Planning for the electrification of the Victorian bus fleet

Should the charging network for Melbourne’s electric bus fleet be depot-based, or a combination of depot-based and en-route charging? Further, in any decisions on this, the planning for the electrification needs to consider the constraints of both the transport and energy sectors. These issues were the focus of the recently complete project, Charging requirements for Melbourne’s electric bus fleet. That project’s final report is available for download below.

The transport sector is responsible for approximately one-fifth of greenhouse gas emissions in Australia. Electrification is the only viable solution to decarbonise the transport sector and reduce its emissions.

In striving for this reduction, Victoria has made the decision that all new public transport buses must be zero emission by 2025. NSW also has an aggressive target, with the plan to have Greater Sydney’s entire bus fleet zero emission by 2035. Other states and territories are at various stages of trials and targets.

Work on the project was carried out by the Centre for New Energy Technologies (C4NET) and RMIT University.

Objectives

The objectives for the project were to:

• build a spatio-temporal charging map of electric buses for the metropolitan Melbourne public transport network
• investigate daytime charging locations (e.g. transit bus stops).

These two main objectives were then divided into two work packages:

1. WP1: Spatio-temporal charging maps: In this the research aims to design an AI-enabled calculation engine for analysing the charging requirement of electric buses (via deep learning or other applicable data-driven machine learning modelling techniques).
2. WP2: Optimal size and location of charging stations: This aim here was to design an integrated optimisation framework to identify and rank non-depot charging stations in the whole transport system.

Energy consumption factors

Levels of energy consumption by electric buses can be determined by three broad groups:

1. Vehicle-related: such as weight, speed, acceleration, braking energy regeneration, and the energy consumption of auxiliary systems (e.g. airconditioning);
2. Environment-related: ambient temperature, wind speed, road and traffic conditions; and
3. Driver-related: driving patterns, charging habits, route planning

Datasets

Data used in this project emanated from a number of sources. Firstly, via the industry standard that is General Transit Feed Specification (GTFS). This information is collated by Public Transport Victoria, and includes data points such as stops, routes, trips, stop times, and fare attributes and rules.

Next was the high resolution data collected from June to November 2022 from Victoria’s Zero Emission Bus Trial. Amongst the data points collected here were trip start and end time, route number, travel time, average speed, distance, air conditioning status, total energy consumption, and energy consumption per kilometre travelled.

This, plus some model generated data from the widely-known powertrain simulator ADVISOR (ADvance VehIcle SimulatOR) – the software developed by the US National Renewable Energy Lab (NREL), added to provide enough detail so as to be adequate for the use of machine learning.

Methodology

In order to conduct a comprehensive analysis of energy consumption, the datasets were then subjected to a diverse set of machine learning (ML) techniques.

These ML methods were chosen to capture different aspects of energy consumption behaviour, such as linear relationships, complex patterns, and interactions between multiple variables. By employing a range of algorithms, including neural networks, support vector machines, linear regression, random forests, decision trees, k-nearest neighbours, an ensemble method is proposed to provide an estimate of the energy consumption of buses on different routes.

Leveraging this model and using a sophisticated graph-theoretic approach, a mathematical optimization framework is proposed to determine the optimal location and size of charging stations. Assuming consistent bus schedules with internal combustion engine (ICE) vehicles, this framework also estimates the required electric bus fleet size for full metropolitan coverage. Furthermore, it allows for flexible maximisation of daytime charging using cost-effective and clean energy, potentially necessitating a larger fleet to accommodate the existing schedule.

Report findings

The datasets and the various ML techniques were then applied to address the two work packages.

For Work Package 1, an online tool was developed, available online, at E-bus Energy Estimator. It allows you to produce a bus route visualisation on a map, along with presenting for that route the energy summary, road information (elevation, gradients, etc).

Work Package 2 investigated the optimal size and location of charging stations across Victoria. It did this via addressing the following three questions:

1. What is the minimum number of battery electric buses (BEB) need to serve the fleet’s transport requirements?
2. What is the minimum number of charging stations needed to keep that fleet operational, without needing BEBs to return to depots for charging?
3. Where are the prospective charging stations located, and how much energy do they demand?

For this task, the research assumed that all stations (that is route start and end points) can potentially be turned into a charger station, enabling charging between trips without needing to return to depots.

Detailed findings are revealed in the report, but key observations include:

• Many of the identified stations are common across both weekday and weekend scenarios, demonstrating the applicability of charger stations throughout the week
• The ten most important locations for charging stations to cover all weekday and weekend trips in Melbourne, that have appeared in different scenarios, are identified at Frankston Station, Chelsea Station, Sunshine station, Melbourne Airport, Mernda Station, Belgrave Station, King Street (Melbourne City – subject to space availability), Airport West SC, Northland SC, and Craigieburn Station. de the Frankston station in the east and Sunshine station in the west.
• Results show that 1730 BEBs in the Melbourne metropolitan area can absorb up to 23MWh energy for daytime charging.
• The model demonstrates that increased clean charging is achievable in most cases without building new charging stations, simply by increasing the capacity of existing ones.

Expected project impacts

This project not only enhances understanding of electric bus charging challenges for operators and planners, emphasizing the crucial need for unified planning between transport and energy sectors, but also underscores the importance of local research in driving smoother electrification. This paradigm shift necessitates the development of new tools and solutions for techno-economic-social studies, planning, and operation, tailored to the Australian context.

As a contribution in this context, proof-of-concept of a data-driven solution for optimal electric bus charging infrastructure is developed in this project to maximise daytime energy use, and significant reductions in emissions and improvements in operational efficiency, all grounded in local data and expertise.

Dr. Ali Moradi Amani, Senior Research Fellow, RMIT University

Download the final report

Download your copy of the final report, Spatio-temporal analysis of charging requirements for Victoria’s Electric Bus Fleet, by clicking the button below.

Webinar

A webinar presenting findings and outcomes from the project was held on 18 March 2025. A recording of the webinar is available at: Optimising electric bus fleet charging