See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/265826759

Optimization Models for Transportation Systems Planning

Article · February 2013

DOI: 10.1002/9781118600283.ch6

CITATIONS READS
0 577

2 authors:

Teodor Gabriel Crainic Michael Florian

School of Management, Université du Québec à Montréal Université de Montréal
367 PUBLICATIONS 17,988 CITATIONS 153 PUBLICATIONS 7,377 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Transportation Systems Planning View project

The Benders Decomposition Algorithm: Literature, Acceleration, Extension, Parallelization View project

All content following this page was uploaded by Michael Florian on 01 November 2019.

The user has requested enhancement of the downloaded file.

___________________________

National Planning Models and

Instruments

Teodor Gabriel Crainic

Michael Florian

June 2008

CIRRELT-2008-20
 National Planning Models and Instruments
Teodor Gabriel Crainic1,2,*, Michael Florian1,3
1
Interuniversity Research Centre on Enterprise Networks, Logistics and Transportation (CIRRELT)
2
Department of Management and Technology, Université du Québec à Montréal, C.P. 8888,
succursale Centre-ville, Montréal, Canada H3C 3P8
3
Department of Computer Science and Operations Research, Université de Montréal, C.P. 6128,
succursale Centre-Ville, Montréal, Canada H3C 3J7

Abstract. This paper focuses on the analysis and planning of multimodal, multiproduct
transportation systems at the international, national, and regional levels, where the movements of
several commodities through the transportation networks and services of several carriers are
considered simultaneously. The main questions relate to the evolution of a given transportation
system and its response to various modifications of its socio-economic, regulatory, and
technological environment. These questions are often part of cost-benefit analyses and
comparative studies of policy and investment alternatives with broad and significant impact, not
only on the transportation system but also on the economy and society as a whole. Canadian
researchers have contributed significantly to this field. They have not only developed models and
methods to perform national planning activities, but also transferred these results to practice
through decision-support software commercially distributed worldwide. This paper presents an
overview of national planning issues together with the main methodological approaches proposed
in the literature, reviews the contribution of Canadian researchers to the science and practice in
this field, and examines current challenges and research trends.

Keywords. National and regional planning, multimodal and multiproduct transportation.

Acknowledgements. Financial support for the current projects mentioned in the paper is provided
by the Natural Sciences and Engineering Research Council of Canada (NSERC) through its
Industrial Research Chair and Discovery grant programs.

Results and views expressed in this publication are the sole responsibility of the authors and do not
necessarily reflect those of CIRRELT.

Les résultats et opinions contenus dans cette publication ne reflètent pas nécessairement la position du
CIRRELT et n'engagent pas sa responsabilité.
_____________________________

* Corresponding author: [email protected]

Dépôt légal – Bibliothèque nationale du Québec,
Bibliothèque nationale du Canada, 2008
© Copyright Crainic, Florian and CIRRELT, 2008
National Planning Models and Instruments

1 Introduction

Transportation is a vital activity for the development, organization, and operations of society.
One is not overstating the case by claiming that the development of transportation, in terms
of geographic coverage, technology, efficiency, structure, and organizational principles, is
closely related to the development of society and civilizations throughout history. There
is hardly any social or economic activity that does not involve transportation and is not
tributary to the efficient, smooth, and timely movement of people, goods, and information.

It is thus not surprising that significant research efforts have been and continue to be
dedicated to issues related to the analysis, planning, management, and operations of trans-
portation systems. The Canadian operations research community is at the forefront of these
efforts and its contributions are many, varied, and important. Applications relate to pas-
senger and freight transportation systems, ranging in scope from a local neighbourhood to
an entire city or region, from a given pick-up and delivery zone to a specific inter-city corri-
dor, from the transportation system of a country to that of a continent. Applications cover
all modes of transportation and all planning levels, from the long term strategic view of
the system, through tactical planning and allocation of resources, to short term planning
of operations and their real-time dynamic management and control. Significant Canadian
contributions have been made to the fundamental models and algorithms supporting all
these activities and many of these contributions have been transferred to practice through
decision-support systems.

This paper focuses on the analysis and planning of multimodal, multiproduct trans-
portation systems. The focus of these models and methods is broad: they apply to strategic
planning issues at the international, national, and regional levels, where the movements of
several commodities through the transportation networks and services of several carriers are
considered simultaneously. The main questions address the evolution of a given transporta-
tion system and its response to various modifications in its environment, e.g., transportation
infrastructure, population distribution, patterns and volumes of production, consumption,
and trade, policies and legislation, technology, and so on and so forth. These questions are
often part of cost-benefit analyses and comparative studies of policy and investment alter-
natives with broad and significant impact not only on the transportation system but also on
the economy and society as a whole.

Canadian researchers have contributed significantly to this field. They have not only
developed models and methods to perform national planning activities, but also transferred
these results to practice through decision-support software commercially distributed world-
wide. The objective of the paper is therefore to present an overview of national planning
issues together with the main methodological approaches proposed in the literature, to re-
view the contribution of Canadian researchers to the science and practice in this field, and
to discuss current challenges and research trends.

The remainder of this paper is organized as follows. Section 2 briefly reviews a number

CIRRELT-2008-20 3
National Planning Models and Instruments

of fundamental concepts related to transportation systems and planning (for more detailed
presentations, one may consult Crainic, 2003; Crainic and Kim, 2007; and Macharis and
Bontekoning, 2004). Section 3 is dedicated to the description of national planning scope,
goals, and main activities, as well as to a succinct review of the associated literature. The
STAN model, planning method, and software system is described in Section 4, and we
conclude in Section 5.

2 Transportation Systems

Movements of people and goods derive from, support, and impact the activities of society.
The demand for freight transportation thus derives from the interplay between producers
and consumers of goods and services. Producers require transportation services to move raw
materials and intermediate products, and to distribute final goods in order to meet customer
demands. The demand for transportation services is met by private firms, and by public and
semi-public organizations that supply transportation infrastructure and services at various
levels of service level and price. Governments supply a significant part of the infrastructure:
roads and highways, as well as significant portions of ports, internal navigation structures,
and rail facilities. Governments also regulate and tax the industry. The interactions among
production and consumption activities, as well as between the corresponding transportation
demand and supply, result in the traffic flows of people, goods, vehicles, and convoys one
may observe in any given region.

