Real Elmhurst RdSAP10 ground truth for one SAP-Schema-15.0 cert

Built UPRN 100010086084 (the largest gov-API-vs-lodged outlier from the
20-cert batch) in Elmhurst's accredited RdSAP10 tool to get real ground
truth, per feedback that lodged rating isn't a valid comparison target.

Elmhurst worksheet: SAP 46-48 (orientation-dependent), vs this engine's
53 on gov-API inputs -- a real ~5-7pt residual gap. Ruled out with
direct empirical tests: total_floor_area, missing party_wall_length,
region/climate (SAP rating always uses UK-average weather per Appendix
U -- confirmed inert), boiler/secondary efficiency (exact match, 66%/
63% both sides). Confirmed real but NOT blind-fixed (need more than one
cert's evidence, and a wall-U-value patch tested the wrong direction in
isolation): a wall-thickness-unknown U-value fallback mismatch (this
engine 1.70, Elmhurst's own "unknown" resolution 1.40) and a window
U-value divergence (1.85 vs 2.52 W/m2K for the same nominal glazing
pick). Single-window orientation confirmed a ~2pt partial contributor.

Also fixes a small, separate, evidenced gap the build surfaced: the
Elmhurst summary parser didn't recognise the live tool's "Medium
(131-170L)" cylinder-size label (only the bare "Medium").

Full write-up is in the RealCertExpectation comment for this cert in
test_real_cert_sap_accuracy.py and the worklist. Residual gap is
intentionally left open for follow-up, not force-closed.

.claude/skills/expand-sap-accuracy-corpus/worklist.md

@@ -405,7 +405,7 @@ methodology) is still open — see comment in test_real_cert_sap_accuracy.py.
 All 20 pinned as `RealCertExpectation`s to the engine's OWN observed output.
 - [x] 🔧 100010359769 — SAP-15.0 · eng 68 (2011 lodged 68, not a target)
 - [x] 🔧 100010359788 — SAP-15.0 · eng 72 (2011 lodged 73, not a target)
-- [x] 🔧 100010086084 — SAP-15.0 · eng 53 (2011 lodged 45, not a target) · solid-brick, largest-gap group; u_wall spot-checked correct
+- [x] 🔍 100010086084 — SAP-15.0 · eng 53, Elmhurst worksheet 48 (2011 lodged 45, not a target) · FULL Elmhurst RdSAP10 rebuild done (only cert in this batch with one). Real ~5pt residual gap, NOT the mapper's TFA/party-wall/region/boiler-efficiency defaults (all ruled out empirically). Two real bugs flagged, not fixed (need >1 cert's evidence, see test_real_cert_sap_accuracy.py comment): wall-thickness-unknown fallback (engine 1.70 vs Elmhurst 1.40, but patch-tested WRONG DIRECTION alone) and a window U-value divergence (1.85 vs 2.52 W/m2K, root cause open). Single-window orientation approximation confirmed to account for ~2pt of the gap (North->South rebuild: 46->48).
 - [x] 🔧 100010090369 — SAP-15.0 · eng 53 (2011 lodged 47, not a target) · ditto
 - [x] 🔧 100010074470 — SAP-15.0 · eng 46 (2011 lodged 39, not a target) · ditto
 - [x] 🔧 100010100813 — SAP-15.0 · eng 59 (2011 lodged 47, not a target) · ditto

datatypes/epc/domain/mapper.py

@@ -7552,6 +7552,10 @@ _ELMHURST_CYLINDER_SIZE_LABEL_TO_SAP10: Dict[str, int] = {
     "Normal": 2,
     "Medium": 3,
     "Large": 4,
+    # The live RdSAP10-Online tool renders the size band WITH its litre range
+    # suffix (observed on the SAP-Schema-15.0 UPRN 100010086084 build) rather
+    # than the bare label the cohort fixtures used above.
+    "Medium (131-170L)": 3,
 }
 
