Optimizing Ballast allocations with AI and Machine Learning

ABOUT AURIZON 

Aurizon (previously One Rail Australia) is a leading intermodal and bulk freight provider in South Australia and the Northern Territory. Its business includes the 2,200-kilometre Tarcoola-to-Darwin railway line, which connects regions rich in resources and agricultural commodities with Darwin, the closest port to Asia.

CURRENT PROCESS – TARCOOLA TO DARWIN  MANUAL BALLAST INSPECTION

The current method of inspecting and prioritising ballast delivery trains is visual track inspection. Areas of depleted ballast are identified through subjective volume estimations with the locations approximated from the nearest km post. This process is open to inconsistencies and is intrinsically unreliable, which can lead to suboptimal and inefficient ballast delivery.

AURIZON’S BUSINESS OBJECTIVE

Aurizon approached Cordel to provide a fully automated capture, processing pipeline and visual interpretation across the 2,200km network. This would allow the Aurizon ballast maintenance team to triage and allocate resources from their central Adelaide office and vastly reduce the hours spent on track manually inspecting ballast condition. Using this information, the replenishment of ballast could be delivered much more efficiently and consistently. In addition to these logistical savings, the minimization of track workers inspecting in the corridor provides a clear safety improvement.

Short-Term Goals:                                                          

• Identify priority areas for immediate ballast replenishment
• Provide volumetric materials estimation
• Measure the depth of ballast under sleepers

 Long-Term Goals:

• Monitor ballast depletion over time
• Optimise the ballast purchasing cycle, thus delivering clear commercial benefit to Aurizon

CORDEL’S AUTOMATED SOLUTION APPROACH

The ARTC Track Geometry Inspection Vehicle (AK Car) has been fitted with Cordel’s “Wave32” scanning equipment for the past three years. During this time, Cordel has captured over 180,000 km of data and synchronised the Lidar, Inertial Navigation, GPS, and Camera Vision technologies. 

This technology delivers data to Cordel’s in-house AI and Machine Learning platform to quantify the location and quantity of ballast deficiencies across the Aurizon network.

With this approach, a task typically requiring track inspectors to collate with other routine inspections was completed in days by utilising Cordel’s hardware installed on the AK car and processed by Cordel’s automated digital platform.

PROJECT ACHIEVEMENTS AND OUTCOME

Cordel’s ballast profiling solution analyzed the entire network in 20m increments. Each 20m cross-section reports measured volume, deficiency, and ballast depths compared to the design standards. These 20m cross-sections can then be rolled up into appropriate-sized sections, such as 1km or 10km, to provide detailed planning information in a format that the Aurizon maintenance team can analyze.

Aurizon has access to all reports, digital twins, and video data through the unique Web Portal, which provides easily searchable and accessible data.

On completion of the initial 6-month engagement, Cordel had delivered;

1. 100% digital twin and 4K video of the network
2. Eliminated track inspector bias for ballas volumes
3. Triagged reports to optomize ballast dump planning

Integrating this solution has improved the effectiveness of the Aurizon ballasting and tamping programs with savings identified in hours spent on track and efficient delivery of materials at the locations in need. The tamping program has also been optimised to ensure adequate ballast is available for the tamping process.

The deployment of Cordel’s LiDAR-based technology, automated processing, and consistent visualization of the outputs via Cordel’s WebViewer delivered 100% of the Aurizon objectives. It has also enabled greater efficiencies in ballast management and replenishment.

Client Testimonial

“Using this technology, Aurizon has been able to implement a cost-effective and accurate method for scoping future ballast drop locations.” 

James Freestone – Infrastructure Maintenance Engineer – Aurizon

“Automating the ballast inspection process is one of many use cases that can turn digital twin information into an invaluable data source within the inspection and maintenance program.” 

Paul Mullins – GM & VP APAC Cordel