Phase 08: an optimized Layer-1 engine on the end of the feed system.
Point it at what the optimizer wrote and fire it::
from feedtwin.engine import Chamber, CEATable, load_engine, injector_legs
design = load_engine("configs/impinging_lox_ch4_8000N_optimal.yaml")
design.summary()
# impinging injector, oxygen/methane, A_t 1398.2 mm^2,
# A_inj ox 124.55 mm^2 / fuel 60.76 mm^2, design 8000 N at 24.1 bar
chamber = Chamber(design.throat_area,
CEATable("output/cache/cea_cache_LOX_CH4_3D.npz",
expansion_ratio=design.expansion_ratio))
ox_leg, fuel_leg = injector_legs(design)
Nothing is retyped. The injector areas come from Layer 1's own geometry through
a **type-aware** extractor -- round jets for an impinging face, orifices and an
annulus for a pintle, core and annulus for a coaxial -- and an unrecognised type
is refused rather than guessed at, because a wrong area is a wrong mass flow and
every number downstream of it stays perfectly plausible.
Combustion is **propellant-aware** for real: :class:`CEATable` reads
EngineDesign's own CEA cache, so ``c*``, chamber temperature, ``gamma`` and
``C_F`` come from the same table both tools use. That is what lets a firing
trace report chamber temperature and O/F drift rather than assuming them.
The engine is a boundary, not a component: chamber pressure depends on the flows
and the flows depend on chamber pressure, and the session closes that loop around
the whole feed system (``feedtwin.session.core.Session._close_chamber``).
1.15.0