norman m. sadeh e-supply chain management laboratory school of computer science
DESCRIPTION
University of Minnesota – Digital Technology Center December 2002 Intelligent e-Supply Chain Decision Support. Norman M. Sadeh e-Supply Chain Management Laboratory School of Computer Science Carnegie Mellon University. Outline. Supply Chain Management: New Context - PowerPoint PPT PresentationTRANSCRIPT
University of Minnesota – Digital Technology CenterDecember 2002
Intelligent e-Supply Chain Decision Support
Norman M. SadehNorman M. Sadeh
e-Supply Chain Management Laboratorye-Supply Chain Management LaboratorySchool of Computer Science School of Computer Science Carnegie Mellon UniversityCarnegie Mellon University
Copyright ©2002 Norman Sadeh
Outline Supply Chain Management: New ContextSupply Chain Management: New Context Agent-Based Collaborative Decision SupportAgent-Based Collaborative Decision Support
MascotMascot Available-To-Promise/Capacity-To-Promise Available-To-Promise/Capacity-To-Promise
FunctionalityFunctionality Empirical ResultsEmpirical Results Dynamic Supply Chain Management PracticesDynamic Supply Chain Management Practices
Early ResultsEarly Results TAC’03: A Supply Chain Trading CompetitionTAC’03: A Supply Chain Trading Competition
Summary and Concluding RemarksSummary and Concluding Remarks
Copyright ©2002 Norman Sadeh
Supply Chain Management
Planning and coordinating procurement, production and Planning and coordinating procurement, production and distribution activitiesdistribution activities From raw material suppliers to manufacturers …to From raw material suppliers to manufacturers …to
distribution centers …to retailers and consumersdistribution centers …to retailers and consumers Trillions of dollars annuallyTrillions of dollars annually Good practices directly impact the competitiveness of Good practices directly impact the competitiveness of
companiescompanies Timely and cost-effective delivery of products to Timely and cost-effective delivery of products to
customerscustomers Extends to product design and configurationExtends to product design and configuration
Copyright ©2002 Norman Sadeh
Why is SCM Difficult? Involves multiple organizationsInvolves multiple organizations Each organization tries to satisfy multiple objectives Each organization tries to satisfy multiple objectives
Cost, timeliness, quality, market share, etc.Cost, timeliness, quality, market share, etc. Each organization operates subject to:Each organization operates subject to:
Internal ConsiderationsInternal Considerations::Finite capacity, existing inventory, etc.Finite capacity, existing inventory, etc.
External ConsiderationsExternal ConsiderationsAvailable suppliers and their capacities, order quantities and due Available suppliers and their capacities, order quantities and due
dates, contractual arrangements, transportation constraints, etc.dates, contractual arrangements, transportation constraints, etc. Numerous sources of uncertaintyNumerous sources of uncertainty
Capacity, supplies, demand, etc.Capacity, supplies, demand, etc.
Copyright ©2002 Norman Sadeh
Historical Perspective
Purchasing Production Sales Distribution
MaterialsManagement
ManufacturingManagement
Distribution CustomersSuppliers
SuppliersInternal
Supply ChainCustomers
Buyers/Sellers
Buyers/Sellers
Buyers/Sellers
Buyers/Sellers
InventoryControl
e-Commercee-Markets/Exchanges
e-Supply Chains
Functional Silos
EnterpriseIntegration
Dynamic Internet-enabled
Supply Chain
Supply ChainIntegration
Copyright ©2002 Norman Sadeh
Beyond the Early eMarket Hype Dynamic business practices are mainly confined to Dynamic business practices are mainly confined to
MRO MRO Suppliers don’t like being evaluated solely based on Suppliers don’t like being evaluated solely based on
pricepriceCovisint, E2open exchanges: more emphasis Covisint, E2open exchanges: more emphasis
on supporting collaborationon supporting collaborationRequires richer environmentsRequires richer environments
• Multiple attributes – not just priceMultiple attributes – not just priceLack of adequate standardsLack of adequate standards
Lack of adequate decision support toolsLack of adequate decision support tools Evaluate a large number of optionsEvaluate a large number of options
Standardization efforts are taking timeStandardization efforts are taking time
Copyright ©2002 Norman Sadeh
Some Open Research Issues
Long vs. Short term contractsLong vs. Short term contracts Information exchangeInformation exchange Collaborative decision supportCollaborative decision support Multi-attribute negotiationMulti-attribute negotiation Peer-To-Peer/local view vs. more lobal viewPeer-To-Peer/local view vs. more lobal view
P2P ChallengeP2P Challenge: Coordinating negotiation across : Coordinating negotiation across multiple tiersmultiple tiers
Challenge for the Global ViewChallenge for the Global View: : Creating the right incentives for information sharingCreating the right incentives for information sharing How global? How often do you clear? etc.How global? How often do you clear? etc.
