CHP Test Matrix

Reference engine: EnergyPlus 26.1 Generator:InternalCombustionEngine, Generator:CombustionTurbine, Generator:MicroTurbine, and Generator:FuelCell with TrackSchedule dispatch matched to the CogenS electric load profile. BOTH engines apply the same OEM-grounded part-load polynomial per CHP class (Fuel Input Ratio + Heat Recovery Ratio), sourced from the production curve_coefficients DB table that the TEA pipeline uses. EP-side curve coefficients derived mathematically from the CogenS DB polynomial via per-generator-type transformations (shaft_power_curve for ICE, BG PL fuel input for CT, eff_plr cubic for MT, Annex42 power-quadratic for FuelCell) so both engines compute identical fuel input at every PLR. Scored against ASHRAE Guideline 14-2023 strict on Reciprocating, Gas Turbine, and Microturbine; FEMP M&V Tier 2 standard (+/-15% annual sum, +/-15% NMBE) on Fuel Cell for the documented internal-energy-bookkeeping architectural difference (AirSupply / FuelSupply / AuxHeater / ElectricalStorage / Inverter parasitic loads as separate fuel-consuming terms in EP vs. lumped electric_efficiency in CogenS).

Reference engine: Two-layer Stage 2 framing. Layer A is an engine contract regression: a first-principles ramp/startup trajectory computed independently from the user-input parameters (ramp_up_rate, ramp_down_rate, min_plr, shutdown_control) via a closed-form recursion. The 0.0% per-step deviation is the desired outcome here — it confirms the engine's _apply_ramp_limits implementation respects the published per-hour ramp-envelope contract, NOT a cross-engine deviation claim. Layer B is an independent OEM-envelope sanity check: the configured ramp rate per class is verified against the EPA CHP Catalog + OEM datasheet published ranges (RICE 20-100 %/h, GT 5-50 %/h, MicroTurbine 50-200 %/h, FuelCell 1-25 %/h union of cold + warm start). EnergyPlus Generator:InternalCombustionEngine, Generator:CombustionTurbine, and Generator:MicroTurbine do not expose transient parameters; FuelCell Power Up / Down Transient Limits enter at Stage 4 control-strategy dispatch. The real-fixture cross-engine extension (10 scenarios) layers a constant-PLR annual KPI cross-val against EnergyPlus 26.1 (RICE/GT) and PySAM 7.1.1 Fuelcell (SOFC/PAFC/MCFC) on top, anchored to the locked CHP shared fixture (Hospital + Chicago + ComEd + Peoples Gas) and the Stage 1 cross-engine constant-efficiency handshake.

Reference engine: Analytical engineering reference from EPA CHP Catalog degradation physics — capacity AND heat rate degrade at the same annual rate. Closed-form formulas (ann_deg(t) = 1 - (t/8760) * annual_deg_pct/100; e_out(t) = min(PLR_command * cap_rated, ann_deg(t) * cap_rated); fuel(t) = e_out(t) / (eta_rated_curve * ann_deg(t)); therm(t) per the engine's two-branch dispatch contract) derived directly from EPA CHP Catalog conventions. NOT a cross-engine deviation: EnergyPlus Generator:CombustionTurbine and Generator:MicroTurbine do not model inter-year ageing, and NREL SAM Generic System uses the same linear ramp formula as CogenS and would be a circular reference. The analytical reference is independent of the engine's fuel_plc * max_in_ava implementation path. The FC restart-degradation real-fixture extension layers a 12-scenario second-channel cross-val (3 FC classes x 4 cycle regimes) against the same closed-form analytical machinery, with PySAM 7.1.1 Fuelcell as an informational third-engine cross-check.

Reference engine: Strategy logic regression. The reference column is the per-hour expected units_operating from each strategy's documented staging rule (HLF: stage when h_load > min_plr * cap_h; Higher / Lower LF / ELF: stage when e_load > min_plr * cap_e; Fixed Output Always-Run: stage every hour). Per-unit dispatch is forced to Electric Load Following in chp_simulator.py:376 regardless of outer mode, so Stage 4 at num_units=1 is fundamentally a STAGING-rule regression test. The engine's per-unit electric output physics is already validated cross-engine in Stage 1 against EnergyPlus Generator:*. NOT a cross-engine deviation; the 0.0% headline confirms the engine's _calc_n_op implementation enforces the published staging-rule contract per strategy. Layer B (true EP cross-validation per strategy via FollowThermal / FollowThermalLimitElectrical / TrackSchedule) is tagged as a follow-up enhancement -- HLF and the composite modes need PlantLoop-wired heat recovery, and Lower LF has no direct EP equivalent.

