The incontrovertible first law of engineering states that energy cannot be created or destroyed, but is transformed or transferred. This is also known as the first law of thermodynamics.
Then there’s Murphy’s Law, that states what can go wrong will go wrong, at the worst possible moment.
You can’t change the laws of physics. But the ancient Romans one-upped Murphy when they said praemonitus praemunitus, or “forewarned is forearmed”.
For Michael Gibson, Executive Director of GTE Group, this all translates to finding out what can go wrong before it does – in particular, before millions of dollars worth of automated mobile bulk-handling machines come crashing down around him.
These enormous automated stackers, reclaimers and ship loaders are critical links in the Australian mining industry’s distribution chain.
“They can be carefully balanced machines,” Gibson told create. “They have a huge counterweight on one side. They’ve got a boom on the other that’s loading or unloading product. And it’s all held together by stays that connect to the central mast. If one of those gets broken, the whole thing can collapse.”
Collisions can and do happen. Gibson said one crash involved a 1300 t stacker and a reclaimer in October. While the damage was limited to an inspection of the walkway, production was nevertheless halted for 36 hours.
“You can’t just press reset and restart,” Gibson said. “You have to identify the triggering issue and repair it. This will ensure the incident won’t happen again on the next shift.”
He added that historical data from the Pilbara region statistically shows that if 100 machines are operating across Australian sites, then there will be 2.14 incidents per year.
Onsite testing can uncover collision risks, but it’s a slow and costly process. This is why ore handlers only do it periodically.
“It was baffling to me that these machines – rated at between $300,000 and $1 million an hour in terms of revenue – need to be taken offline, two at a time, for at least 12 hours each, for manual testing,” Gibson said. “You were looking at something like $17 million in opportunity cost.”
It was a problem looking for a solution.
“One risk is a bug – either intentional or accidental. All of a sudden the site control system may go, ‘I don’t know what to do, I can’t calculate the distance between the two machines anymore’, and it throws an exception that gives a false negative. We could write tests into its memory. We could throw some numbers at it. But that’s only testing the software. There’s much more to anti-collision systems than that.”
The answer involves transplanting the algorithm, together with its sensor encoders, into a “digital shadow”. GTE Group’s trademarked and patent-pending AuditSense procedure directly stimulates an entire bulk handler’s ACS system – without the bulk handler.
Automated stackers, reclaimers and ship loaders get confused. It’s a tough environment, with extremes of heat and cold, and heavy use, and dust can get past a compromised rubber seal and into an encoding mechanism’s cogs and wheels. These are the distance counters and inclinometers that track a bulk handler’s position and orientation.
“That’s the reason why these machines will usually have multiple encoders in multiple positions,” Gibson said. “An onsite control system is always assessing these.”
But issues can be introduced through a simple upgrade. New hardware may not move or do the maths the same way, giving a different result. Sometimes, it gets asked to do something out of the ordinary.
And, despite being isolated, automated bulk loader operating systems are being “hacked” all the time. Malice is unlikely, Gibson said, but possible: an upset employee, a hostile competitor, even someone hoping to manipulate the market.
“Mostly, though, it’s the result of a quick fix.”
A stockpile may be overloaded or misaligned. So the machines have to reach a little bit further.
“It might be two in the morning. It might be a shift sparky, and they’ll go into the code and put in a bridge – a bit of bypass logic – that lets them ignore a boundary. It happens 40 to 150 times a year per machine.”
But, sometimes, that hack isn’t removed. “A few months or years later, those machines can collide with another. ‘Whoops, we forgot to take it out!'”
Then there’s onsite anti-collision testing, which involves deliberately putting multimillion-dollar stackers, loaders and reclaimers in a risky position.
“Sometimes they succeed and crash them into one another.”
And, like the late-night sparky, those doing the ACS test hacks may not properly clean up after themselves. “The mere fact of doing the test has introduced a risk. AuditSense removes that risk completely.”
Auditsense visualisations. Images: GTE Group
“Crashing a digital shadow has zero impact,” Gibson said. “If we crash them, we just see an unwanted overlap with the machines. That’s when we go, ‘Oh, how did that happen?’, figure it out, and advise the machine owners of what changes must be made to prevent a collision.”
Keeping the test as real as possible is the challenge.
