Building on earlier technical enhancements, the roll-out of driverless operation under the AutoHaul project is already starting to deliver capacity and efficiency benefits that have more than justified a decade of development, Rio Tinto’s Shaun Robertson tells Chris Jackson.
A year after Rio Tinto Iron Ore operated its first driverless train on its extensive Pilbara rail network in Western Australia, the company is now running 90% of its heavy haul trains in unattended mode, reports the group’s Principal Advisor, Port & Rail Infrastructure, Shaun Robertson.
The AutoHaul project took almost a decade to complete, and posed a range of challenges to the mining group and its supply sector partners. However, Robertson believes that its successful implementation has established a platform to support a further 10 to 20 years of growth, and he reports that the efficiencies being achieved have made the payback on the investment ‘very attractive’.
As the world’s second biggest iron ore supplier, Rio Tinto expects to produce between 320 and 330 million tonnes this year. That is slightly down on the 338 million tonnes shipped in 2018, as a result of ‘operational problems’ at its Brockman mines earlier this year, but it is still a seven-fold increase on the volumes being shipped as recently as 2000.
The dramatic ramp-up in production over the past two decades has put pressure on Rio Tinto’s heavy haul railway, which now connects 16 mines and 12 loadout points to four ports. The 1 700 track-km network has been welded together from two formerly independent operations: the Hamersley Iron Railway running southeast from Dampier to Mount Tom Price, West Angelas and Hope Downs, and the former Robe River Railway running southwest from Cape Lambert to Mesa A and Mesa J near Pannawonica. The two routes intersect at West Creek Junction to form a broadly X-shaped network. The longest haul from Hope Downs to Cape Lambert is around 500 km, with trains typically completing the 1 000 km round trip in around 36 h.
Today the operator deploys a fleet of 224 diesel locomotives and 13 000 wagons, operating as 57 trainsets. The entire network is controlled from the company’s Integrated Operations Centre in Perth, which also oversees a wide range of automated mining operations.
Productivity imperative
Robertson notes that it took Rio Tinto and its predecessors 30 years to mine its first billion tonnes of ore, but the second billion took 10 years and the third just six. Now the company is extracting a billion tonnes every three years. This has required ‘a step change in productivity’ from the rail network.
He emphasises that AutoHaul is just the latest step in an ongoing cycle of technical innovation, building on research and experience shared through the International Heavy Haul Association. Coded track circuits and CTC were initially introduced to eliminate lineside signals and facilitate single-driver operation. Axleloads were increased from 32 to 36 tonnes in 2008, and several sections of the network have now been double-tracked. Traffic management was transferred to the IOC in 2010, and ECP braking introduced in 2013.
According to Robertson, there were three main objectives behind the decision to adopt driverless operation. The first was to improve productivity and increase capacity by reducing cycle time and eliminating stops for driver changes. More predictable operation would also reduce operating costs, both from savings in fuel consumption and from a smaller establishment of drivers.
The third big benefit came in reducing the safety risk, with the upgrading of level crossings, improved safety processes, condition monitoring and diagnostic systems. However, he says the biggest single safety improvement has come from a significant reduction in the risk of road accidents for drivers travelling by car to and from remote crew change points.
Robertson is keen to point out AutoHaul has been a ‘partnership development’, involving ‘thousands of people’ at seven locations in four countries. Ansaldo STS (now Hitachi Rail STS) supplied the signalling and communications, based on an ETCS Level 2 architecture, while Wabtec provided the ECP braking and New York Air Brake the Driving Strategy Engine as an evolution of its Leader driver advisory system.
Development timeline
Automation had been under consideration since the 1990s, Robertson explains. As various enabling technologies such as ATP and driver-only operation were phased in, initial studies were undertaken in 2006-07, leading to a ‘proof of concept’ trial on the Paraburdoo line.
The project was briefly put on hold as a result of a market downturn in 2009-10, but a full feasibility study was completed in 2011. The next step was to open consultation with the rail operations team and the unions, in order to garner feedback and emphasise that AutoHaul was about increasing productivity rather than reducing headcount.
A major step forward was the conversion of the locomotives and rolling stock to ECP braking during 2013-14, helping to improve train handling. Robertson points to the logistics challenge of cycling 224 locomotives and 13 000 wagons through the railway’s workshops for retrofitting at a time when the railway was running at full capacity to cope with a 30% jump in production from 230 to 312 mtpa.
Installation of ETCS Level 2 began in 2014, with the construction of the radio base stations, linked by an optic fibre communications network. The new signalling was ready for operation by 2016. With testing and commissioning on the main line running in parallel with the final phase of installation, AutoHaul was broadly complete by mid-2017.
