Is Tom's new ute reliable or just capable?

Bill, Tom and the new ute

Tom arrived at Bill’s place in a brand-new ute.

Properly new. Clean tray, clean tyres and enough electronics inside to make Bill suspicious before he had even opened the door.

Dog walked around it once, sniffed a tyre and sat down.

“What do you reckon?” Tom asked.

Bill looked at the ute. “Very shiny.”

“It’ll tow three and a half tonnes.”

“Will it?”

“It says so.”

“Ah.”

Tom knew that particular “ah”.

“It’s in the specifications, Bill.”

“I’m sure it is.”

So they hooked up Tom’s tandem trailer, loaded the little tractor and drove down the road. The ute pulled away cleanly, climbed the hill out of town without trouble and stopped exactly where Tom asked it to.

Back at Bill’s place, Tom was looking pleased.

“There. Three tonnes, near enough. Towed it perfectly.”

Bill nodded. “So it can tow three tonnes.”

“That’s what I’ve been saying.”

“No. You’ve been saying it’s a three-and-a-half-tonne towing ute. What you’ve shown is that it can tow this trailer, on this road, on a cool morning, with an empty tray.”

Tom stared at him.

Dog opened one eye.

Bill shrugged. “You’ve demonstrated capability.”

A month later Tom arrived again.

The ute was no longer quite so shiny.

“I’ve been testing your bloody robustness,” he said.

“My robustness?”

“You know what I mean.”

Tom had towed the tractor on a hot day, hauled fencing gear up a rough road, loaded the tray as well as the trailer and driven into a fairly unpleasant westerly.

“So?” Bill asked.

“Still works. Transmission gets warmer on long climbs. Suspension doesn’t like a heavy tray and trailer at the same time. Fuel economy is offensive.”

“Sounds like a ute.”

Tom had also changed how he loaded it. Heavy gear went in the right place, tyre pressures got checked properly, and he stopped carrying half the workshop just because there was room.

Bill nodded. “Now you’re learning what it will tolerate.”

Tom frowned. “I already knew it could tow.”

“Capability.”

“And now I know it can still do it when conditions change.”

“Robustness.”

Dog wandered over and inspected the pile Tom had removed from the tray: toolboxes, fencing gear, fuel, a compressor and a generator.

Tom looked at him.

“All right. It was a bit much.”

Dog sat down.

Bill leaned against the tray.

“What if one tyre is low?”

“More heat. Worse handling.”

“Trailer brakes poorly adjusted?”

“Longer stopping distance.”

“Hot day, long hill, headwind?”

“Cooling system and transmission work harder.”

“And what if the driver keeps his foot buried because the brochure says three and a half tonnes?”

Tom shook his head. “The driver would be an idiot.”

Dog immediately looked at Tom.

“Don’t start.”

Bill smiled. “That’s the point. The ute isn’t just an engine with a tow rating. It’s the engine, gearbox, brakes, tyres, cooling system, trailer, load, road, weather, maintenance…”

Tom sighed. “And driver.”

Dog sneezed.

“Especially the driver,” Bill said.

A year later the ute had done nearly forty thousand kilometres.

It had towed tractors, firewood, fencing gear and several loads that Bill considered optimistic. Things had gone wrong too. A trailer plug failed, a tyre delaminated, the trailer brakes needed adjustment and one sensor produced a warning that turned out to be nothing much.

The ute had never left Tom stranded.

One afternoon, while unloading posts, Tom said, “So now I can say it’s reliable?”

Bill nodded. “Now you’ve got some evidence.”

“Forty thousand kilometres is some evidence.”

“It is. Reliability takes time.”

Tom put down the post.

“So capability was the first tow. It proved the ute could do the job.”

“Yep.”

“Robustness was finding out whether it could still do the job when the load, heat, road and other conditions changed.”

“Yep.”

“And reliability is what I learn after it’s done all that repeatedly over time.”

“Pretty much.”

Tom thought about that.

“So a successful test doesn’t tell me much about reliability.”

“Not by itself.”

“And a tow rating doesn’t tell me how robust the whole setup is.”

“Not without knowing the conditions.”

Dog wandered over and sat beside the ute.

Tom looked at him. “What if I tow three and a half tonnes everywhere, flat out, in forty-degree heat, with a flat tyre and no trailer brakes?”

Dog looked at Tom, then slowly walked over and sat beside Bill.

