CONCEPT of An Open-Access Travel demand Modeling platform

Abstract

Travel demand models have an increasing role in decision- and policy-making. The prevailing practice of modelling raises some concerns, main limitations being of human and technical nature– adopting advanced models requires skilled professionals and resources. This paper introduces an open-access modelling concept, which is designed to overcome these barriers. The concept centralizes the bulk of model-development and application tasks, leaving only the coding of proposed project scenarios to the user. This way skilled professionals and the resources necessary for data-intensive models can be concentrated on regional model-building. Planners and policy-makers can access a valid model through open interfaces, where they can define scenarios and evaluate them without intensive model-building tasks and large-scale data collection.
This open-access concept − developed by a single organization − provides an access to a theoretically sound model to a wide array of users, where advanced techniques are implemented. Besides overcoming concerns of resources, expertise or data, and eliminating the reluctance to adopt advanced models, another great advantage is that the same model can be used through a controlled user interface by different organizations making results comparable.
Project ETRAFFIC is a conceptual framework for implementing an open-access modelling platform − funded by the Hungarian government. The prototype of the implementation is under development.

Full text

Contents lists available at ScienceDirect
Case Studies on Transport Policy
journal homepage: www.elsevier.com/locate/cstp
Research Paper
Concept of an open-access travel demand modeling platform
Lajos Kisgyörgy
⁎
, Gergely Vasvári
Budapest University of Technology and Economics, Department of Highway and Railway Engineering, Muegyetem rkp. 3., Budapest, 1111, Hungary
ARTICLE INFO
Keywords:
Open-access model
Modelling concept
Modelling framework
Travel demand forecasting
ABSTRACT
Travel demand models have an increasing role in decision- and policy-making. The prevailing practice of
modelling raises some concerns, main limitations being of human and technical nature–adopting advanced
models requires skilled professionals and resources. This paper introduces an open-access modelling concept,
which is designed to overcome these barriers. The concept centralizes the bulk of model-development and ap-
plication tasks, leaving only the coding of proposed project scenarios to the user. This way skilled professionals
and the resources necessary for data-intensive models can be concentrated on regional model-building. Planners
and policy-makers can access a valid model through open interfaces, where they can define scenarios and
evaluate them without intensive model-building tasks and large-scale data collection.
This open-access concept −developed by a single organization −provides an access to a theoretically sound
model to a wide array of users, where advanced techniques are implemented. Besides overcoming concerns of
resources, expertise or data, and eliminating the reluctance to adopt advanced models, another great advantage
is that the same model can be used through a controlled user interface by different organizations making results
comparable.
Project ETRAFFIC is a conceptual framework for implementing an open-access modelling platform −funded
by the Hungarian government. The prototype of the implementation is under development.
1. Introduction
Travel demand forecasting enables formal testing of policy scenarios
and provides decision-makers a basis for comparing costs and benefits
of tested alternatives (Meyer and Miller, 2001). Most decision-makers
and policy-makers recognize travel demand modelling as a crucial
element of the overall transport planning process. Despite the in-
creasing role of travel demand models, outdated models are still in use
or modelling is completely omitted in decision support, usually due to
the lack of resources, expertise or data (Hatzopoulou and Miller, 2009).
The main limitations of a good modelling practice are of human and
technical nature: contemporary transport planning requires skilled
professionals and theoretically sound modelling techniques with com-
petent software implementations (Ortúzar and Willumsen, 2011).
Travel demand modelling practice in some countries raises some
concerns and questions. According to a web-based survey of the
Transportation Research Board the most cited barriers of model im-
provement were staff, budget and reluctance to switch to more com-
plicated and data-intensive tools (TRB, 2007; Hatzopoulou and Miller,
2009). Although policy-making and planning organizations are aware
of the need for using advanced models −as integrated land-use and
transportation models or activity-based models −they have limited
capabilities to apply them on a large scale. They are reluctant to adopt
advanced models because they perceive them as yet unproven
(Davidson et al., 2007). The study of Shepherd et al. (2006) also
highlighted that resources and skills are major concerns for adopting
advanced models.
