaximumshipmentrate:

SRð Þ ¼t MINðDSRð Þt ; MSRð Þt Þ (9)

ThedesiredshipmentrateistheratethatensuresthatordersarefilledwithintheTDD

asseeninEquation(10).Inaleancontext,firmswouldstrivetominimizetheTDD:

B tð Þ

DSRð Þ ¼t (10)

TDD

The maximum shipment rate based on the firm’s present inventory level and the

minimumorderprocessingtimeasspecifiedbyfollowingequations:

FGIð Þt

MSRð Þ ¼t (11)

MOPTð Þt

FGIð Þ ¼t INTEGðIFTHENELSEð ð ÞFRt –SRð Þt X0; FRð Þt –SRð Þt ;0Þ;1Þ (12)

Finished goods inventory Shipment rate

Backlog

Minimum order Desired

Shipment rate

Maximum

shipment rate processing time

Initial backlog

Target delivery delay

Figure3.

(9)

Theminimumorderprocessingtimerepresentstheminimumtimerequiredtoprocessand

shipanorderanditisexpressedasnormalrandomfunctionasshowninfollowingequation:

MOPTð Þ ¼t RANDOMNORMALðMin; Max; Mean; SD; SeedÞ (13)

3.1.3 Qualitysystem.Qualitycontrolleveliscalculatedas thedifferencebetweenthe

qualitycontroloutputrateandqualitycontrolstartrateasshowninfollowingequation

(seeFigure4):

QCð Þ ¼t INTEGðIFTHENELSEðQCSRð Þt –QCORð Þt X0; QCSRð Þt –QCORð Þt ;0Þ;1Þ (14)

Thesamplingplaninthissystemisconductedafterthethirdproductionstagewherea

specifiednumberofsamplesaresentforqualitytestingasshowninfollowingequation:

QCSRð Þ ¼t PR3ð Þt ðsamplesizeÞ (15)

Thequalitycontroloutputrateequalsthequalitycontrolstartratewithafixeddelay

functionaccountingforinspectionandreworktimesasshowninfollowingequation:

QCORð Þ ¼t DELAYFIXEDðQCSRð Þt ; ITð Þt ; QCSRð ÞÞt (16)

Inspectiontimeisexpressedasrandomnormalfunctiontoreflectthestochasticnature

oftheinspectionprocessasshowninfollowingequation:

ITð Þ ¼t RANDOMNORMALðMin; Max;Mean; SD; SeedÞ (17)

The inspectionprocess isdesignedbasedontheacceptablequality level(AQL)and

directlycontrolsthefinishingrate.Ifthenumber ofdefectsislessthantheAQL,the

finishingrateisdelayedbyonlytheinspectioncycletime.However,ifthenumberof

Inspection time Quality Control QC output rate QC start rate Quality rate

Defects Time Unit WIP for

Production rate 3 finishing CT₄ Quality signal

<Machine

Availability> Finishing rate

<production unit_> Finished

Figure4. _goods Delay time

(10)

defectsisgreaterthan AQL,thefinishingrate isdelayed byinspection timeandan

extrareworkdelaytimeasindicatedinfollowingequations:

Productionunit Productionunit MAð Þt

FRðtÞ ¼IFTHENELSE QSX1; ; (18)

CT4ð Þt CT4ð Þþt DT 100

QSð Þ ¼t IFTHENELSEðD tð ÞpAQL;1;0Þ (19)

ThenumberofdefectsisdisplayedinEquation(20)asastochasticfunctionofquality

controloutputrate:

Dð Þ ¼t RANDOMNORMALðMin QCOR; Max QCOR; Mean

QCOR;SD QCOR; SeedÞ TimeUnit (20)

3.1.4 Leannessscoresystem.Thenewleannessscoresystemdevelopedinthismodel

consistsofthreemetrics(showninEquation(21))which are:overallwork-in-process

efficiency(OWE),OEEandoverallservicelevel(OSL)asdisplayedinFigure5.

Thenewassessmentsystemismeanttocapturethreefundamentalleanoutcomesin

anysystem,mainlyproductionstability/leveling(reflectedintheWIPlevel),production

efficiency(reflectedinqualityandavailability)andresponsivenesstomarket(reflected

in service level). The leanness score is presented as percentage ensures that the

assessment is dimensionless for tracking and comparison purposes. In this paper,

thescoreistheaverageofthethreecomponentsandallthethreecomponentshavethe

sameweightwhichcanbechangedinfurtherresearch(wheresomeleanresultscould

beofconcerntothemanufacturingmangersandthuswouldreceivehigherweight).

Thethreecomponentsoftheproposedleannessscoreareexplainedasfollows:

OEEð Þþt OWEð Þþt OSLð Þt

LSð Þ ¼t 100 (21)

3

3.1.4.1 Overall WIP efficiency. OWE is an indicator of the accumulation of

WIP overtime reflecting internal efficiency as well as stability (see Figure 6).

