Supply Chain Visualization:Supply Chain Visualization:System Dynamics ModelSystem Dynamics ModelSimulation Engine

Three Separate Runtime Systems

Process Handbook(Knowledge engine)

Simulator(Simulation Engine /Analyzer)

Tangible UI(Multi-channel I/O,

Presentation management)

Tspaces(“glue”)

Simulator

• Simulation Engine– System Dynamics – Discrete Event

• System Dynamics Models of Intel• Computer aided simulation analysis

SD Modeling Work

• System Dynamics focuses on– Understanding how a business can cause its own

problems• Unstable production• Declining morale• Etc.

– Understanding leverage points, where a small effort can have a large, beneficial effect

SD Requires Hands-on Knowledge

• Mary Murphy-Hoye• Darren Blue• George Brown• David Fanger• Tom Gardos• Brian Kelly• Karl Kempf• Daniel Mckeon

• Gordon McMillan• Gene Meieran• Alison Page• Dean Phillips• Roberta Bailey Roberts• Nick Rose• Michael Waithe• Ann Johnson

SD Focuses on PatternsReference Modes

2000

Demand & Product Life Cycle

19901980 Time

Demand Peak(Million Units)

Product Family

Products in Family

Inventory in Chain

Time

Available Inventory

Fear

Fear

••

1 Yr

2000

Capacity Re lative Des ired

19901980 Time

Hope

Fear

Capacity Relative to Desired

Fear

100Need

• - ••••••

SD Focuses on Endogenous (internal) Causes Feedback Loops

W hole

Picture

In te l Products

Com petitor Products

+

Potentia lApplications+

MarketDem and for

Chips

+

Inte lDem and

+

Inte l Share

+

+-

In te lRev enues

+

Inv estm ent in R&D

+

+

OEMManufacturers+

+

OEM Com petition

+

Com puterPrices

-

Desire for Low-cost Only Chips (i.e . No

W aterfa lling)

+

In te lBrand+

ProductionCapacity

+

CapacityF lexibility

+Inv entories

Shipm ents+

+

-

OrdersFilled -

+

+

Inte l's Ability toRespond to

Dem and Change

+

ProductInv entory

Obsolescence-

-

Pressure toReduce Costs

+

Pressure toMainta in Profit

Marg ins

+

-

+

Materia lInv entory

Obsolescence-

W rite-offs+

-

PriceDiscounts+

-

+

Pressure toReduce OEM

Costs

OEM Desireto Hold

Inv entories

Tim e toCancelOrders

+

-

+

+

+

In te lProfits

+

+ MarketSegm entation

-

+

+

Usefu lInv entories

Inte lCosts

-

+

++

+

+

+

Product L ife Cycle

- -

OEMs

Share`R&DProfits

InvMgt Plant

SegmentationMarket

Cost Pres-sures

Com

petit

ion

Current/Proposed Solutions

Intel Products

CompetitorProducts

+

PotentialApplications+

Market Demandfor Chips

+

Intel Demand

+

IntelShare

+

+-

Intel Revenues

+

Investment inR&D

+

+

OEMManufacturers

+

+

OEMCompetition

+

ComputerPrices

-

Desire for Low-costOnly Chips (i.e. No

Waterfalling)