Overcoming the major barriers −staffexpertise and data resources
−related to adoption of advanced models would improve project
evaluation. Policy-making and planning organizations build their own
model separately, often duplicating each other’s work as different
projects might require similar spatial and temporal scope. Theoretical
consistency of some of the models could also be an issue. Furthermore
forecasts based upon models built on different scientific or technical
assumptions, designed for different geographical areas and time per-
iods, using different resolution and data, are non-comparable.
Financial estimations of a project usually depend on the accuracy of
travel demand forecasts. Thus accuracy of a forecast has considerable
importance in effective allocation of governmental funds. Different
organizations have different models with different spatial coverage,
time span, economic scenario, etc., which makes objective comparison
−consequently fund allocation, project ranking and scheduling −
difficult.
The aim of this paper is to introduce an open-access modelling
http://dx.doi.org/10.1016/j.cstp.2017.06.004
Received 11 September 2015; Received in revised form 27 April 2017; Accepted 15 June 2017
⁎
Corresponding author.
E-mail address: kisgyorgy.lajos@epito.bme.hu (L. Kisgyörgy).
Case Studies on Transport Policy 5 (2017) 453–459
Available online 17 June 2017
2213-624X/ © 2017 World Conference on Transport Research Society. Published by Elsevier Ltd. All rights reserved.
MARK

concept, which provides a framework for the modelling practice in
order to handle the difficulties mentioned above. This open-access
concept −developed by a single organization −provides an access to a
theoretically sound model to a wide array of users, where advanced
techniques are implemented. Besides overcoming concerns of resources,
expertise or data, and eliminating the reluctance to adopt advanced
models, another great advantage is that the same model can be used
through a controlled user interface by different organizations making
results comparable.
The paper is organized as follows. After this introduction, principles
of the open-access modelling concept are presented. Following, com-
ponents of the framework are described in detail and a possible im-
plementation is described. Later, effects of the concept on the modelling
practice are detailed through case studies. Finally conclusions are
drawn.
2. The open-access travel demand modelling concept
2.1. The general concept
The open-access modelling concept presented in this text handles
the modelling process by two modules: model development and model
application (Fig. 1). The idea is that the most demanding tasks of de-
velopment, implementation, calibration and validation, data collection
and model update −which create the strongest barriers in practice −
should be removed from the daily tasks of project evaluation. Model
application is integrated with the model and the synthesis of the two
modules is provided as an open-access service. Users of the service need
not be concerned about these tasks, as they only have to revise the
economic environment of the forecast and define scenarios −mod-
ifications of the base network.
Model development is accomplished by a design team −consisting
of modelling experts. Tasked with the competent implementation of
advanced, theoretically sound modelling techniques, they also update
and maintain the databases, and are responsible for repeatedly ob-
taining all necessary data for model inputs and forecasts. The team is
also responsible for the validity of the model, performing regular va-
lidity checks and evaluations.
This concept gives more value to the users than usual modelling
tools and concepts, because with the modelling implementation it al-
ready contains the majority of input data as well (e. g. land use, po-
pulation, economic variables, road network etc.), thus users do not have
to collect, analyze and code them. Beside the built-in data the concept
gives significant flexibility to the users, because most parameters can be
changed. However, users do not need to parameterize the whole model,
they can focus on the set of variables which are important for the
analysis of the given task.
2.2. Model development
The core of the model development module is a transportation
forecasting model. Depending on the intended use, any types of models
can be implemented herein, e.g. the four-step or activity based travel
demand model, a modal split model etc. however an integrated land-
use and activity based transportation model is suggested.
Different planning or forecasting tasks require different resolution.
Consequently the model should consist of different layers; each layer
having different spatial resolution, a different depth of detail and is
customized to perform different tasks, e.g. local or regional planning.
