Infraday 2009
Consideration of logistics for policy analysis with freight transport modelstransport models
Hanno Friedrich ([email protected])( @ )Gernot Liedtke ([email protected])
15 10 200915.10.2009
AGENDA
• Motivation
• Representation of logistics in freight transportation modelsRepresentation of logistics in freight transportation models
• Lessons learned
1
NEED OF LOGISTICS IN TRANSPORTATION MODELS
• Rising attention for logistics in politics– Masterplan Logistics EU– Masterplan Logistics EU– Masterplan Logistics Germany
• Higher relevance of freight transport in overall transport
• Logistics as natural interface between economic activity system and transportation system
2
DIFFICULTIES IN MODELING FREIGHT/LOGISTICS
• Heterogeneity of actors and data availability
• Freight transport emerges from Definition Logistic Mesostructure:
g p ginteraction between logistic systems using synergies (concave cost functions)
• The routing of freight flows therefore
A logistic mesostructure is an emergent operational structure, that handlesseveral commodity flows. It can be described by how, where and when• The routing of freight flows therefore
is dependent on:– Existence other flows– Networks/tour structures
ythe goods of the commodity flows are transported, reloaded and stored. Amesostructure is the result of an optimization of one or several actors
– Distribution structures – Locations of warehouses and
warehouse structures
punderspecific circumstances. These circumstances include the state of the actorsactorsand the state of their environment.
Need to consider combinations of flows and necessity to model logistic
mesostructures
3
CHOICES – FROM ECONOMIC ACTIVITY TO VEHICLE FLOWS
Company Aspiration Choices
Logistic choices
Company Aspiration Choices (Profits, Growth)
Activity Pattern Choices(P d t Mi M k t V l )(Product Mix, Markets, Volumes)
Business Location Choices (Factory, Distribution Regions)
Sourcing Choices (Suppliers)
Supply Path Choices
Logistic Location Choices(Warehouses, Reloading Points, Network design)
Supply Path Choices
Lot Size / Frequency Choices
Mode Choices
Dispatching Choices (Tours, Actual operational lot size)
Lot Size / Frequency Choices
4
THE MICRO MACRO GAP
Macro Level: Traffic load
Aggre-gated
indicators
Meso Level: Transport meso-t t (T)
Aggregation possible
structures (T)Lorry survey
n no
t pos
sibl
e:
omm
odity
flow
not p
ossi
ble:
m
odity
flow
sf(n*T)=n*f(T)
Logistic meso-structures (L)
ECHOsurvey
Dis
aggr
egat
ion
Sin
gle
trip ≠
co
Agg
rega
tion
Trip
s ≠
com
m
Optimization and/or
Micro Level: Commodity flows (F)
D and/orsimulation
f(n*F) ≠n*f(F)
Micro Level: Commodity flows (F)
Shipper survey
5
AGENDA
• Motivation
• Representation of logistics in freight transportation modelsRepresentation of logistics in freight transportation models
• Lessons learned
6
Reviewed models
OVERVIEW SELECTED MODELS
ry m
odel
DTR
IP
E/S
LAM
Appr
oach
mod
el
ET2.
0
R P ARLO
G
RA
DE
Cal
gar
GO
OD
SM
ILE
ADA
A
Toky
o
EIU
NE
WIV
ER
BVW
P
ASTR
A
Logistic locations
INTE
R
SYN
T
Logistic locations
Transport paths
evel
s
Mode
c ch
oice
le
Lot size
Logi
stic
Tours
Modeled on disaggregate level (actors, flows between actors or vehicles)
7
Modeled on aggregate (flow) level
INTERLOG – MODEL OVERVIEW
Company generationGeneration Company generationGenerationForwarder generation
Set of attributed companies in space
modulesGeneration Forwarder generation
Set of attributed companies in space
modulesGenerationGeneration
Sourcing module: choice of suppliers
Commodity flows between companies
Distribution Sourcing module: choice of suppliers
Commodity flows between companies
DistributionDistribution
Marketinteraction
module
Shippers models
Forwarders models
Market choiceLorry model
Lot-size choiceMarket
interactionmodule
Shippers models
Forwarders models
Market choiceLorry model
Lot-size choiceMarket choiceLorry model
Lot-size choice
Attributed lorry trips on the road network
Aggregation verification
yNetwork model Attributed lorry trips on the road network
Aggregation verification
yNetwork model
yNetwork model
Aggregation, verificationAggregation, verification
8
INTERLOG – ARTIFICIAL INDUSTRY LANDSCAPE
9
INTERLOG – RESULT: VEHICLE FLOWS
10
SYNTRADE – MODEL OVERVIEW
Model Core W h
Input data and generation of commodity flows
Simulation of h
Output
Simulation of l i ti i
Simulation model
General data• Distances • Regions
Model core
• Warehouse structures
• Pallet km (food retailing sector)
• Shipment structure
warehouse structures in the food retailing sector(model core)
logistics in consumer goods distribution (model periphery)
Model core• Food retailing
companies– Outlet structure– Article structure
• S i
• Optimization heuristic for warehouse structure
p(for inbound transport)
Model periphery • Pallet km (consumer
( )
• Bundling of commodity flows
• Lot size optimization on • Sourcing
Model periphery• Artificial industrial
landscape
including – Number– Levels– Locations
Allocation
• Pallet km (consumer goods distribution)
• Shipment structures for interregional transport flows
ptransport links
• Supply path optimization for commodity flowsp
• Imports• LSPs and
Wholesalers • Consumption in
regions
– Allocation• Forward looking
elements: changes in lot sizes and regions
• Generation of good flows to regions
transport paths
11
E i ti h l ti Si l t d h l ti
SYNTRADE – RESULTS: WAREHOUSE LOCATIONS
Existing warehouse locations of food retailing companies in Germany
Simulated warehouse locations of food retailing companies in Germany
12
SYNTRADE – RESULTS: FUEL PRICE SCENARIO
Change in pallet km(Mio. pallet km) Change in warehouse structures
B i
44.335
Base scenario
36.935
7.400
Fuel price scenario (Assumed change of +25% in transport costs )
37.374
44.143
6.769
Via retailer DirectTotal
13
Via retailer warehouses
DirectTotal
AGENDA
• Motivation
• Representation of logistics in freight transportation modelsRepresentation of logistics in freight transportation models
• Lessons learned
14
LESSONS LEARNED FOR FREIGHT TRANSPORT MODELING
Challenges Solution approaches
• Data availability • Artificial generation of disaggregated datay(modeling heterogeneity)
g gg g• New data sources through more detailed
modeling of sectors
• Complexity– Combinatorial problems
• Simplified but realistic heuristics • Modeling markets and market interaction
– Involvement of many actors in decisions
g• Possible simplifications through market
modeling: simplified representation of supply or demand or the overall market outcome
• Reaching “realistic” overall system states through simulation
• Choosing realistic decision scopes and heuristicsthrough simulation and heuristics
• Forward looking elements in simulation
15
CONCLUSIONS FOR POLICY ANALYSIS
• State of the art freight transport models only include basic logistic aspects
• Recent developments try to include more complex logistic structures but are still limited in scopestructures, but are still limited in scope
• Policy analysis on effects of logistics on transportation by freight transportation models is therefore limited
16