Now live: a hybrid process digital twin for amine gas sweetening — it names the fault and the root cause, not just the alarm. Explore the platform →
What-If
Find the Best Operating Point
Test the move before you make it.
Circulation rate, lean temperature, reboiler duty — every one of them is a trade, and the trades pull against each other. The reference model will tell you what the whole train does at a candidate operating point, and the cost model will tell you what that point is worth, before anyone touches a controller.
Enter a candidate set of boundary conditions on the plant diagram and the model returns the expected response at every monitored point: treated gas quality, column temperature profiles, exchanger duties, solvent inventory, pump behaviour. Not a lookup table — the same physics that produces the diagnosis.
02
Every trade is two-sided
More circulation treats better and costs steam and pumping. A hotter reboiler strips harder and shortens solvent life above 127 °C. Colder lean amine absorbs better and moves you towards hydrocarbon condensation. The model holds all of these at once, which is precisely what a spreadsheet and an experienced guess cannot.
03
An objective function that is already priced
Six consequences carry a nominal cost per hour: off-spec gas, amine losses, excess steam, reliability, solvent lifetime and lost throughput. A candidate point can therefore be scored in money rather than argued about in parts per million — and the argument in the control room usually is about which cost matters more.
04
The envelope is calibrated, not assumed
Circulation rate, L/G ratio and load are graded against fixed absolute bands fitted to this train. The model knows when a proposed point has walked outside the envelope it was validated on, and says so instead of extrapolating quietly.
Where to see it
Every number here is on a screen you can open.
Nothing on this page is a calculation done for a brochure. Each line below names the tool and the view that shows it, running on real operating history.
What-if
Run → set the boundary conditions, read the modelled response
Scoring
Detect → the six costed consequences at the current point
Sensitivity
Discover → ranked drivers against steam per tonne of acid gas
Envelope
Circulation, L/G and load graded on fixed absolute bands
Guided, not automatic — and the difference matters
Today this is a guided sweep, not an automatic optimiser. You choose the candidate points and the model scores them; nothing searches the space on your behalf, and nothing writes a setpoint. The two hard parts of an optimiser already exist — a validated plant model and an objective function denominated in money — which is why automatic search is a next step rather than a different product. We would rather say that than let the word "optimal" do work the software has not yet done.