To provide consistency between these layers, upper levels (with less
detail) are aggregates of the bottom layer (highest resolution) thus the
model would require the coding of a single highly detailed layer. All the
other, less detailed layers are produced by the aggregation of the data of
the highly detailed layer, where the methods of aggregation and the
necessary spatial correlations between layers of different resolution are
part of the model. Users will work with the layer suitable to their actual
modelling task. Although network coding is through arbitrary model-
ling tools, it is suggested to implement it with a well-known software as
VISUM or EMME, but any modelling or GIS software can be used as
well.
Since the proposed model of the framework should be capable to
handle transportation planning tasks on national and local levels, it
must incorporate all possible modes of transport available. To provide
adequate basis for a local study temporal resolution also have to be
enhanced to model not just average daily volumes but peak and off-
peak periods or even traffic demand defined by the hour.
The model development module can be continuously upgraded if
necessary, in order to implement the new advances in transportation
modelling theories. This way the model always contains the most ad-
vanced and theoretically sound modelling techniques with competent
implementation.
Calibration and validation of the model is done in this module as
well. Built-in evaluations and checks for every modelling step and for
the overall model assure the forecasting ability of the model.
Continuous validation covers evaluations and checks that are repeated
on an annual basis. Results generated by users are also checked against
TRAVEL
DEMAND
MODEL
RESOURCES, SKILLS AND EXPERTISE
FORECASTING
INTERFACE
Data
Development
ImplementaƟon
CalibraƟon and
validaƟon
Model support (validaƟon, maintenance)
MODEL DEVELOPMENT
MODEL APPLICATION
Scenarios
Maintenance
and u
p
date
Model use
Model details
User
Scenarios
Results
User
Scenarios
Results
Results
Fig. 1. The open-access travel demand modelling concept.
L. Kisgyörgy, G. Vasvári Case Studies on Transport Policy 5 (2017) 453–459
454

the revalidated model. In case problems are revealed with the revised
data, the model is re-calibrated (Fig. 2). This approach ensures that the
model remains valid, overcoming a hindrance of static travel demand
forecasting models, which is the lack of the re-evaluation of the re-
estimation.
The second component of the modelling module is data collection.
Advanced models are very data-intensive, data collection requiring
significant resources. By centralizing the model and the collection of
data, resources can be focused and used more efficiently, compared to
the case when each organization builds its own model. Methodological
issues of data collection are handled the design team of skilled pro-
fessionals. This minimalizes the uncertainties or errors resulting from
sampling, measurement and data interpretation. The data collection
program requires careful planning, thoroughness and accuracy, as the
resulting models and forecasts will be no better than the data used for
the model estimation and validation. Data are actualized on a regular
basis, including the updating of all the changes and expected devel-
opments of the transportation infrastructure and services and of other
data as well.
The third component of the module is forecasting of input variables.
It is a delicate procedure, because a too optimistic or a too pessimistic
approach can yield very different results. This module consists of sev-
eral economic forecasting models. Certain models serve the prediction
of the future economic environment through the prediction of macro-
economic variables, while other models are designed to predict future
values of the input variables based on that future economic environ-
ment. Also this module estimates the future state of the transportation
infrastructure and services, based on development plans.
Creating long-range forecasts for all the input data of a data-in-
tensive advanced model is overwhelming, while being accurate in these
predictions is nearly impossible. The forecasting component handles
this problem on different levels. First the overall economic environment
is modelled through macro-economic variables (e.g. GDP). Then each
input variable is projected by its forecasting model (Fig. 3). These
models take the forecasted macro-economic environment and logical
relations among input variables into consideration. This process pro-
vides a reliable and robust estimation for the future values of input
variables.
Considering the difficulty of a long-range forecast, the forecasting
component provides estimation scenarios instead of providing a single
estimation. One of the scenarios is the most probable estimation of the
future based on our best knowledge of today. Other scenarios are op-
timistic and pessimistic estimations of the future macro-economic en-
vironment. This way the forecasting model takes uncertainties of
forecasting into consideration as well.
The forecasting component also provides predictions for the trans-
portation system. It automatically includes prominent developments
and changes in the network along with relevant transportation policies
into the model. The forecasting models are updated continuously ac-
cording to the changes in trends, policies and development programs.
