Transport and Mobility Laboratory

The spatial and temporal distribution of built-space supply plays an important role in shaping urban form and thus the general travel pattern in an urban area. Within an integrated framework, we are interested in modelling the decisions of a builder in terms of when, where, what type, and how much built-space to build. We present a discrete-continuous model formulation for the built-space supply decisions that are based on expected profit maximization. The framework is applied to estimate a model for supply of new office space in the Greater Toronto Area (GTA) for the duration of 1986 to 2006. To our knowledge, this work is the first that models the where, when, how much, and what type of office space to supply in a single framework at a fairly disaggregate spatial zoning system. The results indicates a risk taker behaviour on the builders‟ part, while market conditions and supply of resources (labour, construction cost etc.) are also found to be important factors in decision making.

This paper presents the hedonic analysis of office space rent in the Greater Toronto Area (GTA). The effects of accessibility, quality, location, and market conditions on rent are explored. Data indicate a high degree of spatial heterogeneity and clustering effects. Spatial analysis techniques are incorporated within the hedonic framework to capture these effects. Results indicate that access to transport infrastructure, distance from CBD, and vacancy rate are significant in explaining the variation in the rent.

Firm location models constitute a significant element of integrated land use and transportation models because the behavior of individuals and households is influenced by employment location. This paper examines office firm location models, both for new and relocating firms, to identify ways to improve their estimation and application. The results of the paper suggest that introduction of anchor points, such as the firm current location and the owner household location, significantly improve the estimation of office location models. A generation procedure for choice sets is then shown to improve the model's ability to replicate the decision behavior of office firms and thus provides a way to increase efficiency in the application of the model.

There are many markets within the transportation industry that exhibit price taking behaviour. Some examples include: airline seats auctions, freight spot markets and public transit infrastructure projects. While this behaviour is ubiquitous, the underlying mechanism can be captured in a single framework. Here, we present a microsimulation based framework of a price taking market that recognizes the generality of this problem. The description of our approach follows.
Consumer agents are interested in one or more differentiated products offered by the producer agents. Before making a selection, consumers establish preferences for each product within their choice set that is based on their consumption behaviour and the products’ attributes. If the choice set and individual product preferences for each consumer are known, clearing this market could be reduced to a graph theoretic problem of finding maximum weighted bipartite matching. The market is represented as a bipartite graph where the consumers and products/producers are the vertices, the choice sets are the edges, and the unidirectional/bidirectional preferences are the weights on the edges.
Maximum weighted bipartite matching problems belong to the class of NP problems that requires exponential time to arrive at a solution. Therefore, we use a highly efficient approximation algorithm that guarantees an optimal solution. Our implementation mechanism iterates through the edges in a bottom-up manner, until market participants that cannot find acceptable matching drop out – thus leaving an optimal one-to-one assignment between the remaining consumers and products/producers in the market. The complexity of the algorithm depends on both the size of the graph and the number of edges adjacent to each vertex. Selecting an adequate step size in the iteration can give a near-linear running time performance.
We developed this price taking market clearing mechanism for the Integrated Land Use, Transportation, and Environment (ILUTE) modelling system, which is an on-going large-scale microsimulation of the Greater Toronto and Hamilton Area, underdevelopment at University of Toronto. The framework is implemented as an abstract class that is general enough so that any price taking market application can be extended from it. At present, marriage, rental, and labour markets are the realizations of this mechanism in ILUTE.
An operational example is the marriage market. In this market, potential husbands and wives define their respective choice sets using a rule-based method. Afterwards, they are matched to each other using a utility maximization framework where the utility for each potential husband-wife pair is defined by the socioeconomic attributes (e.g. income, age). Similar approaches are used for the rental and labour markets.
To overcome the high computational requirements of microsimulating larger markets, we designed and implemented a high performance parallel version of this algorithm that takes full advantage of the readily available multi-core processing technology.

Firm (employment) location influences behavior of individuals and households in the short and long term. This paper presents simulations conducted to test the ability of an office location model to replicate the actual choices of office firms in the Greater Toronto Area. Findings are that including explicit search areas could significantly increase the ability of location models to replicate actual choices by office firms. In addition, simulation procedures themselves have an impact on the forecasting capabilities of the modeling system. Finally, regardless of the suggested improvements, the model system ability to replicate location decisions of specific firms is somewhat limited.

