TwinEdge / Platform · Trying things safelyThe model of the plant
Operating scenarios
Explore the next operating scenario.
Use the connected process model to compare an outage, load change, or different operating point before planning the intervention.
Questions worth asking it
- 01If this pump trips, how long before the tank is a problem?
- 02Can we take this line down on Tuesday and still meet demand?
- 03What does the forecast storm do to the inflow?
- 04If we move this setpoint, what does it cost in energy?
- 05Which single failure hurts us most right now?
§ 01
How it works.
- 01Copy where things stand
Take the plant exactly as it is at this moment — every level, pressure, temperature, and machine state — and work on the copy.
- 02Change one thing
Trip a pump. Move a setpoint. Add the inflow of a storm. Close a valve somebody is about to close for real.
- 03Run it forward
The change travels through everything connected to it, the same way it would in the plant, for as long into the future as you want to look.
- 04Compare against doing nothing
The scenario is shown next to the plant carrying on unchanged, so the difference is the answer rather than something you have to work out.
§ 02
How much to trust the answer.
A model is not the plant. It is right about the things it was given equations for, and silent about everything else — the valve that sticks, the level switch that has always read high, the operator who intervenes at exactly this point every time.
So the useful way to read a scenario is as a well-argued expectation, not a prediction. It is far better than a guess, and it should not be treated as a promise.
The way to build confidence is to check it backwards. Run scenarios against events that already happened and see how close it gets. That number is worth more than any claim on this page.
“Ask it about last winter, when you already know what happened. Then decide how much to trust it about next winter.”
How to judge a simulation
Replay an incident you already had.
Pick a trip or a storm from last year. We set the copy to the state just before it, run it forward, and put the prediction next to what actually happened.