Copyright ©2002 Norman Sadeh
MASCOT: Collaborative Decision Support
Decisions are evaluated in Decisions are evaluated in collaborationcollaboration with with potential business partners potential business partners
Supply chains can be Supply chains can be dynamicallydynamically set up in set up in response to changing market requirementsresponse to changing market requirements
Emphasis on Mixed Initiative Decision Emphasis on Mixed Initiative Decision SupportSupportDon’t try to automate everything!Don’t try to automate everything!
Copyright ©2002 Norman Sadeh
MASCOT Supply Chain Agent
Bidding & Order Mgmt.
Planning & Scheduling
Procurement
BusinessEntity
EnterpriseLevel
MascotAgent
Tier 1Suppliers
eMarket
eMarket
eMarket
Site Level MascotAgent
Site Level MascotAgent
ProspectiveCustomer
Request forQuote
eMarket
Copyright ©2002 Norman Sadeh
MASCOT: Overall Objectives Leverage benefits of finite schedulingLeverage benefits of finite scheduling Rapid and accurate evaluation of partner-dependent Rapid and accurate evaluation of partner-dependent
decisions :decisions :Bids & Requests for QuotesBids & Requests for Quotes
• including real-time ATP/CTPincluding real-time ATP/CTPAlternative product/subcomponent designsAlternative product/subcomponent designsMake-or-buy decisionsMake-or-buy decisions
Customizable mixed-initiative functionalityCustomizable mixed-initiative functionalityCollaborative solution development, workflow Collaborative solution development, workflow
managementmanagement Facilitate integration with legacy systemsFacilitate integration with legacy systems
Copyright ©2002 Norman Sadeh
A Customizable Agent Wrapper
Blackboard
Control KB
Agenda
ControllerControlProfile
GUI
Event Queue
EnterpriseSystem
eMarketPortal
PotentialBusinessPartner
PotentialCustomer
Bidding/Order Mgmt KS
BOM/Flow Config. KS
Demand Explosion KS
Communication KS
Knowledge Sources
Contexts
ExternalSystems
Outgoing messages
Incoming Events
Scheduling/CTP KS
RFQ/Procurement KS
Currentworkingcontext
Copyright ©2002 Norman Sadeh
Main Architectural Features Blackboard Blackboard ContextsContexts: “What-if”: “What-if”
Different assumptions (e.g. demand, resources, Different assumptions (e.g. demand, resources, suppliers) and different solutionssuppliers) and different solutions
Unresolved issues Unresolved issues Extensible set of Knowledge SourcesExtensible set of Knowledge Sources (KSs) (KSs)
Allows for modular & reusable KSsAllows for modular & reusable KSs Provides for easy integration with legacy systemsProvides for easy integration with legacy systems
Mixed Initiative ControlMixed Initiative Control Customizable user profileCustomizable user profile
Copyright ©2002 Norman Sadeh
Unresolved Issues Help keep track of Help keep track of incomplete, inconsistent and incomplete, inconsistent and
unsatisfactoryunsatisfactory aspects of a context solution aspects of a context solution Examples: unprocessed RFQ, insufficient availability Examples: unprocessed RFQ, insufficient availability