To understand the factors that influence the spatio-temporal distribution of built space, and thus population in an urban area, play an extremely important role in our greater understanding of urban travel behaviour. Existing location of activity centres, especially home and work, strongly influences the short-term individual-level decisions such as mode of transportation, and long-term household-level decisions such as change in job and residential location. Conditions in the built space market also affect households’ and firms’ location and relocation decisions, and hence influence the general travel patterns in an urban area. This research addresses a very important, but at the same time, not very widely investigated dimensions that play a key role in the evolution of built space and population distribution: Markets. A disequilibrium based microsimulation modelling framework is developed for the built space markets. This framework is then used to operationalize the Greater Toronto and Hamilton Area’s owner-occupied housing market within Integrated Land Use Transportation and Environment (ILUTE) modelling system. Simulation results captured heterogeneity in the transaction prices, due to type of dwellings and different market conditions, in a very disaggregate fashion.

Analysing the factors that influence the spatio-temporal dynamics of transport, built space, energy, and population in a complex urban engineering system, play an extremely important role in our greater understanding of the evolution of the system and the behaviour of various agents (households, firms, governments, etc.) involved in it. In this context, this book addresses two very important, but at the same time, not very widely investigated dimensions that play a key role in the evolution of complex urban engineering systems: Markets and Decisions. A disequilibrium based microsimulation modelling framework is developed for the urban built space markets. On the decisions side, this book presents a generic multidimensional modelling framework that captures the behaviour of builders in terms of the supply of new built space. The where, when, how much, and what type of supply decisions were incorporated within a single framework. In addition to that, this book also develops a comprehensive hedonic analysis for the asking rent of office space. Spatial analysis techniques are incorporated within the hedonic framework to capture clustering and heterogeneity.

Gathering data about pedestrian origin, destination and route is difficult, in particular indoor and on a large scale. Different sensors could be used (Bluetooth, GPS, cameras, Smartphones, etc.). These data are important for route choice modeling, description of congestion, efficient design of new facilities, and travel guidance and information systems.
Most data collections are device-centric. In this paper, we focus on the network infrastructure and propose to use WiFi traces to generate pedestrian Origin-Destination matrices. WiFi traces allow for the localization of mobile devices and thus of their users, but are scarce and fuzzy.

We collect logs of EPFL campus members from the WiFi access points in collaboration with Cisco. This method of collecting data has several advantages over device-centered processes like the ones using smartphones, since it doesn’t need to have formal acceptance from the users. This process is transparent for the users, and it also works for the users that are not authorized to connect to the system.

This methodology with WiFi traces could help to understand pedestrian behavior with a low cost in data collection for transportation hubs, commercial centres and maybe even for city centres (e.g. shopping street segments) or for mass events, if they are covered by WiFi.

Several modeling approaches in both microscopic and macroscopic scales have been already put forward into the pedestrian modeling literature. The most operational one of these models is the conventional social force model, where forces are firstly defined by physical concepts (Newton’s equation) and then have been applied to pedestrian behaviors. On the contrary, there are also models based on discrete choice modeling concept that go deeper in behavioral aspects of pedestrians’ reactions rather than physical forces. In this paper, firstly we go through the literature in pedestrian modeling domain. Subsequently a real pedestrian simulation experiment conducted in the SV building at EPFL (Ecole Polytechnique Fédérale de Lausanne) is explored. In this building at not a big entrance hall people’s walking behaviors become extremely interesting at certain periods of the day since they take different entrances and exits depending on various destinations that they can reach by passing this hall. We put a special emphasis on model calibration. The simulator we used for calibration and simulation is called VISSIM. The VISSWALK add-on of the software (intended exclusively for pedestrians) has been mainly exploited. This simulator is produced by PTV company and is based on the social force model. There is another dataset stem from a train station which experiences interesting passenger behaviors as a highly frequented multi-modal transportation hub. This second dataset is currently being collected and processed and could be of a hight interest for more development in pedestrian modeling especially in model calibration and validation. Basically, this project of pedestrian flow simulation in a transportation hub is going to take advantage of the same methodology and calibration method used for the SV building. For the SV building experiment various forecasting (demand change) and optimization (supply change) scenarios are also built and simulated. The results are analyzed.