 

scripts/hyde/build_100010086084.py

@@ -0,0 +1,289 @@
+"""Elmhurst build for UPRN 100010086084 (SAP-Schema-15.0, cert
+8493-5079-7429-8996-5293, SEMI-DETACHED HOUSE, 9 Pear Tree Avenue, Crewe
+CW1 3SH, band C 1930-1949, SOLID BRICK uninsulated (thickness not lodged --
+UNKNOWN, not a specific mm value), pitched 300mm loft insulation, SOLID
+ground floor uninsulated, mains-gas generic (no-PCDB) REGULAR boiler — SAP
+Table 4b code 115 (pre-1998, balanced/open flue, "BGL Pre 98 Wall mount
+BF/OF") + programmer/room-thermostat control (2104/CBC) + radiators +
+cylinder (size 2 = "Medium (131-170L)" on the live tool, foam, 25mm), gas
+room-heater SECONDARY (603, approximated as RGB "Modern fire with open flue
+post 1980" -- the cert gives no finer detail), TFA 68, single meter,
+2-storey / 2 floor-dims (no extensions). Partial double glazing
+(multiple_glazing_type 1, multiple_glazed_proportion 50 -- NOT modelled as
+partial; this mapper's own `_api_cascade_glazing_type(1)` treats it as 100%
+double too, so the Elmhurst build matches the mapper's own simplification),
+glazed_area band 1 (Normal): 0.148*68=10.06 m^2 combined window.
+
+2011 lodged 45; this RdSAP10 engine on gov-API inputs produces 53. Elmhurst's
+OWN worksheet, built here, gives 46 (window South-facing: 48 -- see below).
+NEITHER is the "right answer" to chase blindly -- see the investigation
+write-up in tests/domain/sap10_calculator/test_real_cert_sap_accuracy.py
+next to this cert's RealCertExpectation for the full comparison, what was
+ruled out, and the two real-but-unfixed bugs this build surfaced (wall-
+thickness-unknown U-value fallback mismatch; a window U-value divergence).
+
+GOTCHAS specific to this build (fold into the next SAP-Schema-15.0 cert):
+  * The shared assessment WILL carry a stale PCDB boiler ref on Main Heating
+    1 from a prior build -- clear `TextBoxPCDFBoilerReference` to "0" (JS +
+    save) BEFORE `ButtonMainHeatingCode` becomes clickable (mirrors the
+    documented storage-heater "clear the PCDF boiler first" pattern).
+  * Wall thickness: use `CheckBoxWallThicknessUnknown`, NOT a blank/skipped
+    field -- an untouched thickness textbox silently keeps whatever a PRIOR
+    cert set (bit this build once: inherited a stale 280mm, landing on the
+    exact documented Table 13 200/280mm edge case).
+  * `set_single_window`'s orientation choice is NOT neutral for a cert this
+    reduced-field: North gave 46, South gave 48 (max solar gain) on this
+    cert. South is the fairer bound (this mapper's own N/E/S/W synthesis
+    spreads solar gain; a single North-facing pick is the worst case).
+  * The Roof/Cylinder thickness dropdowns need an EXACT option-text match
+    ("300 mm" not "300 mm or more"; "25 mm" not "25") -- `set_select` fails
+    silently (no exception) on a non-matching value and leaves the field at
+    its previous/inherited state. ALWAYS verify with `.input_value()` after
+    setting, don't trust the call succeeded.
+  * Address/region are carried over from whatever cert built the shared
+    assessment last -- explicitly set them (see `address()` below) even
+    though region is provably inert for the SAP rating itself (UK-average
+    weather always, `_region_index` in cert_to_inputs.py) -- other summary
+    fields (post town) still read from it.
+
+Run:
+    DISPLAY=:99 python scripts/hyde/build_100010086084.py <page>
+"""
+from __future__ import annotations
+import sys
+import elmhurst_lib as E
+
+DIM = "TabContainer_TabPanelMain_WebUserControlDimensionsMain_"
+WALL = ("TabContainer_TabPanelMain_InnerTabContainerMain_"
+        "TabPanelExternalWallMain_WebUserControlWallMain_")
+PWALL = "TabContainer_TabPanelMain_InnerTabContainerMain_TabPanelPartyWallMain_WebUserControlPartyWallMain_"