of supplies, missed prior delivery commitmentof supplies, missed prior delivery commitment Automatically updatedAutomatically updated as the solution is modified as the solution is modified Supports flexible mixed initiative Supports flexible mixed initiative workflow managementworkflow management
Associated with KS activations, scripts and goalsAssociated with KS activations, scripts and goals
Copyright ©2002 Norman Sadeh
Three Levels of Problem Solving Knowledge Source ActivationsKnowledge Source Activations
e.g. Demand Explosion (RFQ1)e.g. Demand Explosion (RFQ1) ScriptsScripts
e.g. Evaluate (RFQ1) e.g. Evaluate (RFQ1) 1. Copy_Current_Context2. Incorporate(RFQ1)3. Demand_Explosion (RFQ1)4. Reoptimize_Schedule_with_Net_Demand (RFQ1)5. Procure_Subcomponents_Net_Demand(RFQ1)6. etc.
• Goals: Search among multiple options
Copyright ©2002 Norman Sadeh
Mixed Initiative Workflow Management
U/C:Incorporate event
into Context
Blackboard:Update
Unresolved Issues
U/C:Select UnresolvedIssue to Resolve
U/C:Select Resolution
Method
KS:Execute
Resolution Method
Controller:Activate
Resolution Method
U/C:Modify Assumptions
within Context(“What-if”)
ModifiedContext
ModifiedContext
UnresolvedIssue
SelectedUnresolved
Issue
SelectedResolution
Method
ActivatedAgenda
Item
ModifiedCopy ofContext
U/C = User or Controller
Copyright ©2002 Norman Sadeh
Status Customized to support coordination between a machine Customized to support coordination between a machine
shop and a tool shop at Raytheonshop and a tool shop at Raytheon Over 150 machine centers & over 100 peopleOver 150 machine centers & over 100 people 50% of incoming orders require new tools50% of incoming orders require new tools
alternative BOM & process planning optionsalternative BOM & process planning options Reduced tardiness by 23 percentReduced tardiness by 23 percent
Integration of process planning & schedulingIntegration of process planning & schedulingTighter coordinationTighter coordination
Used to study the benefits of different supply chain Used to study the benefits of different supply chain coordination policies and different order promising policiescoordination policies and different order promising policies
Copyright ©2002 Norman Sadeh
Dynamic Supply Chain Coordination
•orders•requests for bid•bid acceptance/rejection•order cancelation/modification
•products•bid submissions•revised delivery dates
Supply chainentity
Supply chainentity
Supply chainentity
Supply chainentity
Copyright ©2002 Norman Sadeh
The Coordination Challenge Generate Generate robust yet competitive and cost-effective robust yet competitive and cost-effective
promise datespromise dates Multi-tier “capacity-to-promise” functionalityMulti-tier “capacity-to-promise” functionality
Sources of Sources of uncertaintyuncertainty are both internal and external are both internal and external incoming orders, supplies, internal capacity, etc.incoming orders, supplies, internal capacity, etc.
Is it possible, through dynamic coordination, to reap Is it possible, through dynamic coordination, to reap the benefits of the benefits of finite schedulingfinite scheduling, while offsetting the , while offsetting the brittlenessbrittleness of its solutions ?of its solutions ?