2.3. Model application
Model application is the module practical users, planners and
policy-makers can use to make their estimations through an interface.
The model provides the frame for a baseline estimation of the future;
through the interface users can modify this baseline estimate according
to their needs. The interface supports policy-scenario definition, net-
work modifications and adjustment of the economic environment.
However modelling algorithms cannot be changed by users, only data
and assumptions.
Model operation is similar to a black box towards users. They have
no influence on modelling assumptions and algorithms, they only
provide the scenario to be modelled, then receive model results. Good
modelling practice, however, requires the understanding of the prin-
ciples and considerations of the model, because these define the scope
and extents of the model results. Users must be familiar with the un-
derlying methodology of the applied model. The core of this module is
the application interface, which provides access to the model. There are
five major elements of the interface.
The first element contains model information. This element in-
troduces the basics of the model, its available layers and details. Based
on this information, users will be able to decide if any of the available
types and levels of the model are suitable for their task, and further-
more to choose the most appropriate one.
The second element informs users about modelling assumptions −
effects that are taken into account, developments or changes of the
transportation system that are taken into consideration, policies that
are applied. This provides users the necessary insight into modelling
details, enabling them to build accurate prediction scenarios.
The third element is policy definition. Here the future value of the
policy variables (e.g. tolls and fees) can be defined directly. The in-
terface provides policy assumptions of the model, furthermore the ac-
tual and predicted values in the model for all the variables. These as-
sumptions and data can be modified if needed or if their impact needs
to be evaluated.
The fourth element, the transport system scenario includes all the
changes in the transport services or infrastructure. Here users can
valid
invalid
Calibrated and
valid model Validity checks Re-calibraƟon of
the model
New data
Fig. 2. The process of continual validation based on regular validity
checks.
User-deĮned
scenario
EsƟmated economic
environment scenario
Database of model inputs
(current values, trends)
Modeling scenario
(predicted values of inputs)
Built-in forecast of
macroeconomic variables
Forecast models
of input variables
Model esƟmaƟon
Fig. 3. Prediction of the future values of input variables. (White boxes
represent data, scenarios and estimation, blue boxes indicate models. The
dark blue box portrays the model output.).
L. Kisgyörgy, G. Vasvári Case Studies on Transport Policy 5 (2017) 453–459
455

define proposed or expected changes in the transportation system. The
interface provides actual infrastructure and services, and also shows
expected developments and changes for the timeframe of the forecast.
Users can add and delete elements from this list.
The fifth element is the economic modelling interface. A central
problem of both short and long term forecasting is how to handle future
trends, because forecasts are only as accurate as the underlying as-
sumptions. It is more practical to take several plausible assumptions
into consideration rather than a single one, which may later turn out to
be incorrect. (Schnaars, 1987) Thus, instead of a point estimation for
the future trends, an aggregate of estimations is created, based on the
multiple theoretical scenarios.
Modifying the model data in order to create scenarios often requires
deep knowledge of the model architecture. Different users require dif-
ferent levels of control over future values of the modelling environ-
ment. Those without expertise seek a reliable prediction without the
need to reach into the forecast. Those with expertise might want more
control over the variables to analyze certain specific cases. To ensure
good modelling practice data modification should be controlled. For
this purpose the user levels are defined: basic, advanced and expert.
Access of different users of the open-access model would be divided
into three levels:
Basic level users typically have lower expertise in the field of
transportation modelling. This level mostly consists of engineers from
infrastructure-design companies, where cost-benefit analysis of their
plans (e.g. the construction plan of a bypass road) requires input from a
network model. These users are offered with built-in prediction of the
modelling environment. Only basic parameters can be modified here,
e.g. creating pessimistic and optimistic variations of predictions is
possible. Deep modelling experience is not required, but users can
change basic predictions to a small extent, i.e. they can adjust the
transportation system elements and the basic assumptions in their
project.