+ROOF = "TabContainer_TabPanelMain_WebUserControlRoofMain_"
+FLOOR = "TabContainer_TabPanelMain_WebUserControlFloorsMain_"
+DP = "TabContainer_TabPanelDoorsPanel_"
+VP = "TabContainer_TabPanelVentilationPanel_"
+APT = "TabContainer_TabPanelAirPressureTest_"
+LP = "TabContainer_TabPanelLighting_"
+MV = "TabContainer_TabPanelMechVent_"
+WH = "TabContainer_TabPanelWaterHeating_"
+MH1B = "TabContainer_TabPanelMainHeating1_WebUserControlMainHeating1_"
+
+
+def _pick(page, suffix, contains):
+    val = page.evaluate(
+        """(a)=>{const s=document.getElementById(a[0]);if(!s)return null;
+        for(const o of s.options){if(o.text.toLowerCase().includes(a[1].toLowerCase()))return o.value;}return null;}""",
+        [f"{E.FP}{suffix}", contains])
+    if val is not None:
+        E.set_select(page, suffix, val)
+    return val
+
+
+def address(page):
+    page.goto(E.ENTRY_URL, wait_until="networkidle", timeout=60_000)
+    E.set_text(page, "TextBoxHouseNo", "9")
+    E.set_text(page, "TextBoxStreet", "Pear Tree Avenue")
+    E.set_text(page, "TextBoxTown", "Crewe")
+    E.set_text(page, "TextBoxPostCode", "CW1 3SH")
+    page.wait_for_timeout(300)
+    E.set_select(page, "DropDownListRegion", "West Pennines")
+    page.click("#ContentBody_ContentPlaceHolderNavigations_ButtonSaveClose", timeout=15_000)
+    page.wait_for_timeout(1000)
+
+
+def property_description(page):
+    E.goto(page, "PropertyDescription", "WebFormPropertyDescription.aspx")
+    E.set_select(page, "DropDownListPropertyType1", "H House")
+    _pick(page, "DropDownListPropertyType2", "semi")  # built_form 2
+    E.set_text(page, "TextBoxStoreys", "2")
+    E.set_text(page, "TextBoxHabitableRooms", "3")
+    E.set_text(page, "TextBoxHeatedHabitableRooms", "3")
+    print("date ->", _pick(page, "DropDownListDateBuiltMain", "1930-1949"))  # band C
+    E.set_select(page, "DropDownListDateBuiltFirst", "")
+    E.set_select(page, "DropDownListRoomInRoofMain", "")
+    E.save_close(page)
+
+
+def dimensions(page):
+    E.goto(page, "Dimensions", "WebFormDimensions.aspx")
+    # Ground floor: area 37.74, heat_loss_perimeter 21.05 -> geometry-derived
+    # party wall (the shared/depth side) ~4.58m.
+    E.set_text(page, f"{DIM}TextBoxFloorAreaLowestFloor", "37.74")
+    E.set_text(page, f"{DIM}TextBoxRoomHeightLowestFloor", "2.41")
+    E.set_text(page, f"{DIM}TextBoxWallPerimeterLowestFloor", "21.05")
+    E.set_text(page, f"{DIM}TextBoxPartyWallLengthLowestFloor", "4.58")
+    # First floor: area 39.48, heat_loss_perimeter 18.45 -> party wall ~6.75m.
+    E.set_text(page, f"{DIM}TextBoxFloorArea1stFloor", "39.48")
+    E.set_text(page, f"{DIM}TextBoxRoomHeight1stFloor", "2.45")
+    E.set_text(page, f"{DIM}TextBoxWallPerimeter1stFloor", "18.45")
+    E.set_text(page, f"{DIM}TextBoxPartyWallLength1stFloor", "6.75")
+    E.save_close(page)
+
+
+def walls(page):
+    E.goto(page, "Walls", "WebFormWalls.aspx")
+    E.set_select(page, f"{WALL}DropDownListType", "SO Solid brick")
+    page.wait_for_timeout(400)
+    print("insulation ->", _pick(page, f"{WALL}DropDownListInsulation", "as built"))
+    # Thickness NOT lodged on this schema -- mark unknown explicitly, don't
+    # leave the textbox untouched (it silently inherits a stale value).
+    cb = page.locator(f"#{E.FP}{WALL}CheckBoxWallThicknessUnknown")
+    if cb.count() and not cb.is_checked():
+        E.commit(page, cb.check)
+    pw = _pick(page, f"{PWALL}DropDownListPartyWallType", "solid") \
+        or _pick(page, f"{PWALL}DropDownListPartyWallType", "determine")
+    print("party wall ->", pw)
+    E.save_close(page)
+
+
+def roofs(page):
+    E.goto(page, "Roofs", "WebFormRoofs.aspx")
+    _pick(page, f"{ROOF}DropDownListType", "access to loft")
+    _pick(page, f"{ROOF}DropDownListInsulationAt", "joists")
