Talk titles and abstracts



Wednesday 4th


Welcome and Registration


Introduction. Massimo Marrelli, UNINA, Rector, t.b.c.; Bruno Montella, UNINA, Head of DICEA.


Welcome speech. Vincenzo Torrieri, UNINA, High Technology District on Transport and Logistics, DATTILO.


Introduction to COST. Thierry Goger, EU COST Office.


Introduction to the COST Action MULTITUDE. Vincenzo Punzo, UNINA.


The management of Uncertainty in the Transportation Engineering Practice (Mark Brackstone, IOMI)


Handling uncertainty in microsimulation in practice – more than just statistics. Tom Van Vuren, Divisional Director at Mott MacDonald.

I will discuss the experience from my own international application of microsimulation and other transport models, leading to the argument that managing uncertainty should be driven by the decisions to be made based on the model’s outputs, the time-scales involved in the delivery of the decision, and the impacts of getting it wrong.  Too often, uncertainty analyses are determined by software capabilities, which may be interesting theoretically, but adds little to releasing value from the modelling exercise. I will argue for a reduced role of modelling, and a more realistic expectation of models and modellers capabilities.


The management of uncertainty whilst maintaining successful operation of London’s road network. Vladimir Vorotovic, Traffic Modelling Specialist at Transport for London.

Transport for London (TfL) is responsible for the management and operation of London’s 6,000 traffic signals and their accompanying systems, technologies and equipment. We are in charge of continually improving traffic operations whilst supporting Mayoral objectives, using best practice value-for-money solutions at all times. I offer to discuss the approach taken by TfL, in particular in the operational arm of the organisation, when taking projects from the planning/modelling stage to on-street operations. One of the projects which will be discussed is the successful delivery of transport operations during the London 2012 Games, the lessons learnt and the way forward for our organisation. TfL now faces new challenges, including large increases in population and employment, and it is essential that we utilise everything at our disposal to continue to keep London moving no matter what challenges we face.


U.S. Perspectives on Managing Uncertainty in Transportation. Robert Bertini, Portland State University, Past Deputy Administrator at the US DOT RITA. 

Different levels of government play different roles in the financing, planning, design, operations and maintenance of the surface transportation system in the U.S.  How do the federal, state and local transportation agencies deal with uncertainty in forecasting transportation demand and supply and modeling future impacts of transportation developments? What is the role for policy makers, practitioners and practitioners in the public and private sectors? What are the trends in this area and what are the data and performance measurement needs moving forward?


Cognitive Rationality, Public Engagement and Quantitative Analyses in Transportation Decision Making. Ennio Cascetta, UNINA, Past Ministry of Transport of the Campania Region, Italy. 

Failures in transport-related decision making abound in real life in areas ranging from infrastructure design and building to traffic and demand management schemes . Problems in effective transport-related decision making are due to a number of reasons including the gap still existing between the traditional technical approach to transportation planning and real-life processes. The talk discusses how to attain more transparent and participated transportation planning processes integrating cognitive decision-making, Public Engagement and quantitative analyses, and focuses on the new requirements for quantitative methods deriving from the proposed framework.


Coffee break


Round table: The management of Uncertainty in the Transport Decision Making (Mark Brackstone, IOMI) 

Participants: Vassili Alexiadis (Cambridge Systematics), Robert Bertini (Portland State University), Ennio Cascetta (UNINA), Craig Drury (UK Highway Agency), Tom Van Vuren (Mott MacDonald), Vladimir Vorotovic (Transport for London).

Thursday 5th


MULTITUDE Topics and Results (Vincenzo Punzo, UNINA) 


Calibration and Validation Basic Principles. Jaume Barcelo, UPC.

Simulation is a useful technique to provide an experimental test bed to compare alternate system designs, replacing the experiments on the physical system by experiments on its formal representation in a computer in terms of a simulation model. Simulation may thus be seen as a sampling experiment on the real system through its model. The reliability of this experiment depends on the ability to produce a simulation model representing the system behaviour closely enough or, in other words, “how close the model is to reality”.