It is noteworthy that, as illustrated in Figure 1, these processes are not straightforward

and sequential. Rather, strong feedback loops exist among the demand, the supply, and
the regulatory, economic, technological, and political environment of transportation. Thus,
for example, enhancing the transportation infrastructure or services in a given region may
contribute to create new economic opportunities resulting in modifications in the relative
importance of groups of products consumed and produced within a region and to increase
volumes exchanged within the region as well as with other regions. Such modifications
to the region’s demand for transportation will then most probably result in new services
being offered and, eventually, in new investments in infrastructure. Based on such obser-
vations, governments and international development agencies routinely use investments in
transportation infrastructure as an instrument of regional development.

The organizations that generate the demand for freight transportation are usually iden-
tified as shippers, which may be the producers of the goods or some intermediary firm, e.g.,
freight forwarders, third-party logistics enterprises, and brokers. Shippers have requirements
in terms of cost, time, and quality of service, e.g., on-time delivery reliability. These criteria
directly impact the selection of a transportation mode (witness the large market share of
trucking) and the behaviour of the entire system, e.g., the congestion of the road network,
particularly in and around large urban zones. Shippers decide how much of a given product
to send, when to send it, and how to ship it: what mode and carrier or combination thereof,

CIRRELT-2008-20 4
National Planning Models and Instruments

Physical Production and

(conceptual) consumption
infrastructure and of goods and
services services

SUPPLY DEMAND

Legal, social and

economic environment

Movements of people, goods, vehicles = Traffic

Costs/profits, delays, energy, emissions, …

Figure 1: Transportation and society

the latter corresponding to demand for intermodal transportation. Indeed, the movement of
goods often requires the combined use of several modes. Road and rail, and ocean navigation
combined to road and rail are two of the more frequently used transportation-mode combi-
nations. The utilization of particular loading units — the containers — often facilitates the
operation of intermodal transportation systems. People use similar criteria of cost and time
when selecting how to undertake their trips. Passenger transportation demand thus results
from the combined personal decisions on when and how to travel to everyone’s particular
destination.

Carriers supply transportation services. Railroads, maritime shipping lines, motor-

carrier companies, and postal services are examples of carriers. They may contribute to
the infrastructure as well, as illustrated by the North-American railroad companies. Carri-
ers are generally classified into one of two main types: consolidation-transportation systems,
where one vehicle or convoy serves to move freight for different customers with possibly
different initial origins and final destinations, and customized -transportation carriers that
provide dedicated service to each particular customer. Within passenger transportation, the
former corresponds to public transport by bus, train, or plane, while the latter encompasses
transport by private means, car, bicycle, or taxi.

Truckload trucking offers a typical example of customized transportation for freight trans-

CIRRELT-2008-20 5
National Planning Models and Instruments

portation (e.g., Powell, Bouzaı̈ene-Ayari, and Simaõ, 2007). When a customer calls, the
dispatcher assigns a truck and a driver (or driving team for very long movements) to the
corresponding transportation task. The truck moves to the customer-designated location, is
loaded, and then moves to the specified destination to be unloaded. The driver then calls the
dispatcher to give its position and request a new assignment. The dispatcher may indicate
a new load, ask the driver to move empty to a new location where demand should appear in
the near future, or have the driver wait and call later. Maritime navigation services provided
by for-hire ships belong to the same carrier type (Christiansen et al., 2007).

Freight consolidation transportation is performed by Less-Than-Truckload (LTL) motor

carriers, railways, ocean shipping lines, regular and express postal services, etc. Freight
transportation in some countries where a central authority more or less controls a large part
of the transportation system also belongs to this category. Consolidation transportation
carriers and fundamentally all intermodal transportation systems are organized as so-called
hub-and-spoke networks (Crainic and Kim, 2007). In such systems, service is offered between
certain origin-destination points, the local or regional terminals. Their number is significantly
larger than the number of direct, origin to destination services operated by the carrier.
Consequently, and to take advantage of economies of scale, low-volume demands are moved
first to an intermediate point — a consolidation terminal or a hub — such as an airport,
a seaport container terminal, a rail yard, or an intermodal platform. At a hub, traffic is
consolidated into larger flows that are routed to other hubs by high frequency, high capacity
services; more than one service, of possibly different modes, may be operated between hubs.
Lower frequency services, often operating smaller vehicles, are used between hubs and origin-
destination terminals. When the level of demand justifies it, high frequency, high capacity
services may be run between a hub and a regional terminal or between two regional terminals.

Terminals are an essential component of transportation systems. For passenger trans-

portation, rail stations, bus terminals, and airports provide access to mass-transportation
modes. All consolidation-type freight carriers, railroads, LTL motor carriers, intermodal
ocean navigation lines, etc., make an extensive use of terminals to sort and consolidate cargo
into vehicles and, eventually, vehicles into convoys for increased efficiency. Rail trains, barge
trains, and multi-trailer assemblies illustrate the latter case. Intermodal transportation could
not exist without intermodal terminals that provide transfer services between two, or more,
different modes. Terminals, on the other hand, imply additional handling of cargo and ve-
hicles, resulting in delays, costs, and additional risk of damage. They also have limited
capacity, which contributes to the delays experienced by cargo and vehicles.

Various measures are associated to the flows of vehicles and cargo and give indications
on the performance level of the transportation system, e.g., the expected number of vehicles
using a given road during the morning rush hour and the corresponding level of congestion,
the average number of container ships waiting for access to port facilities, the total number of
ton-kilometres of freight moved by a given train type, etc. Many of these system performance
measures, particularly those linked to the capacity of the infrastructure or services, also
translate into efficiency measures for the users of the system: average or total travel time
between given origin and destination points in the network, average time required to pass

CIRRELT-2008-20 6
National Planning Models and Instruments

through given intermodal facilities, total cost of transportation, and so on.

