The heat-pump PSR-extension fix (SAP 10.2 Appendix N2) drops corpus MAE from
0.740 to 0.726 on the 1000-cert RdSAP-21.0.1 sample; within-0.5 holds at
74.1%. Tighten the ceiling and log the slice.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
SAP 10.2 Appendix N2 (PDF p.101, footnote 44/45): for an air/ground/water
source heat pump whose plant size ratio exceeds the record's largest PSR,
the efficiency is reciprocal-interpolated between the largest-PSR value and
100% at twice the largest PSR (100% beyond that); below the smallest PSR the
efficiency is 100%. Our interpolator instead clamps to the top/bottom row.
Anchored to the accredited Elmhurst worksheet for cert 100110101713 (golden
fixture case 56, PCDB 100061): PSR 3.10665 over the record's largest 2.0
gives eta_space,1 = 147.011 -> (206) = 0.95 x 147.011 = 139.660, vs the
clamped 352.0 -> 334.4% that over-rates the dwelling by +18 SAP.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
within-0.5 floor 0.73->0.74 (now 0.741), MAE ceiling 0.762->0.740 (now
0.7397) on the fixed RdSAP-21.0.1 corpus. Log entry appended.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Gate PV generation/credit in cert_to_inputs on gov-API pv_connection:
credit only when ==2 ('connected'); ==1 ('present but not connected to the
dwelling's meter') contributes zero to the dwelling's cost/CO2/PE per
RdSAP 10 §11.1 / SAP 10.2 Appendix M. Non-int (None / site-notes str) keeps
the credit-if-array behaviour, so the Elmhurst/Summary + synthetic paths are
unchanged (no regression).
Corpus: all 5 pv_connection=1 PV certs move inside ±0.5 (e.g. 100051118081
+6.5→+0.5); MAE 0.760→0.740, within-0.5 73.8→74.3%, no regression
(pv_connection=2 certs keep their credit).
Also corrects a now-load-bearing latent bug: the solar-recommendation
overlay tagged recommended arrays pv_connection=1 ('not connected') — which
the new gate would zero. A new install connects to the dwelling's meter, so
it must be 2; pinned by the overlay test.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
RdSAP 10 §11.1 / SAP 10.2 Appendix M: PV is included in a dwelling's
assessment only if connected to the dwelling's own electricity meter. The
gov-API pv_connection enum encodes this — 0=no PV, 1=present-but-not-
connected, 2=connected. Corpus-validated (57 PV certs: pv_connection=1 MAE
4.48->1.22 without credit, 0/5 need it; pv_connection=2 needs it, MAE 0.98
vs 10.29) and Elmhurst-proven (connected SAP 87 vs not-connected 74).
cert_to_inputs currently credits a pv_connection=1 array; the test pins that
it must contribute zero generation. Adds pv_connection to make_minimal_sap10_epc.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
_is_elmhurst_roof_window mis-routes a vertical window (Location "External
wall") to sap_roof_windows when the building part's roof is a PARTY ceiling
(A "Another dwelling above" / NR "Non-residential space above"). A party
ceiling has no external roof, so it cannot host a rooflight.
Surfaced by simulated case 53 (cert 000565 re-keyed as a mid-floor electric-
storage flat, roof "A Another dwelling above"): its External-wall window
(U 2.00) routed to sap_roof_windows -> window area not deducted from the wall
(wall over-counted ~7 W/K) + priced as a roof window -> our SAP 74.0 vs
Elmhurst worksheet 75. Elmhurst lodges it Type "Window", Location "External
wall".
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
A gov-API flat can lodge dwelling_type="Mid-floor flat" while carrying its
own exposed roof — a top-floor flat mislabelled mid-floor. _dwelling_exposure
keyed roof exposure on the dwelling_type label alone, dropping the roof
heat-loss term: space-heating demand under-read ~32%, SAP over-read +7.
Fix: when the main building part lodges a *determined* roof_insulation_location
(an RdSAP integer code, not the "ND" Not-Defined party-ceiling sentinel),
expose the roof regardless of a contradictory label. Structured field, not a
description string and not roof_construction (which the gov-API lodges
building-wide on every unit, so it is not a per-unit signal).
