Skip to content

Phase 44 — Environmental & Subsystem Integration

Summary

Wired 30+ previously built-but-disconnected environment, CBRN, EW, GPS, logistics, and population engines into the battle loop. Zero new source files — all engines already existed. This is pure instantiation + query wiring.

Sub-phases

44a: Weather & Night Effects (12 tests)

  • scenario.py: Instantiate WeatherEngine, AstronomyEngine, TimeOfDayEngine, SeaStateEngine in _create_engines(). Parse weather_conditions.precipitation from scenario YAML. Merge weather_conditions.visibility_m into calibration when not explicitly overridden.
  • engine.py: Fix weather engine call from step(clock) → update(dt) (the hasattr(engine, "step") guard silently skipped it). Same fix for SeaStateEngine (update(dt_seconds)). TimeOfDayEngine is query-only — removed per-tick call entirely.
  • battle.py: Weather Pk modifier table (8 states, 0.55–1.0). Weather visibility caps detection range when worse than calibration. Night visual detection ×0.3, thermal sensors get +thermal_contrast bonus. Sea state Beaufort >4 increases naval target dispersion.

44b: CBRN/EW/Space Effects (9 tests)

  • battle.py: CBRN MOPP detection degradation and fatigue penalty via cbrn_engine.get_mopp_effects(). EW jamming degrades radar detection quality via ew_engine.compute_radar_snr_penalty() (only for weather_independent sensors). GPS CEP factor degrades crew_skill for GPS-guided weapons via gps_engine.compute_gps_accuracy() + compute_cep_factor().

44c: Logistics Engine Wiring (10 tests)

  • scenario.py: Instantiate MedicalEngine and EngineeringEngine in _create_engines(). Add medical_engine, engineering_engine fields to SimulationContext. Add maintenance_engine to get_state/set_state (was missing).
  • engine.py: Call maintenance_engine.update(dt_hours, temperature_c, timestamp) per tick, feeding weather temperature. Call medical_engine.update(dt_hours, timestamp) per tick.
  • battle.py: Equipment readiness gate — readiness <0.3 skips engagement, 0.3–1.0 scales crew_skill (min 0.5).

44d: Population Engine Wiring (6 tests)

  • scenario.py: Conditionally instantiate CivilianManager + CollateralEngine when escalation_config is present. Add collateral_engine field to SimulationContext.
  • engine.py: Placeholder for collateral update in escalation cycle (CollateralEngine is event-driven via record_damage(), no per-tick update needed).

Files Modified

File Changes
simulation/scenario.py _parse_weather_state() helper, env/med/eng/pop engine instantiation, clock parameter, get_state/set_state additions (~80 LOC)
simulation/engine.py Fixed weather/sea_state update calls, added maintenance/medical updates, collateral placeholder (~40 LOC)
simulation/battle.py Weather Pk table + modifier, night/thermal modifiers, sea state dispersion, CBRN MOPP, EW jamming, GPS CEP, readiness gate (~120 LOC)
3 existing test files Updated mocks for new weather/tod API (step→update, query-only)

New Test File

tests/unit/test_phase44_env_subsystem.py — 37 tests

Key Decisions

  1. Weather modifiers computed once per tick: Query weather/illumination/sea state outside the per-unit loop. These don't change within a single tick.
  2. Min of calibration vs weather visibility: visibility_m = min(calibration_vis, weather_vis). Scenarios with explicit calibration_overrides.visibility_m override weather; otherwise weather engine provides the value.
  3. Night detection = 0.3 (70% penalty): Conservative — night without NVG/thermal severely limits detection. Thermal sensors bypass via _WEATHER_BYPASS_TYPES. Thermal gets small bonus at night (+thermal_contrast × 0.3, capped at +0.2).
  4. EW jamming as detection quality degradation: detection_quality_mod *= ew_factor. Cascades into vis_mod and ROE id_confidence.
  5. GPS CEP as crew_skill degradation: No direct CEP parameter in route_engagement(), so degrade crew_skill as proxy.
  6. Engine update method fix: engine.py used hasattr(engine, "step") guards that silently skipped engines whose method was named update. Fixed all environment engine calls.
  7. Readiness < 0.3 skips engagement: Below 30% readiness, unit is combat-ineffective. Between 30-100%, readiness scales crew_skill (capped at 0.5 min).
  8. TimeOfDayEngine is query-only: No per-tick update needed. Removed the call entirely rather than keeping a no-op.

Descoped

  • Campaign-scale logistics: TransportEngine, DisruptionEngine, NavalLogisticsEngine, NavalBasingEngine, PrisonerEngine. Deferred.
  • CommandEngine: Requires HierarchyTree + TaskOrgManager setup. Complex instantiation. Deferred.
  • Population sub-engines: DisplacementEngine, CivilianHumintEngine, InfluenceEngine. Deferred.
  • Strategic air/IADS: AirCampaignEngine, IadsEngine, StrategicBombingEngine. Campaign-scale. Deferred.

Known Limitations

  • Weather Pk modifier is simple table lookup: Does not account for weapon type (e.g., guided missiles less affected by rain than visual-range weapons).
  • Night modifier is binary (day/night): No twilight gradation for visual sensors — twilight treated as night.
  • Maintenance readiness requires equipment registration: Without explicit register_equipment() calls (not done in ScenarioLoader), get_unit_readiness() always returns 1.0. Full logistics wiring deferred.
  • Medical/engineering engines instantiated but have no scenario data: No facilities/projects registered. They run per-tick but produce no effects until scenario YAML includes medical/engineering data.