The process of determining whether the simulation model is close enough to the actual system is usually achieved through the validation of the model, an iterative process involving the calibration of the model parameters and comparing the model to the actual system behaviour and using the discrepancies between the two, and the insight gained, to improve the model until the accuracy is judged to be acceptable.

This lecture overviews of some of the most common definitions, proposes a formal definition providing technical grounds to quantify the concept of closeness between model an reality in terms of distances and statistical methods to measure it, as describes some methodological approaches to implement them.


Sensitivity Analysis of Model Outputs. Biagio Ciuffo, EC JRC.

Traffic simulations have become indispensable tools for academicians and practitioners worldwide. A significant amount of research has gone into improving the quality of the simulators. Nonetheless, model calibration continues to be a key factor in ensuring the accuracy of their outputs. One area of model calibration in which meaningful contributions are still needed is the Sensitivity Analysis (SA) of the input parameters.

The SA explores the relationship between the simulation output and the input parameters. Due to the limitation of time and other resources, most calibration procedures cannot afford to calibrate all parameters in the model. Thus, calibration is carried out for only a limited number of input parameters. As one could imagine, the selection of an incomplete set of parameters for calibration may lead to multiple issues, including but not limited to, model imprecisions, and unrealistic values for the calibrated parameters. Hence, focusing on the incorrect set might have a cascading effect. Therefore, a proper SA, including the initial screening of the parameters, can be very valuable for the subsequent calibration process. Moreover, it may actually reduce the total efforts needed during the actual model calibration. Unfortunately a proper model SA is hardly performed in the common practice. For this reason, the COST Action TU0903 focused on the analysis, application, and development of efficient SA techniques.


Time dependent OD Demand Estimation. Constantinos Antoniou, NTUA.

Demand estimation is a key problem in transportation planning and engineering. One of the activities undertaken within WG3 is the development of a common evaluation and benchmarking platform for OD estimation algorithms. The main goal of this platform is to provide a test-bed in which a number of algorithms can be implemented and tested under the same conditions. The objective is not to conclude that one approach is “best”, but to provide a support comparison in a variety of settings and conditions in order to help determine the particular situations and conditions under which one approach might behave more favorably than another. The design of the platform and results from its application using several OD estimation algorithms will be presented in this session.


Coffee break


Key note speech - The Importance of Traffic Flow Modelling for Improved Traffic Operations. Markos Papageorgiou, TUC.

Mathematical modelling is an important prerequisite for pertinent planning and efficient operation of vehicular traffic systems and networks. Specifically for traffic operations, traffic flow modelling provides a fruitful basis for the design, testing and operation of a series of important general tasks, including monitoring, estimation, surveillance, optimisation and control. The presentation first recalls some basic facts and taxonomy for traffic flow models. This is followed by an outline of model usage within a number of selected applications involving powerful concepts, such as Kalman Filtering, optimal control, feedback control, MPC (model-predictive control), decision support systems. The presentation is finalised with: (i) some critical remarks on the development and usage of traffic flow models; and (ii) highlights on the importance of modelling while designing the traffic system operations of the future.


MULTITUDE Outputs (Ernesto Cipriani, UNI ROMA3)


Traffic Data: enhanced NGSIM, MoCoPo. Marcello Montanino, UNINA and Christine Buisson, IFSSTAR.

The availability of very precise trajectory data is one of the main change in traffic flow data in the last ten year. This talk will permit to make a brief review of what was done with NGSim data (collected 10 years ago) and their limitations. We will thus present two direction towards better data set availability: the first one is to enhance the NGSim data themselves, the second one is to collect a new data set: the MOCoPo trajectories.

The talk will present results from the application to NGSIM data of a multistep reconstruction procedure for vehicle trajectories that aims at eliminating the main inconsistencies and noise while preserving i) the actual driving dynamics (vehicle stoppages, shifting gears, etc.), ii) the internal consistency of trajectories (i.e. the consistency among space travelled, speed and acceleration) and iii) the platoon consistency (i.e. the actual inter-vehicle spacing).

The talk will then explain the MOCoPo data collection and final production procedure, describe the raw trajectory data and their precision, and define an enhancement methodology.


