Navigating climate uncertainty: Strategic planning with imperfect data
by Sam Gill
The answer to that question is not trivial, especially if the project spans across the three eastern coast states of Australia. Currently Australia’s longest rail project, Inland Rail, is a proposed 1600 km rail line that connects Melbourne to Brisbane freight in 24 hours via the States of Victoria, New South Wales and Queensland, with a combination of new rail infrastructure and upgrade of existing infrastructure.
Rail noise across each State is regulated and managed differently with their respective guidelines and policy documents. Victoria and NSW have day and night decibel thresholds, whilst Queensland has a 24-hour exposure threshold. Similarly, for sections where existing rail are being upgraded, all three States have slightly different thresholds which include an absolute threshold in Queensland or a combination of an absolute threshold and a relative increase in noise in Victoria and New South Wales. Furthermore, considerations of factors which affect rail noise such as rail speeds, track joints, level crossing bells and train horns are considered differently across the three States. In this regard, the modelling of future rail noise levels needs to carefully account for these differences and assessed the predicted impacts in each jurisdiction against the respective thresholds.
One important parameter for assessing rail noise impacts is a pass-by maximum noise level (Lmax). This parameter is critical for a freight-dominated project like Inland Rail as the locomotives could be loud as they go past the residences. Typically, this is assessed as a 95th percentile Lmax, which means that any unusually rare and loud events are excluded (as they would fall within the top 5%). However, in Queensland, the criteria is a Single Event Maximum (SEM) defined as the arithmetic average of the 15 loudest pass-by maximum levels within a given 24-hour period. This parameter is challenging to predict, especially for new rail infrastructure where it is not possible to measure the SEM on field.
To overcome this challenge, a prediction method based on a ‘Mote-Carlo’ statistical model was adopted. In this model, rail pass-by noise levels are randomly picked from databases of numerous pass-by noise levels to simulate the noise levels on a given day, and these random values are averaged to obtain the SEM. Given the randomness in this method, this random picking process is repeated several thousand times to obtain a trend and derive the most likely SEM that can be expected on field. This mathematical prediction technique was tested on existing rail lines and found to correlate well with field measurements.
There exists a need to develop a consistent project-wide rail noise criteria that is effective in addressing all the nuanced differences in the criteria, whilst being simple and effective to implement and understand for all stakeholders. We recommend technical assessments and engagement with state authorities early in the project development phase to investigate noise emissions, controls and development of appropriate criteria. Once approved, the project criteria can be used across all sections of the project to ensure residents adjacent to the project would get a consistent outcome.
For further information, please contact Arvind.
Co-authored by Susan Kay
Senior Program Environment Advisor – Acoustics
Australian Rail Track Corporation