Each individual or organization that is part of the transportation system reacts individ-
ually to these measures and adjusts its operations accordingly (e.g., shippers may move to
rail to avoid high environmental taxes on road-based transportation). These decisions and
eventual modifications in shipping or operational policy will impact the entire system, by for
example modifying the traffic conditions experienced by the other participants, which will
then modify their behaviour resulting in further changes in transport conditions, and so on.
Observed traffic and performance measures are thus the result of an equilibrium between the
decisions of many participants reflecting their own decision criteria and general behaviour
as well as the condition of the transportation system at the time the observations are made.

The transportation sector in any given region forms thus an integrated dynamic system
and its analysis and planning should be undertaken accordingly. This is the scope of the so-
called national or regional planning models and methods, which aim to represent and analyze
the system in order to understand and predict its behaviour and that of its components,
considering various social, economic, policy, and technological trends and forecasts.

3 National Planning

National planning methodologies — models, methods, and instruments — aim to build a

comprehensive representation of a multimodal transportation system, its main components
and their interactions. Their goal is to achieve a sufficiently good simulation of the global
behaviour of the system to offer a correct representation of the current situation, and to
serve as an adequate analysis tool for planned or forecast scenarios and policies concerning
modifications in

• infrastructure through construction of new modal — roads, rail tracks, pipelines, wa-
terways — or intermodal — maritime and river ports, land intermodal platforms and
terminals, airports — facilities and expansion of existing ones. Infrastructure is also
modified through the abandon of existing facilities, e.g., regional rail lines;

• technology. On the hardware side, this includes new or enhanced vehicular and infras-
tructure technologies such as new traction modes, improved engines, intelligent cruise
control and other advanced driving assistance systems, larger ships, more efficient load-
ing and unloading equipment, etc. Technology also includes a software component, in
particular Intelligent Transportation Systems for carriers, advanced fleet management,
for example, terminals, e.g., enhanced systems to manage containers in port and rail
yards, and infrastructure, e.g., advanced clearing mechanisms at border control points;

• socio-economic characteristics of the region following changes in the population distri-

bution within the region and the patterns and volumes of production, consumption,

CIRRELT-2008-20 7
National Planning Models and Instruments

and trade within the region as well as with external zones, due to urbanization and
various other factors, such as the exhaustion of resources, industry relocation, the cre-
ation of special economic zones, the emergence of new economic players (e.g., China,
India, etc.), and so on;
• policy and regulation from labour conditions to environment-motivated taxation, such
as energy consumption taxes or dynamic demand management, and regulation (e.g.,
obligation to use rail to transit through certain region), from inter-provincial trade
corridors to multi-state free trade agreements, from tariff barriers to security-related
regulation, and so on.

National planning models represent flow volumes by commodity and transportation

mode, as well as associated performance measures, defined on a network representation
of the transportation system. The methods need to be flexible and adaptable to the scope of
studies encompassing a broad range of geographical dimensions, from the level of a country
or a group of countries, to that of a province, region, or corridor. They should provide the
tools to analyze issues identified “now”, as well as to build new evaluation procedures as
the need arises. The modelling framework should be tractable and produce easily accessible
results. No single formulation may address such a broad scope, and thus, a national plan-
ning methodology is typically a set of models and procedures. The three main components
of such a methodology are:

1. supply modeling to represent the transportation modes, infrastructure, carriers, ser-

vices, and lines; vehicles and convoys; terminals and intermodal facilities; capacities
and congestion; economic, service, and performance measures and criteria;
2. demand modeling to capture the product definitions, identify producers, shippers, and
intermediaries, and represent production, consumption, and zone-to-zone (region-to-
region) distribution volumes, as well as mode choices for transportation. Relations of
demand and mode choice to the performance of economic policies and transportation
system performance are also addressed here;
3. assignment of multicommodity flows (from the demand model) to the multimode net-
work (the supply representation). This procedure simulates the behaviour of the trans-
portation system and its output forms the basis for the strategic analyses and planning
activities. The assignment methodology must therefore be both precise in reproducing
current situations and sufficiently general to produce robust analyses of future scenarios
based on forecast data;
4. a number of data-manipulation tools for the analysis, fusion, validation, and updating of
information, as well as result-analysis capabilities for, e.g., cost-benefit, environmental
impact, and energy consumption policy analyses, complement the methodology.

Figure 2 presents the classical combination of these models and methods into the so-called
four-step planning method. Such an approach starts from the economic, demographic, social,

CIRRELT-2008-20 8
National Planning Models and Instruments

DEMAND SUPPLY
Transportation Network
Generation
Modes, infrastructure, terminals,
Total production & consumption
vehicles, costs, congestion, …
by product for O and D zones

Allocation
OD matrices by product SIMULATION
Assignment
OD flows
Mode Choice performance measures
OD matrices
by product & mode

Analyses
feedback

Figure 2: Four-step transportation planning methodology

and political current or forecast data in a given region. The initial step is to determine the
geographical division of the region into zones and to identify the product groups to be
considered. Notice that all activity — production, consumption, shipping and reception of
cargo — within a zone is represented in aggregated form associated to a single point, the
so-called centroid of the zone. This step thus reflects, and specifies for all the other steps of
the planning procedure, the degree of aggregation of the data available for the region under
study.

The product-specific demand is then generated in two steps: first the total production
and consumption volumes (as well as imported and exported volumes, if relevant) and then,
the distribution of these quantities among the origin-destination (zone-to-zone) pairs of the
region. These phases are called generation and allocation, respectively. The last step of the
demand-generation process corresponds to the determination of the modal choice specifying
the set of modes — type of infrastructure and services — that may move the demand of
each product and origin-destination pair. The assignment step then determines the actual
itineraries used to move the demand for each product, each origin-destination pair and each
mode choice, thus performing the simulation of the global behaviour of the system given the
particular scenario studied.