Bill nodded.

“Dog’s completed the risk assessment.”

Capability tells you the thing can do the job under a particular set of conditions.

Robustness tells you how well it keeps doing the job when the load changes, something wears, the weather turns or part of the system stops behaving perfectly.

Reliability is what repeated real use eventually tells you about whether that robustness was actually there.

And the system is rarely just the machine. It includes the load, the road, the maintenance and, unfortunately, the bloke holding the steering wheel.

Dog reckoned that last bit was obvious.

Robustness Beats Capability

I have spent a lot of time looking at the proposed biosolids processing plant at Glan Devon. The proposal is unusual: biosolids would be brought to the site, dried using heat produced by burning prepared non-recyclable waste, and then processed into a fertiliser product.

Much of the discussion around the project has focused on the waste incinerator and its emissions. That is understandable. The application includes an air-quality model intended to show that predicted concentrations around the site remain below relevant environmental criteria.

The longer I have looked at the proposal, though, the less convinced I am that emissions at one selected operating point are the right place to start. The bigger question is robustness.

By robustness I mean something fairly simple: can the whole plant keep operating within its environmental limits when the feed changes, equipment deteriorates, loads move around, something trips, the plant starts or stops, and all the other ordinary things that happen in a real industrial operation begin to occur?

That is a much harder question than asking whether the technology can produce an acceptable emissions result under a particular test condition.

Capability is not robustness

Engineering studies have to simplify reality. A plant is designed around assumed feed rates, moisture levels, temperatures, equipment capacities and operating conditions. Engineers then calculate what should happen if those assumptions are met.

There is nothing wrong with that. The problem comes when evidence of capability starts being treated as evidence of robustness. Just because a plant can do something under some conditions doesn't mean it will do the same thing under all reasonably expected conditions.

An emissions test can show that an incinerator is capable of producing low emissions under the conditions tested. A dryer test can show that a dryer can handle a nominated amount of material. A design calculation can show that a pollution-control system should achieve a certain performance.

A commercial plant has to do more than reproduce a successful test. It has to keep performing while conditions move around.

Waste changes. Moisture changes. Equipment fouls and wears. Pumps and fans stop. Reagent systems block. Sensors drift. Operators make adjustments. Plants start, stop and restart. Sometimes two parts of the plant move in opposite directions at the same time.

A robust system absorbs that variation without environmental performance leaving the acceptable envelope. A fragile system may work extremely well when everything is close to the design case and become much less predictable when it is not.

Real incinerators show a big difference in robustness

This became much clearer when I looked at Environment Agency records for operating waste incinerators in England.

For four years where the public data allow a reasonably clean comparison of air-emissions compliance, roughly half of the facilities recorded at least one air-emissions permit-condition non-compliance during the year. That does not mean half the plants were constantly running over their limits. Some events were short, different pollutants have different compliance rules, and abnormal-operation provisions complicate the picture.

The useful point is that these were mature, regulated plants with pollution-control equipment, continuous monitoring and experienced operators. Environmental departures still occurred often enough that they cannot reasonably be dismissed as freak events.

The more interesting result appeared when I followed the same 23 plants through all four comparable years. Seven had no recorded air-emissions non-compliance in any of those years. Nine recorded non-compliance in at least three of the four.

So I started looking for the obvious explanation. Perhaps the better plants used different furnace technology, were much newer, burned a different type of waste or simply had fewer combustion lines.

None of those explanations worked particularly well.

The strongest visible pattern was organisational. Six of the seven zero-record plants shared the same wider operator, including several closely related facilities operating under common systems and procedures. That does not prove that the operator caused the better result, and one plant operated by the same company appeared in the recurrent group.

But it strongly suggests that the important thing is larger than the furnace.

Maintenance practices, operator experience, alarm settings, shutdown rules, feed acceptance, equipment redundancy, preventive maintenance and the way knowledge moves through an organisation may all influence whether an ordinary plant problem becomes an environmental breach.

That is what I mean by robustness being a property of the whole operating system.

The failures were mostly ordinary industrial problems

The reports from plants that repeatedly recorded problems are revealing because the causes are so ordinary. Wet or difficult waste, unstable combustion, blocked lime or urea systems, filter failures, failed pumps, fan trips, power faults, interruptions to waste feeding, monitoring problems and startup or restart conditions all appear.

These are not extraordinary disasters. They are the things complex industrial plants have to deal with over years of operation.

