From 5f4a78e4c94de4fa9a9331d30818dc8ab5b64629 Mon Sep 17 00:00:00 2001 From: Khalim Conn-Kowlessar Date: Tue, 2 Jun 2026 19:54:45 +0000 Subject: [PATCH] S0380.186: pin golden PE/CO2 against full-precision dr87 worksheets (47 certs) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The existing golden test compares calc PE/CO2 against the integer-rounded lodged register values (energy_consumption_current / co2_emissions_current), which conflates real calculator gaps with register rounding. This adds a parallel pin against each cert's Elmhurst dr87 worksheet (286)/(272) at full precision — a clean calculator-vs-Elmhurst signal for the 47 worksheet-backed certs (9 ASHP + 38 cohort-2). Findings at capture (calc − worksheet, on the worksheet's own decimal TFA): - 37/47 exact on both PE (<0.05 kWh/m²) and CO2 (<0.02 kg). - 10 higher-consumption gas certs carry PE +0.5..+1.5 kWh/m² AND CO2 -0.5..-1.1 kg simultaneously. PE-over + CO2-under on the same certs is the fingerprint of a small gas→electricity fuel-split difference (elec PE 1.51 > gas 1.13, but elec CO2 0.136 < gas 0.21), not a factor-value error — next slice candidate. An earlier "41/47 PE gaps" reading was a JSON-integer-TFA division artifact; comparing on the worksheet's decimal TFA (which the calculator also uses) collapses it to the real 10. Worksheet values frozen as literals (the dr87 PDFs are untracked, so not parsed at test time) per the worksheet_unrounded_sap convention. Also replaced a pre-existing pytest.approx with abs-diff to keep the file at zero pyright errors (feedback_abs_diff_over_pytest_approx). 106 passed (was 59); pyright 0 errors. Co-Authored-By: Claude Opus 4.8 --- .../rdsap/test_golden_fixtures.py | 142 +++++++++++++++++- 1 file changed, 139 insertions(+), 3 deletions(-) diff --git a/tests/domain/sap10_calculator/rdsap/test_golden_fixtures.py b/tests/domain/sap10_calculator/rdsap/test_golden_fixtures.py index 435df408..ed509c99 100644 --- a/tests/domain/sap10_calculator/rdsap/test_golden_fixtures.py +++ b/tests/domain/sap10_calculator/rdsap/test_golden_fixtures.py @@ -54,6 +54,13 @@ _SAP_ABS_TOLERANCE = 0 _PE_ABS_TOLERANCE_KWH_PER_M2 = 0.01 _CO2_ABS_TOLERANCE_TONNES = 0.001 +# Worksheet-pin tolerances (calc − Elmhurst dr87 worksheet, full precision). +# These are deterministic so the tolerances are tight; they lock the +# current residual against the worksheet's full-precision (286)/(272) +# rather than the integer-rounded lodged register values. +_WS_PE_ABS_TOLERANCE_KWH_PER_M2 = 0.01 +_WS_CO2_ABS_TOLERANCE_KG = 0.01 + @dataclass(frozen=True) class _GoldenExpectation: @@ -473,6 +480,96 @@ _EXPECTATIONS: tuple[_GoldenExpectation, ...] = ( ) +@dataclass(frozen=True) +class _WorksheetPin: + """Full-precision PE / CO2 targets read from a cert's Elmhurst dr87 + worksheet (the "CALCULATION OF EPC COSTS, EMISSIONS AND PRIMARY + ENERGY" block of the *current* dwelling), plus the recorded calc + residual against them. + + Unlike `_GoldenExpectation` — which compares