Reference engine: Multi-unit staging regression. The reference column is the per-hour expected units_operating computed independently from the engine's _calc_n_op implementation, by re-deriving the documented brute-force band-search algorithm in units_operating.py:40-83 (first pass: smallest n with load in [n*factor_low, n*factor_high]; second pass: smallest n with load < n*factor_low minus 1; load-following preference to drop one if total load <= (n-1)*1.1*factor_high). Sinusoidal diurnal load swings system PLR across the [0.5, 1.0] optimum band so the engine's n_op decision varies across the week (typically cycling 1 / 2 / 3 units within a day). NOT a cross-engine deviation; the 0.0% headline confirms the engine's staging algorithm enforces the published band-search contract. EnergyPlus ElectricLoadCenter sums per-generator outputs by construction, so cross-EP comparison would test the same equation; per-unit Generator:* fidelity is already cross-validated in Stage 1.

Reference engine: Two-size dispatch regression. The reference columns are per-hour expected (n1_op, n2_op) computed independently from the engine's min-fuel dispatch algorithm in chp_multisize_staging.py:240-399 calc_min_fuel_dispatch_chp_elec. For each hour the algorithm enumerates every (n1, n2) combo whose available capacity meets the load, evaluates up to 3 load-split scenarios per mixed combo (size-1-maxed-first, size-2-maxed-first, proportional), and picks the (combo, split) minimizing total fuel using each size's electric polynomial. Sinusoidal diurnal load swings the system PLR so the (n1, n2) decision varies across the week. NOT a cross-engine deviation; the 0.0% headline confirms the engine's min-fuel dispatch algorithm enforces the published enumeration-and-fuel-minimization contract. EnergyPlus has no two-size brute-force dispatch primitive; per-size Generator:* fidelity is validated cross-engine in Stage 1, and the multi-unit identical sum is validated in Stage 5.

Reference engine: Two-technology dispatch regression. Same min-fuel enumeration algorithm as Stage 6 (chp_multisize_staging.py:240-399 calc_min_fuel_dispatch_chp_elec) but with HETEROGENEOUS equipment_type per size: each size brings its own electric polynomial coefficients and its own min_plr_e. The analytical reference reads min_plr_1 and min_plr_2 per size and uses them in the min-fuel-penalty floor. Sinusoidal diurnal load swings the system PLR so the (n1, n2) decision varies across the week, with the two technologies actually competing on different fuel-cost-vs-PLR shapes. NOT a cross-engine deviation; the 0.0% headline confirms the engine's min-fuel enumeration enforces the published contract under heterogeneous polynomials. EnergyPlus has no two-technology brute-force dispatch primitive; per-technology Generator:* fidelity is validated cross-engine in Stage 1.

Reference engine: CHP + Boiler coupling regression. The reference columns are per-hour expected CHP thermal output and boiler residual computed independently from the engine's coupling formula. Under Stage 1 anchor (linear thermal poly through origin, no degradation, OADB at ISO ref): CHP h_out = PLR_command * cap_h when both electric and thermal load are above their respective min_plr floors; engine idles at min_plr_h * cap_h when EITHER load falls below the floor (Idle-at-Minimum-Capacity with shutdown_load_priority Either Load). Boiler residual = max(0, thermal_load - CHP_h_out). Sinusoidal diurnal thermal_load swings 0.45-1.25 of CHP cap_h so the residual cycles between 0 (CHP oversized -> waste) and a positive value (CHP undersized -> boiler covers gap) within each day. NOT a cross-engine deviation; the 0.0% headline confirms the engine's CHP thermal output trajectory composes correctly with the documented coupling formula. The downstream boiler module is independently validated against EnergyPlus Boiler:HotWater in the boiler module report (192 scenarios at ASHRAE Guideline 14-2023 strict).