A robust plant does not somehow avoid every disturbance. It contains them.

A reagent pump fails, but the plant has enough warning, redundancy or operating margin to prevent the failure from becoming an emissions event. A troublesome waste load arrives, but the plant rejects it, blends it, reduces feed or adjusts combustion before the problem propagates. A fan trips, and the plant moves safely towards shutdown rather than struggling on in an unstable state.

This also explains why average emissions can be misleading. Two plants may both have very low average emissions, yet one repeatedly stays inside its permit limits when disturbed while the other occasionally crosses them.

The averages can look almost identical. The difference appears in the difficult part of the distribution — the upper tail, the extremes, the times when things are not normal, the startup, the fault, the upset and the recovery.

Glan Devon is a coupled system, not just a furnace

This is particularly important at Glan Devon because the proposal is not simply a waste incinerator.

The proposed operation uses heat from the incinerator to dry biosolids. That creates a coupled industrial system in which both sides can vary.

The waste fuel can change in moisture, heating value, density, chemical composition and how it behaves after preparation. The biosolids can also vary in moisture, solids content and throughput.

Those variations affect different parts of the plant.

If the incoming biosolids are wetter, the dryer needs more heat to produce the same amount of dried material. If the dryer slows down or trips, its demand for heat changes. If the waste fuel has a lower heating value, the combustor may produce less useful heat for the same feed rate. If heat exchangers foul, the relationship changes again.

Somewhere in the plant, those differences have to be absorbed.

There may be perfectly sound engineering solutions: buffers, storage, automatic load control, changes in feed rate, heat rejection, shutdown sequences or other control strategies. I am not assuming the plant cannot manage these interactions.

I am saying that how it manages them is part of the environmental robustness case.

The incinerator cannot sensibly be assessed as though it operates in isolation while the dryer quietly accepts whatever heat happens to arrive.

The coupling also makes startup and shutdown more interesting. The furnace, dryer, heat-transfer system and pollution controls do not necessarily become stable at exactly the same time. If the dryer stops suddenly, the thermal side has to respond. If the combustor trips, the dryer has to respond. A restart may move both systems through a sequence of conditions that looks quite different from normal operation.

Those whole-of-plant transient states belong in the environmental story.

The current evidence is much stronger on capability than robustness

The Glan Devon air assessment relies on Xetrov emissions information derived from a test using polyurethane dust at about 70 per cent load, with the measured results scaled upward to estimate full-load emissions.

That test is useful evidence. It demonstrates that the combustion system achieved particular emissions performance under that test configuration (Like the towing capacity of Tom's ute!)

But the proposed operating fuel is not polyurethane dust. It is heterogeneous non-recyclable waste. The actual operation also includes biosolids drying, heat transfer, pollution controls, process-air systems and the controls tying all of those components together.

So the test tells us considerably more about capability than it does about robustness.

The question is no longer simply, “Can the combustor produce low emissions?” The question is how the complete Glan Devon operation behaves as waste properties change, biosolids moisture changes, heat demand changes, equipment performance changes and the plant moves through different operating states.

That becomes more awkward because the proposed trial work is itself intended to help refine feed blends, operating conditions and controls. There is nothing unusual about trials in developing an industrial process, but the sequence deserves scrutiny. If the trial is still needed to determine how parts of the integrated system should operate, then some of the behaviour needed to establish environmental robustness is still being discovered.

The air model adds another uncertainty problem

This is where I think the air-quality discussion also needs to be more careful.

There are two separate modelling problems.

The first is the plant itself: what does the complete operation actually emit as feed, load, equipment condition, dryer demand and operating state change?

The second is the atmosphere: once those pollutants leave the stack, where do they go and what concentrations occur at surrounding locations?

The Glan Devon assessment uses CALMET and CALPUFF for that second task. These are mathematical models of the atmosphere. CALMET constructs a three-dimensional meteorological field from input data and model relationships, while CALPUFF then represents how pollutant puffs move and disperse through that modelled atmosphere. These are the models expected by the regulator - but they are still models.

Neither is simply observing what the real atmosphere will do.

They contain assumptions, empirical and semi-empirical relationships, choices about terrain and turbulence, meteorological inputs and other model settings. Different credible choices can produce different predictions.

At Glan Devon there is also a particular issue with the meteorological basis. The modelling used one year, 2021, and CALMET was reportedly operated in No-Observations mode, without an on-site meteorological dataset being used to anchor the local wind field.