CIRRELT-2008-20 9
National Planning Models and Instruments

The last two steps require the definition of the transportation supply in the region, in
particular the multi-modal infrastructure and service network available to move the demand
and its attributes in terms of costs, travel and terminal operation times, energy consumption
or level of emissions, etc. The assignment step also requires the specification of the criteria
used to select itineraries and to measure performance, as well as the rules to translate volumes
of demand of given products into vehicle and convoy utilization for each transportation mode
specified in the corresponding mode-choice set. The actual national planning process is
rarely linear, however. Thus, for example, many analyses are usually performed on a given
set of supply and demand data, both current and forecast, by varying the parameters (e.g.,
vehicular technology or system utilization policies) of the future scenarios contemplated.
On a more general scale, feedback mechanisms are used to modify the parameters of the
demand generation steps, including allocation and mode choice, for future scenarios given
the simulated performance of the transportation system.

The prediction of multicommodity freight flows over a multimodal network is an im-

portant component of transportation science and has generated significant interest over the
years. However, perhaps due to the inherent difficulty and complexity of such problems, the
study of freight flows at the national or regional level has not yet achieved full maturity,
in contrast to passenger transportation where the prediction of car and transit flows over
multimodal networks has been studied extensively and several of the research results have
been transferred to practice (Cascetta, 2001; Crainic and Florian, 2006; Florian and Hearn,
1995). The number of applications is increasing, however, and the methodologies used are
continuously more comprehensive and sophisticated as illustrated in the survey of national
planning projects undertaken in various regions of the world by Crainic, Gendreau, and Kun-
cyté (2006). In the following, we review the most frequently used methodologies for freight
planning and point associated references.

3.1 Demand

The modeling of demand attempts to describe the economic activities of a region, its pro-
duction, consumption, import and export of goods. For planning purposes, the output of
demand models is a series of product-specific demand matrices indicating the volumes to be
moved from one zone to another. The process is often completed by the modeling of mode
choice, which specifies for each product and origin-destination combination on what set of
transportation infrastructure or services the demand may be moved.

A number of countries have developed input-output models of their economy which serve
to determine the basic production and attraction of goods (Isard 1951; Casceta 2001 and
references therein). In order to use an input/output model as a demand model, it is necessary
to disaggregate the inputs and outputs by region and zone. This process is complex, and
is usually not integrated with a supply representation and assignment procedure. When an
input/output model is not available, the initial determination of origin-destination matrices
is carried out by using national statistics on production, consumption, imports and exports

CIRRELT-2008-20 10
National Planning Models and Instruments

combined with surveys of particular industrial sectors to complete missing or unreliable

information. This process may be tedious since one has to reconcile data from several sources
which may be collected by using different geographical subdivisions or inconsistent product
definitions. The results of the disaggregated input/output model or the ad hoc estimation
procedures serve for the initial computation of origin-destination matrices for each product
but without a subdivision by mode.

A second class of models that is well studied for the prediction of interregional commodity
flows is the spatial price equilibrium model and its variants (Friesz, Tobin, and Harker, 1983;
Harker and Friesz 1986a,b; and Harker 1987; see also Florian and Hearn, 1995 and Nagurney,
1999). This class of models determines simultaneously the flows between producing and
consuming regions, as well as the selling and buying prices that satisfy the spatial equilibrium
conditions. A spatial equilibrium is reached when for all pairs of supply and demand regions
with a positive commodity flow, the unit supply price plus the unit transportation cost is
equal to the unit demand price; the sum is larger than this price for all pairs of regions
with no exchanges. A simple network (bipartite graph) is generally used to represent the
transportation system. These models rely to a large extent on the supply and demand
functions of producers and consumers, respectively, which are rarely available and quite
difficult to calibrate. There are thus relatively few applications of this class of models for
the determination of demand by products.

3.2 Mode choice

Mode-choice models aim to describe the set of transportation modes or services that may
be used to carry specific products or groups of products. The mode choice definition may
be rather general, e.g., petroleum moves by ship and pipeline, extremely specific indicating
a particular set of single or multimodal paths for a given product, shipper, and origin-
destination pair, or anywhere in between. The level of detail of modal specification need not
be the same for all products or inter-zone trade flows. The specification of mode choice may
be inferred from historical data and shipper surveys or it may result from a formal description
and modeling effort (Winston 1983). The output of this process are either coefficients that
indicate how to split the demand of a given origin-destination pair between the paths of a
given set, or origin-destination demand matrices with particular sets of allowed modes.

Random utility models, developed and largely used for the analysis and planning of person
transportation systems, have been proposed for freight transportation as well, but their use
in actual applications is scarce (Cascetta 2001). The huge number of paths that have to be
explicitly generated and stored, coupled to the challenge of performing this task for forecast
data, may explain this phenomenon. At aggregated levels, mode choices have been specified
for particularly important product flows by explicitly surveying the major logistic chains
used between pairs of macro-regions.

CIRRELT-2008-20 11
National Planning Models and Instruments

3.3 Supply representation and assignment

Once modal origin-destination matrices have been developed, the next step is to assign them
to the supply network model by using some route choice mechanism. The results of such an
assignment model — product flows and performance measures — form part of the input to
demand and cost-benefit modeling and analysis.

One class of assignment mechanisms is based again on the application of random utility
models to the choice of paths defined previously by the mode choice phase. It is notewor-
thy that the attributes of predefined paths are determined by the state of the network at
generation time and are not responsive to assignment results. Thus, for example, congestion
conditions are very difficult to represent. Moreover, the utility and choice models have to be
calibrated, and all paths have to be generated, for each scenario, which is quite difficult to
perform when forecast data is used.

Another class, network optimization models, enable the prediction of multicommodity

flows over a multimodal network that represents the transportation facilities at a level of
detail appropriate for a nation or region. The demand and mode choice are exogenous and
intermodal shipments are permitted. Within the specified mode choice, the optimization
(assignment) engine determines the best (with respect to the specified network performance
measures) multimodal paths for each product and origin-destination pair. The emphasis is
on a proper representation of the network and its different transportation modes, the corre-
sponding intermodal transfer operations, the various criteria used to determine the movement
of freight, the interactions and competition for limited resources captured through the repre-
sentation of congestion effects, and the associated estimation of the traffic distribution over
the transportation system considered to be used for comparative studies or for discrete time
multi-period analyses.

