1419 - AP010 -Transit Operational Strategies

Selected Papers from 1419 - AP010 - Transit Operational Strategies

The Transit Operational Strategies Session describes research on improving operational efficiency of transit services.

Full Details about the Paper Session can be found here: Transit Operational Strategies Papers


A Practical Method to Adjust Bus Routes Based on Transfer Penalties Using Trip-Chain Data and SP Survey

Full Presentation

Authors: Younghun Bahk, University of California, Irvine, Kwangho Baek, Korea Transport Institute (KOTI), Jin-Hyuk Chung, Yonsei University

Bus routes are continuously adjusted in response to a city’s development and structural change. This study focuses on relatively frequent route adjustments in a corridor level rather than restricting the whole system in a city. The effects of route adjustment can be measured by changes in both operator and passenger costs. This paper provides a method to predict transit travel choice reflecting transfer penalties based on trip-chain smart card data and stated preference (SP) survey data. In particular, scheduling elements are collected from trip-chain data and transfer penalties are obtained via SP survey. Based on a bus route adjustment model, an incremental multinomial logit model is used to forecast travel. As a case study, a route separation case in Seoul, Korea, is analyzed. Combining empirical data with the result of survey data, all scheduling elements are obtained. Following the steps suggested in the study with those transfer-penalty-related parameters, bus route adjustment in a city can easily be analyzed by practitioners. Keywords: Transfer penalty, Bus Route, Route Adjustment, Trip-Chain Data, SP Survey, Value of Time.


Public Transport Fleet Replacement Optimization Using Multi-type Battery-powered Electric Buses

Full Presentation

Authors: Chunyan Tang, Xiaoyu Li, Dalian Maritime University, Avishai (Avi) Ceder, Technion Israel Institute of Technology, Xiaokun Wang, Dalian Jiaotong University

To achieve a green and sustainable public transit system, most transit agencies plan to completely replace current diesel and hybrid buses with battery-powered electric buses (EBs) in the decades ahead. Based on performances of EBs in practical operations, this study develops a transit fleet replacement model using multi-type EBs to determine an optimal fleet replacement plan in a cost-effective manner, considering associated diesel-electric replacement rates and in-vehicle crowd costs for passengers. Multi-type EBs include small EBs with fast charging technique, and large EBs with fast and slow charging techniques. The proposed model is applied to a real life, case study of the transit system in Qingdao, China. The results obtained indicate that large EBs with a high price tag are preferentially purchased in the first few years of the analysis period, while small EBs with a low price tag are favored in the latter years. Interestingly, with the increase of passenger demand, a large EB with a fast charging method presents more benefits than others. On the contrary, a small EB has more advantages in a transit system with low demand.


Real-time Transit Control for Transfer Synchronization with Deep Reinforcement Learning

Authors: Xinlian Yu, Alireza Khani, University of Minnesota

In transit system, an optimally designed timetable may not perform as expected due to service disruptions, stochastic running time and fluctuations in demand and other unexpected events. Therefore, real-time operational control strategies are required to remedy these problems. This research develops a real-time control framework that attempts to coordinate the operations of public transportation services to allow smoother transfers between different lines. The problem is formulated as a Markov Decision Process (MDP) with unknown transition probabilities to account for benefit over the full controlling period, taking account of variations in travel delays and passenger demand. The action space consists of a combinations of holding and speed changing decisions. A model-free approach based on deep reinforcement learning is proposed to determine the optimal strategies. With simulations using real-world data, we demonstrate the effectiveness of the proposed method through extensive empirical experiments.


Route Overlap in Multi-modal Urban Transit Route Choice

Full Presentation

Authors: Malvika Dixit, Ties Brands, Niels van Oort, Serge Hoogendoorn, Technische Universiteit Delft, Oded Cats, Delft University of Technology

Capturing unobserved correlation between overlapping routes is a non-trivial problem in transit route choice modelling. Research so far has been inconclusive on whether this overlap is perceived positively or negatively by travelers in case of urban networks. This study investigates how travelers value different types of overlap when making transit route choice decisions. We estimate a series of path size correction logit (PSCL) models to account for alternative specification of route overlap in the context of a multi-modal urban transit networks. Our estimation is performed for smart card data from Amsterdam. In addition to the conventional journey leg-based overlap, alternative definitions of overlap in terms of transfer nodes are proposed for multi-leg journeys. The results show that the overlap between transit routes is valued positively when incorporated using either leg-based or transfer node-based PSC separately. When considered simultaneously, overlap of transfer nodes is found to be valued positively by the travelers, but subsequent overlap of journey legs is valued negatively. The results indicate that travelers prefer having multiple (unique) travel options at a transfer location. Moreover, overlapping legs, especially after the common transfer location, do not add any value in terms of robustness, and therefore reduce the attractiveness of overlapped routes, compared to completely unique routes. This study highlighted the need to consider the overlap in terms of both journey legs and transfer nodes, as ignoring the latter might lead to contrasting conclusion on the overlap of journey legs.