Design

1
Design Optimization in E-Supply Chains

• Doctoral Research
• Roshan Gaonkar
• Supervisor Prof N. Viswanadham
• The Logistics Institute Asia Pacific

2
Agenda

• The Internet and E-Supply Chains.
• Assumptions, Motivation and Contributions.
• Mathematical Models for Planning in E-Supply
  Chains
• Basic LP model for Private Marketplaces.
• Realistic MILP model for Private Marketplaces
• QP and MILP model for Supply Chains with Public
  Trading Exchange.
• Future Work

3
Fundamentals of E-Supply Chains
4
Trends in E-Supply Chains

• Emergence of Electronic Marketplaces
• Private Marketplaces.
• Public Trading Exchanges.
• Virtual Organizations and Extended Supply Chains
• Information-based Supply Chain Managers.
• Alliances and Partnerships
• Outsourced Manufacturing and Logistics.
• Global Supply Chain Networks.

5
Global Extended SC Networks
Source Analysis of Manufacturing Enterprises by
Prof. N. Viswanadham
6
A Typical Scenario
Global
Partner selection based on customer location
7
Extended Supply Chain Planning

• Global optimum in planning, using global
  visibility.

8
Motivation, Assumptions and Contributions
9
Physical Significance

• Dynamic Manufacturing Networks
• Network of companies sharing same destiny.
• Information visibility between partners.
• Contract Manufacturing in the Electronics
  Industry.

10
Hi-Tech Manufacturing

• Dell private marketplace
• Receives orders from customers.
• Global Supply Chain.
• Manufacturing outsourced to contract
  manufacturers and logistics outsourced to 3PLs.
• Constant access to supply chain operational
  information.
• Manages supply chain through superior planning.

11
Motivation

• To understand emerging business models in
  E-Supply Chains.
• Channel Masters.
• 4th Party Logistics.
• Contract Manufacturing.
• To develop planning tools for knowledge based
  businesses Internet-enabled supply chains.

12
Basic AssumptionsPrivate Marketplace

• Controlled by dominant channel master.
• Contract Manufacturers and Logistics Partners.
• High-level of trust exists between partners.
• Global Visibility in the Extended Supply Chain
• Schedules
• Capacities
• Costs
• Inventories
• Profit sharing between partners.

13
Basic AssumptionsPublic Trading Exchange

• Market-maker builds environment of trust.
• Supply-demand information
• Quantity
• Cost
• Delivery Date
• Companies participate in multiple marketplaces

14
Research Contributions

• Defined and formulated specific research problems
  in Internet-enabled extended supply chain
  networks.
• Developed optimization models for systematic
  management of on-line knowledge-based businesses.

15
Research Contributions

• Develop a common framework to analyze various
  supply chain strategies.
• Make-to-Order, Make-to-Stock, New Product
  Development etc.
• Models for partner selection in supply chain
  networks.
• Contract Manufacturers.
• Strategic and Operational Level.
• Inclusion of logistics in supply chain planning.
• Fixed Schedules.
• Transshipment Hubs.
• Synchronization of Manufacturing and Logistics

16
Classification of Models
17
Mathematical Models for Planning in E-Supply
Chains
18
A Basic LP Planning Model for Private Marketplaces
19
Models deployed in the SC
20
Basic AssumptionsPrivate Marketplace

• Controlled by dominant channel master.
• Contract Manufacturers and Logistics Partners.
• High-level of trust exists between partners.
• Global Visibility in the Extended Supply Chain
• Schedules
• Capacities
• Costs
• Inventories
• Profit sharing between partners.

21
Model Formulation

• Activities
• Sub-Assembly Production
• Transport from Suppliers to Manufacturer
• Manufacturing/Assembly
• Transport from Manufacturer to Buyers

• Inventories
• Sub-Assembly inventory at Supplier
• Sub-Assembly inventory at Manufacturer
• Model inventory at Manufacturer
• Model inventory at Buyer

22
Model Features
Supply Chain Information Shared Decisions to be Made
Available to promise Manufacturing Capacity for each Supplier. Fixed Schedules for Transportation Complex Product structure with multiple components, sub-assemblies, brands Inventory costs at multiple levels Transportation costs Production costs Determination of multiple plant schedules Determination of multi-period schedules Allocation of procurement quantities amongst multiple suppliers
Strategic level Partner selection and Operational
level Scheduling
23
Notation

• Parameters
• D buyers demanded quantity
• P cost of production for manufacturer/supplier
  or cost price to buyer
• U unit transportation cost
• C production/manufacturing capacity
• T Transportation capacity
• Variables
• S supplies transported between two parties.
• I inventories at each time period
• Q quantity produced in each time period

• i index used to denote products
• j index used to denote suppliers
• k index used to denote assemblers
• l index used to denote models
• m index used to denote the buyers
• Subscripts
• I set of components.
• L set of finished models
• J set of suppliers.
• K set of Manufacturers
• M set of Buyers

24
Objective

• Maximise Profit
• Profit Revenue (Cost of Production Cost of
  Transportation Cost of Inventory)

