HyFlux Extensions
=================

Capabilities beyond the CASCADE parity perimeter. These are HyFlux-originated
workstreams — they cite CASCADE data where disclosed but make no parity claim.

Superconducting LH₂ + SC propulsion analysis
--------------------------------------------

**Source:** ``src/engine/cascade.ts:computeLCA`` (``scAnalysis``),
``src/engine/fuel-costs.ts:q5Renormalisation``
**Test evidence:** ``tests/fuel-costs.spec.ts``,
``tests/cascade-parity.spec.ts``

The superconducting (HTS) motor lever ``sc`` (%) scales delivered LH₂ CI by
:math:`(1 - sc/100)` and reports an analysis block: power density 5.2 kW/kg
conventional vs 25–35 kW/kg HTS, "free" cryogenic cooling from the LH₂
sink, and programme anchors (Airbus ASCEND 500 kW demo, RTX/Pratt 260 kW →
2.5 MW, Hinetics MW-class). In the cost model the fuel-cell + SC chain
converts fuel to shaft at η = 0.55 vs 0.40 for turbines — enough to invert
the methane-vs-hydrogen cost ranking per unit of propulsive output (see
:doc:`costs`).

Airport infrastructure map groundwork
-------------------------------------

**Source:** ``src/map/`` (``geodesic.ts``, ``demand.ts``, ``airports.ts``,
``provider.ts``, ``AirportMap.tsx``)
**Test evidence:** ``tests/geodesic.spec.ts``,
``tests/airport-demand.spec.ts``

Phase 5 scaffold (DL-031/DL-035): a MapLibre-first airport comparison map
behind a provider abstraction (token-free default tiles, zero recurring
cost per DL-011), with all geometry computed geodesically — haversine
distance, direct-geodesic destination points, Chamberlain–Duquette
spherical polygon area; **never from screen pixels** (Web-Mercator pixel
areas are latitude-distorted). Six airport archetypes (LHR, AMS, INV, SIN,
DXB, MEM) and a demand-driven sizing model (peak-day/peak-hour factors,
reserve-days storage, tank counts, truck movements with a pipeline
threshold) carry a standing **CONCEPTUAL — NOT VALIDATED** label: every
constant is named, editable and literature-approximate, and the outputs
must not be used for design, safety or cost decisions.

Usable-fuel pathways (Workstream A)
-----------------------------------

**Source:** ``src/engine/fuel-pathways.ts:usableFuel``,
``reproduceBoeingUsableFuel``
**Test evidence:** ``tests/usable-fuel.spec.ts``

An independent first-principles chain model: usable fuel = fuel energy
delivered to the aircraft (LHV) ÷ primary input energy, with per-stage
retentions and overheads,

.. math::

   \text{usable} = \frac{\prod_i r_i}{\text{baseInput} + \sum_j o_j}

evaluated at central, all-pessimistic and all-optimistic stage values.
Boeing's published chart values (claim register BC-01…BC-05) fall inside
the independent envelopes for all five pathways — bio-CH₄ 54.3 % vs 54 %,
e-CH₄ 48.1 % vs 43 % (claim digitised/unverified), LH₂ 44.8 % vs 38 %
(Boeing pessimistic-side), bio-SAF 30.2 % vs 30 %, e-SAF 23.8 % vs 22 % —
all classified *Conditionally validated* since Boeing discloses no stage
assumptions. Key boundary finding: e-SAF's 22 % is only reproducible when
non-jet Fischer–Tropsch co-products are treated as losses (no energy
allocation); with co-product credit the chain stays above 28 %. Boeing's
implicit boundary choice is undisclosed (WorkstreamA-UsableFuel-Report).

Pulse curve
-----------

**Source:** ``src/engine/curves.ts:pulseAt``, ``buildPulseSeries``
**Test evidence:** ``tests/pulse-curve.spec.ts``

The CASCADE custom "pulse" curve (cf. CASCADE *Operations*, Custom Curves →
Pulse), implemented exactly from its published formulation (DL-020): two
concatenated sigmoids

.. math::

   y(t) = \begin{cases}
   0 & t \le t_0 \\
   A_{peak}\, S\!\left(\frac{t - t_0}{t_{peak} - t_0}\right)
     & t_0 < t \le t_{peak} \\
   A_{peak} + (A_f - A_{peak})\, S\!\left(\frac{t - t_{peak}}
     {t_f - t_{peak}}\right) & t_{peak} < t \le t_f \\
   A_f & t > t_f
   \end{cases},
   \qquad
   S(t') = \frac{1}{1 + \left(\frac{t'}{1 - t'}\right)^{-a}}

with exact value :math:`A_{peak}` at :math:`t_{peak}`, :math:`A_f` held
beyond :math:`t_f`, and zero-slope knots for :math:`a > 1`. Eleven tests
cover monotonicity, half-amplitude midpoints, continuity at all knots,
zero-duration edges and shape-parameter behaviour. UI exposure remains
deferred: the shared ``CurveEditor`` param model (shape, entry/end years
and values, slope, inflection year, custom points) does not map onto the
pulse parameter set, and no approximate mapping is offered (see
:doc:`aircraft`).

LH₂ CI-driven accounting view (``lh2_ci_driven``)
-------------------------------------------------

**Source:** ``src/engine/cascade.ts:computeLCA`` (``lh2CiDriven``,
``deliveredLH2CIAt``, ``lh2LedgerSharePct``)
**Test evidence:** ``tests/lh2-energy-ledger.spec.ts``,
``tests/cascade-measured-lh2.spec.ts``

An opt-in HyFlux **accounting view** (default off, bit-identical when off):
the LH₂-attributable share of the aircraft wedge (the ``sc × 0.5`` term) is
re-attributed to the energy wedge as a CI-driven term of the same form SAF
and methane use, so LH₂ abatement tracks the *delivered* carbon intensity —
and becomes a visible **penalty** on slow-decarbonising grids.
``lh2_ledger`` is accepted as a deprecated alias (DL-142) and is not
URL-serialisable.

**This is not CASCADE's method.** A live ``orchestrationLayerInterface``
capture (2026-08-01) measured that CASCADE books LH₂ aircraft abatement
wholly in the **Aircraft** wedge, applies delivered-LH₂ CI inside the
Aircraft module, and prevents double counting by *sequential activity
removal* — hydrogen aircraft shrink the residual Jet-A pool that the later
Operations and Energy stages receive; there is no Energy-wedge LH₂ term and
no explicit de-duplication subtraction (see :doc:`methodologyParity`,
"Measured CASCADE LH₂ wedge accounting"). The engine therefore **refuses
the combination**: enabling this view under
``evidence_basis: 'CASCADE_PARITY'`` throws, and no library preset sets the
flag. UI surfaces label it *"HyFlux CI-driven LH₂ accounting view —
experimental HyFlux analysis; CASCADE parity uses sequential Aircraft-stage
LH₂ accounting instead."*

Known deviations / limitations
------------------------------

* SC analysis constants (power densities, programme powers) are literature
  anchors, not validated product data.
* Airport sizing is explicitly conceptual; no geocoded demand data yet.
* Usable-fuel stage values are literature-approximate envelopes, not
  calibrated plant data; the e-methane Boeing value remains unverified
  pending chart digitisation (DL-008).
* Pulse curve is engine-level only (CurveEditor param-model mismatch,
  documented under :doc:`aircraft`).