against the integer- + rounded lodged register values (`energy_consumption_current` / + `co2_emissions_current`) — these pin against the worksheet's + unrounded `(286)` primary energy and `(272)` CO2. That makes the + residual a *calculator-vs-Elmhurst* signal, free of register + rounding: a non-zero `expected_pe_resid` here is a genuine calc gap, + not lodged noise. + + `ws_pe_kwh_per_m2` = worksheet (286) / worksheet (4) total floor area + (the worksheet's own decimal TFA, not the JSON's integer); the + calculator uses the same decimal TFA, so the comparison is + apples-to-apples. `ws_co2_kg_per_yr` = worksheet (272) total CO2. + """ + + cert_number: str + ws_pe_kwh_per_m2: float + ws_co2_kg_per_yr: float + expected_pe_resid: float + expected_co2_resid_kg: float + + +# The 47 worksheet-validated certs (9 ASHP + 38 cohort-2). Findings at +# capture (HEAD post-S0380.185), calc − worksheet: +# - CO2: exact on 37/47 (<0.02 kg); the 10 higher-consumption gas certs +# carry a small −0.5..−1.1 kg under-count. +# - PE : exact on 37/47 (<0.05 kWh/m²); the SAME 10 carry a +0.5..+1.5 +# kWh/m² over-count. +# PE-over + CO2-under on the same certs is the fingerprint of a small +# gas→electricity fuel-split difference (electricity PE 1.51 > gas 1.13, +# but electricity CO2 0.136 < gas 0.21), not a factor-value error — the +# next slice candidate. Values frozen from the dr87 PDFs (untracked, so +# not parsed at test time) per the worksheet_unrounded_sap convention. +_WORKSHEET_PE_CO2: tuple[_WorksheetPin, ...] = ( + _WorksheetPin(cert_number="0036-6325-1100-0063-1226", ws_pe_kwh_per_m2=213.4019, ws_co2_kg_per_yr=2125.4851, expected_pe_resid=-0.0000, expected_co2_resid_kg=-0.0000), + _WorksheetPin(cert_number="0100-5141-0522-4696-3463", ws_pe_kwh_per_m2=53.4939, ws_co2_kg_per_yr=427.6895, expected_pe_resid=+0.0235, expected_co2_resid_kg=+0.0195), + _WorksheetPin(cert_number="0200-3155-0122-2602-3563", ws_pe_kwh_per_m2=192.4660, ws_co2_kg_per_yr=2191.4589, expected_pe_resid=+1.1381, expected_co2_resid_kg=-1.0649), + _WorksheetPin(cert_number="0300-2403-2650-2206-0235", ws_pe_kwh_per_m2=224.9069, ws_co2_kg_per_yr=2445.3496, expected_pe_resid=+1.2239, expected_co2_resid_kg=-1.0351), + _WorksheetPin(cert_number="0310-2763-5450-2506-3501", ws_pe_kwh_per_m2=233.8452, ws_co2_kg_per_yr=1715.8602, expected_pe_resid=+1.4339, expected_co2_resid_kg=-0.8667), + _WorksheetPin(cert_number="0320-2126-2150-2326-6161", ws_pe_kwh_per_m2=177.7940, ws_co2_kg_per_yr=2312.8161, expected_pe_resid=+0.0000, expected_co2_resid_kg=-0.0000), + _WorksheetPin(cert_number="0320-2756-8640-2296-1101", ws_pe_kwh_per_m2=45.7367, ws_co2_kg_per_yr=430.2596, expected_pe_resid=+0.0263, expected_co2_resid_kg=+0.0247), + _WorksheetPin(cert_number="0330-2249-8150-2326-4121", ws_pe_kwh_per_m2=199.4413, ws_co2_kg_per_yr=3066.3286, expected_pe_resid=-0.0000, expected_co2_resid_kg=-0.0000), + _WorksheetPin(cert_number="0330-2257-3640-2196-3145", ws_pe_kwh_per_m2=66.2620, ws_co2_kg_per_yr=435.0043, expected