Code a Manual-Entry electric "Direct acting" boiler to SAP 191 🟩

A PasHub Manual-Entry electric boiler lodges "Direct acting" in its "Heating
System (Boiler):" age-band cell. `_PASHUB_MANUAL_BOILER_AGE_BAND_TO_TABLE_4B`
covers only gas/liquid Table 4b rows, so the descriptor resolved to None and the
boiler stayed uncoded — taking the generic 0.80 gas-boiler seasonal efficiency
AND, on a Dual meter, `_table_12a_system_for_main`'s None-fallback that bills the
whole main-heating load at the off-peak LOW rate. The two errors pull opposite
ways, so an identical mis-mapping over-rated one dwelling and under-rated another
(fixture 497712825571 +9.5 on a Dual meter; 461386632387 -6.2 on a Single meter).

A direct-acting electric "boiler with radiators" is SAP 10.2 Table 4a code 191
(direct-acting electric boiler, efficiency 1.00, §12 Rule 3 tariff). Code it
directly in the manual-boiler resolver (gated on fuel Electricity + no PCDB
product), so the single code value drives the efficiency table, the Table 12a
pricing row and the tariff dispatch together. Both fixtures converge:
497712825571 → +0.6, 461386632387 → +0.4. LRHA within-0.5 52.4% → 53.4%,
MAE 0.835. Renamed `_pashub_manual_boiler_table_4b_code` → `_pashub_manual_
boiler_code` (it now also returns the Table 4a electric code).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

backend/documents_parser/tests/test_pashub_sap_accuracy_lrha_wave3.py

@@ -96,11 +96,17 @@ _MANIFEST_PATH = _FIXTURES_DIR / "manifest.json"
 # code-691 dwellings. Routed the resolved `_rdsap_tariff` into the secondary cost
 # path (mirroring the CO2 path, which already did this) → within-0.5 49.5% →
 # 52.4%, MAE 0.977. Storage-heater dwellings (7-hour by §12) are unchanged.
+# 2026-07-24 (accuracy tail audit): a Manual-Entry electric "Direct acting"
+# boiler was left uncoded (Table 4b covers only gas/liquid), so it took the 0.80
+# gas-boiler default AND, on a Dual meter, 100%-off-peak billing → ±6-9 mis-rate.
+# Coded it to SAP Table 4a 191 (direct-acting electric boiler, eff 1.00): fixture
+# 497712825571 +9.5 → +0.6, 461386632387 -6.2 → +0.4 → within-0.5 52.4% → 53.4%,
+# MAE 0.835.
 # Treat the constants as a regression tripwire, NOT a cohort accuracy figure;
 # re-baseline (coverage growth is not loosening) as further mapper fixes unblock
 # fixtures.
-_MIN_WITHIN_HALF: float = 0.52
-_MAX_SAP_MAE: float = 0.98
+_MIN_WITHIN_HALF: float = 0.53
+_MAX_SAP_MAE: float = 0.84
 
 _KNOWN_GAP_REASON = (
     "pashub `from_site_notes` mapper does not int-code a main-heating "
@@ -161,6 +167,27 @@ def _evaluate(deal_id: str) -> Outcome:
     )
 
 
+@pytest.mark.skipif(
+    "497712825571" not in _BY_ID or "461386632387" not in _BY_ID,
+    reason="direct-acting electric boiler fixtures not in manifest",
+)
+@pytest.mark.parametrize("deal_id", ["497712825571", "461386632387"])
+def test_direct_acting_electric_boiler_codes_to_191(deal_id: str) -> None:
+    """Regression: a Manual-Entry electric "Direct acting" boiler is a
+    direct-acting electric wet system → SAP Table 4a code 191 (efficiency 1.00,
+    §12 Rule 3 tariff). Left uncoded it took the generic 0.80 gas-boiler default
+    AND, on a Dual meter, billed its whole load at the off-peak low rate — two
+    errors that pull opposite ways (fixture 497712825571 over-rated +9.5 on a
+    Dual meter; 461386632387 under-rated -6.2 on a Single meter). Coding it 191
+    fixes both (the code drives the efficiency table, the Table 12a pricing row
+    and the §12 tariff together): 497712825571 +9.5 → +0.6, 461386632387
+    -6.2 → +0.4. The bound is 1.0 (not 0.5) — the fix removes the ±6-9 mis-code,
+    leaving only sub-point residuals within the cohort's normal spread."""
+    outcome = _evaluate(deal_id)
+    assert outcome.diff is not None
+    assert outcome.diff < 1.0, f"direct-acting electric boiler mis-coded: diff {outcome.diff:.2f}"
+
+
 @pytest.mark.skipif(
     "462024626384" not in _BY_ID, reason="fixture 462024626384 not in manifest"
 )

