Uncertainty Quantification
Every modelled figure carries uncertainty; we attach bounds rather than quote single numbers as if they were measured.
Why single numbers mislead
A modelled result quoted as a bare number invites the reader to treat it as measured. We instead attach uncertainty: the plug regime is 166–830x (a range, not a point), availability is 0.86–0.995, plug overstress is 3–3.9x. Ranges are the honest form for a design study.
Where uncertainty comes from
- Model form: which physics the code includes or omits
- Inputs: material properties, cross-sections, geometry tolerances
- Extrapolation distance: larger reach, larger uncertainty
- Integration effects not captured in single-subsystem models
How it shrinks
Uncertainty narrows with data. Coupon tests shrink the magnet range; regime experiments shrink the confinement range; fleet hours shrink availability. The test burner is the largest single reducer of uncertainty before commercial commitment.
Quoting ranges rather than single numbers is a deliberate refusal to let a modelled figure pose as a measured one. The plug regime is 166–830x, availability is 0.86–0.995, plug overstress is 3–3.9x; each range carries its worst case in plain sight. Ranges narrow only as data arrives, and the test burner is the single largest reducer of uncertainty before any commercial commitment.
See V&V methodology.