Transport, Mobility, and Workplace Location : Models and Applications

Abstract: Travel demand analysis is one of the core constituents of transportation studies. Therequired insight to maintain and develop a sustainable transportation system, in additionto learning from previous research globally and locally, is generated from studyingthe effects of previous policies, investigating future possibilities and potential outcomes,and describing the current situation. The objective of the thesis is to use urbanmodeling and decision support methods to contribute to the knowledge that improvesdecision making for a sustainable society.In this thesis, three of the papers focus on implementation and application of adynamic model of movement for prediction and forecast of workplace demand andaccessibility, and for a trip chaining problem. This framework formulates movementthrough a Markov chain and solves it by using the Bellman equation which by theassumption of IID Gumbel error terms turns into a recursive logit, which reflects dynamicand directional nature of time in modeling movement. This approach is usedfor modeling a workplace choice model and accessibility to work (Paper 1), that isapplied for workplace allocation in a scenario planning framework for urban developmentgrowth with a 2040 forecasted synthetic population (Paper 3), and a dynamictrip chaining model with flexible number of trips in the chain (Paper 4). The workplacelocation choice model is unique as it connects the land use and transport modelsin a framework that is consistent with random utility maximization approach whilerespecting forward-looking behavior of individuals and the dynamic and directionalnature of time. This approach allows for evaluations of counterfactual scenarios inPaper 1 and in future workplace growth in Paper 3 for understanding the implicationsof workplace allocations under different urban planning scenarios in terms of travelbehavior, with implied social segregation issues, workplace demand and distributionof welfare. When applied in a trip chaining context (Paper 4), the methodology allowsfor flexibility in number of trips in the chain, with the implication of creating the linkbetween trip chaining problem and the marginal utility of time and the marginal rateof substituition not only for different trip purposes, e.g. work and leisure, but also forany other variables in the model, even in the absence of scheduling, which is novel inthe literature of trip chaining and time valuation.A fundamental aspect of the methodology that is used in this thesis revolves aroundthe utilization of Multivariate Extreme Value (MEV) models. We explore the intricaciesof these models in detail in Paper 2, and take one step against the conventionby using a replicated nest in a multinomial setting. We discuss the empirical implicationof retrieving exchangeability, when it is compromised. This approach gives areminder to respect the complexity of error term in MEV models, while benefitingfrom the generality of its definition.Travel time and costs are among the factors that impact transportation and land useinteractions. In Paper 5, we address this interaction by exploiting a natural experimentto investigate green vehicle owners’ responses to phasing out of their time-varyingtoll exemption in Stockholm through comparison of data in 2012 and 2013, whichwere before and after this policy was implemented. The results show a significantdrop in the total number of green vehicles that crossed the toll stations in 2013, anda significant shift to off-peak crossing time in the toll stations in 2013 compared to2012.