“It’s very important that we don’t introduce any shortcuts to generating the machine position data. In the AuditSense world, we have physical representation.”
The ACS controller software enters a virtual reality environment. It just doesn’t know it. No code is hacked, no fake signal is injected directly into the data stream.
“We’re only simulating the interactions that cause its encoders to generate a signal. We actually have exactly the same units that are on the machines on site in the digital shadow.”
The sensors are physically manipulated by variable-speed motors.
“One might be for travelling up and down the rail. Another for measuring slew – the pivoting of a boom in a horizontal plane. And others will be for luff – pivoting up and down in a vertical plane.
“The encoders will interpret that input and send digital numbers through a whole lot of paths to inform the off-board anti-collision system of each machine’s precise position. So if we bypass all that, we’re not actually testing it.”
The encoders encode. The ACS logic and processor computes. It then sends instructions back to a virtual bulk handler to allow or disallow movement.
“In the real world implementation, if all those inputs don’t line up, then it will fail. It may judge that one encoder is an outlier – the rest add up just fine. Then it will just say, ‘I don’t care about that outlier encoder.’
“If more than one encoder doesn’t align, it might conclude, ‘I don’t know where I am, stop operation!'”
If the ACS output doesn’t match that intended by the system design requirements, AuditSense has done its job and found a fault. And though Murphy’s law holds true – what goes wrong happens in a virtual environment with far less dire consequences.
An iron ore reclaimer can shift up to 17,000 t/hour. At almost $160/t, a collision’s potential impact on revenue can be significant.
Each machine is more than 60 m long and about 25 m high. The loading boom is about 5 m wide. And the whole 3000 t automation moves along rails at about 0.5 m per second (1.8 km/h).
Newton’s second law of motion translates this into a whole lot of momentum.
“At that maximum speed, they will pull up in about one to two metres if they must stop in a hurry,” Gibson said. “You’ve got one reclaimer on a set of rails and another stacking machine on the next set of rails on the opposite side of a stockpile. But their booms can geometrically overlap, so they can collide.”
Testing their anti-collision systems generally involves taking two machines off the production line for up to 12 hours. Potentially, that’s up to $40 million worth of ore sitting still.
“Typically, though, machine availability is anywhere from 50-92 per cent.”
But any stoppage, either deliberate or accidental, represents a substantial impact on turnover. Digital shadows make this concern redundant, Gibson said.
“As AuditSense is being run, we view a flat-panel graphic. So, it’s not just spitting out numbers. We can visualise everything the controller tells the machine.”
Mining giant BHP uses AuditSense at its Jimblebar open pit iron ore mine, 40 km east of Newman in the Pilbara. And it’s rolling the technology out across its fleet of 42 automated bulk handlers.
These will now be tested annually instead of once every three years.
“We also have the ability to continuously monitor software onsite. As soon as the ACS-related site control system is changed, AuditSense initiates a collision-risk check”.
The GTE lab is a “plug-and-play” setup. The exact model and make of a bulk handler’s sensor encoders are paired with manipulators on an easily accessible board. The relay to its ACS is replicated on the same modular unit.
The operator software is a snapshot of that in use onsite.
“This is not a digital twin,” Gibson said. “It’s a digital shadow. There is no bidirectional communication path. It’s a shadow representation of what’s onsite.”
This is about security and safety. Each machine is photographically mapped. This is then converted into a 3D polygon model that can be manipulated through the same three axes of movement.
“We use a mathematical vector approach to determine how close the machines are to one another, which essentially means there is an ‘infinite resolution’ for the nearest distance calculation.”
Testing is done in real time – which is counterintuitive for most digital twin and shadow operators. But Gibson said it’s all about keeping the sensor input as real as possible. They must “move” at the same speed they do onsite.
“The integrity of the test result is utmost,” he said. “If we run it faster, all we’re doing is introducing a potential risk related to timing and related propagation delays, or lack thereof.”
The time benefit is already dramatic. “Customers at the moment audit their machines only once every one to three years. We can audit it every month, if they like. And at a huge cost saving, because their machines will still be producing while we’re carrying out the maintenance in the lab.
“Our ghosts crash into each other so your machines don’t.”
This article was originally published in the May 2025 issue of create with the headline “Sense-check”.