Advanced level users are those who have to analyse more variations
(i.e. multiple scenarios) than bacis level users. Typical applications are
transportation studies. They have more control, but they cannot access
the model directly. Through the interface they can define modifications
and changes they want in the model, but only on a meta-level. A
template provides them the ability to define intended changes without
accessing the model details: e.g. users can insert the intended land use
developments without having to know exact zone boundaries.
Expert level users would be those who require direct access to all
model variables. This level ensures an option when the predefined
toolsets of basic and advanced levels do not fit the modelling task at
hand. Users with expert-level access can set the future value of every
variable, which enables them to make any type of evaluation. It is as-
sumed that expert level users are able to set the necessary scenarios for
their project without introducing theoretical inconsistency or making
modelling mistakes. This option gives great flexibility, but also great
responsibility. Users need to be familiar with all modelling details, and
must be able to track down the effects of changes in the model. Working
with the expert level requires a lot of expertise and precaution as it can
be very complicated. Without a full understanding of the model, the
environment and possible future trends it is easy to create configura-
tions where the model loses validity, and the sheer amount of the input
variables can be difficult to review. The expert level enables highly
detailed fine-tuning, however, at a high level of risk as well. Only the
most experienced and skilled professionals with perfect understanding
of the model should use this level.
3. Implementation of the concept
Aprototype of the open-access modelling concept was implemented
as part of the ETRAFFIC project, funded by the Hungarian government.
The implementation consists of the basic functions regarding both
modelling algorithms and user privileges, and is currently under eva-
luation. In this project a demand model was used, which integrated
land use, activities and travel behaviour as part of a classical four-step
modelling approach. A modular framework was developed for the
model implementation (Vasvári et al., 2017).
The basic concept of the implementation is a client-server relation.
The modelling core and the database run on a server connected to the
internet, which can be accessed by users through browser clients. A
modular architecture was used where the central database provided the
link between modules. Fig. 4 shows the modelling framework.
Input and parameter values are stored in a database, including all
data required for modelling: economic data, socio-demographical data,
transportation infrastructure and relevant features of the network,
model parameters etc. At the early stages of the project the database
contained time series of available input data of the period 2008–2012.
Then as they became available, new data was appended to the database
for the year 2013.
During calibration, validation and forecasting the model used its
parameter values from this database, and inserted calibrated parameter
values and the outputs into it. The model was calibrated and validated
based on the data of an overall household survey in Hungary and of
available socio-economic statistics. The base year of performing the
household survey was 2012. The model was then validated again for
each year in the period of 2008–2013. These checks proved that the
model was valid for the entire period. The implementation contains
automatic checks on validity which are run annually, and the model can
be recalibrated if necessary.
Administrators of the system access both the database and the
modelling environment through a desktop client. They can upload new
data, modify existing data and calibrate, validate the model. With a
higher level of access the administrators can make modifications in the
model itself.
For everyday users the implementation offers so-called dashboards
where basic model forecasts and parameters can be modified, e.g. users
can complement the road network, change parameters of network
elements; or they can change future values of policy variables, etc.
Definition of modelling scenarios is also accomplished here. These
SERVER CLIENT
Database module Scenario management
module
Travel demand
forecasƟng module
Economic
forecasƟng module Dashboard modules
Result visualinjaƟon
module
Fig. 4. Modules of the ETRAFFIC modelling framework.
L. Kisgyörgy, G. Vasvári Case Studies on Transport Policy 5 (2017) 453–459
456

dashboards send the modifications to the server, where the scenarios
are merged into the database, creating a scenario-database (Fig. 5)asan
input for the modelling algorithms. Predictions for the scenario and
model results are sent back to the user interface.
The dashboards provide unrestricted access to policy variables and
variables of the transportation system. Every dashboard has three dif-
ferent levels, according to the expertise or the intentions of the users.
The more detailed dashboards include the functions of the simple
dashboards, providing all functionality for high level users as well.
Levels of the dashboards can be chosen independently, e.g. users can
select advanced options in policy making and basic level option in
economic forecasting at the same time.
