Network Modeling for Walking Infrastructure: Developing Pedestrian Traffic Assignment Methodologies for Large-Scale Footpath Networks
| Autor: | Lilasathapornkit, Tanapon |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2022 |
| Předmět: |
Volume delay function
Traffic assignment problem Travel behavior Mixture of Logit models Macroscopic model Pedestrian flow Dynamic network loading Link transmission model GPS trajectories Path Size Logit Stochastic model Bidirectional pedestrian flow Pedestrian route choice 400512 Transport engineering Dynamic user equilibrium |
| DOI: | 10.26190/unsworks/24453 |
| Popis: | This dissertation focuses on exploring network-wide pedestrian bidirectional dynamics, developing frameworks for both static and dynamic pedestrian network traffic assignment models, and estimating an empirical pedestrian route choice model using revealed preference (RP) data from Sydney, Australia. To capture the bidirectional impact (e.g. lane formation and congestion due to flows from counter direction), four different types of pedestrian volume delay functions (pVDFs) are proposed and calibrated against controlled experimental data. The proposed pVDF is an essential component to develop a user equilibrium traffic assignment problem (UE-pTAP) framework. The proposed UE-pTAP framework has been applied to both a small-scale toy network and a largescale real-world network to demonstrate the impact of the bidirectionality of walking streams on the network assignment solution. To capture the pedestrian travel behavior in a dynamic context, this dissertation also proposes a pedestrian dynamic network traffic assignment (DTA) model based on the link transmission model (LTM). The formulated dynamic UE (DUE) adopts a bidirectional fundamental diagram that takes into account properties of bidirectional pedestrian streams such as self-organization. The applicability and validity of the model are demonstrated in a small grid network, a long corridor, and a large-scale real-world network. To better understand the pedestrian route choice decision-making in an urban environment, four types of logit-based route choice models, C-logit (CL), path size logit (PSL), and error component (EC), are also estimated using revealed preference (RP) data passively collected from smartphones in Sydney, Australia. The cross-validation results show that the four route choice models perform equally well. However, the PSL has a slight edge over other models because it has the least variability and requires much less computational resources than the EC models. Overall, the quantitative values for various attributes e.g., distance, the number of turns, maximum gradient, the number of points of interest (POI) along a path, and green view index (GVI), are comparable to previous findings in the literature. Overall, this dissertation offers valuable and new insights into the study of pedestrian bidirectional dynamics and the development of pedestrian network models in both static and dynamic contexts. The development of stochastic route choice models using smartphone data also provides an in-depth knowledge of decision-making behavior influenced by various built environment factors beyond travel distance. Results of this research help with the strategic planning of walking infrastructure in cities as well as estimating foot traffic across footpath networks for operational purposes. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|