Vous êtes sur la page 1sur 37

Transportation Management

Fundamental Concepts
Chris Caplice ESD.260/15.770/1.260 Logistics Systems Nov 2006

Agenda
Introduction to Freight Transportation Levels of Transportation Networks

Physical Operational Service

Impact of Transportation on Planning

MIT Center for Transportation & Logistics ESD.260

Chris Caplice, MIT

Case Study: Shoes from China


How should I ship my shoes from Shenzhen to Kansas City?

Shoes are manufactured, labeled, and packed at plant ~4.5M shoes shipped per year from this plant 6,000 to 6,500 shoes shipped per container (~700-750 FEUs / year) Value of pair of shoes ~$35

Map showing Shenzhen, China and Kansas City, US removed due to copyright restrictions.

MIT Center for Transportation & Logistics ESD.260

Source: Chiu MLOG 2005

Chris Caplice, MIT

Pallets vs Slipsheets

Images of pallets removed due to copyright restrictions.

48 x 40 in. pallet is most popular in US (27% of all palletsno other size over 5%) 1200 x 800 mm "Euro-Pallet" is the standard pallet in Europe

MIT Center for Transportation & Logistics ESD.260

Chris Caplice, MIT

Containers
Characteristics

Airtight, Stackable, Lockable TEU (20 ft)


Volume 33 M3 Total Payload 24.8 kkg
Images removed due to copyright restrictions.

International ISO Sizes (8.5 x 8)


FEU (40 ft)


Volume 67 M3 Total Payload 28.8 kkg

Domestic US (~9 x 8.25)


53 ft long
Volume 111 M3 Total Payload 20.5 kkg

MIT Center for Transportation & Logistics ESD.260

Chris Caplice, MIT

Inland Transport @ Origin


3 Port Options

Shekou (30k)
Truck

Yantian (20k)
Rail Truck

Hong Kong (32k)


Rail Truck Barge

In Hong Kong

9 container terminals

Figure by MIT OCW. MIT Center for Transportation & Logistics ESD.260

Chris Caplice, MIT

Ocean Shipping Options


40 shipping lines visit these ports each w/ many options Examples:

APL APX-Atlantic Pacific Express Service


Origins: Hong Kong (Sat) -> Kaohsiung, Pusan, Kobe, Tokyo Stops: Miami (25 days), Savannah (27), Charleston (28), New York (30)

CSCL American Asia Southloop


Origins: Yantian (Sat) -> Hong Kong, Pusan Stops: Port of Los Angeles (16.5 days)

Canada
CHINA
Kansas City DC

United States

New Kowloon

Mexico
Shenzhen Plant

Victoria

MIT Center for Transportation & Logistics ESD.260

Figure by MIT OCW.


Chris Caplice, MIT

Inland Transportation in US

Port of New York

United States
Kansas City DC Port of Los Angeles / Long Beach.

New York

Kansas City

Los Angeles

Port of Savannah

Savannah

Port of Miami

Mexico

Miami Figure by MIT OCW.

MIT Center for Transportation & Logistics ESD.260

Chris Caplice, MIT

Port of New York / New Jersey


Maher Terminal

Express Rail II NS RR
Double stack thru: Harrisburg, Pittsburgh,

MN

Montreal
MI WI

ME

Minneapolis/ St. Paul

Massena Toronto Syracuse Albany


NY VT NH

Ayer

Milwaukee
IA

MI

Cleveland, Ft. Wayne, to Kansas City

Chicago Indianapolis
IL

Fort Detroit Buffalo Wayne PA Toledo Harrisburg Pittsburgh


OH IN

Worcester
CT RI

MA

NJ DE

Columbus
WV VA

NEW YORK/ NEW JERSEY

CSX RR (5-10 days)


Double stack thru: Philadelphia, Baltimore,

Cincinnati

Washington D.C.
MD

Kansas City

MC

East St. Louis Nashville

KY

Norfolk

Knoxville
TN

Washington, Pittsburgh, Stark, Indianapolis, to Kansas City

Memphis Little Rock


AR

Charlotte

NC SC

Birmingham Atlanta Meridian


MS AL GA

Columbia Charleston Savannah Jacksonville


FL

CSX Norfolk Southern Canadian Pacific Canadian National Connections

Truckload (2.5 3 days)


NJ Turnpike to I-78W, I-

LA

Mobile New Orlears

81S, I-76/70 to Kansas City

Tampa
0 0 500 Km 500 Mi

Miami

Figure by MIT OCW. MIT Center for Transportation & Logistics ESD.260

Chris Caplice, MIT

Truck & Intermodal Operations


Over the Road Truck Power Unit & 53 Trailer
Photographs removed due to copyright restrictions.

