Multi-EchelonNetworkEvaluationandInventoryStrategy

Boxi XuPatrickScottMITMasterofSupplyChainManagementClassof2017

Overview• Thesissponsor– Majorglobaloilfieldservicecompany

• Objective• Assessdecentralizedvs.centralizedmaterialsupplymodel• CurrentNetwork– decentralizedacross30locations• Proposednetwork– centralizedacross3globaldistributioncenters

2

Overview• EvaluationMetrics• OperationsEfficiency• CostEfficiency

• Scopefocusandnarrowing• Purchaseitems(85%ofallmaterials)

• 3mainsegments– drilling,testingandwireline.

3

InventoryLevel

InventoryCost

InventoryTurns

UnitInventory

Cost

Absolute

Relative

WorkingCapitalEfficiency

DollarSpent

Illustrationofcurrentandproposednetworks

4

CurrentNetworkDecentralizedCoordination

ProposedNetworkCentralizedCoordination

ManufacturingSite

DistributionCenter

FieldServiceCenter1

FieldServiceCenter2

FieldServiceCenter3

FieldServiceCentern

……

Supplier

DistributionCenter

FieldServiceCenter1

FieldServiceCenter2

FieldServiceCentern

……

ManufacturingSite1

ManufacturingSite2

ManufacturingSiten

……

Supplier

DemandProfile

11885

4415

23901559 1115 826 677 465 360 297 227 316

0

2000

4000

6000

8000

10000

12000

14000

1 2 3 4 5 6 7 8 9 10 11 12Parts

MonthswithDemand

EMSDemandFrequency• IntermittentDemand• EMS&FieldLocations• Reflectsnatureofmanufacturingprocessandindustry

• ProbabilityDistribution• SomesegmentsarestrongcandidatesforPoissondistribution

Method• Modelproposedmode• Basestockreplenishmentmodel• WeeklyReviewforreplenishment• TotalInventory=SafetyStock+PipelineInventory• PipelineInventory=AverageDemand/DayxLeadTimebyDay• SafetyStock

Example• DemandoverL+R~20• Demandfrequency~6

50% 70%

NormalDistributionDemandoverL+R

SafetyStock

MethodInputs• Distributionoverleadandreview

time• Poissoniflessthan10• Normalifgreaterthan10

• ServiceLevelSegmentation• HighRunner– 85%• Runner– 70%• Stranger– nosafetystock

0

2000

4000

6000

8000

10000

12000

14000

1 2 3 4 5 6 7 8 9 10 11 12Parts

MonthswithDemand

EMSDemandFrequency

HighRunnerRunner

2.86.01.0

1.82.7

2.7

0

5

10

15

20

25

30

CurrentState ProposedState

PipelineInventory(MillionUSD)

8

Result– InitialEvaluation

8.4

1.3

0.6

0.6

17.0

11.0

0

5

10

15

20

25

30

CurrentState

ProposedState

SafetyStock(MillionUSD)

DSC

Field

EMS

• Reductioninsafetystockoutweighsincreaseinpipelineinventory

9

Result– InitialEvaluation

26.0

8.7

6.5

10.4

CURRENTSTATE

FUTURESTATE

TOTALINVENTORY(MILLIONUSD)

PipelineInventory

SafetyStock

8158

CURRENTMODE PROPOSEDMODE

DAYSOFINVENTORYONHAND

• Roughly40%reductionintotalinventory

Result– InitialEvaluation

10

1.80.5

0.8

0.8

2.9

2.7

-

1.0

2.0

3.0

4.0

5.0

6.0

InvHoldingCost(current) InvHoldingCost(proposed)

CostConsideration(MillionUSD)

DSC

Field

EMS

• ManagerialCostAssumptions• Personnelcostremainsconstant• OrderandReviewcostsremainconstant

• Toogoodtobetrue?• Validatingthemodel

Result– Discussion• Safetystockreductionof13.1millionUSD…really?