Training Schools and STSMs. Winnie Daamen, TUDelft.

One of the aims of the Action is to disseminate knowledge on the use of simulation models, calibration and validation to practitioners as well as academics. This is done by providing training during training schools and Short Term Scientific Missions. In total three training schools have been performed during the Action, as well as multiple Scientific Missions by young researchers. During this presentation, a short overview will be given on each of the summer schools and the results of some of the Scientific Missions will be addressed.


State of the Art on Traffic Simulation. Christine Buisson, IFSSTAR.

First task placed on the MULTITUDE agenda 4 years ago, the state of the art of simulation is now (quasi) a reality. Its title will be: “Traffic simulation and data: Validation methods and application”. The book will be published by Taylor and Francis, thus making the results of this project accessible for a large public. The state of the art identifies which methods and procedures should be developed to support the use of traffic simulation models. It has been developed of both simulation research and practice. An elaborate overview has been produced on how to deal with traffic data (data collection as well as state estimation techniques and methods to process and enhance data), frameworks for calibration and validation, sensitivity analyses, calibration and validation techniques. The presentation will focus on three main points: the content of the book, the authors of the various chapters and the publication planning.


A Case for Guidelines. Peter Vortisch, KIT.

While much of the activity in MULTITUDE has focused on technical issues a top level review has also been undertaken on guidance available on how to perform calibration and validation worldwide, and this has resulted in the publication of a Case for Guidelines (CfG). The document firstly has assessed user need through the performance of a web survey and ranked key issues in terms of the degree of satisfaction that end users currently have with the content and scope of existing documents.

Following this, the CfG has examined how well these issues have (or have not) been currently addressed and proposes how these could be improved, not only through the provision of better guidance, but also the adoption of several technical steps that would allow end users to conduct ‘better calibration’.


Open Discussion




Key note speech - Challenges in Pedestrian Flow Modeling. Serge Hoogendoorn, TU Delft.

Pedestrian flow dynamics show fascinating individual interaction behaviours and emergent collective phenomena. Capturing these in a practical simulation model turns out to be hard to do. In the past 15 years or so, researchers from various disciplines have, with varying success, tried to come up with microscopic and macroscopic modelling paradigms that try to capture pedestrian flow dynamics to the extend that the resulting simulation models are sufficiently realistic.

This talk will start by given an overview of the fascinating key phenomena that are observed in pedestrian flow dynamics. After motivating why these phenomena are crucial to reproduce, and which are the characteristics that need reproduction, different modelling paradigms will be presented and briefly discussed. In particular, continuum pedestrian flow models (the equivalent of the well know kinematic wave model for car traffic) and microscopic pedestrian flow models based on differential game-theory will be presented.

Having discussed the different modelling approaches, the different steps of establishing pedestrian model validity will be discussed (face validity, construct validity and predictive validity) and the implications for the presented modelling frameworks will be presented. In doing so, the issue of pedestrian data collection will be discussed, showing that next to the validity of the model, also the validity of the data plays a crucial role.

The talk will end with presenting the key challenges that still remain within this fascinating research domain.


Calibration and validation in theory (Peter Wagner, DLR)


7 sources of Data Uncertainties in model calibration and validation. Matthew Karlaftis, NTUA.

Calibrating and validating a transportation simulation model depends on data collected from actual traffic systems. This collection is frequently complex, and leads to increased uncertainties in the models. Recent developments in technology and the widespread use of powerful computers and mathematical models allow researchers an unprecedented opportunity to account for such uncertainties and direct work in 7 challenging directions. It is these challenges we discuss and offer modeling suggestions.


From component calibration to system prediction, or why the sum may be less than the whole. Hani Mahmassani, Northwestern University.

This conceptual discussion provides a framework for analyzing the sources of uncertainty in model calibration, and the effect that these have on overall prediction errors when using a model system with multiple independently calibrated components. Examples discussed include (1) the effect of correlated driver behaviors in micro-simulation models, (2) meso-model applications at the network level, and (3) approaches for recognizing sources of uncertainty in real-time, simulation-based predictive environments.