Container on Flat Car (COFC) Double Stack

Trailer on Flat Car (TOFC)


MIT Center for Transportation & Logistics ESD.260

10

Chris Caplice, MIT

Air Freight

Photographs removed due to copyright restrictions.

MIT Center for Transportation & Logistics ESD.260

11

Chris Caplice, MIT

Transport Options
So how do I ship shoes from Shenzhen to Kansas City? What factors influence my decision? Consider different types of networks

Physical Operational Strategic


12
Chris Caplice, MIT

MIT Center for Transportation & Logistics ESD.260

Transportation Networks
UNITED STATES

Atlantic Ocean

Example: Two clients ship product to Rotterdam. There are rail and truck options to multiple ports with various ocean carrier options.

Legend:
Truck Rail Ocean

Figure by MIT OCW. MIT Center for Transportation & Logistics ESD.260

13

Chris Caplice, MIT

Transportation Networks

UNITED STATES

ATLANTIC OCEAN

Example: Two clients ship product to Rotterdam. There are rail and truck options to multiple ports with various ocean carrier options.

Legend:
Truck Rail Ocean

Physical

Operational

Strategic

Figure by MIT OCW. MIT Center for Transportation & Logistics ESD.260

14

Chris Caplice, MIT

Three Layers of Networks


Physical Network: The actual path that the product takes from origin to destination. Basis for all costs and distance calculations typically only found once. Operational Network: The route the shipment takes in terms of decision points. Each arc is a specific mode with costs, distance, etc. Each node is a decision point.

Strategic Network: A series of paths through the network from origin to destination. Each represents a complete option and has end to end cost, distance, and service characteristics.
MIT Center for Transportation & Logistics ESD.260

15

Chris Caplice, MIT

The Physical Network


Guideway

Free (air, ocean, rivers) Publicly built (roads) Privately built (rails, pipelines) Publicly built (ports, airports) Privately built (trucking terminals, rail yards, private parts of ports and airports) Public (roads, air space, rivers) Private (rail, pipelines)

Terminals

Controls

The physical network is the primary differentiator between transportation systems in established versus remote locations.
MIT Center for Transportation & Logistics ESD.260

16

Chris Caplice, MIT

Operational Network
Four Primary Components

Loading/Unloading Local-Routing (Vehicle Routing) Line-Haul Sorting


Collection Main Sort

HUB

Distribution

Pick up cycle
Local Sort

Line-haul

LOC LOC
Line-haul Main Sort Local Sort

Delivery cycle

HUB

Node & Arc view of network Each Node is a decision point


MIT Center for Transportation & Logistics ESD.260

17

Chris Caplice, MIT

Strategic Network
Path view of the Network Used in establishing overall service standards for logistics system Summarizes movement in common financial and performance terms used for trade-offs
Air: 3 days, $??/pair

Shenzhen

All Water Route: 35 days, $0.65/pair

KC

Land Bridge: 28 days, $0.75/pair


MIT Center for Transportation & Logistics ESD.260

18

Chris Caplice, MIT

Transportation Impact on TC
D D Q TC (Q) = vD + A + rv + k L + BSO Pr[ SO] 2 Q Q

How does transportation impact our total costs?


Cost of transportation
Value & Structure

Lead Time
Value & Variability & Schedule

Capacity
Limits on Q

Miscellaneous Factors
Special Cases

MIT Center for Transportation & Logistics ESD.260

19

Chris Caplice, MIT

Simple Transportation Cost Functions


Pure Variable Cost / Unit
$/shipment

Modify unit cost (v) for Purchase Cost


units

$/unit

units

Pure Fixed Cost / Shipment


$/shipment

Modify fixed order cost (A) for Ordering Cost


units

$/unit

units

Mixed Variable & Fixed Cost


$/shipment $/unit

Modify both A and v


units

units MIT Center for Transportation & Logistics ESD.260

20

Chris Caplice, MIT

More Complex Cost Functions


Variable Cost / Unit with a Minimum
$/shipment $/unit

algorithm
units

units

Incremental Discounts
$/shipment $/unit

algorithm
units

units

Note that approach will be similar to quantity discount analysis in deterministic EOQ

MIT Center for Transportation & Logistics ESD.260

21

Chris Caplice, MIT

Lead Time & Lead Time Variability


Source 3PL US Pac Rim EU xL 55 25 85 75 L 45 25 35 40 What is impact of longer lead times? What is impact of lead time variability? What are the sources of the lead time & variability?