Copyright ©2002 Norman Sadeh
Real-Time Promising(RTP):General Considerations Net DemandNet Demand: Inventory Allocation & Demand Explosion: Inventory Allocation & Demand Explosion SchedulingScheduling
Available vs. modified capacityAvailable vs. modified capacity Schedule around prior commitments vs. reoptimizationSchedule around prior commitments vs. reoptimization Schedule ReoptimizationSchedule Reoptimization
Assess impact on prior commitmentsAssess impact on prior commitmentsCosts & Priorities: order priorities, late delivery Costs & Priorities: order priorities, late delivery
penalties, inventory costs, etc.penalties, inventory costs, etc.Other Tradeoff: Speed versus “optimality”Other Tradeoff: Speed versus “optimality”
Assess desirability & decide whether to submit quoteAssess desirability & decide whether to submit quote Micro-Boss RTP moduleMicro-Boss RTP module: real-time reoptimization - : real-time reoptimization - user user
specifies desired response time specifies desired response time (Sadeh et al. ‘94-99)(Sadeh et al. ‘94-99)
Copyright ©2002 Norman Sadeh
RTP: Further Refinements
Profitable-To-PromiseProfitable-To-Promise Selective RTP ValidationSelective RTP Validation
Copyright ©2002 Norman Sadeh
Profitable-To-Promise Overall Profit = Total_Revenue - Total_CostsOverall Profit = Total_Revenue - Total_Costs
Total_Revenue: Total_Revenue: Sum over all ordersSum over all orders Total_Costs: Total_Costs: Production costs, inventory costs (raw Production costs, inventory costs (raw
materials, in-process, finished goods), late delivery materials, in-process, finished goods), late delivery penalties, etc.penalties, etc.
Takes into account impact on prior commitmentsTakes into account impact on prior commitmentse.g. late delivery penalty when another order gets e.g. late delivery penalty when another order gets
bumpedbumped Bid only if overall profit increasesBid only if overall profit increases
Other variations can be considered Other variations can be considered e.g. strategic customers, market share considerationse.g. strategic customers, market share considerations
Copyright ©2002 Norman Sadeh
Empirical Study: Multiple RFQ Processing Policies
Response:Response: Always bid - no due date negotiationAlways bid - no due date negotiation Only submit a bid if overall profit increasesOnly submit a bid if overall profit increases Bid conditional on acceptance of possibly relaxed Bid conditional on acceptance of possibly relaxed
promise datepromise date Capacity-To-Promise ComputationCapacity-To-Promise Computation
Leadtime-basedLeadtime-based Local finite capacity scheduling & supply Local finite capacity scheduling & supply
leadtimesleadtimes Coordinated finite capacity schedulingCoordinated finite capacity scheduling
Copyright ©2002 Norman Sadeh
Empirical Study: Assumptions A lot-for-lot make-to-order environmentA lot-for-lot make-to-order environment
Internal sources of uncertainty at each tier due to resource Internal sources of uncertainty at each tier due to resource breakdowns and variations in processing timesbreakdowns and variations in processing times
Stochastic order arrivalStochastic order arrival
Finite capacity schedules regenerated dailyFinite capacity schedules regenerated daily
Micro-Boss scheduling system Micro-Boss scheduling system
JIT objective: minimize sum of tardiness & inventory costsJIT objective: minimize sum of tardiness & inventory costs
Execution priority in accordance with the latest released Execution priority in accordance with the latest released scheduleschedule
Copyright ©2002 Norman Sadeh
Evaluation Criteria
Number of bids refused or rejectedNumber of bids refused or rejected Number of tardy orders Number of tardy orders Average utilization of the most utilized resourceAverage utilization of the most utilized resource Average supply chain leadtimesAverage supply chain leadtimes Average due-date adjustment (as part of bid negotiation)Average due-date adjustment (as part of bid negotiation) Profit (sales revenue minus costs)Profit (sales revenue minus costs)
Total in-system inventory costs (WIP and finished goods)Total in-system inventory costs (WIP and finished goods) Total tardiness costsTotal tardiness costs
Promise date accuracyPromise date accuracy
Copyright ©2002 Norman Sadeh
Basic Supply Chain Configuration
Agent 1
Agent 2
Agent 3
Agent 4
Agent 5
Agent 9
Agent 10
Agent 6
Agent 7
Agent 8
Externalcustomers
Supply chain
Tier3 Tier2 Tier1
Copyright ©2002 Norman Sadeh
Benefits of Dynamic Finite Capacity Coordination
0
3
6
9
12
15
18
Leadtime-based Local FCS Coordinated FCS
Av
era
ge
le
ad
tim
e (
da
ys
) (i
n b
ars
)
-2000000
-1000000
0
1000000
2000000
3000000
4000000
Pro
fit
(in
hig
h-l
ow
-lin
es
)
Case with competition and negotiable promise dates
Copyright ©2002 Norman Sadeh
Benefits of Dynamic Coordination - Contd.