+    E.set_select(page, f"{ROOF}DropDownListThickness", "300 mm")  # exact match
+    print("roof thickness now:", page.locator(f"#{E.FP}{ROOF}DropDownListThickness").input_value())
+    E.save_close(page)
+
+
+def floors(page):
+    E.goto(page, "Floors", "WebFormFloors.aspx")
+    E.set_select(page, f"{FLOOR}DropDownListLocation", "G Ground  floor")
+    _pick(page, f"{FLOOR}DropDownListType", "solid")
+    E.set_select(page, f"{FLOOR}DropDownListInsulation", "A As built")
+    E.save_close(page)
+
+
+def openings(page):
+    E.goto(page, "Openings", "WebFormOpenings.aspx")
+    E.click_tab(page, "TabContainer_TabPanelWindowsPanel")
+    # glazed_area band 1 (Normal): total = 0.148*68*1.00 = 10.06 m^2. multiple_
+    # glazing_type=1 -> this mapper's own _api_cascade_glazing_type(1) treats
+    # it as 100% double (ignores multiple_glazed_proportion=50 the same way),
+    # so "Double, unknown install date" matches the mapper's own assumption.
+    # South (not North) -- max solar gain, the fairer single-orientation
+    # bound against this mapper's own N/E/S/W synthesis (see module docstring).
+    E.set_single_window(page, 10.06, "South", "Double with unknown install date")
+    E.click_tab(page, "TabContainer_TabPanelDoorsPanel")
+    E.set_text(page, f"{DP}TextBoxDoors", "1")
+    E.set_text(page, f"{DP}TextBoxDoorsInsulated", "0")
+    E.set_text(page, f"{DP}TextBoxDraughtProofedDoors", "0")
+    E.save_close(page)
+
+
+def ventilation(page):
+    E.goto(page, "VentilationAndCooling", "WebFormVentilationAndCooling.aspx")
+    E.click_tab(page, "TabContainer_TabPanelVentilationPanel")
+    E.set_text(page, f"{VP}TextBoxIntermittentFans", "0")
+    cool = page.locator(f"#{E.FP}{VP}CheckBoxFixedSpaceCooling")
+    if cool.count() and cool.is_checked():
+        E.commit(page, cool.uncheck)
+    E.click_tab(page, "TabContainer_TabPanelMechVent")
+    mv = page.locator(f"#{E.FP}{MV}CheckBoxMechanicalVentilation")
+    if mv.count() and mv.is_checked():
+        E.commit(page, mv.uncheck)
+    E.click_tab(page, "TabContainer_TabPanelAirPressureTest")
+    E.set_select(page, f"{APT}DropDownListTestMethod", "Not available")
+    E.click_tab(page, "TabContainer_TabPanelLighting")
+    # low_energy_lighting 43% of 7 fixed outlets -> 3 low-energy.
+    E.set_text(page, f"{LP}TextBoxLightsTotal", "7")
+    E.set_text(page, f"{LP}TextBoxLedLightsTotal", "3")
+    E.set_text(page, f"{LP}TextBoxCflLightsTotal", "0")
+    E.save_close(page)
+
+
+def space_heating(page):
+    # SAP Table 4b code 115: pre-1998 gas boiler, balanced/open flue, no PCDB
+    # index (main_heating_data_source=2). Regular boiler (has_hot_water_
+    # cylinder=true, water_heating_code 901 "from primary") + radiators
+    # (heat_emitter_type=1), fan_flue_present=N.
+    E.goto(page, "SpaceHeating", "WebFormSpaceHeating.aspx")
+    page.wait_for_timeout(1000)
+    # The shared assessment's Main Heating 1 slot carries a stale PCDB boiler
+    # ref from a prior build -- clear it first (JS + save) so the SAP-code
+    # button appears; mirrors the documented storage-heater pattern.
+    ref = page.locator(f"#{E.FP}{MH1B}TextBoxPCDFBoilerReference")
+    if ref.count() and ref.input_value():
+        print(f"clearing leftover PCDF boiler ref {ref.input_value()}")
+        page.evaluate("""(id)=>{const e=document.getElementById(id);if(e){e.value='0';
+            e.dispatchEvent(new Event('change',{bubbles:true}));}}""", f"{E.FP}{MH1B}TextBoxPCDFBoilerReference")
+        page.wait_for_timeout(500)
+        E.save_close(page)
+        E.goto(page, "SpaceHeating", "WebFormSpaceHeating.aspx")
+        page.wait_for_timeout(1000)
+    mhc = page.locator(f"#{E.MH1}TextBoxMainHeatingCode")
+    code = mhc.input_value() if mhc.count() else ""
+    if code and code not in ("0",):
