Utility Tariff and Metering Test Matrix

Reference engine: NREL System Advisor Model PySAM 7.1.1 Utilityrate5 with ur_ec_tou_mat populated as a single period at the flat rate, demand charges disabled (ur_dc_enable=0), monthly fixed charge zeroed out. Bill output read from Outputs.utility_bill_w_sys_year1 (scalar annual) and Outputs.charge_w_sys_ec_ym (12-tuple monthly). Both engines collapse to load times rate by construction.

Reference engine: Same NREL SAM PySAM 7.1.1 Utilityrate5 contract as Stage 1, run through the parallel annual-driver entry point. Confirms the annual driver and plumbing that subsequent Stages 2B-12B would use.

Reference engine: NREL SAM PySAM 7.1.1 Utilityrate5 with ur_ec_sched_weekday[12][24], ur_ec_sched_weekend[12][24], and ur_ec_tou_mat[N][6] derived from the same TariffSchedule abstraction that drives CogenS. Post-init coverage check verifies every (month, hour, daytype) cell is claimed by exactly one HourBlock. Calendar pinned to Monday January 1 2018 to match the internal SAM weekday/weekend mapping.

Reference engine: NREL SAM PySAM 7.1.1 Utilityrate5 with ur_ec_tou_mat rows of [period_idx, tier_idx, max_usage, units='kWh', buy_rate, sell_rate] per tier. Cumulative-monthly threshold model: tiers reset on the month boundary; each hour's kWh allocated to whichever tier the running cumulative position falls into.

Reference engine: NREL SAM PySAM 7.1.1 Utilityrate5 with ur_dc_enable=1 plus ur_dc_flat_mat (flat NCD) and ur_dc_tou_mat plus ur_dc_sched_weekday / ur_dc_sched_weekend (TOU demand) derived from the same DemandSchedule abstraction that drives CogenS. The translator merges all-hours weekday-plus-weekend NCD pairs into a single DemandPeriod to match SAM's combined-daytype monthly peak.

Reference engine: NREL SAM PySAM 7.1.1 Utilityrate5 with both the energy-rate inputs (Stage 2 path) and the demand-rate inputs (Stage 4 path) populated jointly. The composition stage; no new SAM contract beyond the two predecessor stages.

Reference engine: Year 1: live NREL SAM PySAM 7.1.1 Utilityrate5 (matched handshake on the Stage 5 combined bundle). Years 2 through 10: NIST Handbook 135 closed-form escalation formula, applied as the analytical reference because SAM Utilityrate5's bill output for years 2 through N is a flat repetition of year 1 (the rate_escalation input is consumed downstream by SAM Cashloan and SingleOwner, not by Utilityrate5's bill output itself).

Reference engine: NREL SAM PySAM 7.1.1 Utilityrate5 with ur_metering_option=0 (kWh-netted, paired with ur_nm_yearend_sell_rate for the year-end true-up compensation rate) or ur_metering_option=1 (dollar credits, paired with ur_sell_eq_buy=1). PV generation injected via ur.SystemOutput.gen as a deterministic noon-bell synthetic profile sized to 30% of annual load - byte-identical to what CogenS sees.

Reference engine: NREL SAM PySAM 7.1.1 Utilityrate5 (Stage 7 base) plus an explicit NBC computation on the gross hourly imports applied identically on both engines. Both engines see byte-identical gross imports by construction (load and PV are identical, so max(load - pv, 0) is identical) so the NBC adder closes to the penny.

Reference engine: NREL SAM PySAM 7.1.1 Utilityrate5 with ur_metering_option=2 (Net Billing) or ur_metering_option=3 (Net Billing with carry-over). Sell rate carried in the sixth column of ur_ec_tou_mat (per-period sell rate); ur_sell_eq_buy=0 so SAM uses the encoded sell rates rather than mirroring the buy rate.

Reference engine: NREL SAM PySAM 7.1.1 Utilityrate5 with ur_metering_option=4 (Buy-All Sell-All) and ur_sell_eq_buy=0. The two energy flows live on separate meters: full load bills at retail with no on-site-generation offset, full generation sells at wholesale as a separate revenue stream.

Reference engine: NREL SAM PySAM 7.1.1 Utilityrate5 (ur_metering_option=1, ur_dc_enable=1) consuming the same hourly net-grid-import profile that the CogenS linear-programming dispatch solver produces. The LP coordinates CHP electric output (eta_elec=0.35, eta_th=0.45), BESS charge/discharge (RTE=0.90), boiler (eta=0.85), and TESS (RTE=0.95) against the same combined utility bundle both engines see. NEM 1.0 with year-end true-up applied on both sides.

Reference engine: Two-layer cross-validation. Layer A (all 9 scenarios): an independent closed-form analytical reference - standard NPV formula, scipy.optimize.brentq for the IRR root-finder, linear-interpolation simple payback against the cumulative cash-flow zero-crossing, and the published IRS Publication 946 MACRS GDS percentage tables (5-year property for PV and BESS, 7-year property for CHP, half-year convention) for the depreciation tax shield - both sides use identical published formulas so agreement to floating-point precision is by construction. Layer B (3 representative scenarios, Hospital/Office/Retail x PG&E B-19): REAL PySAM Singleowner 7.1.1 (NREL SAM C++ kernel via PySAM bindings) using a weighted Depreciation.depr_custom_schedule that aggregates the mixed-asset MACRS schedules into a single per-year array, with 30% PV ITC applied as TaxCreditIncentives.itc_fed_amount in year 1. Agreement at exactly 0.0000% adjusted-NPV deviation with a documented savings-tax architectural disclosure (the CogenS engine treats bill savings as tax-free; PySAM correctly treats them as taxable income because they reduce a tax-deductible operating expense - the implicit-tax discounted gap matches PySAM exactly).