Studies in the 1970s used rather simple network representations (e.g., Jones and Sharp,
1977; Sharp, 1979). Several studies also attempted to extend spatial equilibrium models to
include more refined network representations and to consider congestion effects and shipper-
carrier interactions. Friesz, Gottfried, and Morlok (1986) present a sequential model which
uses two network representations: detailed separate networks for each carrier, and an ag-
gregate, shipper-perceived network. On each carrier network commodities are transported
at the least total cost. On the shipper-perceived network, traffic equilibrium principles are
used to determine the carriers that shippers choose to move their traffic. This approach
was quite successful in the study of logistics of products where a very limited number of
shippers and carriers interact and strongly determine the behaviour of the system. A typical
example is the coal market between electric utilities in the United States and their suppliers
in exporting countries. Friesz and Harker (1985), Harker and Friesz (1986a,b), Harker (1987,
1988), and Hurley and Petersen (1994) present more elaborate formulations. This line of
research has not, however, yielded practical planning models and tools yet, mainly because
the formulations become too large and complex when applied to realistic situations. For a
more detailed review of these efforts see Guélat, Florian, and Crainic (1990) and Crainic,

CIRRELT-2008-20 12
National Planning Models and Instruments

Florian, and Léal (1990).

Models based on more sophisticated representations of the supply network were intro-
duced by a group of Canadian researchers from the Centre for Research on Transportation
(CRT) in Montreal in the second half of the 1980s (Florian and Crainic 1989; Crainic,
Florian, and Leál 1990; Guélat, Florian, and Crainic, 1990). They proposed a modelling
framework of a multimodal network, made up of modes, nodes, links, and intermodal trans-
fers, on which multiple products are to be moved by specific vehicles and convoys between
given origin and destination points according to a generalized cost combining several criteria,
most notably, cost and time. Later, Jourquin and Beuthe (1996) proposed a methodology
based on the same general ideas and used it to perform several research projects in Europe
(e.g., Jourquin and Beuthe, 2003, 2006, and Geerts and Jourquin, 2001). We present this
Canadian contribution to national planning and the developments that followed in the next
sections.

4 The STAN Model and Method

Our contribution to the field takes the form of scientific results, instrumental developments,
and a worldwide commercial distribution of the resulting decision-support system called
STAN (Strategic Transportation ANalysis). We briefly review these contributions in this
section.

The primary goal when developing STAN was a methodology independent of particular
applications. The aim was and continues to be to offer a modelling framework sufficiently
general and flexible to provide the means to address a wide variety of cases in terms of
geographical dimensions and richness of details in the supply, demand, and decision criteria
representation. A second and closely related objective was to provide a method and software
robust with respect to their utilisation by planners without a deep knowledge of optimisation.
These objectives explain a number of design choices which have been a posteriori justified
by the industrial utilisation of the methodology.

4.1 The STAN methodology

The STAN modelling framework of the supply side of a national transportation system is
that of a multimodal network, made up of modes, nodes, links, and intermodal transfers.
Here, a mode is a means of transportation having its own characteristics, such as vehicle type
and capacity, as well as specific cost measures. Depending on the scope and level of detail
of the strategic study, a mode may represent a carrier or part of its network representing
a particular transportation service, an aggregation of several carrier networks, or specific
transportation infrastructures such as ports.

CIRRELT-2008-20 13
National Planning Models and Instruments

The network consists of nodes N , links A, modes M, and transfers T that represent all
possible physical movements on the available infrastructure. To capture the modal charac-
teristics of transportation, a link a ∈ A is defined as a triplet (i, m, j), where i ∈ N is the
origin node, j ∈ N is the destination node, and m ∈ M is the mode. Parallel links are
used to represent situations where more than one mode is available for transporting goods
between two adjacent nodes. This network representation enables easy identification of the
flow of goods by mode, as well as various cost functions (e.g. operating cost, time delay,
energy consumption, emissions, noise, risk, etc.) by product and mode.

To model intermodal shipments, one must allow for mode transfers at certain nodes of
the network and compute the associated costs and delays. Intermodal transfers t ∈ T at a
node of the network are modeled as link to link, hence mode to mode, allowed movements.
A path in this network then consists of a sequence of directed links of a mode, a possible
transfer to another mode, a sequence of directed links of the second mode, and so on.

On the demand side, a product is any commodity (or collection of similar commodities)
— goods or passengers — that generates a link flow. Each product p ∈ P transported over
the multimodal network is shipped from certain origins o ∈ N to certain destinations d ∈ N
within the network. The demand for each product is exogenous and is specified by a set of O-
D matrices. The mode choice for each product is also exogenous and is indicated by defining
for each O-D matrix a subset of modes allowed for transporting the corresponding demand.
Shipper behaviour is assumed to be reflected in these O-D matrices and associated mode
choice. Let g m(p) be a demand matrix associated with product p ∈ P, where m(p) ⊆ M is
the subset of modes that may be used to move this particular part of product p.

The flows of product p ∈ P on the multimodal network are the decision variables of the
model. Flows on links a ∈ A are denoted by vap and flows on transfers t ∈ T are denoted by
vtp ; v stands for the vector of all product flows. Vehicle and convoy (e.g., train) movements
are deduced from these flows. Cost functions are associated with the links and transfers of
the network. For product p, the respective average cost functions spa (v) and spt (v) depend
on the transported volume of goods. Then, the total cost of product p on arc a is spa (v)vap ,
and it is spt (v)vtp on transfer t. The total cost over the multimodal network is the function
F , which is to be minimized over the set of flow volumes that satisfy the flow conservation
and non negativity constraints:
!
X X X p
F = spa (v)vap + st (v)vtp . (1)
p∈P a∈A t∈T

m(p)
Let Lod denote the set of paths that for product p lead from origin o to destination
d using only modes in m(p). The path formulation of the flow conservation equations are
then: X m(p)
hl = god o, d ∈ N , p ∈ P, m(p) ⊆ M, (2)
m(p)
l∈Lod

CIRRELT-2008-20 14
National Planning Models and Instruments

m(p)
where hl is the flow on path l ∈ Lod . These constraints specify that the total flow moved
over all the paths that may be used to transport product p must be equal to the demand for
that product. The non-negativity constraints are:
m(p)
hl ≥ 0, l ∈ Lod , o, d ∈ N , p ∈ P, m(p) ⊆ M. (3)

The relation between arc flows and path flows is vap = l∈Lp δal hl , a ∈ A, p ∈ P, where
P
Lp is the set of all paths that may be used by product p, and δal = 1 if a ∈ l (and 0,
otherwise) is the indicator function
P which identifies the arcs of a particular path. Similarly,
p
the flows on transfers are vt = l∈Lp δtl hl , t ∈ T , p ∈ P, where δtl = 1 if t ∈ l (and 0,
otherwise). Then, the system optimal multiproduct, multimodal assignment model consists
of minimizing (1), subject to constraints (2) and (3). The optimality principle ensures that in
the final flow distribution, for each product, demand matrix, and origin-destination pair, all
paths with positive flows will have the same marginal cost (lower than on the other paths).
When required, a decision criterion equilibrating the average generalized cost on the paths
of certain products may be also used.

