Modelling Rail Transit Dwell Time Using Automatically Collected Passenger Data

This paper presents challenges and opportunities of using automatically collected passenger data for modeling dwell time in rail transit. While past studies on rail dwell time mainly relied on manual studies with limited spatial and temporal observations, the authors are using a large collection of dwell observations on the Massachusetts Bay Transportation Authority (MBTA) Red Line, collected using automatic vehicle location (AVL) and automatic fare collection (AFC), to explore the applicability using automated data for dwell time monitoring and estimation. The authors confirm findings from previous literature such as the effect of passenger activity and passenger flow directionality, but have also observed heterogeneity among stations on a single line. A robust weighted ordinary least squares regression with bi-squared error function is applied to estimate dwell time model coefficients. The authors have also explored identification of unpredictable nature of dwell time in large dwell time data sets for factors other than passenger activity such as train congestion, passenger incidents, and operator behavior. Namely, a model for identifying train congestion related dwell time delays with the use of AVL data is discussed to improve the robustness of dwell time modelling. The paper concludes with discussion on the importance of system-wide dwell time modelling and monitoring that incorporates heterogeneity between stations for reliable headway operations along the rail line.

• Supplemental Notes:
  • This paper was sponsored by TRB committee AP000 Public Transportation Group.
• Corporate Authors:

  Transportation Research Board

  ,    
• Authors:
  • Wolofsky, Gabriel
  • Saidi, Saeid
  • Attanucci, John
  • Salvucci, Frederick P
• Conference:
  • Transportation Research Board 98th Annual Meeting
  • Location: Washington DC, United States
  • Date: 2019-1-13 to 2019-1-17
• Date: 2019

Language

• English

Media Info

• Media Type: Digital/other
• Features: Figures; References; Tables;
• Pagination: 5p

Subject/Index Terms

• TRT Terms: Automatic fare collection; Automatic vehicle location; Data mining; Dwell time; Mathematical models; Rail transit; Regression analysis
• Identifier Terms: Massachusetts Bay Transportation Authority
• Subject Areas: Data and Information Technology; Operations and Traffic Management; Passenger Transportation; Planning and Forecasting; Public Transportation;

Filing Info

• Accession Number: 01697554
• Record Type: Publication
• Report/Paper Numbers: 19-03817
• Files: TRIS, TRB, ATRI
• Created Date: Mar 1 2019 3:51PM