Major Packaged Consumer Goods Manufacturer

Avg (std dev)

Total Lead Time = 56 days 2 (2) 2 (1) 26 (5)


Ocean Transport
22

15 (12)
Production Time

5 (2)
Unload & Clear Port

6 (3)
Inland Transport
Chris Caplice, MIT

Inland Transport

Load & Delay @ Port

MIT Center for Transportation & Logistics ESD.260

Lead Time Variability


Demand ~U(1,3) Lead Time 3 weeks 6 units Demand ~U(1,3) Lead Time ~U(3,6) 11 units 6 days

4 units

6 units 3 days

6 units

7 units 4 days

MIT Center for Transportation & Logistics ESD.260

23

Chris Caplice, MIT

Lead Time Variability


This is essentially the random sum of random numbers

D~(xD, D) items demanded / time, iid L~(xL, L) number of time periods

We want to find the characteristics of a new variable, y:

y = di = d1 + d 2 + d3 + d 4 + ... + d L
i =1

Note that any observation of demand, di, consists of both a deterministic and a stochastic component:

% di = E[ D] + d % ] = 0 and 2 = 0 + 2 where E[d % D d


MIT Center for Transportation & Logistics ESD.260

24

Chris Caplice, MIT

Lead Time Variability


First, lets find the expected value, E[y]
E [ y ] = E[d1 + d 2 + d3 + ...d L ] = E ( E [ d1 ] + E [ d 2 ] + E [ d3 ] + ...E [ d L ]) + % +d % +d % + ...d % = E + LE D E d [ ] 2 3 L 1 = E [ D] E [ L] + 0 E [ y ] = E [ D] E [ L]

( d% + d%
1

% + ...d % + d 2 3 L

MIT Center for Transportation & Logistics ESD.260

25

Chris Caplice, MIT

Lead Time Variability


What is the impact of lead time variability? Assumptions

Lead Time and Demand are independent RVs DLeadtime =Demand over lead time Leadtime= Standard deviation of demand over L

E ( DLeadtime ) = E ( L) E ( D)
2 Leadtime = E ( L) + ( E ( D) ) L 2 D 2

Questions we can answer: 1. What is the impact of lead time variability on safety stock? 2. What is the trade-off between length of lead time and variability?
MIT Center for Transportation & Logistics ESD.260

26

Chris Caplice, MIT

Transportation Options
When is it better to use a cheaper more variable transport mode?

Air higher v, smaller Rail lower v, larger


TRC = TRC a TRC r where, TRC a = 2 Aa Dva r + k a L,a va r + Dva TRC r = 2 Ar Dv r r + k r L,r v r r + Dv r

Pick mode with smaller TRC


MIT Center for Transportation & Logistics ESD.260

27

Chris Caplice, MIT

Transport vs. Inventory Costs


Multiple Modes
Mode 1 Mode 2 Mode 3

cv+rvtm
Transport Cost per Shipment

cf
Shipment Size
MIT Center for Transportation & Logistics ESD.260

28

Chris Caplice, MIT

Time Space Diagram


Consider a simple Production to Consumption Network
Space

Consumption
m ng i v o

holding

Storage Production

holding

Primary Tradeoffs:
handling Moving vs. Holding Time

MIT Center for Transportation & Logistics ESD.260

29

Chris Caplice, MIT

Mode Comparison Matrix


Truck Operational Cost Market Coverage Degree of competition Traffic Type Length of haul Capacity (tons)
Moderate Pt to Pt Many All Types Low Terminal to Terminal Few Low to Mod Value, Mod to High density Medium Long 50 12,000

Rail
High

Air

Water
Low Terminal to Terminal Few Low value, High density Med - Long 1,000 6,000

Terminal to Terminal Moderate High value, Low density Long 5 12

Short Long 10 25

MIT Center for Transportation & Logistics ESD.260

30

Chris Caplice, MIT

Mode Comparison Matrix


Truck Speed Availability Consistency (delivery time) Loss & Damage Flexibility
Moderate High High Low High