The policy dashboard provides information on the policies included
in the model and their considered values. Basic level users can include
or exclude policies, or can change them generally, e.g. they can increase
estimated parking fees with 10%. For advanced level users the dash-
board offers more customized policy defining, e.g. redefining the
parking zones and their fees. High level users can directly access the
policy variables, but they can also make changes through the basic or
advanced level dashboards.
The transportation network dashboard supports definition of the
evaluated transportation system. It displays the actual transportation
network and services in the model, along with considered developments
and modifications planned in the future. Basic level users can add new
elements to the network, exclude elements from the existing network
and omit developments. Users can also modify the characteristics of
transport services, e.g. public transport service frequency. Advanced
level users can influence the characteristics of the transport components
and services, e.g. cost functions. Expert level users, as usual, have direct
access to all variables.
The economic dashboard is designed for the forecasting of the
economic environment. It informs the users about the basic assump-
tions applied by the model. The first level is for basic level users, who
do not require wide customization or who do not have the economical
skills to directly predict future values of input variables. For them the
dashboard offers a choice among the most realistic forecast and its very
optimistic,optimistic,pessimistic and very pessimistic versions. With these
scenarios users can model different economic environments of the fu-
ture. For advanced level users the second level offers more possibilities:
besides the basic scenarios they can manipulate major macro-economic
and demographic variables in the forecast, such as the GDP. This pro-
vides more flexibility but means more liability as well, as the possibility
of making bad forecasts increases. The third level is for the expert level
users, who can directly define the future value of any input variable.
4. Future impact
The open access travel demand concept is capable of significantly
changing the modelling practice. Not only does it overcome the major
problems and challenges, and effectively focus on resources and ex-
pertise, but it also offers access to the implementation of the most ad-
vanced and validated theoretical models for everyone. This way it
promotes the overall usage of travel demand modelling.
The implementation of the concept requires an organization whose
main task is to build, validate and maintain models. This organization
employs experienced and skilled modelling professionals, and is re-
sponsible for the collection, update and maintenance of data required
by the models. Its task is the constant monitoring of the modelling
practice and the continuous improvement and update of the models.
The expected effects of the open-access concept on modelling
practice and the subsequent changes are shown through three case
studies presenting three typical organizations and their practices.
4.1. Leading organizations
4.1.1. Background
Leading organizations play a significant role in the policy-making of
a state or a region. They are professionally and financially determining
players, and they have a large engineering staff. They have their own
modelling experts and travel demand models for different tasks. Such
organizations usually build their own models within the framework of
government-financed projects. These models are later used for other
policy-making or planning activities for third parties as well.
Generally there are two or three leading organizations in a region, a
state or country; these usually compete for the policy and strategy-
making projects.
4.1.2. Problems
Model development and related data collection and management
are financed by public money. When considering multiple projects −
assigned to different organizations −for the same area these tasks are
funded redundantly. Later on the underlying data and intermediary
results of these projects might be used by the assigned companies for
other projects having separate budgets, which makes the policy ques-
tionable. Also the updating of these data-intensive models requires
significant financial resources, which often leads to the use of out-of-
date models.
Besides the public funding of model development and data collec-
tion tasks being repetitive, the fact that all leading organizations use
different models is also a problem, as their results are non-comparable.
For example, the travel demand forecasts of different leading
Economic forecast
dashboard
Policy dashboard
TransportaƟon
network
TransportaƟon
system scenario
Economic environ-
ment scenario
Modeling scenario
Forecast of macro-
economic variables
IntegraƟon of
scenarios
Forecast of input
variables
Model informaƟon and
assumpƟons
Fig. 5. Process of scenario definition.
L. Kisgyörgy, G. Vasvári Case Studies on Transport Policy 5 (2017) 453–459
457

organizations for a ring motorway around the capital of Hungary,
Budapest showed a difference of approx. 300% (partly caused by hectic
changes in the long-time development plans, heavily influenced by
current lobbies, which significantly influenced the results.)
The third problem is that these model results might be manipulated,
which is morally unacceptable, but still there is a chance for such in-
stances, as the modelling professional directly serves the client and
might bend the results towards the interest of the client. When the
viability of a politically promoted project is in question, such biased
results may arise.