+

IntelBrand+

ProductionCapacity

+

CapacityFlexibility

+Inventories

Shipments+

+

-

OrdersFilled -

+

+

Intel's Ability toRespond to Demand

Change

+

Product InventoryObsolescence

-

-

Pressure to ReduceCosts

+

Pressure toMaintain Profit

Margins

+

-

+

Material InventoryObsolescence

-

Write-offs+

-

Price Discounts+

-

+

Pressure toReduce OEM

Costs

OEM Desire toHold

Inventories

Time to CancelOrders

+

-

+

+

+

IntelProfits

+

+ MarketSegmentation

-

+

+

Useful Inventories

Intel Costs

-

+

++

+

+

+

Product LifeCycle

--

44

44

33

22

11

55

66

66

1717

1818

8899

101077

1313

12121111

1414

1919

20202525 2727

2626

1515

1616

2222

2323 2424

2828

22

Model Detail

AssemblyInflow

AssemblyStart Rate

FinishedGoods

Inventory Shipments toChannel

AssemblyWIP

Net AssemblyCompletion

AssemblyRejects

AssemblyInventory

Simulation Model

FabricationInventory

Finished GoodsInventoryProduction

Start RateNet

ProductionRate

Sales toChannel

Desired FinishedGoods Inventory

ExpectedChannel

Demand forIntel Chips

DesiredFabricationInventory

Desired FinishedGoods Inventory

Coverage

Finished GoodsInventory

Adjustment

FabricationInventory

Adjustment

DesiredProductionStart Rate

ManufacturingCycle Time

FabricationAdjustment

Time

DesiredAssembly

StartsTime to UpdateChannel Orders

-

-

+

++

+

-

Time to AdjustFinished Goods

Inventory

+ +

-

+

+

-+

-

+

+

+

AssemblyInventory Net

AssemblyCompletion

+

Time to CompleteAssembly

Desired NetAssembly

Completion

DesiredAssemblyInventory

AssemblyInventory

Adjustment

Time to AdjustAssemblyInventory

+

+

+

+

-

-

InitialAssemblyInventory

InitialFabricationInventory

AssemblyStart Rate

Line Yield

Die PerWafer

Die PerWafer Yield

Unit to DieYield

ProductionRejects

<ChannelDemand forIntel Chips>

+

+

-+

+

++

AssemblyRejects

GrossAssembly

Completion

++

+-

Desired NetProduction

Rate

--+<Line

Yield>

<BacklogAdjustment>

+

BacklogSwitch

DesiredGross

AssemblyCompletion +

+ +

-

Desired GrossProduction

Rate

+

Max GrossProduction Rate

MaximumGross

AssemblyCompletion

+

-

+

MaximumShipment

Rate

Minimum OrderProcessing Time Table for Orde

r Fulfillment

OrderFulfillment

Ratio

Safety Stock Coverage

+

Intels MarketShare

Perceived IntelFraction Orders

Filled

Time toPerceive

Fraction OrdersFilled

+

ChannelDemand

+

Table forAttractiveness

ReferenceFraction

Orders Filled

CompetitorsAttractiveness

TotalAttractiveness

Intels Attractivenes

s+

+

+

+

-

-

Initial FractionOrder FilledChannel Order

BacklogChannelDemand forIntel Chips

OrderFulfillment

Rate

DesiredShipment

Rate

Target DeliveryDelay

AcceptableBacklog Backlog

Adjustment

Time to Adjust Backlog

+

-

InventoryCoverage+

<Target DeliveryDelay>Perceived

PresentDemand

Change inPerceived Present

Demand -

HistoricalDemand

Change inHistorical Demand

-

Time to PerceivePresent Demand

Time Horizon forHistoricalDemand

-

-

+Indicated

Trend

-

+-

+ForecastDemand

+

+

ForecastHorizon

+

AvailableCapacity

CapacityAcquisition Delay

Average Life ofCapacity

IndicatedCapacity

+

-

+

<Die PerWafer>

-

+

CapacityAcquisition

CapacityObsolescence

CapacityReplacing

CapacityAdjustment

AnticipatedTotal Yield

<Unit to DieYield>

<Die Per WaferYield>

Forecast WaferDemand

Brown - don't know much aboutBlack - Sure Equations

Blue - Guess

Pink- Parameters

<Desired ProductionStart Rate>

<AvailableCapacity>

Model Behavior (output)

baseFabrication Inventory48,65337,71126,76915,826

4,88412.3 24.1 36

Time (Month)

baseChannel Demand for Intel Chips5.637 M5.004 M4.371 M3.738 M3.105 M

12.3 24.1 36Time (Month)

BUT ….

In this ModelIntel’s Demand is Endogenous

(internally generated)

Eigenvalue Description

Second Eigen Pair

-2

-1

0

1

2

Time 18.75 19.75 20.75 21.75 22.75 23.75 24.75 25.75 26.75

RealE3 RealE4 ImagE3 ImagE4

The Way Its Supposed to Work

Backlog

Marketshare

Demand+

+

Shipping

-

FinishedInventory

-Pull

AssemblyInventory-

Pull

FabInventory

+

+ Push

Stockout: Pull � Push

Backlog

Marketshare

Demand

FinishedInventory

Shipping FabInventory

AssemblyInventory

+

+

+

+

-

-

+Push

Pull

Push

Fraction oforders shipped

+

-

Pull Pull � Push Push

Backlog

Marketshare

Demand

FinishedInventory

Shipping FabInventory

AssemblyInventory

+

+

Fraction oforders shipped

+

-

+

+

+

-

Push Push

+

Push

Stockouts Really Are a Problem

• Variable demand makes chain harder to manage, causing …

• Stockouts, which …• Convert the system from pull to push …• Causing variations in demand

Solutions

• Hold more inventory?• Internal markets?

Methodological Insights

• Eigenvalue analysis– Analysis is faster– Analysis is deeper

• TUI brings a new dimension to interacting with a model

• “Periodic table” of molecules• Handbook and SD …

Opportunities

• Further development of Intel SD models• Exploration of combined Discrete event / SD

/ Agent-based modeling• Simulating the dynamic impact of internal

markets on supply chain• Making eigenanalysis easier, integrating with

TUI and Handbook

Summary

• Simulation engine: SD and DE• System Dynamics: How we do it to ourselves• Intel models: Counter-intuitive causes of

undesirable behavior• Computer-aided simulation analysis allows

computer-aided thinking