Validation of a conventional metro network model using real data

Abstract

This article describes the validation of a conventional metro network model using real data. The metro system in Hong Kong is used as a case study. The study is characterised by (1) using information retrieved from ticketing gates to construct station-to-station O-D trip matrices, (2) using operator supplied transit line schedules to describe vehicle movements, (3) using site surveys to determine the time spent by passengers walking between station ticketing gates and platforms or between station platforms, and (4) using link counts to validate the trip assignment results. The study aims to assess how accurately the performance of an existing metro system can be predicted by using a conventional network assignment model. It also explores the differences between modeling metro systems as compared to bus systems, and investigates how conventional transit network models should be improved for the modeling of metro systems. The validation results indicate that the maximum error in individual cross-harbor link flows is approximately 12%. As the assignment results are found to be sensitive to the setting of relative weightings attached to in-vehicle time, walking time, waiting time, and line change penalty, it is considered that the model accuracy can be further improved by carrying out surveys to determine the route choice behaviors of different types of passengers and using more complex assignment algorithms to account for the variability in the path choice behaviors of passengers. Copyright © Taylor and Francis Inc.