4.1.3. Effect
The open-access concept is unfavourable for leading organizations,
as they face the chance of losing their leading position on the travel
demand forecasting market, and they will not be able to carry out po-
litical orders brought in order to gain the desirable results. Furthermore
such organizations would lose the necessity to employ their modelling
staffas well. On the other hand, with the open-access model they can
create reliable forecasts for their core policy-making and planning ac-
tivities. In a certain sense this open-access concept can be perceived as
the outsourcing of the modelling activity from a policy-making and
planning organization.
For the government the concept has multiple advantages. The
multiple financing of model building and data collection becomes un-
necessary. By creating a modelling organization, where modelling ex-
perts build theoretically secure and advanced models with competent
implementation, the resources for modelling are used in a more effi-
cient way. Furthermore the access fees from third parties provides a
sufficient source for the modelling organization to fund model main-
tenance and development activities, ensuring that the most advanced,
valid model will be accessible at all times. Another advantage might be
that the results will be much more difficult to manipulate.
4.2. Medium size organizations
4.2.1. Background
Medium size organizations have some kind of modelling software
and a few professionals who can use this software. They do not have
large-scale models due to the lack of data and resources. The general
activities of such organizations are planning, and the modelling results
are typically used for feasibility studies.
4.2.2. Problem
A new model is built for every single project, and the extent and
resolution is defined and financed by the specific project. The funding
cannot cover extensive data collection and model development, so a
minimal solution will be used. This results simple and less valid models
with low a reliability of the forecast.
4.2.3. Effect
The open-access concept enables these organizations to eliminate
the modelling activity and to focus on their core activities. Travel de-
mand forecasting will be cheaper and more reliable for them, and they
will be able to communicate their results more effectively, make them
clear and achieve acceptance. This improves their competitiveness in
their major fields of profession, even against leading organizations.
4.3. Small size organizations
4.3.1. Background
Small size organizations have no models and no stafffor modelling.
Their general activities consist of smaller scale planning or designing.
4.3.2. Problem
Small size organizations have absolutely no resources for modelling.
Their own forecasts are usually very rough with a very low level of
reliability considering the future, so alternatively they might assign
their modelling tasks to other organizations. However, as the latter is
too expensive for their usual projects, often the rough and unreliable
predictions are used. Involving potential competitors in the project also
increases the risk that later the client might decide to contract the
modelling organization instead.
4.3.3. Effect
The open access concept is beneficial for small organizations. They
can achieve reliable predictions for their projects for a far more rea-
sonable price than before. Competitiveness is significantly improved;
they can offer better prices for a higher quality of work. The risk of
losing future contracts is eliminated as well.
5. Conclusions
The concept provides a theoretically sound and valid model which is
up-to-date, and in which the most advanced techniques are im-
plemented. Skilled professionals and available resources are focused on
expensive, data- and expertise-intensive, methodologically challenging
solutions of model development and model-implementation. However
providing centralized models for a region is not a novelty. The main
benefit is in providing secure modelling techniques with competent
implementations for everyone, creating the capabilities to run advanced
models on a large scale.
Using the concept policy-making and planning organizations can
overcome the lack of resources, expertise or data, as all these are pro-
vided to them by the open-access model. There is no need for a high
number of modelling professionals, as there is no need to perform in-
tensive modelling tasks −a single person with enough knowledge to
interpret the results is enough. Similarly there is no need to collect a
large amount of data, as the model is continuously maintained and
validated, thus always up-to-date. The concept can eliminate the gen-
eral reluctance to adopt advanced models as well. The only task left
with such organizations is the creating of scenarios. This makes the
whole modelling process simple, easy, efficient, effective and theore-
tically sound.
The concept improves the overall quality of modelling. For infra-
structure-planning organizations where modelling is not their high-
profile, instead of using their own −more or less valid −models or no
models at all, they can make estimations with a valid and reliable tool.
