A $750,000 Lesson in Crush Risk from Removed Guarding

In January 2026, WorkSafe WA confirmed a $750,000 fine against MLG Oz Limited after a diesel mechanic suffered a serious crush injury at the Mungari gold mine near Kalgoorlie.

The incident did not involve an explosive failure or a rare mechanical fault, it involved a running conveyor, removed guarding, no enforced isolation, and a task that had quietly drifted into normal work without being treated as high risk.

This episode is another example of how preventable injuries continue to occur in high-consequence environments.

During my time as an Inspector, I saw this pattern repeatedly. Guards removed to address production jam points. Access panels left off around known fault areas. Temporary fixes that became permanent because they helped keep things moving. They were almost always implemented with good intentions and rationalised as low risk because nothing had gone wrong yet. It was concerning how often these conditions were treated as acceptable.

What Actually Went Wrong?

A senior mechanic was adjusting conveyor belt tracking on an operating stacker. The guarding that had been installed during commissioning had been removed, and no task-specific safe work procedure existed for clearing or adjusting while the plant was energised. No lockout or isolation was applied.

Clothing was caught in a nip point, and the worker was crushed, sustaining a fractured humerus.

The Real Failure Was Systemic

It is tempting to frame this as a one-off lapse or an experienced worker taking a shortcut. That narrative may seem comforting, but it is incorrect.

The failure sat in predictable places that I have seen time and again during inspections.

First, guarding removal was tolerated. Once guards are removed and the plant continues to operate, the organisation implicitly accepts the risk. No one may say it out loud, but the decision has been made.

Second, task creep set in. Adjusting belt tracking while running became normalised because it worked yesterday and nothing bad happened. This is classic outcome bias, judging the safety of a task by its previous result rather than its inherent risk.

Third, isolation became optional. Lockout existed on paper but not in behaviour. This is where heuristics kick in, with experienced workers relying on mental shortcuts like “I’ve done this a hundred times” or “I’ll just be quick.”

Fourth, supervision and verification were absent. No one was checking whether the controls designed on drawings and procedures still existed on the plant itself.

Why Experience Did Not Save This Worker

Experience does not protect individuals from hazards such as stored energy, pinch points, or rotating machinery. Often, it increases exposure when systems rely on judgement instead of physical (engineered) controls.

Skilled workers are often the ones asked to just tweak it while it is running. Familiarity bias creeps in, the task feels routine, the risk feels manageable, and vigilance drops.

When workplaces rely on memory, discretion, and goodwill rather than engineered barriers, they are placing unrealistic expectations on human performance.

Human Factors Were Always in Play

This incident sits squarely at the intersection of cognitive bias and system design.

Normalisation of deviance allowed unsafe conditions to persist. Optimism bias led people to believe that serious injury was unlikely. Time pressure and production focus narrowed attention to the immediate problem rather than the latent risk.

None of these are personal failings. They are known human limitations. Safe systems are built by acknowledging those limitations and designing around them, not by expecting people to overcome them through willpower alone.

Practical Controls That Actually Prevent Recurrence

Start with guarding discipline. If guarding is removed for maintenance, the plant does not run. No exceptions, no productivity arguments, no grey areas.

Enforce isolation by design. Isolation points must be obvious, accessible, and physically applied before access. Group lockboxes and personal locks are not bureaucracy, they are controls that compensate for human fallibility.

Treat non-routine tasks as high risk by default. Belt tracking, clearing build-up, and fault-finding around conveyors require task-specific procedures and permits, even if they have been done hundreds of times.

Design out the temptation. If a task is commonly performed while the plant is running, redesign the system so it physically cannot be done that way. Engineering controls will always outperform administrative intent.

Verify controls in the field. Supervisors must physically confirm guarding, isolation, and access arrangements. Paper compliance without verification creates a false sense of security.

The Leadership Test

This case is not about rules. It is about leadership under pressure.

Someone knew the guarding was gone. Someone knew the stacker was running during adjustments. Someone accepted that risk, likely believing they were making the best decision at the time.

Safety leadership is the willingness to stop work when controls are missing, even when production pressure is loud and the task feels routine.

Systems fail quietly long before people get hurt. By the time an injury occurs, the organisation has already made many small decisions that allowed it.

The Takeaway

Crush injuries around conveyors are not accidents. They are the predictable outcome of tolerated risks, removed controls, and systems that fail to account for human behaviour.

If you want to prevent the next one, stop relying on experience and good intent, and start building systems that make unsafe work difficult, awkward, or impossible.

The fine was $750,000. The true cost was nearly the loss of a life.