Aircraft — Fleet Renewal & Future Aircraft#

Implementation status: Fleet renewal Equivalent (equation-level parity benchmarked, DL-028); Future Aircraft Simplified — all five §26 review actions implemented (opt-in), proprietary reference tables substituted (coverage matrix rows Fleet Renewal, Future Aircraft) Source: src/engine/curves.ts:fleetRenewalFraction, src/engine/curves.ts:insertionCurveAt, src/engine/futureAircraft.ts (whole module) Test evidence: tests/fleet-renewal-parity.spec.ts, tests/fleet-insertion.spec.ts, tests/future-aircraft.spec.ts

Fleet renewal (piecewise quadratic phase-out)#

The fraction of the legacy fleet still in service at year \(t\) follows a piecewise quadratic — flat, quadratic ramp-down over \(T_{ramp}\), linear steady state, mirrored quadratic to zero at \(t_{endProd} + T_{out}\) (cf. CASCADE Fleet Renewal, eqs. 103–109):

\[\begin{split}f(t) = \begin{cases} 1 & t \le t_{start} \\ 1 + a\,(t - t_{start})^2 & t_{start} < t \le t_{start} + T_{ramp} \\ m\,t + b & \text{linear interior} \\ -a\,(t_{start} + T_{out} - t)^2 & t \ge t_{start} + T_{out} - T_{ramp} \\ 0 & t \ge t_{start} + T_{out} \end{cases}\end{split}\]

with

\[a = \frac{1}{2\,T_{ramp}\,(T_{ramp} - T_{out})}, \qquad m = \frac{1}{T_{ramp} - T_{out}}, \qquad b = 1 + a\,T_{ramp}^2 - m\,(t_{start} + T_{ramp})\]

and \(t_{start} = t_{endProd}\) (defaults: end of production 2035, ramp 5 yr, phase-out 25 yr). A pre-start shortening rule re-anchors the curve when \(t_{endProd}\) precedes the model start year: \(T'_{out} = T_{out} + (t_{endProd} - t_{start})\) (DL-028).

Note

RST-internal finding F3. CASCADE’s published closed form for the intercept \(b\) omits the “+1” and fails continuity with its own piecewise definition by exactly 1 at \(t = t_{endProd} + T_{ramp}\). HyFlux’s continuity-derived \(b\) (above) is used; the discrepancy is queued on the Boeing evidence-gap list (DL-028, claim register BC-14).

The aircraft wedge in the aggregate engine is gated by this curve: the wedge scales with \(1 - f(t)\), i.e. only the share of the fleet that has actually turned over receives the new-technology improvement.

Future Aircraft module#

futureAircraft.ts implements the route-level fleet-insertion model (cf. CASCADE Future Aircraft). The mechanisms, as implemented:

1. Distance correction. Flown distance adds a fixed operational correction to the great-circle distance (eq. fa_distance_correction):

\[d = d^* + 51\ \mathrm{km}\]

2. Reference range-energy tables with 1D interpolation. Per-flight tank-to-wake energy \(e^*_{ref}(d)\) is linear-interpolated over flown distance between table knots (eq. fa_energy_consumption_1D_linear_interp), clamped to the end knots (no extrapolation):

\[e^*(d) = e_i + \frac{d - d_i}{d_{i+1} - d_i}\,(e_{i+1} - e_i)\]

3. Custom-aircraft energy rescaling. A user efficiency delta \(\Delta_{cust}\) rescales the reference energy (cf. CASCADE Future Aircraft, “Defining Energy Consumption of a Single Future Aircraft”):

\[e_{fut} = e_{ref}\,\frac{1 + \Delta_{cust}}{1 + \Delta_{ref}}\]

with \(\Delta_{cust} \in [-0.30, +0.70]\) enforced (scenario bounds; violations raise RangeError). Note the bounds discrepancy: CASCADE’s own bounds page (April 2026 snapshot) states −30 %/+60 % — the module keeps the wider driving-prompt bound and the divergence is recorded (DL-025, DL-029).