Causality and stochasticity in traffic models. Ben Heydecker, UCL.

Macroscopic models are well established as descriptions of traffic flow under a range of conditions. However, for a model of this kind to be used most effectively for control and management, it should be structured to reflect both the systematic influences among the variables and the stochastic variability that remains beyond that. This presentation will develop these general principles in the context of traffic flow models and so address issues such as their formulation and calibration: this leads on to discussion of driver behaviour and the phenomenon of capacity drop.


Coffee break


Calibration and validation in practice 1 (Constantinos Antoniou, NTUA)


Bridging the gap between methodology and practice with traffic modeling. Jordi Casas, TSS.

The main outcome of the MULTITUDE project has been a focus on new methodologies for the calibration and validation of traffic simulation models, such as sensitivity analysis and dynamic OD estimation. However, another relevant outcome is the overview of the survey related to the use of “validation and calibration” tools or techniques considering traffic simulation model practitioners. This paper presents some reflections on the possible role of traffic simulation developers, in this case developers of Aimsun, in connecting the so-far quite separate worlds of the scientific community and real-life practitioners.


Calibration of Roundabouts. Axel Leonhardt, PTV.

Roundabouts have gained popularity over the last decades, and today they are integral part of many upgrade projects and development schemes.

Simulation is used if design manuals like the Highway Capacity Manual are not applicable, for example because the roundabout is part of a larger network simulation, demand dynamics need to be captured, geometry is non-conform or simply because highly detailed output is needed. Roundabout capacity is subject to driving behavior and huge differences can be observed for similar geometries in different places. Calibration of these behavior elements hence is important to gain reliable results. This presentation will show some examples how to calibrate these behavioral elements.


Vehicle Emissions Modelling using Microsimulation. Malcolm Calvert, SIAS.

The impact of vehicle emissions on our atmosphere is a critical problem in today's world. Transport models can be used to estimate vehicle emissions in a given location but how certain can we be about the results? Traditional methods of modelling aggregate levels of vehicle emissions based on average vehicle speeds. By contrast, the vehicle speed and acceleration profiles, which can be derived from microsimulation models provide a much greater degree of detail....but does that mean we have more certainty in the results?

This talk will present a real world microsimulation model which was used to examine options for improving air quality on a congested network. A number of the challenging elements of the project will be discussed as well as the most important question... how certain can we be about the results?


Detailed Simulation Study of the German freeway A92 with SUMO. Peter Wagner, DLR.

This presentation will discuss a detailed simulation of a longer stretch of the German freeway A92 north of Munich. By comparing in a detailed manner the simulation results with the measured one minute data, differences to reality will be worked out and recommendations will be drawn that are related to the further development of  SUMO.

In addition, it is demonstrated how such a simulation can be made correct with respect to the boundary condition, a topic whose complexity is often underestimated.


Social dinner

Friday 6th


Key note speech - Global sensitivity analysis: an ingredient of modeling. Stefano Tarantola, EC JRC.

Mathematical modellers and regulatory agencies worldwide share the belief that sensitivity analysis is a key ingredient of the quality of a model-based study. According to the European Commission . The term interaction is also used in the guidelines for modellers of the US environmental Protection Agency: << [sensitivity analysis] methods should preferable be able to deal with a model regardless to assumptions about a model’s linearity and additivity, consider interaction effects among input uncertainties […], and evaluate the effect of an input while all the other inputs are allowed to vary as well>>.

In spite of this call for the use of global tools for sensitivity analysis, plenty of cases are found in the literature where local sensitivity analysis is employed. In local analyses factors’ importance is investigated by derivative of the output with respect to that input; such derivatives are informative only at the base point where they are computed, but do not provide for an exploration of the rest of the space of the input factors unless some conditions are met in the mathematical formulation under analysis. In this talk I will illustrate the properties of global sensitivity analysis approaches and will discuss variance-based methods, the best practice available today.


Calibration and validation in practice 2 (Tomer Toledo, Technion)


Addressing Uncertainty in Transportation Analysis. Lessons learned from the new Guidelines sponsored by the US Department of Transportation. Vassili Alexiadis, Cambridge Systematics.