-6000000
-4000000
-2000000
0
2000000
4000000
6000000
0,40 0,60 0,80 1,00 1,20
Nominal load
Av
era
ge
pro
fit
Leadtime-basedLocal FCSCoordinated FCS
Copyright ©2002 Norman Sadeh
Dynamic Supplier Selection
A manufacturer has a given set of customer orders A manufacturer has a given set of customer orders to satisfyto satisfy
Each order has a required delivery date along with Each order has a required delivery date along with a penalty for missing that datea penalty for missing that date
The manufacturer’s capacity is finiteThe manufacturer’s capacity is finite Each order requires a number of components for Each order requires a number of components for
which suppliers have submitted bidswhich suppliers have submitted bids Supply bids include a price and delivery dateSupply bids include a price and delivery date
Copyright ©2002 Norman Sadeh
Supplier Bid Selection
Component 11
Order 1
Order 2
Component 12
Component 21
Component 22
Component 23
Supplier BidSupplier BidSupplier Bid
Supplier BidSupplier BidSupplier Bid
Supplier BidSupplier BidSupplier Bid
Supplier BidSupplier BidSupplier Bid
Supplier BidSupplier BidSupplier Bid
(Price, Delivery Date)
(Delivery Date, Late Penalty)
Copyright ©2002 Norman Sadeh
Trading Agent Competition
TAC Classic: Travel Agent ScenariosTAC Classic: Travel Agent Scenarios About 20 entries in the pastAbout 20 entries in the past
TAC’03: Supply Chain Trading CompetitionTAC’03: Supply Chain Trading Competition Agents compete for supplies and demandAgents compete for supplies and demand Fixed Assembly CapacityFixed Assembly Capacity RFQs from customers – delivery date and RFQs from customers – delivery date and
tardiness penaltytardiness penalty RFQs to suppliersRFQs to suppliers Interests on money borrowed from the bankInterests on money borrowed from the bank
Copyright ©2002 Norman Sadeh
Office hoursOffice hours
00:00 00:0000:00
Negotiationand planning
08:00 17:00 08:00 17:00
a day (24 h)
Production
Negotiationand planning
Production Production
Productionand delivery
schedule
Deliveryof PCs to
customers
RFQsto
suppliers
Responsesfrom
suppliers
RFQs andorders fromcustomers
Componentsfrom
suppliers
14 game seconds Zero game seconds
A TAC Day
Copyright ©2002 Norman Sadeh
Summary e-SCM is about more open and more dynamic business practices e-SCM is about more open and more dynamic business practices Mascot: Mascot:
Rapid evaluation of partner-dependent decisionsRapid evaluation of partner-dependent decisions Mixed initiative decision supportMixed initiative decision support Coordinated real-time Profitable-To-Promise functionalityCoordinated real-time Profitable-To-Promise functionality
Ongoing work: Ongoing work: Combine e-SCM and multi-attribute negotiation – together Combine e-SCM and multi-attribute negotiation – together
with the Univ. of Michiganwith the Univ. of Michigan Dynamic Supplier SelectionDynamic Supplier Selection Trading Agent CompetitionTrading Agent Competition
Copyright ©2002 Norman Sadeh
Q&A