A manufacturing unit must balance WIP to maintain productivity for known and

futurerequirements. So,organizationsshould establish itstargetWIP andcompare

theiractualWIPwiththistargetasshowninfollowingequation:

TGWIPð Þt TWIPð Þt

OWEð Þ ¼t IFTHENELSE TGWIPð Þt pTWIPð Þt ; ;

TWIPð Þt TGWIPð Þt (22) Overall WIP Leanness score efficiency Overall service level OEE Figure5. Leanness score system

(11)

Figure6. Overall WIP efficiency Figure7. Overall equipment effectiveness

Target WIP level is calculated based on little’s law multiplying the demand rate

(assuming that in the ideal lean state it should equate the production rate) by

thetheoreticalcycletime.ActualWIPlevelistheaverageofallWIPaccumulatedover

thedifferentproductionstages(seefollowingequations):

TGWIPð Þ ¼t DRð Þt THCT (23) TWIPð Þ ¼t P4 i¼1WIPið Þt (24) 4

3.1.4.2 OEE.OEEisameasureforleanmanufacturingenvironments,anditshouldbe

viewedasa“ContinuousImprovementEngine”thatprovidesarobustoutlineforthelean

journey(see Figure7). OEE consistsof three factors:performance efficiency, machine

availabilityandqualityrate.Itcanbecalculatedasdisplayedinfollowingequation:

MAð Þt PEð Þt Rð Þt

OEEð Þ ¼t (25)

100 100 100

TheavailabilityportionoftheOEEmetricrepresentsthepercentageofscheduledtime

that the operation is available to operate and it is calculated as Equation (26).

Theavailabilitymetricisapuremeasurementofuptimethatisdesignedtoexcludethe

WIP for WIP for

finishing stage 1 stage 2 _{Production rate 2} stage 3 Production rate 3

Production rate 1 Overall WIP efficiency Total WIP Target WIP THCT Demand rate Demand rate OEE Performance

efficiency Theoretical output

Machine _<Production rate

<Quality rate> Availability Net operating time 3>

Actual output

<Production rate Net Available

1>

Time

Unplanned <Production rate

Planned downtime

downtime 2>

Scheduled production time

(12)

effects of quality, performance and scheduled downtime events. The losses due to

wastedavailabilityarecalledavailabilitylosses:

NOTð Þt

MAð Þ ¼t 100 (26)

NAT

Net available time is calculated as the scheduled production time subtracted by

planneddowntime(seefollowingequation):

NAT¼SPT PDT (27)

Net operating time is the unplanned downtime subtracted from net available time

(seeEquation(28)).Unplanneddowntimeistheunscheduleddowntimeeventsanditis

calculated asa random functiontoreflect thestochasticuncertainty ofmachinesin

productionenvironmentsasshowninEquation(29):

NOTð Þ ¼t NAT UPDTð Þt (28)

UPDTð Þ ¼t RANDOMNORMALðMin; Max; Mean; SD; SeedÞ SPT (29)

Theperformanceportion oftheOEEmetricrepresentsthespeed atwhich thework

centerrunsasapercentage ofitsdesignedspeed. Theperformancemetricisapure

measurementofspeedthatisdesignedtoexcludetheeffectsofqualityandavailability.

Thelossesduetowastedperformancearealsooftencalledspeedlosses.Itiscalculated

as thepercentage betweenactual outputand theoreticaloutput (see Equation (30)).

Deviation from target output indicates cases of either lower production or

overproductionandbothareconsideredwastesfromleanperspective:

THOUTð Þt

PEð Þ ¼t IFTHENELSE AOUTð Þt XTHOUTð Þt ;

AOUTð Þt

100; 100 (30)

THOUTð Þt

Actualoutputisthenetoperatingtimetimestheminimumofallthethreeproduction

stagesratesasshowninfollowingequation:

AOUTð Þ ¼t NOTð Þt MINðMINðPR3ð Þt ;PR2ð Þt Þ;PR1ð Þt Þ (31)

Theoretical(ortarget)outputisthemultiplicationofscheduledproductiontimeandthe

demandrate(seefollowingequation):

THOUTð Þ ¼t SPT DRð Þt (32)

The quality portion of the OEE metric represents the good units produced as a

(13)

Figure8.

Overall service level

measurementofprocessyieldthatisdesignedtoexcludetheeffectsofavailabilityand

performance.Thelossesduetodefectsandreworkarecalledqualitylosses:

QCð Þt Dð Þt

QRð Þ ¼t 100 (33)

QCð Þt

3.1.4.3 OSL. OSL reflects the level at which the customer orders is filled on time

(seeFigure8).ItistheratiobetweentheTDDandactualdeliverydelayasshownin

Equation(34).Theactualdeliverydelayistheratiobetweenthebackloginthesystem

andthefilledorderrateasseeninEquation(35):

TDD OSLð Þ ¼t (34) DDð Þt Bð Þt DDð Þ ¼t (35) FORð Þt

4. Significance of demand dynamics on leanness level

Theanalysisinthispaperisbasedonthedataofanindustrialcasestudyofakitchen

equipmentmanufacturer.ThecompanyisoneoftheleadingcompaniesinEgyptinthe

field of central kitchen equipment since 1980. Their business scope is importing,

designing,manufacturingandsupplyingallcentralkitchensequipmentandlaundries

forhotels,touristicvillages,restaurantsandhospitals.Thecompany’sdatausedbythe

developedmodelisshowninTableAI.

Beforeinvestigatingtheimpactofdemanddynamicsonthedevelopedleannessscore,

thesignificanceofdemandamongotherparametersinthepresentedmodelwasexplored.

ADOEtechniquewasemployedforthatpurposewherethesixstochasticparametersof

the15inputparameterswhereselectedfortheDOEanalysis.Excludingthedeterministic

parameterswasbasedonourinteresttoexplorethedynamicsgeneratedbyuncertainty

andvariability.ThesixparametersareshowninTableI.