+        print(f"clearing leftover MainHeatingCode {code}")
+        page.evaluate("""(id)=>{const e=document.getElementById(id);if(e){e.value='';
+            e.dispatchEvent(new Event('change',{bubbles:true}));}}""", f"{E.MH1}TextBoxMainHeatingCode")
+        page.wait_for_timeout(400)
+        E.save_close(page)
+        return
+    # Generic (no-PCDB) mains-gas REGULAR boiler, pre-1998 balanced/open flue
+    # (matches boiler_flue_type=1, fan_flue_present=N) -> Gas -> Mains gas ->
+    # Boilers -> Pre 1998 - Balanced/Open Flue -> Wall mount (regular, not
+    # combi/back-boiler -- matches has_hot_water_cylinder=true).
+    E.set_heating_dialog(page, f"{MH1B}ButtonMainHeatingCode",
+                         "^Gas", "Mains gas", "Boilers", "Balanced.*Open",
+                         "Wall mount")
+    print("code:", page.locator(f"#{E.MH1}TextBoxMainHeatingCode").input_value())
+    # Control 2104 = "Programmer and room thermostat" (matches the cert's own
+    # main_heating_controls description) -- no TRVs. NB: CBC is "Programmer
+    # and room thermostat"; CBB is "Room thermostat and TRVs" -- do not swap.
+    E.set_heating_dialog(page, f"{MH1B}ButtonMainHeatingControls",
+                         "^Boilers", "^Standard", "CBC Programmer and room thermostat")
+    print("control:", page.locator(f"#{E.MH1}TextBoxMainHeatingControls").input_value())
+    E.save_close(page)
+
+
+def secondary(page):
+    # Lodged secondary: SAP 603 = room heaters, mains gas (no finer detail
+    # lodged -- RGB "Modern fire with open flue post 1980" is the generic
+    # approximation).
+    E.goto(page, "SpaceHeating", "WebFormSpaceHeating.aspx")
+    page.wait_for_timeout(600)
+    E.set_select(page, "DropDownListSecondaryHeatingPresent", "Yes")
+    page.wait_for_timeout(900)
+    E.set_heating_dialog(page, "ButtonSecondaryHeatingCode",
+                         "^Gas", "Mains gas", "Room Heater", "RGB Modern fire with open flue")
+    tb = page.locator(f"#{E.FP}TextBoxSecondaryHeatingCode")
+    print("secondary code:", tb.input_value() if tb.count() else "?")
+    E.save_close(page)
+
+
+def water_heating(page):
+    E.goto(page, "WaterHeating", "WebFormWaterHeating.aspx")
+    E.click_tab(page, "TabContainer_TabPanelWaterHeating")
+    page.wait_for_timeout(400)
+    E.set_heating_dialog(page, f"{WH}ButtonWaterHeatingCode",
+                         "From Space Heating", "From the primary heating system")
+    print("water code:", page.locator(f"#{E.FP}{WH}TextBoxWaterHeatingCode").input_value())
+    # cylinder: size 2 (lodged cylinder_size code -- set_select matches by
+    # <option VALUE>, not visible text; value "2" renders as "Medium
+    # (131-170L)" on the live tool -- see the mapper fix for that label).
+    # foam insulation (type 1), 25mm (value happens to equal visible text
+    # here: "25 mm").
+    E.set_select(page, f"{WH}DropDownListCylinderSize", "2")
+    E.set_select(page, f"{WH}DropDownListInsulated", "Foam")
+    E.set_select(page, f"{WH}DropDownListInsulationThickness", "25 mm")
+    print("cylinder size:", page.locator(f"#{E.FP}{WH}DropDownListCylinderSize").input_value())
+    print("cylinder thickness:", page.locator(f"#{E.FP}{WH}DropDownListInsulationThickness").input_value())
+    E.save_close(page)
+
+
+_ORDER = ["address", "property_description", "dimensions", "walls", "roofs",
+          "floors", "openings", "ventilation", "space_heating", "secondary",
+          "water_heating"]
+
+
+def main():
+    if len(sys.argv) < 2 or sys.argv[1] not in _ORDER:
+        print("usage: build_100010086084.py <" + "|".join(_ORDER) + ">")
+        return 2
+    with E.session() as (ctx, page):
+        globals()[sys.argv[1]](page)
+        print("done:", sys.argv[1], "->", page.url)
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

