Knowledge
People trained in the subjects this rests on: process technology and thermodynamics, absorption and solvent chemistry, machine learning and statistics. The kind of knowledge you can derive from first principles rather than look up.
About QemOPS
Powering Decarbonization Through Operational Wisdom
Process technology in the energy field, artificial intelligence, data science, and operations and maintenance. Four disciplines that rarely sit at the same table, and a platform that only works where they meet. What it is for is simpler than what it is: less energy burned, less carbon released, and a credible road to net zero on units that already exist.
The people behind it
Four things, and they are not interchangeable.
They are usually spoken of as one word. In a team, which one is missing tells you exactly how the work will fail.
Expertise
Engineers who have designed, commissioned and troubleshot gas treating and capture units, working with people whose profession is the science behind them. Expertise is knowing which method fits a question — and which one only looks impressive.
Experience
Years on running plants: night shifts, turnarounds, and faults that had to be settled before the morning meeting. Years, too, of delivering software other people have to trust, which teaches you what not to promise.
Innovation
People who work across four fields that rarely meet — process technology, AI, data science, operations and maintenance — and build what none of them delivers alone. That habit is the team, not a line on a slide.
What the founders bring
Four disciplines, and the join between them.
Any one of these alone builds something familiar: a simulator nobody calibrates, a model nobody can explain, a dashboard nobody acts on. The product is the join.
The product is the small square in the middle.
Process technology, in energy
Gas treating, carbon capture and the utilities around them: absorbers, regenerators, solvent loops, exchanger trains and the steam that drives all of it. This is what fixes the structure of the model — the equations are engineering, and they are never fitted to make a curve look better.
Artificial intelligence
Used where it earns its place and nowhere else. The diagnosis is physics and evidence, not a classifier; the language model sits on top as an assistant that answers from the plant knowledge base and refuses when the value is not in front of it. AI that cannot show its reasoning has no business in a control room.
Data science
Calibration, residuals, indicator grading, back-testing and the discipline of keeping training data separate from the data an answer is produced on. It is also what tells you when a fit has no skill — which is how the forecast came to stop at six hours instead of guessing at days.
Operations management and maintenance
What a shift can actually do with an answer: when the window for a cleaning is, what it costs per hour to wait, which findings are worth a work order and which are worth watching. Every diagnosis carries a price for this reason — ranking work is a maintenance decision before it is an engineering one.
What it is for
Lower energy, lower carbon, and an honest route to net zero.
We are not going to claim a tonne of CO₂ we have not measured. What we will claim is this: on a running unit, most of the available reduction is not a new technology. It is waste nobody can currently see.
- 1
Find the energy nobody is charged for
A fouled exchanger, a degraded solvent or an over-circulating pump does not raise an alarm. It raises reboiler duty, quietly, for months. The model computes what the steam should be for the conditions you are running, so the waste has a number on it instead of being absorbed into the monthly total.
- 2
Put the carbon next to the euros
Excess steam is excess fuel, and excess fuel is emissions. The consequence costing already carries an emissions bucket beside off-spec gas, amine losses, reliability and throughput, so an energy fault and its greenhouse gas footprint are read off the same screen rather than reconciled a quarter later.
- 3
Fewer failures is fewer emissions
Unplanned shutdowns, flaring, venting and solvent replacement are all carbon events as well as reliability events. Catching a fault while it is still a drift is the cheapest abatement available on most units, and it needs no capital project to unlock.
- 4
Net zero is an accounting problem before it is a technology problem
A site cannot reduce what it cannot measure, and most of the measurement it needs already exists in the historian. Our part of net zero is unglamorous: make the waste visible, price it, and give the crew the evidence to act this week rather than in the next capital cycle.
Where it came from
Science that survived contact with a real plant.
Four origins, and one thing they are all pointed at.
Industry and academia
Years on operating gas treating and capture units, alongside people whose profession is the science of it. The plant people know what breaks and when. The academics know why, in equations that hold outside the range you happened to observe.
Knowledge first, software second
The unit was documented before a line of the platform was written: 26 equipment items, 161 tags, 22 faults, 19 root causes, 54 indicators, 247 causal edges. The software is a way of using that. Without it there is nothing to compute.
Reviewed, not just built
The indicator set went through expert review and came back changed. One fault rule was removed entirely — it was reading the operator’s own antifoam dose back as evidence of foaming. That review is why we trust the output.
Tested against a real plant
Every figure on this site comes from a calibrated model of an operating amine train with thirteen days of one-minute history. Calibration is fitted on training data; diagnoses run on data the model has never seen.
A greener plant
Less energy burned, less carbon released.
Where we stand today
Early, and specific about it.
We are not going to describe one calibrated train as a global install base. Here is the actual position, and every line of it is countable in the delivered system.
- Calibrated on
- One amine gas sweetening train — absorber, regeneration, filtration and solvent loop
- Live capabilities
- Run · Diagnose · Prognose · Explore · Train · Gage · Ask
- In development
- Tools built to a customer’s question, on the same engine output
- Knowledge base
- 22 faults · 19 root causes · 54 indicators · 6 costed consequences
- Declared blind spots
- 12, published with what to watch instead
- Focus today
- Amine gas sweetening and post-combustion carbon capture
How we work
The rule the whole platform is built around.
- We will not print a number the data cannot support. The forecasting tool gives no days-to-failure figure, because the back-test showed the fit has no skill beyond about half a day. The assistant refuses to answer when the value is not in front of it, rather than estimating. Twelve blind spots are published instead of hidden.
- Every answer shows its reasoning. A diagnosis that cannot be checked is a diagnosis nobody should act on. Faults carry the indicators that fired, the ones that corroborate and the ones that would have exonerated them.
- The plant stays in charge. The engine diagnoses, forecasts and prices. It does not write to a control system, and we do not intend to change that.
If your unit looks like this one, we should talk.
The quickest way to find out whether the model fits your plant is to send us a tag list and a datasheet. We will tell you honestly either way.
Get in touch