25
Constraints

• Capacity Constraints
• Production Capacity
• Transportation Capacity

26
Constraints

• Inventory Flow Constraints
• Tracking of inventory level at each time period
• Consumption and addition to inventory

27
Constraints

• Availability of Raw Materials

• Fulfillment of Order

28
Experiments

• Dynamic Supply Chain Network Configuration for
  different orders.
• Quantifying the Impact of Information Sharing.
• Make-to-Order
• Make-to-Stock (modeled by inventory holding)

29
Data
30
Dynamic SC ConfigurationPartner Selection
31
Quantifying the Impact of Information Sharing

• No information sharing
• Need to rely on forecasting.
• Need to keep safety stock.
• Make-to-stock.
• Information sharing
• Synchronization of activities.
• JIT manufacturing and delivery.
• No inventory.
• Make-to-order.

32
Constraints modeling MTS

• Stock level constraints
• Enough components to meet same production level
  as last n periods.
• Enough finished goods to meet same demand as last
  n periods.

33
The Value of Sharing Info
34
Impact on the Capacity of the Network

• Minimal warehousing requirements for
  make-to-order SC.
• Bull-whip effect.

Profit Increase of 380 at a cost increase of
only 12
35
A Realistic MILP Planning Model for Private
Marketplaces
36
Additional Features

• Fixed costs
• Production
• Transportation
• Can be used to model international trade tariffs.
• Transportation Lead-times
• Air Sea
• Transshipment Hubs and Merge-in-Transit
• Customer Service Levels

37
Additional Notation

• d index to denote transportation mode (1
  Air 2 Sea).
• D Set of Transportation modes.
• h index to denote transshipment hub.
• H Set of Transshipment hubs.
• g index to denote shipment package.
• G Set of shipment packages.

• Parameters
• TFC Fixed cost of Transportation.
• PFC Fixed cost of Production.
• TL Transportation lead-time.
• CSL Customer Service Level.
• LSC Cost of Lost Sale.
• BD Buyer Demand.
• Variables
• S Supplies received at the destination.
• BS Qty sold to Buyer.

38
Objective Maximize Profit
Production
Transportation
Fixed Costs
39
Capacity Constraints

• Capacity Constraints with Fixed Costs
• Production Capacity
• Transportation Capacity

40
Transportation Constraints
41
Customer Service Level
42
Transshipment Hub

• Model scenario where suppliers may be preferred
  for procurement, if they are already supplying
  other components.
• Model merge-in-transit and cross-docking centers.
• In-coming inventory, Packaging and Outgoing
  inventory

43
Transshipment Hub Constraints
44
Computational Complexity

• Production planning problems with fixed cost are
  NP hard.
• Using Branch and Bound
• Network flow problems with fixed cost do not
  converge fast enough.
• Hence, need to develop tighter formulations.

45
Tighter Formulation

• Implication Constraint

• Zero-Production Nodes

46
Experiments

• Dynamic Supply Chain Network Configuration for
  different orders.
• Effect of Transshipment Hubs.
• Analysis of Supply Chain Costs.
• Managing Multiple Generations of Products.

47
Dynamic SC Network Configuration
48
Dynamic SC Network Configuration
49
Dynamic SC Network Configuration
50
Dynamic SC Network Configuration

• Selection of partners based on location of buyer.
• Total landed cost of fulfilling the order.
• Logistics congestion can result in underutilized
  manufacturing plants.
• Synchronization of manufacturing with the
  logistics schedules.
• In combined planning manage trade-off
• In savings from joint procurement against the
  need to procure from more expensive suppliers.

51
Transshipment Hubs
52
Transshipment Hubs

• Existing suppliers are preferred for procurement
  of other sub-assemblies.
• Sub-assembly suppliers down to 3 from 4, Contract
  manufacturers down to 2 from 3.
• Results in supplier rationalization.

53
Analysis of Supply Chain Costs
54
Analysis of Supply Chain Costs

• Decreasing demand and Seasonal-up
• More expensive suppliers and transportation to
  meet large demands early on.
• Ascending demand and Seasonal-down
• Inventory costs are higher because of need to
  store goods to meet late demand.

55
Managing Multiple Generations of Products
56
Managing Multiple Generations of Products
57
Managing Multiple Generations of Products

• Time-to-market vs. Product Introduction cost.
• Trade-off between savings from joint procurement
  for two different generations and expenses for
  procurement from expensive suppliers.

58
QP and MILP model for SC with Public Trading
Exchange
59
Models deployed in the SC
60
Models for PTX

• Quadratic Programming
• Dynamic Pricing based on Supply Demand.
• Chooses qty and price in both marketplaces.
• Mixed Integer Linear Programming
• Combinatorial auction.
• Chooses winning bids in both marketplaces.