A fullfactorial DOE method wasimplemented where each factor hadtwo levels

selected as a minimumandmaximum. Thevalues foreach of theparameterswere

based onthe selected casestudy historicaldata. The 26¼64 runs werecarriedout

whilemonitoringtheresponse(leannessscore).Fullfactorialwasusedtoidentifymain

effects andfactorinteractions withoutanyconfounding orambiguity (Montgomery,

2005).AllrunsandtheirresponseareshowninTableAII.

Thesignificanceofmainfactorsandthetwofactorsinteractionswereinvestigated.

ThenormalplotforstandardizedeffectsisshowninFigure9.

Backlog

Delivery delay Overall service

level Target delivery

(14)

From Figure 2, the most significant factors are demand rate (highest), defects,

changeover time with demand rate interaction, demand rate with unplanned

downtime interaction, defects with demand rate interaction, unplanned downtime,

minimumorderprocessingtime,changeovertimeandtheyareshowninFigure10

withtheirrelativesignificance.

Itisshownfromtheseresultsthatdemandratehasamajoreffectontheleanness

score.Thesignificance DOEanalysiswasusedtoconfirm andpavetheroad tothe

importance of exploring the impact of demand dynamics on the leanness level of

manufacturingsystems.Understandingdemandvariabilityanditsimpactonleanness

level and how it relates to other internal parameters is vital in successful lean

implementationandthiswhatwillbepartiallyexaminedinthenextsections.

5. Impact of demand dynamics on leanness score

Afterexaminingtheparameters’significanceandillustratingthatthemostsignificant

parameteronleannessscoreinthedevelopedmodelisdemandrate,inthissection,the

leannessscoreanditscomponentsareinvestigatedunderdifferentdegreesofdemand

variability.Demandvariabilityiscapturedusingcoefficientofvariation(CV).CVrefers

toastatisticalmeasureofthedistributionofdatapointsinadataseriesaroundthemean.

Parameter Minimum Maximum

COT D DR IT MOPT UPDT 0.003 (5/100) _×QCOR 10 0.005 6 0.016 _× SPT 0.005 (10/100) _× QCOR 20 0.007 7 0.03 _× SPT TableI. Parameters values 99 95 90 80 70 60 50 40 30 20 10 5 1

Normal Plot of the Standardized Effects

(response is Response, = 0.05) 0 –200 –400 –600 –800 –1,000 –1,200 –1,400 CF CE BF BE BC AC AB F E C B A Not Significant Significant Effect Type COT B C D E F Factor Name A Defects DR IT MOPT UPDT Figure9. Significance plot % Standardized Effect

(15)

Figure10.

Pareto chart for significance

TableII.

Data for low demand coefficient of variation with various means values

Pareto Chart of the Standardized Effects

(response is Response, = 0.05) 2 Te rm C B AC CF BC F E A AB CE BF BE AF AE EF DE CD AD DF D BD 0 200 400 600 800 1,000 1,200 1,400 Factor Name A COT B Defects C DR D IT E MOPT F UPDT Standardized Effect

Itrepresentstheratioofthe standarddeviationtothemean.Thedataaretakenfrom

thecompanyhistoryandgatheredfromthelast10years.

First,theleannessscoreanditscomponents’willbeexaminedunderlowlevelsof

demandvariation(CV¼0.33,0.5,0.22)withvaryinglevelsofdemandmeans.Then,the

dynamicsassociatedwithhigherlevelsofdemandvariation(i.e.CV¼0.5,1.5,1)with

constantmeanwillbeinvestigated.

5.1 DynamicsoflowdemandCVwithvaryingmeans

ThebehaviorofthemodelisinvestigatedinthissectionunderoveralllowlevelofCV.

However,theaverageandthestandarddeviationoftheselecteddemandpatternsare

changingatdifferent magnitudestobeabletoseethebehaviorofthesystemwhen

excited under different means of demand. Table II shows the different values of

demandmean,standarddeviationandtheirCV.

Inthisanalysis,variousdemandpatternsareconsidered.Thefirstpatternisoneof

highdemandaverageandhighstandarddeviation(thehighestCV)anditisreferredto

asDR1,thenextpatternisthelowdemandaverageandlowstandarddeviationanditis

referred toas DR2. Finallythe system isinvestigated under whatwe labeledas the

normalaveragedemandreferredtoasDRwhichfitsbetweentheprevioustwopatterns.

Suchanalysiswillenableleanpractitionerstoenvisiontheimpactofdifferentdemand

excitation patterns to the systems and also draw a clearer picture of how demand

variabilityingeneral(goingfromlowtonormaltohighandalsovaryingbothaverage

CV Demand rate value

DR 0.33 RANDOM NORMAL (10, 20, 15, 5, 1)

DR1 0.5 RANDOM NORMAL (21, 63, 42, 21, 1)

(16)

andstandarddeviation)canhelpbetterunderstandhowaleansystemreactsandthus

developoptimaldemandmanagementpoliciesaswellasinternaloperationalplans.

5.1.1 TheimpactoflowdemandCVsontheOEE.Figure11showstheeffectofthe

considereddemandpatternsontheOEE.