Estimates of project and policy impact are built on the same theoretical
basis, thus decision-makers can rely on the results during project se-
lection, even if the estimations were made by different organizations.
The open-access modelling concept improves the communication
between professionals and decision-makers. Modelling is too complex a
task for decision-makers, consequently modellers are advisors of trust to
them. The open-access modelling concept provides the possibility and
the resources to develop this trust, and help decision-makers truly un-
derstand and utilize the results.
The open-access modelling concept has its drawbacks and potential
pitfalls as well. To avoid them some protection is built into the concept.
The non-expert user can only use the resolutions set by the expert staff
and can only make small changes in the model, which can not destroy
the validity of the model. Also there are some check functions in the
concept −the validity of user-defined scenarios before and after the
run of the model.
Limitations of the current implementation are its unimodality (i.e.
having only one mode of transport). The user interfaces are still in early
stages. It also must be admitted that companies with an expert mod-
elling team might find the proposed framework somewhat rigid, lacking
the multitude of customizations they need to their tasks but these
companies are small in number compared to the whole of the possible
user-base in a national level. We recognize this inflexibility as the cost
of providing maintained common basis for all transportation modelling
tasks for a country.
L. Kisgyörgy, G. Vasvári Case Studies on Transport Policy 5 (2017) 453–459
458

Acknowledgements
This research was made in the frame of “Etraffic: implementation of
an open-access travel demand model”project. The project identification
number is KTIA_AIK_12-1-2013-0062, and the project was funded by
the Innovation Fund of Research and Technology, operated by the
Hungarian Government.
Vincent Váczi gave us very valuable help with the English language.
Many thanks for him.
János Tóth and István Fi contributed to the paper by proof reading
and giving useful advice.
References
Davidson, W., Donnelly, R., Vovsha, P., Freedman, J., Ruegg, S., Hicks, J., Castiglione, J.,
Picado, R., 2007. Synthesis of first practices and operational research approaches in
activity-based travel demand modelling. Transp. Res. Part A 41, 464–488. http://dx.
doi.org/10.1016/j.tra.2006.09.003.http://www.sciencedirect.com/science/article/
pii/S0965856406001121.
Hatzopoulou, M., Miller, J., 2009. Transport policy evaluation in metropolitan areas: the
role of modelling in decision-making. Transp. Res. Part A 43 (2009), 323–338. http://
dx.doi.org/10.1016/j.tra.2008.11.001.http://www.sciencedirect.com/science/
article/pii/S0965856408002048.
Meyer, M.D., Miller, E.J., 2001. Urban Transportation Planning: A Decision-Oriented
Approach, second ed. McGraw-Hill, New York.
Ortúzar, J.D., Willumsen, L.G., 2011. Modelling Transport, 4th edition. John
Wiley & Sonspp. 1–3.
Schnaars, S.P., 1987. How to develop and use scenarios. Long Range Plann. 20 (1),
105–114. http://dx.doi.org/10.1016/0024-6301(87)90038-0.http://www.
sciencedirect.com/science/article/pii/0024630187900380.
Shepherd, S.P., Timms, P.M., May, A.D., 2006. Modelling requirements for local transport
plans: an assessment of English experience. Transp. Policy 13, 307–317. http://dx.
doi.org/10.1016/j.tranpol.2005.12.002.http://www.sciencedirect.com/science/
article/pii/S0967070X05001617.
TRB, 2007. Metropolitan travel forecasting: current practice and future direction. com-
mittee for determination of the state of the practice in metropolitan area travel
forecasting. Transportation Research Board Special Report 288. TRB, National
Research Council Washington DC, pp, pp. 7–8 (pp.63).
Vasvári, G., Bocz, P., Kisgyörgy, L., feb. 2017. Etraffic–an open access transportation
model. Period. Polytech. Civil Eng. [S.l.] 61 (3), 564–568 ISSN 1587-3773. Available
at: < https://pp.bme.hu/ci/article/view/9355 > .
L. Kisgyörgy, G. Vasvári Case Studies on Transport Policy 5 (2017) 453–459
459