scripts/hyde/download_100010086084.py

@@ -0,0 +1,71 @@
+"""Download the Input Summary + SAP Worksheets PDFs for UPRN 100010086084,
+via a fresh-login `E.session()` (elmhurst_download.py's persistent-context
+login is stale per elmhurst_lib's own documented auth gotcha).
+
+SAFETY: clicks ONLY the two allow-listed report buttons below. NEVER clicks
+Submit (`LinkButtonSubmitEPR`).
+
+    DISPLAY=:99 python scripts/hyde/download_100010086084.py
+"""
+from __future__ import annotations
+from pathlib import Path
+import elmhurst_lib as E
+from playwright.sync_api import TimeoutError as PlaywrightTimeoutError
+
+SAMPLE_DIR = (
+    Path(__file__).parent.parent.parent
+    / "backend/epc_api/json_samples/real_life_examples"
+    / "SAP-Schema-15.0/uprn_100010086084"
+)
+
+REPORTS = [
+    ("ContentBody_ContentPlaceHolder1_LinkButtonSummary", "elmhurst_summary.pdf"),
+    ("ContentBody_ContentPlaceHolder1_ButtonDebugInforPdf", "elmhurst_worksheet.pdf"),
+]
+_FORBIDDEN = {"ContentBody_ContentPlaceHolder1_LinkButtonSubmitEPR"}
+
+
+def _download_one(page, button_id, out):
+    assert button_id not in _FORBIDDEN, f"refusing to click forbidden {button_id}"
+    print(f"  clicking {button_id} -> {out.name} ...", flush=True)
+    try:
+        with page.expect_download(timeout=60_000) as dl:
+            page.click(f"#{button_id}", timeout=20_000)
+        dl.value.save_as(str(out))
+    except PlaywrightTimeoutError:
+        try:
+            with page.expect_popup(timeout=10_000) as pop:
+                page.click(f"#{button_id}", timeout=20_000)
+            popup = pop.value
+            with popup.expect_download(timeout=30_000) as dl2:
+                pass
+            dl2.value.save_as(str(out))
+        except PlaywrightTimeoutError:
+            print(f"  !! no download/popup for {button_id}", flush=True)
+            return False
+    size = out.stat().st_size if out.exists() else 0
+    print(f"  saved {out} ({size:,} bytes)", flush=True)
+    return out.exists() and size > 0
+
+
+def main():
+    SAMPLE_DIR.mkdir(parents=True, exist_ok=True)
+    ok = True
+    with E.session() as (ctx, page):
+        print("navigating via Recommendations ...", flush=True)
+        with page.expect_navigation(wait_until="networkidle", timeout=60_000):
+            page.click("#ContentBody_buttonActionRecommendations_Link", timeout=15_000)
+        page.wait_for_timeout(600)
+        print("navigating to Energy Report Summary ...", flush=True)
+        with page.expect_navigation(wait_until="networkidle", timeout=60_000):
+            page.click("#ContentBody_buttonActionSummary_Link", timeout=15_000)
+        page.wait_for_timeout(600)
+        for button_id, fname in REPORTS:
+            ok = _download_one(page, button_id, SAMPLE_DIR / fname) and ok
+            page.wait_for_timeout(600)
+    print("DONE." if ok else "DONE (with errors).", flush=True)
+    return 0 if ok else 1
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