The algorithm developed for this problem exploits the natural decomposition by product
and results in a Gauss-Seidel-like procedure (Figure 3 gives a schematic description of the
algorithm), which allows the solution of large size problems in reasonable computational
times, even though functions and functional derivatives must be computed numerically to
account for user-defined functions (Guélat, Florian, and Crainic, 1990).

Initialization. k = 0, v 0 is the initial solution.

Major cycle. k = k + 1; For each product p ∈ P do

Descent Direction. Compute the descent direction dk = y k − v k ,

where y k is the “all-or-nothing” solution to the linearized problem

Min F (v k ) + 5F (v k )(v − v k ) subject to v ∈ Ωp , (4)

and Ωp stands for constraint set (2) - (3) for product p ; This involves solving a
shortest marginal cost path problem for each OD pair.
Line Search. Find the step length λpk solution of Min0≤λ≤1 F (v k + λdk ).
Flow Update. v k+1 = v k + λpk dk .

Stopping Criteria. The algorithm stops either when the relative gap is below a given
threshold or when a predefined number of major cycles have been performed.

Figure 3: The STAN assignment algorithm

CIRRELT-2008-20 15
National Planning Models and Instruments

4.2 The STAN software

The initial STAN development took place at the Centre for Research on Transportation
(CRT), a multi-disciplinary and multi-university institution initiated by and located on the
campus of the Université de Montréal. (The CRT has recently evolved into the Interuni-
versity Research Centre on Enterprise Networks, Logistics and Transportation - CIRRELT.)
The interest of the CRT researchers, professors and students in the area of national freight
planning was motivated by a project in Brazil that lasted from 1983 to 1989. The aim of
the project, which was funded by the International Development Research Centre (IDRC),
Canada, was to develop a strategic planning tool for the development of the main export
corridors of goods from Brazil. Other than the group of professors and graduate students
from the CRT, the team of the project consisted of personnel from GEIPOT, the transporta-
tion planning organization of the Ministry of Transportation of Brazil, and professors from
the Catholic University of Rio de Janeiro (PUC-RJ).

A number of important issues became apparent early into the project: a great variety
of national planning projects may arise, even when a single country or region is considered,
requiring significant flexibility in the representation of the transportation system; a hard-
coded model could not address this challenge; planners require many tools to handle data
and results, perform studies, and produce the associated analyzes; planners are not, and
do not need to be, operations researchers or computer science specialists. This motivated
the development of a flexible methodology to address the various issues identified in the
project. This methodology then served as the basis for the STAN software package, which
shared many of its components with the EMME/2 software dedicated to the planning of
multimodal urban passenger transportation systems (Spiess 1984; Babin et al., 1982; the
current version of the software is called Emme 3).

A set of technical reports (Florian and Crainic, 1989) documents the project and its
results, including the first applications to various Brazilian cases. Guélat, Florian, and
Crainic (1990) then present the general model and assignment methodology, Crainic, Flo-
rian, and Leál (1990) focus on the analysis of national, multimodal rail transportation
systems, while Crainic et al. (1990) describe the STAN decision-support system. INRO
(http://www.inro.ca) was assigned the commercialization rights to the STAN software and
continued its development since 1986. A number of these developments were performed in
collaboration with CRT researchers (Crainic, Florian, and Larin, 1994; Crainic et al., 1999,
2002).

STAN, the interactive-graphic software system, is made up of a large number of tools to

define and compare scenarios, input, display, analyze, modify, and output data, as well as
define the supply network (according to the indicated modelling framework), generate de-
mand, specify the assignment model parameters, and perform the corresponding simulation.
Matrix-based computing tools may be used to implement a whole gamut of mode choice
and demand generation and allocation models. Similarly, a network calculator can be used
to combine network data to implement various performance and analysis models, including

CIRRELT-2008-20 16
National Planning Models and Instruments

connectivity studies, incident impact analyzes for hazardous goods, and environment impact
studies (emissions, noise, etc.) for given network regions and vehicle fleets. Moreover, the
path-analysis capability offers the possibility to interactively select and examine paths ac-
cording to generalized, user-defined criteria, to perform select element (link or intermodal
transfer) studies, and to compute demand or performance matrices for selected products and
OD pairs (Crainic et al., 1999, 2002). A macro language can be used to program complex
operations and procedures. See Larin et al. (2000) for a detailed description of the current
version (6.2) of the STAN system, components, interfaces, and tools.

The STAN software was adopted for strategic freight planning in several countries, in-
cluding Finland, Sweden, Norway, Italy, Spain, Germany, the United Kingdom, China, and
Mexico, and has been applied successfully for scenario analysis and planning. Currently, 52
institutions in 18 countries use STAN. A few of the numerous studies undertook with the
help of STAN are mentioned in the references indicated above. Most applications, however,
have been carried out by public agencies and private organizations around the world. While
the results of these analyses are not available for public scrutiny, they witness the success of
Canadian operations research applied to transportation.

5 Perspectives

We have presented an important contribution of the Canadian operations research commu-

nity to transportation science, the STAN methodology and decision-support system for the
analysis and planning of multimode, multiproduct national transportation systems.

