Logistics Network Configuration

1
Logistics Network Configuration

• Jerry Banks

2
The Logistics Network

• Facilities
• Suppliers, warehouses, DCs, and retail outlets
• Goods
• Raw materials, WIP and FGI that flow between the
  facilities

3
(No Transcript)
4
Key strategic decisions

• Determining the appropriate number of warehouses
• Determining the location of each warehouse
• Determining the size of each warehouse
• Allocating space for products in each warehouse
• Determining which products customers will receive
  from each warehouse

5
Objective

• Minimize total cost consisting of
• Production/procurement costs
• Inventory holding costs
• Facility costs (storage, handling, and fixed
  costs)
• Transportation costs
• Subject to meeting a specified service level

6
Increasing the number of warehouses

• Yields
• Improvement in service level
• Average travel time is reduced
• Increase in inventory costs
• Increased safety stocks at each warehouse
• Increase in ordering costs
• More orders are placed to keep the entire system
  operating

7
Increasing the number of warehouses

• Yields
• Reduction in outbound transportation costs
• From the warehouses to the customers
• Usually, an increase in inbound transportation
  costs
• From the suppliers to the warehouses

8
Supply Chain Modeling Ford Case Study

• The Logistics Institute
• Jarod Goentzel
• Paul Griffin
• Don Ratliff
• Ford Motor Company
• Glenn Collier
• Les Ellis

9
Supply Chain
10
(No Transcript)
11
Ford Customer Service Division

• Distribute Ford/Motorcraft parts to U.S. dealers
• Service is priority
• Sales over 3 billion annually
• Stock 220,000 parts
• Regional distribution centers carry 60,000 parts
• National Depot carries remainder

12
North American supply chain

• 1,800 suppliers
• Packagers
• Two Replenishment Centers
• National Depot (slow movers)
• 13 Regional DCs
• 6,000 Ford dealers
• Motorcraft distributors
• Wholesalers

13
Supply chain model

• Transportation
• RC to DC
• DC to dealer
• Routes
• Direct
• Material Handling
• Regional DC
• Inventory
• Regional DC

14
Geocode sites

• U.S.--5 digit zip
• Canada--zip
• Mexico--city names
• Spelling
• Multiple locations

Juarez
Juarez
Juarez
15
North American sites
16
North American network
17
DC inbound transportation cost

• Dedicated truck and rail used from RC
• Use contract costs by Origin-Destination pair
• Fix ratio of truckrail
• Container utilization

18
DC to dealer transportation

• Route contract carriage
• Direct LTL, parcel ground, express
• Mixed sometimes on a route, sometimes direct
• Mode varies by shipment size and priority

19
Current Routes
20
Model outbound routes

• Assign dealers to a route
• Create a point to represent the route
  (RoutePoint)
• Could locate at centroid of all stops
• Smaller network 4,300 demand points to 235
• Calculate route distance instead of direct
  distance
• Does not matter where RoutePoint is located

DC1
21
RoutePoints

22
Benchmark
23
Option
24
End of Ford Example
25
Large amount of data is available

• Location of customers, retailers, existing
  warehouses and distribution centers,
  manufacturing facilities and suppliers
• All products including volumes, special transport
  requirements (such as refrigeration)
• Annual demand for each product by customer
  location

26
Large amount of data is available

• Transportation rates by mode
• Warehousing costs
• Shipment sizes and frequencies for customer
  delivery
• Order processing costs
• Customer service goals

27
Data aggregation

• Too much data
• 1000s of customers for a Coca-Cola DC
• Wal-Mart has 1000s of products
• Aggregation is needed

28
Aggregate customers located in close proximity to
each other

• Replace all customers in a certain cluster as a
  single customer located at the center of the
  cluster

29
Aggregate items into product groups

• Distribution pattern
• All products picked up at the same source and
  destined to the same customers are aggregated
• Product type
• Aggregate variations of the same product into one
  product group

30
Impact of aggregating

• Loss of accuracy versus needless complexity
• Consider the number of aggregated zones
• Researchers say that aggregating into 150 to 200
  zones results in an error lt 1
• More accuracy when customers are distributed
  uniformly within the zone

31
Other considerations when aggregating

• Make sure that each cluster or zone has equal
  demand
• Aggregate products into 20 to 50 product groups

