CRYOPULSE P4.4 — CONDITIONAL DIGITAL PROTOTYPE
Physical measurements: 0. Vehicle not built.

This release is a reproducible requirements model and a developing CAD layout.
It is not a qualified vehicle, measured performance result, safety certification
or statistical demonstration of superiority over a Bugatti.

The selected configuration is design/selected-candidate.json. Output/audit.json
records the numerical gates and their limits. The current primary time step is
0.5 ms; the independent comparison uses 0.25 ms. Earlier coarse-step failures
and the subsequent correction are retained in the research history.

SELECTED DRIVING BUILD
Same fitted effective propulsion-engine curve and reference tyre law; added
mass and finite electrical assistance are counted. The installed floor budget
is 150 kg; supported feeds use a 28 kg allocation. The pack is raised 30 mm,
and the upstream air-volume budget is 0.40 m3. The drive uses a source-derived
500 A continuous-current envelope with approximate voltage scaling and the
conservative unregulated loaded-bus bound. The event windows are explicitly
16 MJ chemical / 12.8 MJ DC, larger than P4.3. Low-voltage controls have a
separate finite auxiliary reserve. No cold module improves this driving score.

The 0.65 m2 external drag target, installed main/floor airflow maps, tyre law,
road seals, cell OCV/resistance, cooling and complete structures are unqualified
requirements. A fitted reference is not an independently validated Chiron twin.
The selected model is slower over 0–200 and 0–300 km/h. Passing a challenge means
it meets that declared numerical completion/control test; it does not mean the
case beats the Chiron performance benchmark.

RUN THE SELECTED NUMERICAL MODEL (Linux / Python 3.12 / C++17 compiler)
Keep all sibling cryopulse-* folders together. They contain inherited numerical
kernels and frozen inputs; folder version names are not separate current scores.
Install requirements.txt into a virtual environment. Set OPENBLAS_NUM_THREADS=1
and OMP_NUM_THREADS=1. From cryopulse-p44:
  python scripts/selected_events.py nominal
  python scripts/selected_events.py half
  python scripts/selected_events.py boost-off
  python scripts/current_challenges.py
  python scripts/current_extra_checks.py surface
  python scripts/current_extra_checks.py direct
  python scripts/speed_policies_v44.py prepare
  python scripts/speed_policies_v44.py
  python scripts/check_launch_clutch.py
  python scripts/requirements_budget.py
  python scripts/audit_current.py
The compiled drive/wheel kernels are built from the supplied C++ sources into
an operating-system temporary directory. No proprietary simulator is required.
The frozen vendor grid is supplied as numeric input with its source URL/hash.
No vendor workbook macros are executed; the original XLSM is not redistributed.

The top-speed modes are separate scalar continuations with carried stores.
The slower reserve-battery launch cannot be combined with the fastest sprint.
There is no software speed limiter in this screen; gearing, engine extrapolation,
tyres, cooling and high-speed dynamics still require qualification.

CAD
cad/p44-L5750-front-v7-complete-layout-study.step is dimensional layout geometry.
STEP declares millimetres; the complete x span is 5750 mm. GLB uses SI metres.
Declared components, sampled steering/travel/growth clearances and hot routing
are checked. Engine, transmission and expansion-machine shapes include unverified
installation allocations. Rib/laminate joints, full chassis/crash design, service
paths and occupant qualification remain open. The separately exported closed-flow
STEP is an intermediate body/wheel exclusion; it is not a final aerodynamic
surface including all external appendages. No whole-car CFD drag credit is earned.

COLD-FLOW BENCH
web/lab/cold-flow contains the separate CF-1.1 demonstrator and its public files.
The supplied measurement file is empty. Running the analysis on the empty record
returns NO_EVIDENCE. Synthetic fixtures test software only and can never become
physical evidence. Read protocol.json, the parts requirements and START-HERE.txt
before fabrication review. The first test uses 20 C and 0 C walls at low pressure;
it does not require liquid nitrogen and does not validate a cryogenic vehicle.
Flow change, air heat removal and electrical input are distinct quantities.

CFD INPUTS — UNDER NUMERICAL REVIEW
The D-series cold-flow initial fields are saved numerical states, not measurements.
Their input dictionaries and fixed initial fields are supplied; rebuild blockMesh
with OpenFOAM v2412, then run all topology/geometry mesh checks. Do not relax them.
The mesh itself is regenerated rather than bundled. For a clean case:
  blockMesh -noFunctionObjects
  checkMesh -allTopology -allGeometry
  decomposePar
  GXX=g++ mpirun --bind-to none -np <numberOfSubdomains> rhoPimpleFoam -parallel
Use the rank count in system/decomposeParDict. The full-gap/half-span domain
requires the declared factor of two for a full physical channel. No 38-channel
manifold benefit is credited. The grid, time-step, physical-time storage and heat
criteria in design/coldflow-numerical-acceptance.json govern acceptance. Interim
snapshots and provisional flow ratios are not accepted results or physical proof.

REPRODUCTION AND INTEGRITY
FILE-SHA256.json hashes every included file. The release ZIP is immutable.
The separate clean-source replay report records archive verification and replay
agreement. Rebuilding the CAD additionally needs gmsh, vtk and Pillow. Full web
presentation/deployment is provided separately as the static website ZIP.

Designed by Genyra LLC — contact@genyra.com — https://cryopulse.org/