It is shown from the figure that the highest effectiveness for equipment, in the

selected settings, was maintained with the normal demand rate DR which has a

mediumCV.Themainreasonbehindthisperformancewasthatthecurrentproduction

levelaveragewasclosetothemeanofthisdemandpatternleadingtohighproduction

efficiency.DR1which hasthehighest CVshows thelowestOEEdueit’s meanhigh

deviationfromthecurrentproductionrateandalsoduetohighlevelofvariabilitythat

impactbothquality andproductionavailability.Ontheotherhand,DR2;whichhave

thelowestCVhas a relativemoderate OEEperformance betweentheprevious two

demand patterns. The cases of higher or lower values of demand means than the

production levelwillhave anegative effectonthe OEEmetric.Thisis because the

theoreticaloutputdependsonthedemandrate,sohigherorlowervaluesofdemand

willaffecttheratiobetweenthetheoreticaloutputandtheactualoutputnegativelyand

inturntheperformanceefficiencyandOEE.Thevariabilitylevel(reflectedintheCV

values)wasnotthemajorcontributortotheOEEperformance.

5.1.2 The impact of low demand CVs on the OSL. The effect of the considered

demandpatternsontheOSLisshowninFigure12.

ItcanbeobservedthatDR2patternhasthehighestandfastedlevelofOSLperformance,

whileoverlongertime(64hoursoreightdays)theOSLforDRpatternreachesthesame

100percentlevel.Incontrast,producingtochasetheDR1patternwillgettheOSLlevelto

decreaseovertimetillitreachesnearzeroattheendofthesimulationperiod.

Theimpactofvariabilityisclearinthisanalysis.SinceDRpatternhadahigherCV

thanDR2pattern,thisvariationcausedthebacklogtoaccumulatetilltheshipmentrate

wasincreasedviamoreproductionandfinallyreachthe100percentservicelevelover

longertime.ThesystemattheDR1pattern(whichhasthehighestvaluefordemand

meanandthehighestCV)showedthelowestservicelevelperformance.Thismanifests

OEE 1 0.75 0.5 0.25 0 0 16 32 48 64 80 96 112 128 144 160 Time (Hour) OEE : DR2 OEE : DR OEE : DR1 Dmnl Figure11.

OEE with low demand variation and various means

(17)

Overall service level 1 0.75 0.5 0.25 0 Dmnl 0 16 32 48 64 80 96 112 128 144 160 Figure12.

OSL with low demand variation and various means

Time (Hour)

Overall service level : DR2

Overall service level : DR1

Overall service level : DR

that boththedemand volumeandvariabilitycontributed tothedeteriorationofthe

OSLintheconsideredcasestudy.Thissuggeststhatimprovingservicelevelsinthese

typesofdemandpatternsrequiresinvestmentsandupgradesinthecurrentproduction

capabilities.ItalsosuggestsrelaxingtheTDDtobeabletocopewithsimilardemand

patternsasademandmanagementpolicy.

5.1.3 The impactof lowdemand CVs onthe overallWIP efficiency. Overall WIP

efficiencybehavioraccordingtothedifferentdemandpatternsisshowninFigure13.

Since production started with zero WIP, the WIP efficiency was initially zero.

Afterrunningtheproduction,theWIPstartedtoaccumulatetillitreachedthedesired

targetWIPleadingtoanOWEof100percent.However,withtheconsidereddemand

patterns,theWIPcontinued toaccumulate duetotheimplemented pushproduction

Overall WIP efficiency 1 0.75 Dmnl 0.5 0.25 0 0 16 32 48 64 80 96 112 128 144 160 Figure13.

OWE with low Time (Hour)

demand variation Overall WIP efficiency : DR2

and various means Overall WIP efficiency : DR1

(18)

policyandwithnoWIPcontrolleanpolicies(likesupermarkets)leadingtoadropinthe

OWEtillitreachedclosetozerooverextendedtimeThevariabilitylevel(CVvalue)for

eachdemand patternaddstotherelativespeed ofthedeteriorationofitsrespective

OWEbehavior.Inthesescenarios,companieshavetoworkonsmoothingtheflowof

productionutilizingdifferentpullingtechniquesfromdownstream.

5.1.4 TheimpactoflowdemandCVsontheleannessscore.Figure 14 displays the

effectoftheselecteddemandpatternsonthenewdevelopedleannessscore.Thefigure

integratesthe OEE,OSLand OWEdynamicbehaviorsintotheleannessscoremetric.

Theleannessscoreforthe consideredcasestudyatthe DRpattern(withthe average

demand mean and average CV) starts with a low performance then it increases till

reaches the highest performance level among the three patterns (about 60 percent).

AlthoughtheLeannessscoreforthesystematDR2pattern(lowestdemandmeanwith

lowestCV)islowerthanwiththeDRpattern(averagedemandlevelandaverageCV),it

showsamoresteadyperformancethanwiththeDRpatternduetotheOSLbehaviorthat

reacheditsmaximumlevelquickly.Leannessscoreforthe systemattheDR1 pattern

(withthehighestdemandmeanandhighestCV)startsatanaverageperformancethen

decreases to less than 20 percent as the impact of its high volume and variability

manifestthemselvesindeterioratingboththeOSLaswellasOWEwithinthesystem.

5.2 DynamicsofhighdemandCVwithconstantmeans

Themodelisstudiednextunderhigherdegreesofdemandvariability(CV)whilekeeping

samemeanfordifferentdemandpatterns.TableIIIshowstheselecteddemandpatterns.