32
Aggregating demand over years

• Customer 1 and Customer 2 on next slide
• Demand over the past five years
• Note that aggregate CV lt CV of each customer

33
Aggregating demand over years
34
Transportation rates

• Next step is to estimate transportation costs
• Rates are almost linear with distance
• Internal rates and external
• External
• TL
• LTL

35
US transportation rates for TL

• Zones
• Every state is a zone
• Except for some large states such as FL
• Two zones in FL
• Zone to zone table costs
• Determine mileage between two cities
• Then, multiply by cost/mile for the origin and
  destination zone
• Cost from Zone A to B not the same as Zone B to A

36
UPS 2nd day by noon 30319 to 22201
37
UPS 5 pounds30319 to 22201

• Next day by 830 AM US57.02
• Next day by 1030 AM US29.32
• Next day by 300 PM US25.54
• 2nd day by Noon US12.74
• By end of 2nd day US11.43
• By end of 3rd day US7.70
• Ground US4.71

38
Frugal customers are taking the air out of
express delivery (WSJ,6/18/02)

• Express delivery is a US25 billion a year
  business in the US
• Dominated by FedEx and UPS
• Packages are moved by air and ground
• Typical ground delivery brings in about US6
• Typical air delivery brings in about US13

39
Frugal customers are taking the air out of
express delivery (WSJ,6/18/02)

• FedEx, UPS and Airborne have spent millions over
  the past few years to make their ground
  deliveries as reliable and high-tech as their air
  counterparts
• Shipments lt500 miles will get there by the end of
  next day by ground

40
Frugal customers are taking the air out of
express delivery (WSJ,6/18/02)

• Boston to New York, 214 miles
• UPS charges US19.14 by 1030 AM next-day
• Air
• UPS charges US4.74 by end of next-day
• Most likely by truck

41
Frugal customers are taking the air out of
express delivery (WSJ,6/18/02)

• About 1/3rd of last years packages marked Air
  never saw the inside of an airplane
• So says U.S. Xpress
• They have a contract with the major airfreight
  carriers to haul by truck

42
LTL Freight Rates

• CNF
• Class 50, 55, 60, , 100, 110, , 400, 500
• 2000 pounds shipped from 30319 to 22201
• Class 50 costs US36.29/100 pounds
• Class 100 costs US63.87/100 pounds
• Class 500 costs US319.34/100 pounds
• So, 2000 pounds of Class 100 shipped from Atlanta
  to Washington would cost 1277.40

43
LTL classes

• Depend on
• Density
• Difficulty in handling
• Liability of carrier

44
Mileage estimation

• Straight line distance (Dab) between a and b
• Approximation

45
Example

• ATL is at lat 33.64 and lon -84.44
• DCA is at lat 38.85 and lon -77.04
• So the approximate straight line distance is
• 69 SQRT(-84.4477.04)2 (33.64-38.85)2
• Or, 624 miles
• The approximate road distance is 624r
• Or, 711 miles where r, the circuitry factor, is
  1.14 in the continental US

46
Example

• Note
• Mapquest says the distance is 642 miles

47
Mileage estimation

• To account for the earths curvature over longer
  distances use

48
Facility costs

• Warehouses and DCs
• Handling costs
• Labor and utilities
• Proportional to the flow of goods
• Fixed costs
• Capital costs, administration
• Not proportional to the flow of goods

49
Facility costs

• Warehouses and DCs
• Storage costs
• Inventory holding costs
• Proportional to average inventory level
• Inventory turnover ratio (l)
• (Annual sales)/(Average inventory level)
• Or, Average inventory level (Annual sales)/l

50
Facility size

• Warehouse or DC must be sized for the maximum,
  not the average
• Maximum inventory level
• 2(Average inventory level)
• Approximation
• But, space must be left for aisles, material
  handling systems, place to put an empty pallet
• So, use a factor of 3 (rather than 2)

51
Facility size

• Annual flow through the warehouse 1000 units
• Turnover ratio 10
• Average inventory level 100
• Size of unit 20 square feet
• Square feet required 100203 6000

52
Facility location problem

• Consider
• Geographical conditions
• Infrastructure condition
• Natural resources
• Labor availability
• Taxation
• Public interest

53
Facility location problem

• A limited number of locations will meet all of
  the considerations
• Facility location problems are generally
  difficult to solve