4. Five-phase insertion curve. The market share of a future type grows 0 → quadratic ramp-up → linear steady state → quadratic ramp-down → 1 (cf. CASCADE Future Aircraft, addressable market share equations):

\[\begin{split}x(t) = \begin{cases} 0 & t \le t_{EIS} \\ a\,(t - t_{EIS})^2 & t_{EIS} < t \le t_{EIS} + T_{ramp} \\ r_{ss}\,t + b & t_{EIS} + T_{ramp} < t \le t_{EIS} + 1/r_{ss} \\ 1 - a\,(t_{EIS} + t_{sat} - t)^2 & \text{else, until } t_{EIS} + t_{sat} \\ 1 & t \ge t_{EIS} + t_{sat} \end{cases}\end{split}\]
\[a = \frac{r_{ss}}{2\,T_{ramp}}, \qquad b = -r_{ss}\left(t_{EIS} + \frac{T_{ramp}}{2}\right), \qquad t_{sat} = T_{ramp} + \frac{1}{r_{ss}}\]

with the EIS-year decrement \(t_{EIS} = t^*_{EIS} - 1\) and the constraint \(r_{ss} \le 1/T_{ramp}\). Deviation: CASCADE truncates violating rates to \(1/T_{ramp}\); HyFlux’s module contract is fail-fast — invalid specs are rejected with RangeError (the truncating variant is also exported as truncateInsertionRate / insertionCurveAt for the aggregate engine). The proportional insertion rate from a global production rate (eq. proportional-insertion-rate) is provided:

\[r_{ss} = \frac{R_{ss}}{n_{in} + (1 + T_{ramp}/2)\,\Delta n_{in}}\]

5. Addressable market cap. Shares per route satisfy \(\sum_{ac} x(ac, rt, t) \le 1\). Deviation: CASCADE freezes each type’s share at the year the sum first reaches 100 % (\(t_{marketSat}\)); HyFlux applies pro-rata normalisation \(x^*_i = x_i / \sum_j x_j\) when the sum exceeds 1 (documented module decision).

6. Flight conservation. Assignment conserves flight count exactly per route:

\[n_{fut,i} = x^*_i\,n_{in}, \qquad n_{exist} = \left(1 - \sum_i x^*_i\right) n_{in}, \qquad n_{fut} + n_{exist} = n_{in}\]

CASCADE’s payload-capacity flight-count normalisation (capacity-ratio rescaling) is not implemented — the synthetic dataset assumes same-class replacement.

7. Eligibility. A type is assignable iff its size class is eligible for the route and route distance ≤ usable range, with usable range = 0.8 × design range — a HyFlux reserve-margin assumption the RST does not specify (USABLE_RANGE_FRACTION, editable).

8. Battery-electric adjustment. Electricity drawn per unit onboard energy (eq. fa_battery_emissions):

\[E^*_{elec} = E_{elec}\,\frac{1}{\eta_{batt}}\left(1 + \frac{\varepsilon_{batt}}{T_{batt}}\right) = E_{elec} \times \frac{1}{0.97}\left(1 + \frac{60}{3000}\right) \approx 1.0515 \times E_{elec}\]

(one recharge cycle per flight cycle; constants per the CASCADE data table). In the aggregate engine the electric wedge applies this to the regional class only (opt-in via el_ceiling under ac_eis_gating; electric range bounded to regional operations), with emissions

\[g_{elec} = E_{displaced}\,(1 - r_{energy}\, m_{batt}\, CI_{elec}(y)/CI_{JetA})\]

where \(r_{energy}\) (el_energy_ratio, default 0.45) is onboard MJ electric per MJ jet and \(m_{batt}\) the multiplier above (DL-029).

Reference-table fidelity limitation#

CASCADE’s reference range-energy tables are generated by Boeing’s proprietary aircraft sizing and mission tools and are not public; exact value parity for the interpolation equation is therefore not achievable and is never claimed (Phase-0 §12, DL-025). HyFlux substitutes transparent synthetic tables built from a documented per-class energy model:

\[e^*(d) = N_{cap}\,(f_{fix} + f_{cruise}\, d) \times k_{carrier}\]

calibrated to order-of-magnitude public figures (ATR-72-class ≈ 0.85 MJ/seat-km cruise; A320neo-class ≈ 0.55; widebody-class ≈ 0.70; freighter ≈ 4.5 MJ/t-km; Jet-A LHV 43.1 MJ/kg), with carrier factors H₂ ×0.95, CH₄ ×1.00, electricity ×0.30 vs Jet-A, and illustrative reference deltas (−0.10/−0.15/−0.12/−0.35). All tables are injectable — every consumer accepts a tables override so licensed data can drop in without touching the model (routes.ts:DEFAULT_REFERENCE_TABLES).