Leanness score 80 60 40 20 0 0 16 32 48 64 80 96 112 128 144 160 Time (Hour)

Leanness score : DR2 Leanness score : DR

Leanness score : DR1

Dmnl

Figure14.

LS with low demand variation and various means

CV Demand rate value TableIII.

Data of higher

DRL 0.5 RANDOM NORMAL (10, 20, 15, 7.5, 1) demand variation

DRH 1.5 RANDOM NORMAL (21, 63, 15, 22.5, 1) with stable means

(19)

Figure15.

OEE with higher CV and stable means of demand

5.2.1 The impactof higher demandCV on the OEE. Figure15 shows the effectof

higherCVswithstablemeansofdemandrateontheOEEmetric.

ThefigureshowsthatthesystemrespondingtothedemandratewiththelowestCV

(DRL)hasthebestOEE(averageof80percent)whilerespondingtothedemandrate

withthehighestCV(DRH)leadtotheworstOEEperformance(averageof45percent).

Althoughthesystemisrespondingtothesamedemandaverage,however,increasing

the variability of the demand will affect negatively both the quality level and the

performance efficiencyleading tothedeteriorationofOEEperformance. Thisaligns

withthecontinuouseffortoftoday’sprocessimprovementapproaches(e.g.LeanSix

Sigma)tocapture,reduceandeliminatevariationswithinmanufacturingsystems.

5.2.2 TheimpactofhigherdemandCVontheOSL.Theeffectofthesamedemand

patternsontheOSLisshowninFigure16.

Theperformanceofthesystem’sOSLwiththeDRMdemandpatternwhichhasthe

mediumlevelofvariation(CV)isbetterthanitsperformancewithdemandpatternwith

lowCV(DRL)whichalsoreaches100percentlevelbutafterlongertime.Thisisbecause

within these parametersettings, therangeofvaluesfor DRL ishigherthanthat for

DRM which caused the system to experience some higher orders insome instants

leadingtohigherbackloginthecaseofDRL.Theseinstantsresemblecaseslikesudden

increaseinvolumeorrushorders.Thesystem’shadtheworstOSLwhenexperienced

thedemandwiththehighestCVleadingtoanearzeroperformanceovertimereflecting

the disability of the system to cope with this level of demand variation while

maintainingthetargetduedatesduetointernalturbulence.

5.2.3 TheimpactofhigherdemandCVontheoverallWIPefficiency.Thebehaviorof

overallWIPefficiencyfortheselecteddemandpatternsisshowninFigure17.

TheperformanceoftheOWE ofthe systemwithalldemandpatternsdeteriorates

overtime.ThisisbecauseWIPlevelsusuallyreflectsthestabilityofthesystem,andwith

demandpatternsthatexhibitvariations,WIPwilltendtoincreaseinordertostabilizethe

systemresultinginlowOWElevelsandleadingtohavingtheactualWIPtobehigher

thanthetargetWIP.Intheconsideredcasestudy,theaverageproductioncycletimeis

OEE 1 0.75 0.5 0.25 0 0 16 32 48 64 80 96 112 128 144 160 Time (Hour) OEE : DRM OEE : DRL OEE : DRH Dmnl

(20)

Overall service level 1 0.75 0.5 0.25 0 Dmnl 0 16 32 48

Overall service level : DRM

64 80 96

Time (Hour)

112 128 144 160 _Figure_16.

OSL with higher CV and stable means of

Overall service level : DRH _demand

Overall service level : DRL

Overall WIP efficiency 1 0.75 0.5 0.25 0 Figure17.

Time (Hour) _{OWE with higher}

Overall WIP efficiency : DRM _{CV and stable means}

Overall WIP efficiency : DRH _{of demand}

Overall WIP efficiency : DRL

closetotheaveragedemandmeanofallpatterns,however,thevariabilitywithinthese

demandpatternsleadtotheaccumulationofWIPbetweenproductionstages.

5.2.4 Theimpactofhigher demandCVontheleannessscore. Theleannessscore

dynamicsagainsttheselecteddemandpatternsisdisplayedinFigure18.

Thehighestleannesslevelforthe systemwaswiththedemandhavingthelowest

variability(DRL),however,itreachesitsbestperformanceaftersometimeduetotheOSL

lowperformanceasdiscussedearlier.Theleannesslevelofthesystemwiththemedium

variabilityisclosetothepreviousoneyetwithbetterconsistency.Thisconsistency is

duetoexperiencinglessinstants ofsuddenincreaseinvolume orrushorders.Inthis

specificsettings,thecompanyisnotinneedtoputefforttoreducethedemandvariation

level from medium tolow as it does not pay back from leanness score perspective.

Dmnl

(21)

Leanness score

Dmnl

Figure18.

LS with higher CV and stable means of demand 80 60 40 20 0 0 16 32 48 64 80 96 112 128 144 160 Time (Hour)

Leanness score : DRM Leanness score : DRL

Leanness score : DRH

Itisalsoclearthatthe systemwhenexperiencingthehighestvariationwiththe DRH

patternwillhavetheworstleannessprofile.Insuchdemandscenarios,decisionmakers

should pursue more effort indemand management to avoid or reduce this level of

variationoradjustthesystemcapabilitiestocopebetterwithsuchlevels.