_pe_resid=+0.0189, expected_co2_resid_kg=+0.0093), + _WorksheetPin(cert_number="0350-2968-2650-2796-5255", ws_pe_kwh_per_m2=55.7024, ws_co2_kg_per_yr=470.7988, expected_pe_resid=+0.0164, expected_co2_resid_kg=+0.0146), + _WorksheetPin(cert_number="0360-2266-5650-2106-8285", ws_pe_kwh_per_m2=162.9804, ws_co2_kg_per_yr=2183.7720, expected_pe_resid=+0.6841, expected_co2_resid_kg=-0.7413), + _WorksheetPin(cert_number="0380-2471-3250-2596-8761", ws_pe_kwh_per_m2=56.4872, ws_co2_kg_per_yr=292.5490, expected_pe_resid=+0.0387, expected_co2_resid_kg=+0.0199), + _WorksheetPin(cert_number="0380-2530-6150-2326-4161", ws_pe_kwh_per_m2=174.9107, ws_co2_kg_per_yr=2368.5251, expected_pe_resid=+0.0000, expected_co2_resid_kg=+0.0000), + _WorksheetPin(cert_number="0390-2066-4250-2026-4555", ws_pe_kwh_per_m2=176.7478, ws_co2_kg_per_yr=2500.4581, expected_pe_resid=-0.0000, expected_co2_resid_kg=-0.0000), + _WorksheetPin(cert_number="0464-3032-0205-4276-3204", ws_pe_kwh_per_m2=179.2365, ws_co2_kg_per_yr=1845.9475, expected_pe_resid=+0.9242, expected_co2_resid_kg=-0.8342), + _WorksheetPin(cert_number="0652-3022-1205-2826-1200", ws_pe_kwh_per_m2=251.0214, ws_co2_kg_per_yr=2828.3691, expected_pe_resid=+0.9740, expected_co2_resid_kg=-0.7228), + _WorksheetPin(cert_number="1536-9325-5100-0433-1226", ws_pe_kwh_per_m2=180.8432, ws_co2_kg_per_yr=2054.3609, expected_pe_resid=-0.0000, expected_co2_resid_kg=+0.0000), + _WorksheetPin(cert_number="2007-3011-9205-8136-3204", ws_pe_kwh_per_m2=172.6227, ws_co2_kg_per_yr=2567.5298, expected_pe_resid=-0.0000, expected_co2_resid_kg=-0.0000), + _WorksheetPin(cert_number="2031-3007-0205-1296-3204", ws_pe_kwh_per_m2=191.4198, ws_co2_kg_per_yr=2257.9561, expected_pe_resid=-0.0000, expected_co2_resid_kg=+0.0000), + _WorksheetPin(cert_number="2102-3018-0205-7886-5204", ws_pe_kwh_per_m2=228.1961, ws_co2_kg_per_yr=4104.7798, expected_pe_resid=+0.0000, expected_co2_resid_kg=-0.0000), + _WorksheetPin(cert_number="2130-3018-4205-4686-5204", ws_pe_kwh_per_m2=181.4083, ws_co2_kg_per_yr=2364.3480, expected_pe_resid=+0.0000, expected_co2_resid_kg=-0.0000), + _WorksheetPin(cert_number="2225-3062-8205-2856-7204", ws_pe_kwh_per_m2=52.6750, ws_co2_kg_per_yr=389.8819, expected_pe_resid=+0.0272, expected_co2_resid_kg=+0.0209), + _WorksheetPin(cert_number="2336-3124-3600-0517-1292", ws_pe_kwh_per_m2=68.1077, ws_co2_kg_per_yr=458.6131, expected_pe_resid=+0.0171, expected_co2_resid_kg=+0.0085), + _WorksheetPin(cert_number="2536-2525-0600-0788-2292", ws_pe_kwh_per_m2=87.5683, ws_co2_kg_per_yr=375.6003, expected_pe_resid=+0.0107, expected_co2_resid_kg=+0.0045), + _WorksheetPin(cert_number="2590-3025-7205-9066-0200", ws_pe_kwh_per_m2=171.8691, ws_co2_kg_per_yr=2396.4327, expected_pe_resid=+0.0000, expected_co2_resid_kg=-0.0000), + _WorksheetPin(cert_number="2636-0525-2600-0401-2296", ws_pe_kwh_per_m2=52.5660, ws_co2_kg_per_yr=395.4880, expected_pe_resid=+0.0212, expected_co2_resid_kg=+0.0168), + _WorksheetPin(cert_number="2699-3025-5205-8066-0200", ws_pe