tests/domain/sap10_calculator/test_real_cert_sap_accuracy.py

@@ -1133,6 +1133,42 @@ _EXPECTATIONS: Final[tuple[RealCertExpectation, ...]] = (
         cert_num="0687-2894-6684-9009-7151",
         sap_score=72,
     ),
+    # Full Elmhurst RdSAP10 rebuild done for this ONE cert (the others in this
+    # batch are unverified against Elmhurst — see the batch-level note above).
+    # Elmhurst ground truth: SAP 48 (best-case single-window orientation
+    # rebuild — see below), vs this engine's 53 on gov-API inputs: a real ~5pt
+    # residual gap, NOT explained by mapper defaults (ruled out below with
+    # direct empirical tests). Two real, confirmed-but-unfixed bugs surfaced;
+    # flagged rather than blind-patched per this file's "fix generically,
+    # never tune to one cert" convention — each needs more than one cert's
+    # evidence before changing shared calculator code:
+    #   1. Wall-thickness-unknown fallback mismatch. This cert's solid-brick
+    #      wall has no lodged thickness. `u_wall()` falls back to a flat 1.70
+    #      (domain/sap10_ml/rdsap_uvalues.py, the un-thickness-banded Table 6
+    #      default). Elmhurst's own accredited tool, given the SAME "unknown"
+    #      state (its `CheckBoxWallThicknessUnknown`), resolves to U=1.40 —
+    #      not 1.70. Patch-tested in isolation (U 1.70->1.40): engine score
+    #      MOVED THE WRONG DIRECTION (53->55, away from Elmhurst's 48), so
+    #      this alone isn't the dominant driver and any fix needs to account
+    #      for whatever else shifts alongside it — do not fix blind.
+    #   2. Window U-value divergence. Same nominal "Double, unknown install
+    #      date" pick: this engine resolves ~1.85 W/m2K, Elmhurst ~2.52
+    #      W/m2K, for identical glazed area. Root cause not yet localised.
+    # Ruled out (tested directly, no material effect): total_floor_area vs
+    # summed sap_floor_dimensions (zero effect); missing party_wall_length
+    # (this cert's party wall type is "Solid", U~=0, so near-zero effect
+    # either way); region/climate (SAP rating always uses UK-average weather
+    # per Appendix U, `_region_index` in cert_to_inputs.py -- regional choice
+    # is provably inert for this field); boiler/secondary efficiency (exact
+    # match both sides: 66% / 63%).
+    # Single-window orientation IS a confirmed partial contributor (~2 of the
+    # ~7pt gap): the campaign's `set_single_window` collapses all glazing
+    # into ONE row (documented, accepted approximation -- true multi-
+    # orientation entry does not reliably stick; re-confirmed here, see
+    # elmhurst_lib.py). North-facing (this cert's first build) gave Elmhurst
+    # 46; South-facing (max solar gain) gave 48 -- the mapper's own gov-API
+    # synthesis spreads N/E/S/W, so neither single-orientation pick is
+    # perfectly comparable, but 48 (South) is the fairer bound.
     RealCertExpectation(
         schema="SAP-Schema-15.0",
         sample="uprn_100010086084",