datatypes/epc/domain/mapper.py

@@ -8578,14 +8578,24 @@ def _pashub_main_heating_control_code(
     return code
 
 
+# SAP 10.2 Table 4a code 191 = "direct-acting electric boiler" (efficiency 1.00,
+# §12 Rule 3 tariff dispatch). A Manual-Entry electric boiler lodges "Direct
+# acting" in its "Heating System (Boiler):" age-band cell; Table 4b (gas/liquid
+# only) does not cover it, so without an explicit code it stayed uncoded → billed
+# at the generic 0.80 gas default AND, on a Dual meter, 100% off-peak low-rate — a
+# large mis-rate in BOTH directions (over-rate on Dual, under-rate on Single).
+_PASHUB_DIRECT_ACTING_ELECTRIC_BOILER_AGE_BAND: Final[str] = "Direct acting"
+_SAP_DIRECT_ACTING_ELECTRIC_BOILER_CODE: Final[int] = 191
+
 # PasHub Manual-Entry boiler descriptor → SAP 10.2 Table 4b sub-row code
 # (`table_4b.py`), keyed on the surveyed "Heating System (Boiler):" cell
 # (`boiler_age_band`), which carries both the boiler type and the pre/post-1998
 # age band. A Manual-Entry boiler has no PCDB `product_id`, so its efficiency
 # must come from Table 4b, not the generic 0.80 gas default (#1649 A). Only the
-# cohort's observed gas/liquid descriptors are mapped; an unmapped descriptor
-# (or an electric "Direct acting" boiler, which Table 4b does not cover) resolves
-# to None, preserving the pre-existing behaviour (no block, no regression).
+# cohort's observed gas/liquid descriptors are mapped; an unmapped gas/liquid
+# descriptor resolves to None, preserving the pre-existing behaviour (no block,
+# no regression). The electric "Direct acting" boiler is handled separately
+# (Table 4a code 191, above) since it is not a Table 4b row.
 _PASHUB_MANUAL_BOILER_AGE_BAND_TO_TABLE_4B: Dict[str, int] = {
     # Gas boilers 1998 or later (Table 4b 101-109).
     "1998 or later - Condensing combi": 104,  # 84/75 winter/summer
@@ -8599,14 +8609,22 @@ _PASHUB_MANUAL_BOILER_AGE_BAND_TO_TABLE_4B: Dict[str, int] = {
 }
 
 
-def _pashub_manual_boiler_table_4b_code(main: PasHubMainHeating) -> Optional[int]:
+def _pashub_manual_boiler_code(main: PasHubMainHeating) -> Optional[int]:
     """Resolve a Manual-Entry boiler's surveyed descriptor to its SAP 10.2
-    Table 4b code (#1649 A). Returns None for a PCDF-Search boiler (product_id
-    lodged — efficiency is the SEDBUK lookup) and for any descriptor the Table
-    4b map does not cover (e.g. an electric boiler), leaving the calculator's
-    existing cascade untouched for those."""
+    main-heating code. Returns None for a PCDF-Search boiler (product_id lodged
+    — efficiency is the SEDBUK lookup) and for any gas/liquid descriptor the
+    Table 4b map does not cover, leaving the calculator's existing cascade
+    untouched. An electric "Direct acting" boiler is a direct-acting electric
+    wet system → Table 4a code 191 (efficiency 1.00), NOT a Table 4b gas/liquid
+    row; without this it stayed uncoded and was billed at the 0.80 gas default
+    and (on a Dual meter) the off-peak low rate."""
     if main.product_id > 0:
         return None
+    if (
+        main.boiler_age_band == _PASHUB_DIRECT_ACTING_ELECTRIC_BOILER_AGE_BAND
+        and main.fuel == "Electricity"
+    ):
+        return _SAP_DIRECT_ACTING_ELECTRIC_BOILER_CODE
     return _PASHUB_MANUAL_BOILER_AGE_BAND_TO_TABLE_4B.get(main.boiler_age_band)
 
 
@@ -8645,9 +8663,10 @@ def _pashub_main_heating_system_code(main: PasHubMainHeating) -> Optional[int]:
         return code
     if system in _PASHUB_BOILER_SYSTEM_TYPES:
         # PCDF-Search boiler → None (SEDBUK product efficiency); Manual-Entry
-        # boiler → its Table 4b code (or None for an uncovered/non-Table-4b
-        # descriptor, keeping the pre-existing cascade) (#1649 A).
-        return _pashub_manual_boiler_table_4b_code(main)
+        # boiler → its Table 4b code, Table 4a 191 for an electric direct-acting
+        # boiler, or None for an uncovered descriptor (keeping the pre-existing
+        # cascade) (#1649 A).
+        return _pashub_manual_boiler_code(main)
     return None