• Fundamentaldifferencesbetweenthetwosystems• Levelofdemandaggregation

11

118

215 1 1 0.5 0.2 0.04 0.03

86

26 34

0

50

100

150

1 2 3 4 5 6 7 8 9

UpperEchelonMaterialFlowConcentration(MillionUSD)

CurrentStateviaEMS FutureStateviaDSC

Result– Discussion

12

Average InternalLeadTime Current Mode Proposedmode

EMS part 0 18days

Field part 28days 7days

• Trackingdemandvalue• EMSdemand(63%)vs.FieldDemand(37%)• Averagepartvalue– EMS(22USD)vs.Field(6USD)

• Trackingtransittime

Result– Compare“In-Theory”• Modelcurrentoperations• Sameapproachastheproposedstate• Removeexcessinventoriesdueto

inefficienciesfromcomparison

• Compareproposedmodewiththe“In-Theory”safetystockforcurrentmode• CurrentMode=13.1millionUSD• ProposedMode=12.9millionUSD• Reductionisnow0.2millionUSDor2%

13

11.1

1.3

1.4

11.0

0.6 0.6

0.0

2.0

4.0

6.0

8.0

10.0

12.0

14.0

CurrentMode ProposedMode

SafetyStock

EMS DSC Fields

Result– Compare“In-Theory”• Initialassumptionof2xslowermaterials

coordinationforproposedmode

• Iftheproposedmodecanprocessasfast…• Proposedmodeincreasespipelineinventoryby

0.8million• Thisisduetothelongerinternaltransittime

withEMSparts,whichcomprisemajoritydemand

14

2.8 2.8

1.0 1.8

2.72.7

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

CurrentState FutureState

PipelineInventory(MillionUSD)

Conclusion• Proposedmodecouldpotentiallyreducethesafetystockby2%(0.2million

USD),butincreasethepipelineinventoryby12.3%(0.8millionUSD)

• However,thereisinefficiencyandroomtoimprovethecurrentpractice• “InTheory”safetystockisonly13.1millioncomparedtoactual26million• Recommendedfurtherstudiesinclude:• Inventorypoliciessuitableforslowandinfrequentmovingdemand,e.g.

Poissondistributionforextremelylowdemandparts• Costimpactofincreasinguseofairfreighttransport

15

Q&A

BackupOtherinsights

17

• Lead time reduction is critical• Pipeline inv. outweighs safety stock• Pipeline Inv. = Demand x Lead Time

• EMS consume more expensive but slow moving parts

FutureStateAverageLeadTime(days) CurrentState

AverageLeadTime(days)

DSCProcessTime 15 EMSProcessTime 7DSCtoField 18 EMStoDSC 7DSCtoEMS 7 DSCtoField 18

Avg PartValue(USD) %High Runner %RunnerEMS 22 9% 36%Field 6 14% 77%

BackupScenarioAnalysis1– ReduceSupplierLeadTime• Safetystockreductionatupperechelonifsupplierleadtimesareshorter• Morereductionimpactwithcurrentmode• Opportunitytoimprovecurrentmode

18

CurrentModeLeadTimeReduction

EMSSafetyStock

AbsoluteReduction %Reduction

0% 11,077 - -3% 10,810 267 2%5% 10,567 509 5%

10% 10,388 689 6%15% 10,188 889 8%20% 9,938 1,139 10%30% 9,218 1,859 17%40% 8,691 2,385 22%50% 8,004 3,072 28%

ProposedModeLeadTimeReduction Houston Dubai Rotterdam Total Absolute

Reduction%Reduction

0% 4,087 4,124 2,772 10,983 - -3% 4,009 4,085 2,741 10,836 147 1%5% 4,016 4,088 2,721 10,825 158 1%

10% 3,905 4,054 2,668 10,628 355 3%15% 3,821 3,720 2,614 10,156 827 8%20% 3,608 3,704 2,559 9,871 1,112 10%30% 3,529 3,541 2,443 9,512 1,471 13%40% 3,251 3,227 2,319 8,797 2,186 20%50% 2,951 3,048 2,185 8,184 2,799 25%

BackupScenarioAnalysis2– SimplifyPartStratification

• LevelofsafetystockvariationatDSClevelifpartstratificationchangedfromthreebucketstotwobuckets

19

PartStratification HR Runner Stranger

ThreeBuckets Demandoccurs9monthsorabovelastyear (ServiceLevel85%)

Demandoccurs4monthsorabovelastyear (ServiceLevel70%)

Demandoccurs3monthsorbelowlastyear (Nosafetystock)

TwoBuckets N.A. Demandoccurs6monthsorabovelastyear (ServiceLevel85%or70%)

Demandoccurs5monthsorbelowlastyear (Nosafetystock)

• Change DSC part stratification in proposed state• Proposed state DSC service both EMS and fields • Impact to DSC safety stock depends on the service level defined for runners• Minimum change to safety stock (+0.25%) to keep service level at high runner level (85%) for “two buckets”

• Change DSC part stratification in current state• Current state DSC service only fields • Impact to DSC safety stock depends on the service level defined for runners• Same level of safety stock between “three buckets” and ”two buckets” if service level for runners defined at 82.5%