61
Basic AssumptionsPublic Trading Exchange

• Manufacturers participate in Multiple PTX
• Participants share supply and demand information
  during negotiations.
• More information ascertained with each round of
  negotiations.
• Information
• Supply-Demand Curves or Qty-Price Bids
• Delivery Date

62
Quadratic Programming Model
63
Features of the Model

• Dynamic Pricing responsive to market
• Selection of Partners
• Selection of Optimal Price
• Selection of Optimal Quantity
• Synchronization of Manufacturing and Logistics
  Schedules.

64
Supply-Demand Curves
65
Notation

• Parameters
• A Slope of supply/demand curve
• B Intercept of supply/demand curve
• C Maximum availability of components
• CM Production capacity.
• T Transportation capacity
• CI Inventory capacity
• SL Service Level
• B Buyers demanded quantity.
• P cost of production
• LT Transportation lead-time
• Variables
• S supplies transported between two parties.
• I inventories at each time period
• M Qty produced by manufacturer
• O Qty of components procured

• i index used to denote comp.
• j index used to denote suppliers
• k index used to denote assemblers
• l index used to denote models
• m index used to denote the buyers
• Subscripts
• I set of components.
• L set of finished models
• J set of suppliers.
• K set of Manufacturers
• M set of Buyers

66
Objective

• Maximize Profit
• Profit Revenue (Cost of Procurement Cost
  of Production Cost of Transportation
  Cost of Inventory)

67
Constraints

• Procurement Marketplace

Marketplace Capacity
Component Supplier Inventory
68
Constraints

• Manufacturing Facilities

69
Constraints

• Finished Models Marketplace

70
Constraints

• Logistics Marketplace
• Warehousing

71
Constraints

• Logistics Marketplace
• Transportation

Transport Capacity
72
Experiment
73
Solution

• Determines optimal quantities and corresponding
  prices.
• The solution of the model also provides schedules
  for manufacturing and logistics.
• QP provides integrated strategic-level dynamic
  pricing and partner selection tool and low level
  operational scheduling tool.

74
MILP Model
75
Combinatorial Auctions

• Sellers quote prices for bundles of components.
• Buyers place bids on bundles of finished models.
• All bids provide
• Qty - q1,q2,q3,q4,q5
• Due Date - 0,0,0,0,1,0,0,0,0
• Price - 123.
• Manufacturer needs to choose optimal seller bids
  and accept optimal buyer bids.

76
Features of the Model

• Combinatorial Auctions in Multiple PTX.
• Selection of Partners.
• Selection of Optimal Bids.
• Production Scheduling.

77
Notation

• Parameters
• SQ Qty being sold of components
• SD Date on which bid will deliver
• SP Quoted selling price of component
• BQ Qty demanded of models
• BD Date on which bid needs to be fulfilled
• BP Quoted buying price of models
• R Units of components required for 1 unit of
  the model
• T Production lead-time
• P Production cost
• W Inventory holding cost
• Variables
• S Accept bid
• I inventories at each time period
• M Qty produced by manufacturer

• i index used to denote comp.
• j index used to denote suppliers
• l index used to denote models
• m index used to denote buyers
• n index used to denote bids
• Subscripts
• I set of components.
• L set of finished models
• J set of suppliers.
• N set of bids
• M set of buyers

78
Objective

• Maximize Profit
• Profit Revenue (Cost of Procurement Cost
  of Production Cost of Inventory)

79
Constraints

• Manufacturer Constraints

80
Future Experiments

• To study impact of dumping on supply chain.
• To study impact of sudden shortages on the supply
  chain.

81
Future Work
82
Future work

• To develop a multi-layer adaptive control for
  supply chain planning
• Based on SC performance can plan to buy or sell
  additional capacity
• To develop risk management models for SC

83
Academic Papers
84
Journal Papers

• Journal Paper
• N. Viswanadham and Roshan Gaonkar, Internet-based
  Collaborative Scheduling in Global Contract
  Manufacturing Networks, Submitted to the IEEE
  Transactions on Mechatronics.
• Journal Paper in Revision
• N. Viswanadham and Roshan Gaonkar, Partner
  Selection and Synchronized Planning in Dynamic
  Manufacturing Networks, Submitted to the IEEE
  Transactions on Robotics and Automation.

85
Conference Papers

• Conference Papers
• N. Viswanadham, Roshan S. Gaonkar and
  V.Subramanian, Optimal configuration and partner
  selection in dynamic manufacturing networks,
  Proceedings of the IEEE International
  Conference on Robotics and Automation, Seoul, May
  2001, pp 854-859.
• Roshan S. Gaonkar and N. Viswanadham,
  Collaborative scheduling model for supply hub
  management, Third AEGEAN International conference
  on Analysis and Modelling of Manufacturing
  Systems, Tinos Island, Greece, May 16-20, 2001.
• Roshan S. Gaonkar and N. Viswanadham, Systematic
  Design of Electronic Marketplaces, Proceedings of
  the Total Enterprise Solutions Conference,
  Singapore, June 2001.
• N. Viswanadham and Roshan S. Gaonkar, Foundations
  of E-supply chains, Int. Conf. on Port and
  Maritime R D and Technology, Singapore, Oct
  29-31, 2001.