54
Complexity increases with

• Number of customers
• Number of products
• Number of possible facility locations
• Number of facilities to be located

55
Network design problems

• Much easier to solve
• Several products are produced at several plants
• Plants may have a limited capacity stated
• There is a known demand for each customer
• Transshipment may be permitted through DCs
• There may be an upper bound on total throughput
  at the DCs

56
Solution techniques

• Mathematical
• Heuristics
• A good solution
• Not necessarily optimal
• Exact
• Optimal, i.e., least cost
• Simulation
• To evaluate a specified design

57
Problem statement

• Single product
• Plant p1 has unlimited capacity
• Plant p2 has an annual capacity of 60 units
• The plants have the same production cost
• Warehouses w1 and w2 have the same handling cost
• Customers with demands c150, c2100 and c350
• Distribution costs shown on the following slide

58
Distribution costs
59
Heuristic 1

• w2 dominates w1 for costs to supply c1, c2, and
  c3
• Thus, choose w2 to supply c1, c2, and c3
• Distribute the maximum from p2
• Set p2 60
• Then, p1 provides (5010050) - 60 140
• Total cost (250) (1100) (250)
  (5140) (260) 1120

60
Heuristic 2

• Consider inbound and outbound costs for each path
• p1 to w1 to c1 costs 0 3 3
• p1 to w2 to c1 costs 5 2 7
• p2 to w1 to c1 costs 4 3 7
• p2 to w2 to c1 costs 2 2 4
• means lowest cost

61
Heuristic 2

• Consider inbound and outbound costs for each path
• p1 to w1 to c2 costs 0 4 4
• p1 to w2 to c2 costs 5 1 6
• p2 to w1 to c2 costs 4 4 8
• p2 to w2 to c2 costs 2 1 3
• means lowest cost

62
Heuristic 2

• Consider inbound and outbound costs for each path
• p1 to w1 to c3 costs 0 5 5
• p1 to w2 to c3 costs 5 2 7
• p2 to w1 to c3 costs 4 5 9
• p2 to w2 to c3 costs 2 2 4
• means lowest cost

63
Heuristic 2

• Consider c1
• Assign 50 from p1 to w1 to c1
• Cost 350 150
• Consider c2
• Assign 60 from p2 to w2 to c2 (max 60)
• Cost 360 180
• Assign 40 from p1 to w1 to c2 (2nd lowest)
• Cost 440 160

64
Heuristic 2

• Consider c3
• Assign 50 from p1 to w1 to c3 (p2 maxed)
• Cost 550 250
• Total cost 150160180250 740

65
Optimization

• Uses linear programming
• Optimal 740
• Same as Heuristic 2
• But, this does not always happen
• Many computer codes are available
• Real problems are much more complex
• They require integer programming
• Very difficult to solve

66
Transportation simplex solution
67
What is a feasible solution and how much does it
cost?
68
BFS using NW corner rule US640
69
How do we know if this solution is optimal?
70
Compute ui and vj values

• For all basic variables
• u1 0
• u1 v1 3
• etc.
• Compute ui and vj values
• v1 3
• v2 2
• etc.

71
Compute shadow prices

• For non-basic variables
• sp13 u1 v3 - c13 -2
• Any shadow prices that are lt 0, ignore
• If all shadow prices are lt 0, optimality has been
  reached
• Allocate as much as possible to cell with maximum
  sp
• sp21 sp32 2

72
2nd iteration US580
73
3rd iteration US540
74
4th iteration US525
75
5th iteration US495
76
6th iteration is optimal US475
77
Simulation models

• Mathematical optimization models are static
• Average or annual demand
• Simulation models are dynamic
• Characterize performance for a given design
• Simulation model may include
• Trace driven inputs
• Various inventory policies
• Material handling systems

78
Simulation models

• Demo of AutoMod material handling simulation

79
Simulation models

• Take time to build
• Take time to run

80
The reject-retooling industry

• Read the article Happy returns Heres what
  happens to Many lovely gifts after Santa rides
  off WSJ, 12/26/01
• Perform an economic analysis
• From the reject-retoolers perspective
• From the manufacturers perspective
• From the technicians perspective
• Be sure to include all logistics costs
• Make assumptions if necessary

81
End