The challenge in the building of a national strategic planning tool is the provision of
the large amounts of data required. One needs to have information about the input-output
structure of the economy and reliable data on shipments of all products in order to build
demand models that would produce the product based origin-destination matrices that are
required for the freight network assignment model. One also needs information on the
economic profiles of shippers and carriers, as well as accurate description of the infrastructure
and services provided, or planned, for the region. The strategic level of planning described
in this paper does not take into account the logistic aspect of freight transportation at the
company level nor does it consider the multiple levels of such logistic systems. Nevertheless,
the countries that used STAN found it very useful, in particular for the planning of strategic
freight corridors as well as various policy and investment analyses at the system level. It is
worth noting that, compared to applications of urban planning models, strategic national
freight methods, decision support systems and models do not abound.

Work is still continuing in this area. A first group of projects targets new models and
procedures to enhance the performance of the system simulation (assignment) step in terms
of problem dimensions and computational efficiency (e.g., Crainic, Damay, and Gendreau,
2008), as well as to provide the means for increased accuracy and refinement in the represen-

CIRRELT-2008-20 17
National Planning Models and Instruments

tation of the behaviour of transportation systems under various system configurations and
data availability (e.g., Crainic, Damay, and Gendreau, 2008). A second group of projects
aims to provide answers to emerging needs in terms of particular studies and elements to
include in the representation of the transportation systems, such as the integration of lo-
gistics chains in the system modelling and simulation (e.g., Crainic, Florian, and Noriega,
2007) and the representation of the impact of Intelligent Transportation System technologies
on the behaviour of the transportation system. Contemplated issues of interest concern the
links between transportation policies, system performance, and the environment, as well as
the relationships between transportation policies and the performance of logistics chains.
We expect to report on some of these developments in the near future.

Acknowledgments

While writing this paper, Michael Florian was Emeritus Professor, Department of Computer
Science and Operations Research of the Université de Montréal, and Teodor Gabriel Crainic
was NSERC Industrial Research Chair on Logistics Management, École des sciences de la
gestion, UQÀM, Director of the Intelligent Transportation Systems Laboratory of CIRRELT
and Adjunct Professor, Department of Computer Science and Operations Research, Universié
de Montréal,. Financial support for the current projects mentioned in the paper is provided
by the Natural Science and Engineering Council of Canada (NSERC) through its Industrial
Research Chair and Discovery grant programs.

References
[1] Babin, A., Florian, M., James-Lefebvre, L., and Spiess, H. EMME/2: An Interactive
Graphic Method for Road and Transit Planning. Transportation Research Record, 866:1–
9, 1982.

[2] Cascetta, E. Transportation Systems Engineering: Theory and Methods. Kluwer Aca-
demic Publishers, Dordrecht, The Netherlands, 2001.

[3] Christiansen, M., Fagerholt, K., Nygreen, B., and Ronen, D. Maritime Transportation.
In Barnhart, C. and Laporte, G., editors, Transportation, volume 14 of Handbooks in
Operations Research and Management Science, pages 189–284. North-Holland, Amster-
dam, 2007.

[4] Crainic, T.G. Long-Haul Freight Transportation. In Hall, R.W., editor, Handbook
of Transportation Science, pages 451–516. Kluwer Academic Publishers, Norwell, MA,
second edition, 2003.

[5] Crainic, T.G., Damay, J., and Gendreau, M. A Comparative Study of Nonlinear
Traffic Assignment Methods for Multi-commodity Multi-modal Transportation Sys-

CIRRELT-2008-20 18
National Planning Models and Instruments

tems. Publication CIRRELT-2008, Centre interuniversitaire de recherche sur les réseaux

d’entreprise, la logistique et le transport (CIRRELT), Université de Montréal, Montreal,
2008.
[6] Crainic, T.G., Damay, J., and Gendreau, M. An Integrated Evaluation Platform for
National/Regional Planning Methods of Multi-modal Transportation Systems for Multi-
commodity Multi-mode transportation Systems. Publication CIRRELT-2008, Centre
interuniversitaire de recherche sur les réseaux d’entreprise, la logistique et le transport
(CIRRELT), Université de Montréal, Montreal, 2008.
[7] Crainic, T.G., Dufour, G., Florian, M., and Larin, D. Path Analysis in STAN. In C.
Zopounidis and D. Despotis, editors, Proceedings of the 5th International Conference
of the Decision Sciences Institute, pages 2060–2064. New Technologies Publications,
Athens, 1999.
[8] Crainic, T.G., Dufour, G., Florian, M., and Larin, D. Path Recovery/Reconstruction
and Applications in Nonlinear Multimodal Multicommodity Networks. In Gendreau, M.
and Marcotte, P., editors, Transportation and Network Analysis: Current Trends Mis-
cellanea in Honor of Michael Florian, pages 2060–2064. Kluwer Academic Publishers,
Norwell, MA, 2002.
[9] Crainic, T.G. and Florian, M. Modèles d’optimisation pour la planification des systèmes
de transport. In Paschos, V., editor, Optimisation combinatoire : applications, pages
117–152. Hermès Lavoisier, Paris, 2006.
[10] Crainic, T.G., Florian, M., Guélat, J., and Spiess, H. Strategic Planning of Freight
Transportation: STAN, An Interactive-Graphic System. Transportation Research
Record, 1283:97–124, 1990.
[11] Crainic, T.G., Florian, M., and Larin, D. STAN: New Developments. In Al S. Khade
and R. Brown, editors, Proceedings of the 23rd Annual Meeting of the Western Decision
Sciences Institute, pages 493–498. School of Business Administration, California State
University, Stanislaus, CA, 1994.
[12] Crainic, T.G., Florian, M., and Léal, J.-E. A Model for the Strategic Planning of
National Freight Transportation by Rail. Transportation Science, 24(1):1–24, 1990.
[13] Crainic, T.G., Florian, M., and Noriega, Y. Computing Shortest Paths with Logis-
tics Constraints. Publication CIRRELT-2007-06, Centre interuniversitaire de recherche
sur les réseaux d’entreprise, la logistique et le transport (CIRRELT), Université de
Montréal, Montreal, 2007.
[14] Crainic, T.G., Gendreau, M., and Kuncyté, R. Prototypage d’une enquête-chargeur
sur les flux de déplacements de marchandises aux Québec. Publication CRT-2006-05,
Centre de recherche sur les transports, Université de Montréal, Montreal, 2006.
[15] Crainic, T.G. and Kim, K.H. Intermodal Transportation. In Barnhart, C. and La-
porte, G., editors, Transportation, volume 14 of Handbooks in Operations Research and
Management Science, pages 467–537. North-Holland, Amsterdam, 2007.