Rail
Slow Moderate Moderate High Low

Air
Fast Moderate Moderate Low Moderate

Water
Slow Low Low Moderate Low

Truck Rail BTU/ Ton-Mile Cents / Ton-Mile Avg Length of Haul Avg Speed (MPH)
MIT Center for Transportation & Logistics ESD.260

Air
21.90 1000 400

Water Pipeline
490 0.27 300 5
Chris Caplice, MIT

2,800 7.50 300 40

670 1.40 500 20


31

42,000 680 0.30 1000 10

Lead Time Variability


This is essentially the random sum of random numbers

D~(xD, D) items demanded / time, iid L~(xL, L) number of time periods

We want to find the characteristics of a new variable, y:

y = di = d1 + d 2 + d3 + d 4 + ... + d L
i =1

Note that any observation of demand, di, consists of both a deterministic and a stochastic component:

% di = E[ D] + d % ] = 0 and 2 = 0 + 2 where E[d % D d


MIT Center for Transportation & Logistics ESD.260

32

Chris Caplice, MIT

Lead Time Variability


First, lets find the expected value, E[y]
E [ y ] = E[d1 + d 2 + d3 + ...d L ] = E ( E [ d1 ] + E [ d 2 ] + E [ d3 ] + ...E [ d L ]) + % +d % +d % + ...d % = E + LE D E d [ ] 2 3 L 1 = E [ D] E [ L] + 0 E [ y ] = E [ D] E [ L]

( d% + d%
1

% + ...d % + d 2 3 L

MIT Center for Transportation & Logistics ESD.260

33

Chris Caplice, MIT

Lead Time Variability


Finding V[y]
% +d % +d % + ...d % = L E [ D] + d 1 2 3 L % +d % +d % + ...d % y = E [ d1 ] + E [ d 2 ] + E [ d3 ] + ...E [ d L ] + d 1 2 3 L

Both terms are independent random variables note that E[D] is a constant, and V[aX] = a2V[x] and that V[X+Y] = V[X]+V[Y], so that
2 % +d % +d % + ...d % ) + V ( d = ( E [ D ]) L L 2 3 1 2 y 2

Substituting

% +d % +d % + ...d % =d 1 2 3 L

We get

2 2 = ( E [ D ]) L + 2 y 2
2

By definition, we know that


2 2

2 2 = E X X ( E [ X ])

Which gives us = E E [ ]
MIT Center for Transportation & Logistics ESD.260

2 2 0 E = E =
Chris Caplice, MIT

34

Lead Time Variability


Substitute in so that, Recalling that,
2 % % % % = + + + = E E d d d ... d 1 2 3 L 2 2

( x1 + x2 + x3 + ... + xn )
2 2

= x + x2 + x3 + ... + xn + 2
2 1 2 2 2

i =1 j = i +1

xx
i

L L %2 % 2 % 2 2 % % % We get that, = E = E d1 + d 2 + d3 + ...d L + 2 di d j i =1 j = i +1 L L % % 2 2 2 2 % % % % = E d1 + d 2 + d3 + ...d L + 2 E di d j i =1 j =i +1

Recalling that if random variables X and Y are independent, then E[XY]=E[X]E[Y], and the E(d~)=0, the second term goes to 0, thus,
2 2 %2 +d % 2 +d % 2 + ...d % 2 =E = E d L 2 3 1

= E [ 1 + 2 + 3 + ... L ] % 2 = E [ L] E [ ] = E [ L] E d i
MIT Center for Transportation & Logistics ESD.260

%2 where i = d i

Which, again, is a random sum of random numbers! (I substituted in the to make it read easier)
Chris Caplice, MIT

35

Lead Time Variability


Starting with,
2 % 2 = E [ L] E d i
2 X 2 =X + E[X ] E 2 We recall that for any 2 X 2 E[X ] X = E random variable X,

or

We get, So that,

2 2 2 2 2 % % = d% + E d = d E d 0 + = % D

2 % 2 = E [ L ] 2 = E [ L] E d D i
2 2 = ( E [ D ]) L + 2 y 2 y 2 2 2 2 + E [ L ] D = ( E [ D ]) L

Combining terms,

MIT Center for Transportation & Logistics ESD.260

36

Chris Caplice, MIT

Questions?

Vous aimerez peut-être aussi