What-If Scenario Simulation

Starting from the twin's current state, what-if simulation explores futures the real machine can experience only once, and safely.

Rehearsing the future

A physical machine lives each moment once. A twin, anchored to the machine's current estimated state, can branch that moment into many hypothetical futures and run each forward. This is what-if scenario simulation: exploring the consequences of an action, a disturbance, or a fault before, or instead of, letting it happen on the real hardware.

What the questions look like

• If we raise the heating on this schedule, does any limit get crossed?
• If a coolant pump degrades, how long until a temperature limit is reached?
• If this coil trips, what is the safest recovery sequence?
• How much operating margin remains before a chosen constraint binds?

Anchoring to the present

The value comes from starting each scenario at the twin's real, assimilated state, complete with the machine's current wear and calibration, not at a generic design point. A scenario that begins where the machine actually is gives an answer relevant to this unit, now. Running many scenarios also needs fast surrogates, since a useful study may sweep thousands of cases.

Carrying uncertainty forward

An honest what-if propagates the uncertainty in the starting state, so the answer is a distribution of futures, not a single line. If the twin is unsure where the machine is, it must be unsure where each scenario leads, and it should say so. This connects what-if studies to ensemble methods and uncertainty quantification.

In the Kronos twins

Because both machines are pre-construction, what-if simulation today runs entirely on the design models, and its role is to shape operating procedures, control strategies, and safety cases before hardware exists. After the Hyperion breeder operates near 2030, the same machinery will answer live operational questions anchored to real state. What-if studies also feed operator training and maintenance planning. See operator training simulators and ensemble forecasting.