github — System Prompt

Date: February 2, 2026 Duration: 4 weeks (Weeks 1-4) Status: ✅ READY TO START Previous Phase: Phase 6 (✅ COMPLETE - 100% validation, 4/4 scenarios, 51/51 tests) ---

Objective: Advance from baseline validation to comprehensive sensitivity analysis, uncertainty quantification, and expanded fire scenarios. Success Criteria: • ✅ Sensitivity analysis framework complete (Week 1) • ✅ 6+ fire scenarios validated (Week 2) • ✅ Three-zone model implemented (Week 3) • ✅ Uncertainty quantification complete (Week 4) Expected Outcome: Production-grade fire prediction system with quantified confidence intervals and expanded scenario coverage. ---

Tier 1: Sensitivity Analysis Framework (Week 1) Goal: Understand parameter impact on predictions New Module: vfep::phase7::SensitivityAnalysis Deliverables: • SensitivityAnalyzer class (parameter sweeps) • ParameterRange data structure • SweepTool executable • Sensitivity CSV reports for 4 current scenarios Key Parameters to Sweep: • Heat Release (50-200 kJ/mol): How does combustion intensity affect predictions? • Wall Heat Loss (0.5-10 W/m²K): Room insulation impact • Room Volume (20-120 m³): Space scaling effects • Pyrolysis Rate (0.5-2.0 kg/s): Fuel generation variation Success Metrics: • Zero compiler warnings • All Phase 6 tests still pass (51/51) • ≥5 new sensitivity tests • Sweep tool produces valid CSV output • NIST and ISO 9705 sweeps complete --- Tier 2: New Fire Scenarios (Week 2) Goal: Expand validation portfolio with 3+ new fire scenarios New Module: vfep::phase7::ScenarioLibrary Deliverables: • Ship Fire scenario (ISO 15653) • Tunnel Fire scenario (FDS reference data) • Industrial Fire scenario (NFPA guidance) • Validation results against literature New Scenarios Detail: Ship Cabin Fire `cpp // ISO 15653: Ship fire in confined space Geometry: 14 m³ cabin Heat Release: 500 kJ/mol (marine materials) Wall Material: Steel (high conductivity) Ventilation: ACH = 2.0 (mechanical) Duration: 300 seconds Acceptance Criteria: ±8% peak temperature, ±10% time to 60°C Literature Baseline: T_peak = 687 K ` Tunnel Fire `cpp // FDS validation case: Tunnel fire Geometry: 240 m³ tunnel section Heat Release: 2500 kJ/mol (high intensity) Wall Material: Concrete (moderate conductivity) Ventilation: ACH = 4.0 (forced ventilation) Duration: 600 seconds Acceptance Criteria: ±12% hot layer temp, ±15% interface height Literature Baseline: T_upper = 850 K, T_lower = 350 K ` Industrial Warehouse Fire `cpp // Large space fire modeling Geometry: 2000 m³ warehouse Heat Release: 1500 kJ/mol (organic materials) Wall Material: Gypsum board (low conductivity) Ventilation: ACH = 1.0 (natural ventilation) Duration: 1800 seconds (long-term burn) Acceptance Criteria: ±15% long-term trends Literature Baseline: Sustained T_peak = 600 K ` Success Metrics: • 3 new scenarios fully integrated • Each validates within ±8-15% acceptance range • ValidationSuite shows 7/7 scenarios passing • All numeric tests still pass (51/51) --- Tier 3: Three-Zone Model (Week 3) Goal: Improve temperature stratification predictions New Module: vfep::phase7::ZoneModeling Deliverables: • ThreeZoneModel class • Dynamic zone interface tracking • Improved stratification accuracy Three-Zone Architecture: ` Zone 1 (Hot Upper): T_u, high particulate concentration Zone 2 (Interface): Transition layer with gradient Zone 3 (Cold Lower): T_l, ambient conditions ` Physics Improvements: • Current: Single-zone with post-hoc two-zone correction • Phase 7: Dynamic three-zone model during simulation • Better treatment of warm layer below hot layer (intermediate zone) • Zone interface height tracking over time Success Metrics: • Stratification scenarios pass (±6% vs current ±9.26%) • All 7 scenarios still validate • No regression in other predictions • Three-zone model code ≤1000 lines --- Tier 4: Uncertainty Quantification (Week 4) Goal: Quantify prediction confidence with ±95% intervals New Module: vfep::phase7::UncertaintyQuantification Deliverables: • MonteCarloUQ class (Latin Hypercube sampling) • Confidence interval calculator • Parameter distribution definitions • UQ analysis for all 7 scenarios Parameter Distributions: `cpp // Uncertain parameters with statistical distributions heat_release: Normal(μ=100 kJ/mol, σ=15 kJ/mol) h_W: LogNormal(μ=2.0, σ=0.8) [W/m²K] room_volume: Uniform(min=20 m³, max=120 m³) pyrolysis_rate: Gamma(α=2.0, β=0.5) [kg/s] wall_conductivity: LogNormal(μ=1.5, σ=0.5) [W/mK] ` Monte Carlo Approach: ` For each scenario: For n = 1 to 100 (samples): • Sample parameters from distributions • Run simulation with sampled parameters • Record peak temperature, time-to-criterion • Store trajectory Compute statistics: • Mean, std dev • Percentiles (5%, 50%, 95%) • Confidence intervals • Sensitivity indices (Sobol/Morris) ` Success Metrics: • ±95% confidence intervals computed for all scenarios • Monte Carlo sampling validates against analytical results • UQ adds <5% to total computation time • Uncertainty results documented with physical interpretation ---

Each Phase 7 work session follows this pattern: ` Pre-Session (5 min): • Read session template (PHASE7_SESSION1_LOG.md format) • Review previous findings • Plan today's tasks Session Start (30 min): • Build verification: cmake --build build-mingw64 • Test verification: NumericIntegrity.exe + ValidationSuite.exe • Update session log "Morning Status" Main Work (3-4 hours): • Implement planned features • Test incrementally • Document findings • Commit working code: git add && git commit Session End (30 min): • Final test run • Update session log with findings • Plan next session • Commit session notes `