6. Summary and recommendations

Thispaperpresentedanewmetrictoassessthedegreeofleannessofamanufacturing

systemunderdynamicconditionsandwasfurtherdemonstratedusingarealcasestudy.

The newmetric integrated OEE, servicelevel and WIP performanceto measurethe

system’s leanness degree. This integration offers a more comprehensive leanness

assessment that reflects system efficiency, responsiveness tocustomers and internal

stability.Suchintegrationiscriticalwithinleansystemsthatfocusonvaluecreationto

customersaswellasinternalsystemperformance.Furthermore,themodelcapturedthe

dynamicsassociatedwithdifferentmanufacturinguncertaintiesandtheirrelationtothe

developedleannessscore(includingforthedemandvolumeandvariabilitydynamics).

A DOE analysis was carried out to examine the most significant parameters

affectingtheproposedleannessscorewhichwasshown,giventheselectedinputdata

ofthecasestudy,tobethedemandrate.Thiswasfollowedbyananalysistoexamine

theimpactofdemandvolumeandvariabilityontheleannesslevelanditscomponents.

The following leanmanagementpolicies andrecommendations (within thescopeof

similarmanufacturingsystemstotheconsideredcompany)wereobtained:

TheOEEdynamicswasnegativelyaffectedinalldemandscenariosbyvariability.

Thisisassociatedtotheadverseeffectvariationhasonqualityandproductivity.

However, it wasshownthatincases oflowvariation(lower valuesofCV),the

higherimpactonOEEwasthesynchronization betweendemandmeanandthe

current production level average. The results suggest that to improve OEE,

demand management policies should reduce variability passed on from the

demandpatternstoproductionandthatproductionplannersshouldworktoalign

(22)

_The_service_level_in_the_presented_model_was_sensitive_to_high_level_of_variation_in

demandshowingverylowperformancewithhighCV.Asvariabilityisreduced,

servicelevelincreasesovertimedependingontheproductionabilitytocatchup

andclearbacklog.Inaddition, instantsofsuddenincreaseinvolumeandrush

orders contributed tothereductioninthe OSL performance intheconsidered

casestudy. Theresponsiveness levelindynamicdemand environmentcanbe

improved (in addition to the typical variation reduction recommendation) by

maintainingadynamicproductioncapacitythatcanscaleupproductivitywith

fastrampupprofiletoclearbacklogaccumulatedduetodemandvariability.

_The_overall_WIP_efficiency _in_the_presented_analysis_demonstrated_how_demand

variation at any level will lead to high levels of accumulated WIP. WIP

accumulation is an expected internal production reaction in order to maintain

stabilityandefficiency.Thisphenomenonwasfurthermanifestedduetothepush

production policies employed by the selected case study. Results confirm the

importanceofabsorbingdemandvariationthroughdifferentleantechniqueslikea

balancedHeijunkaboxesaswellasmovingtowardcontinuousflowifpossibleorat

leastemployingdifferentpulltechniquestoavoidhighlevelofWIPaccumulations.

_The_new_leanness_score_was_demonstrated_to_be_a_successful_assessment _metric

forthesystem’s leannesslevelinthisdynamicenvironment.Resultshighlighted

howthemetricreflectedthegeneraldeteriorationtrendoftheleannesslevelofthe

systemunder high levelof variations evenwith smallervolumes ofdemand. In

addition,thenewscorewasusefulinshowinghowtheleannesslevelofthesystem

usingthethreeconsideredcomponents(andgiventheselectedcasestudy)isvery

closeincasesoflowandmediumvariations.Thiscanhelpleanmangersindeciding

whethertheinvestmentinmanagingdemandtoreducethelevelofvariationfrom

mediumtolowand/ortoincreaseproductioncapacitywillpayoffornot.

The presented work will be extended to include more industrial applications. In

addition,theimpactofstochasticchangeovertimeandinspectiontimeontheleanness

scoreperformancewillbeinvestigated.Furthermore,theeconomicperspectiveoflean

implantationwillbestudiedbyaddingthecostelementintotheproposedmodeland

metric.Finally,changingtheweightforeachoftheleannessscorecomponentswillbe

studiedtoreflectdifferentmanagerialpoliciesandpriorities.

References

Abdulmalek,F.andRajgopal,J.(2007),“Analyzingthebenefitsofleanmanufacturingandvalue streammapping via simulation: a processsector case study_”, International Journal of ProductionEconomics,Vol.107No.1,pp.223-236.

Ali,R.andDeif,A.(2013),“Investigatingtheimpactofthedynamicsassociatedwithincreasing responsiveness level on leanness”, Proceedings of Industrial and Systems Engineering ResearchConference,PuertoRico.

Ali,R.andDeif,A.(2014),“Dynamicleanassessmentfortakttimeimplementation”,Procedia CIRP,Vol.47No.1,pp.2212-8271.

Almomani,M.,Abdelhadi,A.,Mumani,A.,Momani,A. andAladeemy, M.(2014),“Aproposed integratedmodelofleanassessmentandanalyticalhierarchyprocessforadynamicroad mapofleanimplementation”,InternationalJournalofAdvancedManufacturingTechnology, Vol.72No.1,pp161-172.

(23)

Behrouzi, F. and Wong,K.(2011),“Lean performanceevaluation of manufacturing systems: adynamicandinnovativeapproach”,ProcediaComputerScience,Vol.3No.1,pp.388-395. Black, J.(2007),“Design rulesfor implementingtheToyotaproductionsystem”,International

JournalofProductionResearch,Vol.45No.16,pp.3639-3664.