Reference engine: CHP + BESS coupling regression. The reference columns are per-hour expected CHP electric / thermal / fuel + BESS charge / discharge / SoC computed independently from the engine. CHP under Stage 1 anchor (linear curves, no degradation, OADB at ISO ref) runs Electric Load Following clamped at [min_plr_e * cap_e, cap_e]; thermal = (eta_h/eta_e) * electric, fuel = electric / eta_e. BESS dispatch (documented in chp_stage9 _simulate_bess): charges from CHP surplus at sqrt(rt_eff) efficiency, discharges to fill deficit at sqrt(rt_eff) efficiency, SoC tracked across hours within [0, capacity]. NOT a cross-engine deviation; the 0.0% headline confirms the engine's CHP output composes correctly with the documented BESS coupling formula. NREL SAM Generic System + Battery is cited as the industry reference in customer materials but EnergyPlus has no CHP + BESS coupled primitive; per-component physics is validated cross-engine in Stage 1 (CHP Generator:*) and the BESS module Stage 1 (ElectricLoadCenter:Storage:Simple). The GT+2xRICE+BESS+Boiler real-fixture extension layers a true cross-engine 4-equipment plant validation: CogenS multisize CHP simulator + analytical BESS + analytical boiler residual vs EnergyPlus 26.1 3-IDF reference (GT + 2-unit RICE + Boiler) + closed-form analytical.

Reference engine: Full microgrid coupling regression. The reference columns are per-hour expected CHP electric/thermal/fuel + PV schedule + BESS charge/discharge/SoC computed independently from the engine. Under Stage 1 anchor (linear curves, no degradation, OADB at ISO ref), CHP under Electric Load Following clamped at [min_plr_e*cap_e, cap_e]; thermal = (eta_h/eta_e)*electric, fuel = electric/eta_e. PV is from the documented hourly schedule. BESS dispatch: charges from net surplus (CHP + PV - load > 0) at sqrt(rt_eff), discharges to fill deficit at sqrt(rt_eff), SoC tracked across hours within [0, capacity]. Residual closes to grid_import / grid_export. NOT a cross-engine deviation; the 0.0% headline confirms the engine's CHP output composes correctly with the documented PV + BESS + grid coupling. NREL SAM Microgrid is cited as the industry reference in customer materials but EnergyPlus has no integrated CHP + BESS + PV primitive; per-component physics is validated cross-engine in Stage 1 (CHP), BESS Stage 1 (BESS), and PV via standard irradiance schedules. The GT+2xRICE+BESS+Boiler+PV full-microgrid real-fixture extension layers a true cross-engine 5-element microgrid validation: CogenS multisize CHP + analytical BESS + PV + boiler residual vs EnergyPlus 26.1 3-IDF reference (GT + 2-unit RICE + Boiler) + closed-form analytical, with the PV layer using a synthetic Chicago profile calibrated to NREL PVWatts default capacity factor 0.20.

Reference engine: Year-by-year cash-flow regression. The reference columns are per-year analytical operating PV (NIST Handbook 135 EndOfYear discounting), cumulative NPV, IRR (bisection root of the constant-cash-flow NPV polynomial), and IRS Publication 946 MACRS GDS 5-year depreciation. The CogenS financial engine (LifecycleCostAnalyzer + IRR + DepreciationCalculator) is technology-agnostic and was already cross-validated against EnergyPlus LifeCycleCost at +/-0.0000% in Boiler Stage 6 and Chiller Stage 6 — re-invoking it for CHP cash-flow streams produces the same closed-form values by construction, which is why the headline lands at 0.0000% across all four gates.

Reference engine: Year-by-year EaaS revenue regression. The reference columns are per-year analytical electric revenue (NIST Handbook 135 geometric escalation series, esc applied year-over-year from year 1 un-escalated), flat thermal revenue, flat capacity payment, and flat waste-heat capture revenue. cogens side is the output of CogenS calculate_eaas_revenue (eaas_calculator.py:27-33) — the rate-based EaaS contract carried over from Matlab Fin_PPARate.m parity. The 0.0000% headline confirms the rate-based EaaS revenue formula matches the documented closed-form on every year of every scenario.