Train the crew on the faults they have never seen.
OTS runs the reference model as a training rig. Change one input and watch the whole train follow, minute by minute. Or inject a fault and let the trainee work out from the trends what broke.
MVPEvery figure above is counted in the minimum viable product: the amine train modelled for gas sweetening and post-combustion carbon capture.
What it does
Pick one input — feed rate, lean amine, reboiler steam, a temperature set-point — give it a new value and a time to get there. The whole train follows on the same model that Run uses. Cause is known; the lesson is the effect, and how long it takes.
Foaming, fouling, poor stripping, filter plugging and more. Each fault ramps in with a set severity and timing, and the trainee has to read the trends and name it. Effect is given; the lesson is the cause.
Each injected trend is labelled by how it was made: calculated by the physics, approximated, drawn from the fault signature, or replayed from a real episode. A drawn line is never presented as a calculated one. Four faults the model cannot reproduce are left out and say why.
Faults the unit has actually had can be replayed from its own history, so a crew trains on the episodes their plant produced rather than on a generic script.
How it works
Change to teach effects, Inject to teach diagnosis. The two stay separate on purpose.
For Change: the input, the new value and the ramp time. For Inject: the fault, its severity and its timing, or a real episode to replay.
The calibrated reference model steps the train forward minute by minute, with each stream responding at its own measured speed.
The trainee watches the tags an operator is judged on: treated-gas H₂S and CO₂, solvent loading, throughput and the key temperatures.
The instructor can compare the run with a healthy baseline and show which signs pointed to the answer.
Where it applies
The platform is focused on acid gas removal — gas sweetening and post-combustion carbon capture — because that is where the model is calibrated and where the fault library is real. The method underneath is not amine-specific: a physics reference model, residuals against expected-healthy, fuzzy diagnosis over graded indicators, and costed consequences. Extending it to another unit means new datasheets and a new calibration, not a new platform.
The platform runs on a real amine train with thirteen days of one-minute history behind it. Nothing on these pages is a mock-up.
Open the platform