Christopher, M. (2000), “The agile supply chain competing in volatile markets”, Industrial MarketingManagementJournal,Vol.29No.1,pp.37-44.

Deif, A.(2012),“Dynamicanalysis of alean cell under uncertainty”,International Journalof ProductionResearch,Vol.50No.4,pp.1127-1139.

Deif,A.and ElMaraghy,H.(2014),“Costperformancedynamicsin leanproductionleveling”, JournalofManufacturingSystems,Vol.33No.4,pp.613-623.

Detty,R.andYingling,J.(2000),“Quantifyingbenefitsofconversiontoleanmanufacturingwith discrete event simulation: a case study”, International Journalof ProductionResearch, Vol.38No.2,pp.429-445.

Elnadi,M.andShehab,E.(2014),“Aconceptualmodelforevaluatingproduct-servicesystems leannessinUKmanufacturingcompanies”,ProcediaCIRP,Vol.22No.1,pp.281-286. Elsayed,N.,Jondral,A.,Greinacher,S.,Dornfeld,D.andLanza,G.(2013),“Assessmentofleanand

green strategies by simulation of manufacturing systems in discrete production environments”,CIRPAnnals–ManufacturingTechnology,Vol.62No.1,pp.475-478. Forrester,J.W.(1961),IndustrialDynamics,MITPress,Cambridge,MA.

Georgiadis,P.andMichaloudis,C.(2012),_“Real-timeproductionplanningandcontrolsystemfor job-shopmanufacturing:asystemdynamicsanalysis”,EuropeanJournalofOperational Research,Vol.216No.1,pp.94-104.

Helo,P. (2000),_“Dynamicmodelingof surgeeffectand capacitylimitationinsupplychains_”, InternationalJournalProductionResearch,Vol.38No.17,pp.4521-4533.

Huang,M.,Yang,M.,Zhang,Y.andLiu,B.(2012),_“Systemdynamicsmodeling-basedstudyof contingentsourcingundersupplydisruptions”,SystemsEngineeringProcedia,Vol.4No.1, pp.290-297.

Montgomery,D.(2005),IntroductiontoStatisticalQualityControl,JohnWiley&Sons,NJ. Pakdil, F. and Leonard, K. (2014), “Criteria for a lean organization: development of a lean

assessmenttool”,InternationalJournalofProductionResearch,Vol.52No.15,pp.4587-4607. Poles, R. (2013), “System dynamics modelling of a production and inventory system for remanufacturing to evaluate system improvementstrategies”, International Journal of ProductionEconomics,Vol.144No.1,pp.189-199.

Sanchez,A.andPerez,M.(2001),“Leanindicatorsandmanufacturingstrategies”,International JournalofOperations&ProductionManagement,Vol.21No.11,pp.1433-1451.

Savsar,M.andAl-Jawini,A.(1995),“Simulationanalysis of just-in-timeproductionsystems”, InternationalJournalofProductionEconomics,Vol.42No.1,pp.67-78.

Staats,B.,Brunner,D.andUpton,D.(2011),“Leanprinciples, learning,andknowledgework: evidencefromasoftwareservicesprovider”,JournalofOperationsManagement,Vol.29 No.5,pp.376-390.

Sterman,J.(2000),BusinessDynamic – SystemsThinkingAndModelingForaComplexWorld, McGraw- Hill,NewYork,NY.

Susilawati,A.,Tan,J.,Bell,D.andSarwar,M.(2015),“Fuzzylogicbased methodtomeasure degreeof lean activityin manufacturing industry”,Journalof ManufacturingSystems, Vol.34No.1,pp.1-11.

Taj, S. (2008),“Lean manufacturing performancein China:assessment of 65manufacturing plants”,JournalofManufacturingTechnologyManagement,Vol.19No.2,pp.217-234.

(24)

Wan,H.,Chen,F.andRivera,L.(2007),“Leannessscoreofvaluestreammaps”,Proceedingsofthe IndustrialEngineeringResearchConference,Nashville,Tennessee.

Wikner,J.,Towill,D.andNaim,M.(1991),“Smoothing supplychain dynamics”,International JournalofProductionEconomy,Vol.22No.3,pp.231-248.

Womack,J.andJones,D.(2003),LeanThinking,FreePress,NewYork,NY.

Yang,T.,Kuo,Y., Su,C.andHou,C. (2015),“Leanproductionsystem designfor fishing net manufacturing using lean principles and simulation optimization”, Journal of ManufacturingSystems,Vol.34No.1,pp.66-73.

Zhang, Y., Wang, Y. and Wu, L. (2012), “Research on demand-driven leagile supply chain operation model: a simulation based on anylogic in system engineering”, Systems EngineeringProcedia,Vol.3No.1,pp.249-258.

Furtherreading

Alireza,A.,Zulkifli,N.,Yusuff,R.,Hojjati,S.andIsmail,Y.(2011),“Aproposeddynamicmodel foraleanroadmap”,AfricanJournalofBusinessManagement,Vol.5No.16,pp.6727-6737. Womack,J.,Jones,D.andRoos,D.(1990),TheMachinethatChangedtheWorld,NewYork,Free

Press,NY.