For many railroad stations, capacity limits are reached at peak hours and pedestrian congestion can be observed. A sound understanding of pedestrian flows within a train station can help improve overall level of service, customer experience and safety.

In the deployment of microscopic pedestrian
ow simulation, dynamic origin-destination (OD) table estimation is one of the key components. Pedestrian OD demand changes abruptly whenever a train arrives or is about to depart. With longer platforms and higher train capacities projected, step changes in flow pattern will become more pronounced. Furthermore, as especially interregional trains arrive and leave in close succession, such articial pedestrian waves often overlap, leading to complex interactions and, as the case may be, congestion.

The goal of this work is to give an overview of existing literature on dynamic OD table prediction methods, to identify and answer key questions concerning e.g. discretization of time, and to lay out groundwork for the development of a dedicated pedestrian OD demand model for train stations.

It turns out that classic OD estimation procedures for car trac1 are mostly concerned with predicting OD tables based on measured link flows. Models used for aircraft and gate scheduling, where the goal is e.g. to minimize walking between connecting flights, tend to neglect interactions between multiple passenger flows and usually do not consider time as a separate dimension. For train stations, it is favorable to develop a functional, time-dependent relation between the OD table of the complete train station on one hand side and OD tables of incoming and outgoing trains, the corresponding time table and track assignment on the other hand side.

This work is part of a comprehensive project employing pedestrian tracking, flow modeling and infrastructure optimization. Therefore, all ideas and model concepts can be tested, parametrized and validated on a real scenario.

In the future, simulation-based infrastructure and operation planning will allow to better manage pedestrian flows in train stations. We would like to contribute towards an algorithm that estimates dynamically OD tables as a function of infrastructure and passenger demand, which we believe is at the core of such optimization techniques.

Land use and population evolution play an important role in shaping up the travel patterns of an urban area. Various dynamic microsimulation models are developed within the ILUTE framework to capture these complex interactions. In this presentation the development of main models and processes that are managing the evolution of land use and population in ILUTE simulation is discussed. These processes and models are a mix of econometric, stochastic, Monte Carlo, and rule based techniques. A full population simulation for the Greater Toronto and Hamilton Area for a period of 20 years from 1986 until 2006 is in the process of running. The results are validated using Census, Transportation Tomorrow Survey (TTS), housing data, and various other sources. This presentation also presents some of the initial results.

Spatial and temporal distribution of built-space supply plays an important role in shaping up the urban form and thus the general travel pattern in an urban area. Within integrated framework, we are interested in modelling the decisions of a builder in terms of when, where, what type, and how much of a built-space to build. This talk presents a discrete-continuous model formulation for the built-space supply decisions that are based on expected profit maximization. The framework is applied to estimate a model for supply of new office space in the Greater Toronto Area (GTA) for the duration of 1986 to 2006. The results indicates a risk taker behaviour on the builders’ part, while market conditions and supply of resources (labour, construction cost etc.) are also found to be important factors in decision making.

Since 2008, for the first time in the history of mankind, urban areas of the world are host to more than 50% of its population. The importance of urban regions is pivotal as they are analogous to concentrated mass. It is where the growth, technological progress, intellectual vibrancy, and political power are concentrated. For a sustained and competitive urban growth, it is very important to understand the behaviour of its population and treat urban region as a complex system of various interconnected urban infrastructures, providing services to its residents. An important dimension in this context is the understanding of the evolution of urban built space (e.g. housing, office space, etc.) as a direct result of decisions made by various agents (e.g. builders, households, etc.), their participation in built space markets (e.g. housing market), and the two way interactions between various urban systems (e.g. transportation influencing land use and vice versa). The main focus of this talk is the research done in this context, so as to conceptualize the problem and develop modelling methodologies. Discussion is further focused on the microsimulation based operationalization for the Greater Toronto and Hamilton Area (GTHA), validation process, and implications in terms of policy making for a sustained growth in the GTHA.