[Paper Review] Activity-based and agent-based Transport model of Melbourne (AToM): an open multi-modal transport simulation model for Greater Melbourne

Afshin Jafari, Dhirendra Singh+5 more|arXiv (Cornell University)|Dec 16, 2021

Transportation Planning and Optimization31 references4 citations

TL;DR

This paper presents AToM, an open, multi-modal, agent-based and activity-based transport simulation model for Greater Melbourne, calibrated to real-world data using MATSim. It successfully replicates observed mode shares, travel times, distances, and road volumes, offering a flexible, reproducible workflow for simulating walking, cycling, driving, and public transport at city scale.

ABSTRACT

Agent-based and activity-based models for simulating transportation systems have attracted significant attention in recent years. Few studies, however, include a detailed representation of active modes of transportation - such as walking and cycling - at a city-wide level, where dominating motorised modes are often of primary concern. This paper presents an open workflow for creating a multi-modal agent-based and activity-based transport simulation model, focusing on Greater Melbourne, and including the process of mode choice calibration for the four main travel modes of driving, public transport, cycling and walking. The synthetic population generated and used as an input for the simulation model represented Melbourne's population based on Census 2016, with daily activities and trips based on the Victoria's 2016-18 travel survey data. The road network used in the simulation model includes all public roads accessible via the included travel modes. We compared the output of the simulation model with observations from the real world in terms of mode share, road volume, travel time, and travel distance. Through these comparisons, we showed that our model is suitable for studying mode choice and road usage behaviour of travellers.

Motivation & Objective

• To develop a scalable, open-source, multi-modal transport simulation model for Greater Melbourne that includes walking, cycling, driving, and public transport.
• To address the lack of comprehensive, open, and calibrated models for active transport (walking and cycling) at city scale.
• To create a flexible, reproducible workflow using open data and tools for building large-scale agent-based transport models.
• To calibrate mode choice parameters to match real-world mode shares from Census 2016 and VISTA 2016–18 surveys.
• To enable future scenario analysis for active transport interventions and integrated health and transport impact assessments.

Proposed method

• Construct a synthetic population based on 2016 Census data, with individual attributes (age, gender, occupation, household structure) and daily activity schedules from Victoria’s 2016–18 Travel Survey.
• Integrate a detailed road network including all public roads accessible to pedestrians, cyclists, and motor vehicles, with attributes such as slope and bikeway type.
• Implement a MATSim-based simulation framework to model individual agent behavior, including mode choice and route choice decisions.
• Calibrate mode choice parameters using a multinomial logit model to match observed mode shares for work and education trips.
• Validate model outputs against real-world data, including mode share, road volume, travel time, and travel distance during peak hours.
• Develop a modular, open-source workflow hosted on GitHub to support reproducibility and future model extension.

Experimental results

Research questions

• RQ1 Can an open, multi-modal, agent-based transport simulation model accurately replicate real-world travel behavior in a large urban area like Melbourne?
• RQ2 How well does the model reproduce observed mode shares, travel times, distances, and road usage patterns for driving, public transport, cycling, and walking?
• RQ3 To what extent can the model’s mode choice calibration reflect real-world preferences for different travel modes, particularly for active transport?
• RQ4 What are the limitations of the current model in representing active transport behavior, especially regarding road attributes and their influence on mode choice?
• RQ5 How can the model be extended to support integrated health impact assessments and scenario-based policy evaluation?

Key findings

• The AToM model successfully replicates real-world peak-hour car traffic volumes and public transport station usage distributions with high fidelity.
• Mode shares for driving, public transport, cycling, and walking are accurately captured, with calibration achieving close alignment to observed data from Census 2016 and VISTA 2016–18.
• Travel times and distances simulated by the model are realistic and consistent with observed values, indicating sound route choice and congestion modeling.
• The model demonstrates strong performance in simulating motorized and public transport behavior, though accuracy for cycling and walking is lower due to unincorporated road attributes.
• The model’s open workflow and use of open data enable reproducibility and scalability for future research and policy analysis.
• Limitations include the absence of dynamic public transport interactions (e.g., delays due to congestion) and unmodeled impacts of road attributes (e.g., slope, bikeway type) on active transport mode choice.