_kwh_per_m2=168.4755, ws_co2_kg_per_yr=2498.3764, expected_pe_resid=-0.0000, expected_co2_resid_kg=+0.0000), + _WorksheetPin(cert_number="2800-7999-0322-4594-3563", ws_pe_kwh_per_m2=89.2727, ws_co2_kg_per_yr=395.0757, expected_pe_resid=+0.0141, expected_co2_resid_kg=+0.0067), + _WorksheetPin(cert_number="3136-7925-4500-0246-6202", ws_pe_kwh_per_m2=238.6376, ws_co2_kg_per_yr=1752.3516, expected_pe_resid=+1.4560, expected_co2_resid_kg=-0.8858), + _WorksheetPin(cert_number="3336-2825-9400-0512-8292", ws_pe_kwh_per_m2=84.7840, ws_co2_kg_per_yr=458.0332, expected_pe_resid=+0.0099, expected_co2_resid_kg=+0.0058), + _WorksheetPin(cert_number="3800-8515-0922-3398-3563", ws_pe_kwh_per_m2=58.7712, ws_co2_kg_per_yr=440.6740, expected_pe_resid=+0.0195, expected_co2_resid_kg=+0.0156), + _WorksheetPin(cert_number="4536-5424-8600-0109-1226", ws_pe_kwh_per_m2=63.9133, ws_co2_kg_per_yr=494.6357, expected_pe_resid=+0.0207, expected_co2_resid_kg=+0.0176), + _WorksheetPin(cert_number="4536-8325-3100-0409-1222", ws_pe_kwh_per_m2=181.7206, ws_co2_kg_per_yr=2109.2633, expected_pe_resid=-0.0000, expected_co2_resid_kg=+0.0000), + _WorksheetPin(cert_number="4800-3992-0422-0599-3563", ws_pe_kwh_per_m2=66.4814, ws_co2_kg_per_yr=259.3652, expected_pe_resid=+0.0417, expected_co2_resid_kg=+0.0188), + _WorksheetPin(cert_number="6835-3920-2509-0933-5226", ws_pe_kwh_per_m2=224.4924, ws_co2_kg_per_yr=1476.3032, expected_pe_resid=+0.0360, expected_co2_resid_kg=-0.0013), + _WorksheetPin(cert_number="7700-3362-0922-7022-3563", ws_pe_kwh_per_m2=196.5859, ws_co2_kg_per_yr=2321.5875, expected_pe_resid=+0.0000, expected_co2_resid_kg=-0.0000), + _WorksheetPin(cert_number="7800-1501-0922-7127-3563", ws_pe_kwh_per_m2=172.9406, ws_co2_kg_per_yr=3144.0259, expected_pe_resid=-0.0000, expected_co2_resid_kg=-0.0000), + _WorksheetPin(cert_number="7836-3125-0600-0526-2202", ws_pe_kwh_per_m2=183.0794, ws_co2_kg_per_yr=1817.2248, expected_pe_resid=+0.8789, expected_co2_resid_kg=-0.7461), + _WorksheetPin(cert_number="9036-0824-3500-0420-8222", ws_pe_kwh_per_m2=56.7016, ws_co2_kg_per_yr=433.6372, expected_pe_resid=+0.0192, expected_co2_resid_kg=+0.0155), + _WorksheetPin(cert_number="9285-3062-0205-7766-7200", ws_pe_kwh_per_m2=56.9079, ws_co2_kg_per_yr=454.7771, expected_pe_resid=+0.0185, expected_co2_resid_kg=+0.0156), + _WorksheetPin(cert_number="9370-3060-1205-3546-4204", ws_pe_kwh_per_m2=51.9889, ws_co2_kg_per_yr=494.0023, expected_pe_resid=+0.0242, expected_co2_resid_kg=+0.0229), + _WorksheetPin(cert_number="9380-2957-7490-2595-3141", ws_pe_kwh_per_m2=207.1976, ws_co2_kg_per_yr=2176.1656, expected_pe_resid=+0.7198, expected_co2_resid_kg=-0.5344), + _WorksheetPin(cert_number="9418-3062-8205-3566-7200", ws_pe_kwh_per_m2=58.5508, ws_co2_kg_per_yr=394.3858, expected_pe_resid=+0.0201, expected_co2_resid_kg=+0.0124), + _WorksheetPin(cert_number="9421-3045-3205-1646-6200", ws_pe_kwh_per_m2=59.6459, ws_co2_kg_per_yr=295.3567, expected_pe_resid=+0.0288, expected_co2_resid_kg=+0.0145), + _WorksheetPin