1. "Guidance on the Level of Effort Required to Conduct Traffic Analysis Using Microsimulation."  Chapter 6 in this guide presents a statistical analysis methodology linking variability in data sources and simulation results.

2. "Guidebook on the Utilization of Dynamic Traffic Assignment in Modeling."

3. "Analysis Modeling and Simulation for Integrated Corridor Management."  This is a Guide on how to conduct simulation for the analysis of ITS-type technologies, with good practical suggestions as to how to avoid mistakes and conduct model calibration for different operational conditions.  This guide also covers research topics such as a) how to model different levels of traveler information, b) managed lanes, c) congestion pricing, and d) the interaction of highway, public transportation, and parking in simulation.


Modelling flow breakdown on motorway links, the challenges for modelling the impacts of severe congestion on average journey times. Craig Drury, UK Highway Agency.

Model validation points to a need to better understand congested motorways. Congestion is not a simple function of flow and capacity, therefore a new approach is needed. The talk will cover the challenges for microsimulation to represent the day-to-day variations in speed and flow and the progress that has been made with stochastic methods.


Coffee break


Promising results and open issues (Peter Vortisch, KIT)


A simulation framework to investigate uncertainty in traffic modelling. Vincenzo Punzo, UNINA.

In microscopic traffic simulation, inverse analysis is customary applied to get an estimate of both parametric and non-parametric inputs e.g. car-following parameters and OD demand. Error sources include traffic measurements and the estimation process itself. Further, when model components are estimated separately on disaggregate data (i.e. on trajectories) modelling errors arise in simulation from the stochastic interactions of the individual (driver) models. How such different sources of uncertainty affect performances of traffic simulation is a vastly under-investigated issue. To this end a framework is presented which aims to investigate impacts on model outputs of uncertainty sources like accuracy of trajectory data, parameter calibration, parameter correlation and model simplification assumptions.


Co-validation of Expert Judgements and Complex Agent-Based Models. Eddie Wilson, University of Bristol.

Much of the work in MULTITUDE has been about working hard at better parametrisation of our ABMs (e.g. car-following models) using more-or-less traditional scientific method. But in any large-scale transport planning decision, the hard model outputs are inevitably convolved with expert judgements about whether the model output is correct / realistic, and also factors (such as behaviour change, political elements) which are outside the scope of the model. In this talk I will propose a framework for a closer integration of the expert judgements with the model parametrisation itself - the new approach is to attempt a co-validation of the model's parametrisation with the skill of the experts who are judging it.


The MULTITUDE heritage (Ernesto Cipriani, UNI ROMA3)


Traffic Simulation Guidelines. Mark Brackstone, IOMI, and Robert Bertini, Portland State University.

This talk will firstly examine the recommendations made by the project as regards steps that need to be taken to develop guidelines, both in terms of technical issues that need to be advanced and also in terms of overarching actions that need to be taken moving forward in order to enable not only these technical advances, but development of simulation modelling as a whole. Secondly, we will discuss the potential linkages between these recommendations and other existing materials produced and planned by other organizations. In a broader context, how do we sustain and encourage the development of a "community" around the creation, maintenance and evolution of guideline materials. What are the best formats, media, and funding mechanisms to ensure that this effort continues to grow and develop. We will welcome input and comment from all participants.


Beyond MULTITUDE. Constantinos Antoniou, NTUA and Vincenzo Punzo, UNINA.

After almost four years of intense work from a number of researchers and practitioners throughout Europe and beyond, MULTITUDE has achieved some concrete results in the fields related to the management of uncertainty in traffic simulation. This work, however, has also highlighted the need for further work on these topics and has identified subsequent relevant tasks. These directions and plans for follow-up activities of MULTITUDE will be discussed in this talk.


Round table: Merging the views (Mark Brackstone, IOMI)

Participants: Vassili Alexiadis (Cambridge Systematics), Michael Mahut (INRO), Hani Mahmassani (Northwestern University), Markos Papageorgiou, (TU Crete).




Historical Centre and Metro Art Stations Tour