So putting a precise emissions number into CALPUFF does not somehow turn the future environmental impact into a known quantity. The uncertainty carries through from one model to the next.

There is uncertainty in what the plant will emit, and there is another layer of uncertainty in how the model represents the atmosphere carrying those emissions towards neighbours, Barkers Creek, Mondure Crossing and other surrounding areas.

A result below a guideline is therefore evidence from a particular modelled case. The stronger question is whether the conclusion remains acceptable when reasonable variation in both the industrial source and the atmospheric model is considered.

That is another robustness test.

“Worst case” needs to be earned

The same problem applies to claims of “worst case” emissions.

Calling a source term worst case does not make it one. To demonstrate that something is a reasonable bound, you first need to understand the range of conditions being bounded.

For Glan Devon that could include waste moisture and composition, calorific value, particle characteristics, feed rate, combustion air, load, dryer demand, startup, shutdown and pollution-control condition.

Different pollutants may also respond differently to the same process change. Research on direct cyclone combustion, for example, has shown cases where changing combustion air reduced carbon monoxide while increasing nitrogen oxides.

There may therefore be no single physical plant state that is “worst” for every pollutant at once.

The point is not that modelling adverse cases is impossible. It is that a worst-case claim needs a demonstrated relationship to the operating envelope rather than simply a large number selected from available test results.

Conditions are not the same as robustness

One response to uncertainty is to impose conditions. Set emission limits, require monitoring, specify odour requirements and tell the operator what must happen if a limit is exceeded.

Those controls are necessary, but they serve a different purpose.

A condition saying “do not exceed X” defines the boundary. It does not demonstrate that the plant has enough operating margin, control, redundancy and understanding to reliably remain inside it.

There is a significant difference between using conditions to manage the remaining risk of a system whose behaviour has already been reasonably demonstrated and using conditions while important parts of the actual operating behaviour are still unknown.

In the second case, some of the learning occurs after the plant begins operating.

That is where the issue becomes more than an engineering argument.

Who carries the cost of finding out?

Before approval, uncertainty should mainly be the applicant's evidentiary problem. The applicant is seeking permission to introduce a new industrial activity, so the applicant should demonstrate why it is acceptable.

After approval, the balance shifts.

If an unexpected odour occurs, a neighbour may experience it before the regulator hears anything about it. If an intermittent plume affects Mondure Crossing, someone fishing or swimming there may simply leave. If a plant disturbance produces an emissions event, the regulator may have to reconstruct what happened later using monitoring records, complaints and weather information.

The operator has the plant data, engineers, controls and maintenance records. People outside the fence generally do not.

Queensland's environmental regulator has enforcement powers, so this is not an argument that an operator can simply ignore conditions. The issue is more basic: if important uncertainty is left unresolved until operation, some of the cost of discovering whether the original assumptions were wrong can fall on people who did not choose to take that risk.

Approval does not make uncertainty disappear.

It can shift where that uncertainty is eventually resolved.

A better question for Glan Devon

I do not think the sensible approval question is simply whether the Xetrov combustor can produce low emissions. There is evidence that it can under selected conditions.

Nor is it enough to ask whether one CALPUFF run produces numbers below environmental criteria. The atmospheric model has its own assumptions and uncertainty, just as the plant model does.

The broader question is whether the complete Glan Devon system has been shown to remain environmentally acceptable across the range of conditions it can reasonably be expected to encounter.

That includes the waste fuel, biosolids, dryer, heat transfer, combustion controls, pollution controls, monitoring, operators, equipment faults, startup, shutdown and the atmospheric conditions used to predict off-site effects.

Capability tells us that acceptable performance can be achieved.

Variability tells us how the system moves around.

Robustness tells us whether those movements remain contained.

Reliability is what years of real operation eventually tell us about whether the robustness was actually good enough.

For a mature industrial system with well-known inputs and years of operating history, much of that evidence may already exist. For a new combination of technology, variable waste fuel and coupled biosolids drying, it needs to be demonstrated much more carefully.

That is the gap I still see at Glan Devon. The public material provides evidence about selected operating conditions and selected atmospheric modelling conditions. I am not yet convinced it demonstrates the robustness of the whole system and the environmental conclusion when real industrial and atmospheric variability are allowed back into the picture.

© 2026. All rights reserved.