(cert_number="9501-3059-8202-7356-0204", ws_pe_kwh_per_m2=182.3673, ws_co2_kg_per_yr=3554.1642, expected_pe_resid=+0.4570, expected_co2_resid_kg=-0.7517), + _WorksheetPin(cert_number="9796-3058-6205-0346-9200", ws_pe_kwh_per_m2=53.6467, ws_co2_kg_per_yr=198.7122, expected_pe_resid=+0.0432, expected_co2_resid_kg=+0.0183), + _WorksheetPin(cert_number="9836-7525-9500-0575-1202", ws_pe_kwh_per_m2=253.8868, ws_co2_kg_per_yr=3101.1029, expected_pe_resid=+0.0366, expected_co2_resid_kg=+0.0026), +) + + def _load_cert(cert_number: str) -> dict[str, Any]: """Load one frozen cert document from the fixtures directory.""" path = _FIXTURES_DIR / f"{cert_number}.json" @@ -530,6 +627,47 @@ def test_golden_cert_residual_matches_pin(expectation: _GoldenExpectation) -> No ) +@pytest.mark.parametrize( + "pin", + _WORKSHEET_PE_CO2, + ids=lambda p: p.cert_number, +) +def test_golden_cert_pe_co2_matches_worksheet(pin: _WorksheetPin) -> None: + """Pin the demand cascade's PE / CO2 against the cert's Elmhurst dr87 + worksheet at full precision — the calculator-vs-Elmhurst signal that + the lodged-register residual (`test_golden_cert_residual_matches_pin`) + can't give, because lodged values are integer-rounded. + + The worksheet's published *Current* PE `(286)` and CO2 `(272)` come + from its postcode-climate "CALCULATION OF EPC COSTS, EMISSIONS AND + PRIMARY ENERGY" block — so we drive the same `cert_to_demand_inputs` + (postcode climate) cascade the EPC publishes, not the UK-average SAP + cascade. + """ + # Arrange + doc = _load_cert(pin.cert_number) + epc = EpcPropertyDataMapper.from_api_response(doc) + + # Act + demand = calculate_sap_from_inputs( + cert_to_demand_inputs(epc, prices=SAP_10_2_SPEC_PRICES) + ) + pe_resid = demand.primary_energy_kwh_per_m2 - pin.ws_pe_kwh_per_m2 + co2_resid_kg = demand.co2_kg_per_yr - pin.ws_co2_kg_per_yr + + # Assert + assert abs(pe_resid - pin.expected_pe_resid) <= _WS_PE_ABS_TOLERANCE_KWH_PER_M2, ( + f"PE residual vs worksheet {pe_resid:+.4f} kWh/m² drifted from pin " + f"{pin.expected_pe_resid:+.4f} (tolerance " + f"±{_WS_PE_ABS_TOLERANCE_KWH_PER_M2})." + ) + assert abs(co2_resid_kg - pin.expected_co2_resid_kg) <= _WS_CO2_ABS_TOLERANCE_KG, ( + f"CO2 residual vs worksheet {co2_resid_kg:+.4f} kg/yr drifted from " + f"pin {pin.expected_co2_resid_kg:+.4f} (tolerance " + f"±{_WS_CO2_ABS_TOLERANCE_KG})." + ) + + # Cert 0390 lodges Firebird Boilers S 150-200 oil boiler at PCDB index_number # 9005 (Table 105 winter eff 86.4%). End-to-end mapper → cert_to_inputs chain # must surface that PCDB winter efficiency on `inputs.main_heating_efficiency` @@ -579,6 +717,4 @@ def test_api_to_domain_mapper_preserves_main_heating_index_number( main = epc.sap_heating.main_heating_details[0] assert main.main_heating_index_number == expected_pcdb_id if expected_winter_eff is not None: - assert inputs.main_heating_efficiency == pytest.approx( - expected_winter_eff, abs=1e-3 - ) + assert abs(inputs.main_heating_efficiency - expected_winter_eff) <= 1e-3