Aggregate-engine insertion machinery#

The same curve family drives the opt-in EIS-gated aircraft wedge in cascade.ts (DL-026/028/029): per size class \(c \in \{$regional, single-aisle, widebody, freighter$\}\),

\[\text{adoption}(y) = \sum_c w_c \cdot \text{ceiling}_c \cdot \text{adoption}_c(y;\, eis_c - 1)\]

with editable fuel-burn share weights (defaults 5/43/42/10 %, literature-approximate; CASCADE weights by class traffic, which HyFlux does not resolve) and either the insertion curve (when ac_prod_rate is set, \(r_{ss} = 12 \times \text{rate} / \text{addressable fleet}\), default fleet 25,000) or a legacy S-curve over \([eis-1, 2050]\).

Future Aircraft lever UI (AircraftModal)#

Implementation status: Wired (opt-in; engine-level tests, no DOM) Source: src/components/modals/AircraftModal.tsx (FutureAircraftSection, 🛫 Future Aircraft tab) Test evidence: tests/future-aircraft-ui.spec.ts (15 tests)

The EIS-gated insertion machinery is exposed behind a master toggle (ac_eis_gating, default off — the legacy single-curve adoption stays active, every setting below inert, and results are bit-identical to legacy). With gating on, the tab offers:

  • Per-size-class table (regional / single-aisle / widebody / freighter): EIS year, 2050 ceiling slider (0–100 %) and fuel-burn share input per class. The engine defaults mirror the module (\(EIS_c\) falls back to the base eis, ceilings to 100 %, shares 5/43/42/10). Shares need not sum to 100 — the engine normalises pro-rata so \(\sum w = 1\), and the UI warns when they do not. UI input bounds clamp EIS to 2025–2050 while the engine accepts any year and starts insertion at EIS−1 (RST rule); ceilings outside [0, 100] % are engine-clamped. Both clamps are surfaced as inline warnings, as is the all-ceilings-zero collapse of the wedge.

  • Production-rate mode — ac_prod_rate (aircraft/month, steady state). 0/empty clears the field and falls back to per-class S-curve adoption; a positive rate activates the quad-linear-quad insertion curve with \(r_{ss} = 12 \times \text{rate} / \text{addressable fleet}\) (ac_addressable_fleet, default 25,000) and \(t_{sat} = EIS{-}1 + T_{ramp} + 1/r_{ss}\) (ac_prod_ramp_years, 1–15). The panel previews \(r_{ss}\) truncated at \(1/T_{ramp}\) per the RST cap rule.

  • Live verification — a readout of the 2050 aircraft wedge (Mt CO₂e) recomputed from computeLCA on every patch; it reads the 2050 slice explicitly so the readout stays correct when end_year extends the horizon (Traffic Growth).

UI-wiring tests pin: explicit-off ≡ flag-absent bit-identity; monotonic wedge response to the per-class EIS stagger (at 2040 in S-curve mode — every class saturates by 2050 under hold-not-stretch — and at 2050 in production-rate mode, where the insertion curve has not saturated); production-rate ≠ S-curve fallback; ceiling-0 classes contribute nothing; and URL-schema round-trip of every ac_* field the tab writes.

Note

Pulse curve remains engine-only. The shared CurveEditor param model (shape, entry year/value, end year/value, slope, inflection year, custom points) has no field mapping onto the pulse curve’s two-sigmoid parameter set (\(t_0, t_{peak}, t_f, A_{peak}, A_f, a\) — HyFlux Extensions). Rather than offer an approximate mapping, UI exposure is deferred; the engine implementation (curves.ts:pulseAt) is fully tested.

Known deviations / limitations#

  • Pro-rata market cap instead of CASCADE’s \(t_{marketSat}\) freeze.

  • Fail-fast spec validation instead of rate truncation (module level).

  • Payload-capacity flight normalisation (the fleetRenewal.rst substitution + freighter/passenger capacity corrections and the seven-step RCTK conservation proof) is implemented standalone in src/engine/fleet-renewal-route.ts (DL-136, tests/suite-k-fleet-renewal-route.spec.ts) but runs at shadow stage only — the production engine still renews at aggregate level with same-class replacement assumed.

  • Usable-range fraction 0.8 is a HyFlux assumption, not CASCADE data.

  • Existing-aircraft energy uses the route’s typical-class Jet-A table as a stand-in for CASCADE’s SET models.

  • Electric wedge restricted to the regional class; ≤1.3 %-of-gross overlap with the generic aircraft wedge is a documented approximation (DL-029).

  • Synthetic reference tables: no exact value parity possible or claimed.