Appendix1

Parameter Value Unit

AQL (1/100) × THOUT

COT RANDOM NORMAL (0.003, 0.005, 0.004, 0.001, 1) CT1 RANDOM NORMAL (0.05, 0.06, 0.055, 0.005, 1)-COT

CT2 RANDOM NORMAL (0.03, 0.04, 0.035, 0.005, 1)-COT

D RANDOM NORMAL ((5/100) × QCOR, (10/100) × QCOR, (7.5/100) × QCOR, (2.5/100) × QCOR, 1) × Time Unit

DT 0.005

DR RANDOM NORMAL (10, 20, 15, 5, 1)

IT RANDOM NORMAL (0.005, 0.007, 0.006, 0.001, 1) MOPT RANDOM NORMAL (6, 7, 6.5, 0.5, 1)

PDT (2/100) × SPT QCSR PR3× (20/100)

TDD 2

THCT 1

UPDT RANDOM NORMAL (0.016, 0.03, 0.023, 0.007, 1) × SPT

Parts Hour Hour Hour Hour Hour Parts Hour Parts/hour Hour Hour Hour Parts/hour Hour Hour Hour TableAI. Company data

(25)

Appendix2

COT Defects DR IT MOPT UPDT Response

0.003 0.05 10 0.007 7 0.03 54.74315 0.005 0.05 20 0.005 6 0.016 33.72218 0.005 0.1 10 0.007 7 0.016 53.10217 0.005 0.1 10 0.007 7 0.03 53.35221 0.005 0.05 10 0.005 6 0.03 54.39086 0.005 0.05 20 0.007 7 0.016 33.30398 0.003 0.1 20 0.005 6 0.03 29.28129 0.005 0.05 10 0.005 7 0.016 53.78052 0.005 0.1 20 0.007 6 0.016 31.36727 0.003 0.05 20 0.005 6 0.016 32.35737 0.005 0.1 10 0.005 6 0.016 53.43671 0.003 0.1 10 0.005 6 0.03 54.41546 0.005 0.1 20 0.007 7 0.03 29.98033 0.003 0.1 10 0.007 6 0.03 54.41546 0.003 0.05 20 0.007 6 0.03 31.08528 0.005 0.05 10 0.007 7 0.016 53.78052 0.005 0.1 20 0.007 6 0.03 30.18623 0.005 0.1 20 0.005 6 0.03 30.18623 0.005 0.05 20 0.007 6 0.016 33.72218 0.003 0.05 10 0.007 6 0.016 54.84606 0.003 0.1 20 0.005 6 0.016 30.4111 0.005 0.1 20 0.005 6 0.016 31.36727 0.003 0.05 20 0.007 7 0.016 32.0476 0.003 0.05 10 0.007 6 0.03 55.12858 0.005 0.1 20 0.005 7 0.016 31.14493 0.003 0.1 20 0.007 6 0.016 30.4111 0.003 0.05 20 0.005 7 0.03 30.8047 0.005 0.1 10 0.005 6 0.03 53.70631 0.003 0.1 10 0.005 6 0.016 54.13935 0.003 0.05 20 0.007 6 0.016 32.35737 0.005 0.05 20 0.007 6 0.03 32.33182 0.003 0.1 20 0.005 7 0.03 29.1067 0.005 0.05 10 0.007 7 0.03 54.03677 0.003 0.1 20 0.007 6 0.03 29.28129 0.005 0.05 10 0.005 7 0.03 54.03677 0.003 0.1 20 0.005 7 0.016 30.22431 0.005 0.1 10 0.007 6 0.03 53.70631 0.005 0.05 10 0.007 6 0.016 54.11505 0.005 0.05 10 0.007 6 0.03 54.39086 0.003 0.05 10 0.005 6 0.03 55.12858 0.003 0.05 20 0.007 7 0.03 30.8047 0.003 0.1 20 0.007 7 0.016 30.22431 0.005 0.05 20 0.005 7 0.03 31.96293 0.005 0.1 10 0.007 6 0.016 53.43671 0.003 0.1 10 0.007 7 0.03 54.03003 0.003 0.05 20 0.005 6 0.03 31.08528 TableAII. 0.003 0.05 10 0.005 7 0.016 54.48055 Parameters significance experiment (continued)

(26)

COT Defects DR IT MOPT UPDT Response 0.005 0.05 20 0.005 6 0.03 32.33182 0.003 0.05 10 0.007 7 0.016 54.48055 0.003 0.1 10 0.007 7 0.016 53.77384 0.003 0.1 10 0.005 7 0.016 53.77384 0.003 0.1 10 0.007 6 0.016 54.13935 0.005 0.1 10 0.005 7 0.016 53.10217 0.003 0.1 20 0.007 7 0.03 29.1067 0.003 0.05 10 0.005 7 0.03 54.74315 0.003 0.05 10 0.005 6 0.016 54.84606 0.005 0.05 10 0.005 6 0.016 54.11505 0.005 0.1 20 0.007 7 0.016 31.14493 0.005 0.1 10 0.005 7 0.03 53.35221 0.003 0.1 10 0.005 7 0.03 54.03003 0.005 0.1 20 0.005 7 0.03 29.98033 0.005 0.05 20 0.007 7 0.03 31.96293 0.003 0.05 20 0.005 7 0.016 32.0476 0.005 0.05 20 0.005 7 0.016 33.30398 TableAII. Correspondingauthor