CALC-064

The two summary ranges, and why they are not the same measurement

Feeds CLM-0378 and CLM-0379. SHARED - two claims, one section. Answers QST-004. Basis: dimensionless ratios over GBP2024/MWh figures at 20-yr life.

Script. restate-lcoe.py SECTION 6, section_summary() (line 570). Sub-heading (i) is emitted at line 584 and (ii) at line 605.

Why it was needed. The exercise can report its output two ways and they differ by a factor of four. Reporting only the wider one would overstate the disagreement; reporting only the narrower one without saying what was excluded would understate the exercise's own scope. Both are recorded so that neither can be quoted as the other.

Inputs. CALC-062's and CALC-063's outputs, unchanged.

Working, transcribed.

-- (i) FULL RANGE, every restated estimate including bookends (CLM-0378) --
 3 per cent real   GBP  21.1  to GBP 2,647.6 /MWh   = 125.8 x
10 per cent real   GBP  36.8  to GBP 4,626.6 /MWh   = 125.8 x

-- (ii) CORE comparison, FOAK BASELINE figures only (CLM-0379) --
Frazer-Nash BEIS FOAK        28.1
Frazer-Nash ESA FOAK p10     32.4
Frazer-Nash ESA FOAK p90     63.0
NASA RD1 (rebuilt)          564.0
NASA RD2 (rebuilt)          921.7      [all GBP2024/MWh, 3 per cent, 20 yr]

max/min against RD2    32.8 x (3 per cent)   36.5 x (10 per cent)
max/min against RD1    20.1 x (3 per cent)   19.0 x (10 per cent)

Result: 125.8x on the full range, 19.0-36.5x on the core comparison.

State what each ratio is OF, because this is exactly where the two get confused. - 125.8x is the ratio of the single most expensive restated figure (Caltech's "current technology" tier) to the single cheapest (Frazer-Nash BEIS NOAK 5th unit). It compares one study's own pessimistic bookend against another study's own optimistic bookend. It is a property of how wide the studies' internal scenario ranges are, not a measure of institutional disagreement. - 19.0-36.5x is the ratio of the most to least expensive FOAK baseline figure, which is the same set CLM-0360 compared before any restatement. It is the only one of the two that answers the question the exercise was set.

The ratio-invariance in CLM-0378 is worth transcribing carefully rather than passing over. The claim states 125.8x at both 3 and 10 per cent and says it is "ratio-invariant here because both extremes scale proportionally under this pairing". That is a property of this particular max/min pair, both of which are rescaled by the same CRF ratio; it is not a general result, and CLM-0379's core comparison demonstrates that immediately - 32.8x at 3 per cent becomes 36.5x at 10 per cent, because its extremes do not share a native rate.

Conclusion. CLM-0379's 19-37x is the answer to the question posed; CLM-0378's 125.8x exists to stop that figure being misquoted upward. The claim's own notes say it was recorded "to pre-empt a likely misreading of this restatement's own output", which is the right reason to mint a claim and the right reason to keep it.

Where this fails. The core comparison excludes Caltech on size grounds and Roland Berger on non-rescalability grounds. Both exclusions are defensible and both are choices, and a different pair of choices gives a different headline. The exclusions are recorded at CLM-0387 and CLM-0372 (low confidence) respectively rather than being buried in the ratio.

Cite this