CIRRELT-2008-20 19
National Planning Models and Instruments

[16] Florian, M. and Crainic, T.G. (editors). Strategic Planning of Freight Transportation in
Brazil: Methodology and Applications; Part I Project Summary; Part II Methodology:
The New Contributions; Part III The STAN System; Part IV Specifying Data for the
STAN System: Requirements, Concepts, Sources and Estimation Procedures; Part V
Applications. Publications 638 – 642, Centre de recherche sur les transports, Université
de Montréal, Montréal, QC, Canada, 1989.

[17] Florian, M. and Hearn, D. Network Equilibrium Models and Algorithms. In M. Ball,
Magnanti, T.L., Monma, C.L., and Nemhauser, G.L., editors, Network Routing, vol-
ume 8 of Handbooks in Operations Research and Management Science, pages 485–550.
North-Holland, Amsterdam, 1995.

[18] Friesz, T.L., Gottfried, J.A., and Morlok, E.K. A Sequential Shipper-Carrier Network
Model for Predicting Freight Flows. Transportation Science, 20:80–91, 1986.

[19] Friesz, T.L. and Harker, P.T. Freight Network Equilibrium: A Review of the State of the
Art. In A.F. Daughety, editor, Analytical Studies in Transport Economics, chapter 7.
Cambridge University Press, Cambridge, MA, 1985.

[20] Friesz, T.L., Tobin, R.L., and Harker, P.T. Predictive Intercity Freight Network Models.
Transportation Research Part A: Policy and Practice, 17A:409–417, 1983.

[21] Geerts, J.-F. and Jourquin, B. Freight TransportationPlanning on the European Multi-
modal Network: The Case of the Walloon Region. European Journal of Transport and
Infrastructure Research, 1(1):91–106, 2001.

[22] Guélat, J., Florian, M., and Crainic, T.G. A Multimode Multiproduct Network Assign-
ment Model for Strategic Planning of Freight Flows. Transportation Science, 24(1):25–
39, 1990.

[23] Harker, P.T. Predicting Intercity Freight Flows. VNU Science Press, Utrecht, The
Netherlands, 1987.

[24] Harker, P.T. Issues and Models for Planning and Regulating Freight Transportation
Systems. In L. Bianco and A. La Bella, editors, Freight Transport Planning and Logistics,
pages 374–408. Springer-Verlag, Berlin, 1988.

[25] Harker, P.T. and Friesz, T.L. Prediction of Intercity Freight Flows I: Theory. Trans-
portation Research Part B: Methodological, 20B(2):139–153, 1986.

[26] Harker, P.T. and Friesz, T.L. Prediction of Intercity Freight Flows II: Mathematical
Formulations. Transportation Research Part B: Methodological, 20B(2):155–174, 1986.

[27] Hurley, W.J. and Petersen, E.R. Nonlinear Tariffs and Freight Network Equilibrium.
Transportation Science, 28(3):236–245, 1994.

[28] Isard, W. Interregional and Regional Input-Output Analysis: A Model of a Space-

Economy. The Review of Economics and Statistics, 33:318–328, 1951.

CIRRELT-2008-20 20
 National Planning Models and Instruments

[29] Jones, P.S. and Sharp, G.P. Multi-Mode Intercity Freight Transportation Planning
for Underdeveloped Regions. In Proceedings of the 18th Meeting of the Transportation
Research Forum, pages 523–531, 1977.

[30] Jourquin, B. and Beuthe, M. Transportation Policy Analysis with a Geographic Infor-
mation System: The Virtual Network of Freight Transportation in Europe. Transporta-
tion Research Part C: Emerging Technologies, 4(6):359–371, 1996.

[31] Jourquin, B. and Beuthe, M. Multimodal Freight Networks, Analysis with NODUS:
Methodology and Applications. In Dullaert, W., Jourquin, B., and Polak, J., edi-
tors, Across the Border. Building upon a Quarter Century of Transport Research in the
Benelux, pages 163–184. De Boek, Antwerp, The Netherlands, 2003.

[32] Jourquin, B. and Beuthe, M. A Decade of Freight Transport Modeling with Virtual
Networks: Aquired Experiences and New Challenges. In Reggiani, A. and Nijkamp,
P., editors, Spatial Dynamics, Networks and Modelling, pages 181–200. Edward Elgar
Publishing, Cheltenham Glos, UK, 2006.

[33] Larin, D., Crainic, T.G., Simonka, G., Dufour, G., and Florian, M. STAN On-line
Documentation, Release 6.2. INRO Consultants, Inc., Montreal, 2002.

[34] Macharis, C. and Bontekoning, Y.M. Opportunities for OR in Intermodal Freight Trans-
port Research: A Review. European Journal of Operational Research, 153(2):400–416,
2004.

[35] Nagurney, A. Network Economics: A Variational Inequality Approach. Kluwer Aca-

demic Publishers, Boston, second edition, 1999.

[36] Powell, W.B., Bouzaı̈ene-Ayari, B., and Simaõ, H.P. Dynamic Models for Freight Trans-
portation. In Barnhart, C. and Laporte, G., editors, Transportation, volume 14 of Hand-
books in Operations Research and Management Science, pages 285–365. North-Holland,
Amsterdam, 2007.

[37] Sharp, G.P. A Multi-Commodity, Intermodal Transportation Model. In Proceedings of

the 20th Meeting of the Transportation Research Forum, pages 399–407, 1979.

[38] H. Spiess. Contributions à la théorie et aux outils de planification des réseaux de trans-
port urbains. PhD thesis, Université de Montréal, Montreal, 1984.

[39] Winston, C. The Demand for Freight Transportation: Models and Applications. Trans-
portation Research Part A: Policy and Practice, 17:419–427, 1983.

CIRRELT-2008-20 21

View publication stats