The next step was a period of ‘attended operation’ during 2017, where trains ran in automated mode but with drivers on board to supervise and intervene where necessary. Once the necessary regulatory approval had been obtained, the first driverless train operated from Tom Price to Cape Lambert on July 10 2018 — a date that Robertson describes as ‘a momentous occasion’.
The subsequent ramp-up was relatively rapid, and by February 2019 Rio Tinto had achieved 90% autonomous operation with 51 out of 57 trainsets operating in unattended mode. The other six used on the Robe River line have not been automated because of the limited residual life of the rolling stock.
Multiple subsystems
AutoHaul integrates a number of onboard and wayside subsystems which are brought together by a data radio network that also supports the ETCS Level 2 train control.
Onboard systems include ATP-DIVA automatic train protection and ATO-C, which controls the train in line with the instructions of the Driving Strategy Engine. Lineside subsystems include the radio base stations and their connections to the existing interlockings. A Vital Safety Server in the operations centre manages train movements in conjunction with the RBCs and the ATP, while a separate Automation Server runs the ATO systems.
All level crossings have been upgraded, with the active crossings being fitted with obstacle detection systems as well as flashing lights and barriers. CCTV cameras switch on automatically as a train approaches the crossing, allowing train controllers in the IOC to monitor the area visually. In the event of an obstruction being detected, the movement authority of an approaching train is automatically cut back to bring the train to a stand before the crossing. Once the obstruction has been cleared, a new movement authority is issued so that the train can proceed.
Every locomotive has been fitted with an onboard collision detection system and video recorder. If the impact detector is triggered, the controller can remotely review the recorded images before determining whether a train should be allowed to proceed.
Hot axlebox, hot wheel and dragging equipment detectors have been incorporated into the AutoHaul system, as there is no longer a driver able to respond to any alert. New protocols include an automatic stop if dragging equipment is detected, while in the event of a hot axlebox the train will be restricted to 40 km/h until the next defined stopping point.
Because the brownfield installation was being undertaken on a busy railway, Rio Tinto decided that the new signalling should be installed as an overlay on the legacy equipment. A changeover switch in the loco cab allowed drivers to select legacy mode — where the movement authority is transmitted through the coded track circuits — or enhanced mode using the Level 2 movement authorities issued by radio.
Automatic operation
Rio Tinto decided from the outset that automatic operation should be limited to the main line and mine areas. Trains operating at the ports and coastal yards still have a driver in charge. Drivers are also deployed to take trains through the loading station at two mines, and to provide banking assistance on steep grades serving three mines.
Once a train has been prepared for the trunk haul, the set-up driver contacts the control centre, establishes a data connection and switches over to automatic operation. The driver then leaves the cab and enables the ATO using an external switch, leaving the control system to dispatch the train at the appropriate time. To ensure safety, locomotives operating automatically display blue lights on the roof, while small green and red lights adjacent to the cab steps indicate to ground staff whether or not it is safe to board.
The onboard equipment provides for four operating modes:
Passive — trains are still driven manually, but using the new ATP/VSS signalling and communications network;
Driver Assist — manual driving, but with advisory prompts from the DSE to optimise efficient operation;
ATO-Attended — trains driven by AutoHaul, but with a driver onboard to supervise and intervene where necessary;
Autonomous — fully driverless operation with no crew on board.
Rio Tinto has redeployed some drivers to form ‘rapid response’ teams strategically located around the network. These are ready to intervene if a train should be brought to a stand on the main line, checking for any faults, making minor repairs and liaising with the control centre to reset the ATO, or driving the train back in manual mode.
One critical issue has been to ensure the safety of people working on the track while automated trains are operating. Robertson points out that on-track staff have been provided with electronic protection for 20 years, and this has been integrated into the AutoHaul control system. Satellite positioning systems are also provided as a back-up, giving a second method of tracking the location of both trains and staff.
Early results
AutoHaul has already started to deliver operational benefits, with Robertson emphasising that the system is delivering much more consistent performance. For one given section with an average run time of around 25 min, he says the variation between drivers was typically ±5 min, whereas the variation for ATO has been just 30 sec.
Because the Driving Strategy Engine is able to ‘learn’ best practice from different drivers, and apply the optimum train handling methods for each section of line, cycle times have become much more predictable, he explains. This has provided the train controllers with a platform to optimise scheduled operations, for example ensuring that trains meet consistently at the passing loops on the remaining single-track sections.
Commenting that ‘data volumes are off the chart’ already, Robertson suggests that further analysis of the train running and condition monitoring information could deliver a great many incremental benefits, as well as facilitating a move to proactive maintenance strategies based on live data.
‘AutoHaul technology has revolutionised how we run our railway’, he confirms. ‘It has unlocked opportunities we’d not dreamed of a decade ago, and set us up for the future.’
This article first appeared in the November 2019 issue of Railway Gazette International magazine.