On the RdSAP-21.0.1 corpus roof_insulation_location separates the classes with
zero disagreement: all 190 party-ceiling flats lodge "ND"; the 4 mid/ground
flats this exposes all move toward lodged, 0 away. within-0.5 73.3% -> 73.6%,
MAE 0.774 -> 0.761 (ratchets tightened). Verified end-to-end on the same
block: 715363 (location 6, RHI 2694) 81 -> 74 = lodged; genuine mid-floor
sibling 715395 (location ND, RHI 1024) stays party at 83 = lodged.
The override is additive (only ever exposes a label-dropped roof) and reads
the main part, so multi-part flats with a party main ceiling stay party.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
A gov-API cert can lodge dwelling_type="Mid-floor flat" while carrying a
real exposed roof element (roof_construction != 7 "dwelling above") over a
"(another dwelling below)" floor — i.e. a top-floor flat mislabelled
mid-floor. Property 715363 (uprn 6027561) + sibling 715395 (6027563) do
exactly this; the correctly-labelled top-floor sibling 715871 (6027574),
same block + same flat roof, already computes the lodged SAP 74.
_dwelling_exposure keys roof exposure on the dwelling_type label alone, so
it drops the roof heat-loss term, under-reading space-heating demand ~32%
(calc 1833 vs lodged RHI 2694) and over-reading SAP +7 (81 vs 74).
Pins the fix: a mid-floor label + lodged exposed roof must expose the roof
(floor stays party). Also corrects the existing mid-floor fixture to lodge
the party-ceiling code 7 (the default 4 is an exposed pitched roof).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Pin the hit-list properties: each property's real Landlord Override set still
resolves to overlays and trips physical_state_changed, and the rebaseliner
adopts the calculator output (reason physical_state_changed/both) rather than
echoing the lodged headline — so the displayed baseline and the modelled plan
agree. Scores a real cert through the live Sap10Calculator.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add Property.physical_state_changed (true on Site Notes / Landlord Overrides
/ Prediction — trigger (b)/(c)) and pass it from the
PropertyBaselineOrchestrator into the Rebaseliner. So an overridden or
predicted SAP>=10.2 property now stores calc(effective) as its Effective
baseline instead of echoing the lodged headline — closing the "81 in the DB"
divergence between the displayed baseline and the modelled plan.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
A SAP>=10.2 cert whose physical state was changed by Landlord Overrides or
Prediction must rebaseline off the calculator (the accredited lodged figure
no longer describes the dwelling) and tag physical_state_changed / both, and
must NOT log divergence (the calc IS adopted, nothing to validate against).
This is the D1 fix: the stored Effective baseline stops echoing the lodged
headline for overridden properties.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The deterministic calculator reads sap_ventilation.extract_fans_count (which
already round-trips); the top-level epc.extract_fans_count is its mirror (the
mapper sets both from one source). Reconstruct it from the same column so
EpcPropertyData round-trips complete, dropping the allow-list exception.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
community heating fuel + CHP fraction, alt-wall is_sheltered, wall
insulation thermal conductivity, pv_diverter_present, measured cylinder
volume, AP50 air permeability — all calculator-read, all silently dropped on
save. FE columns now live; assert deep-equal round-trip and drop their
coverage-guard allow-list entries so the guard enforces reconstruction.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The fixture lodges is_exposed_floor=True, which was silently dropped on
save->reload before this branch — so the orchestrator (reads the DB) modelled
the floor as not-exposed. Now it round-trips, the exposed floor is honoured,
and the ASHP-led max-gain fallback on this gas dwelling is bill-neutral
(marginally negative) rather than bill-positive: large energy saving, but the
gas->electricity fuel switch offsets the GBP saving. Same optimal package,
same telescoping; only the bill-savings sign assumption changed.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The ADR-0036 guard only inspected EpcPropertyData's top-level fields, so a
dropped field on a NESTED object (the PV-array list, the floor heat-loss
flags) slipped straight through. Generalise it to walk every domain
dataclass reachable from EpcPropertyData and check each field is
reconstructed by a _compose/_to_* mapper or allow-listed (per-field or
whole-class), keyed by Class.field.
Surfaced 14 pre-existing nested gaps the old guard was blind to: 7 are
calculator-read with no FE column (scoring-relevant silent-drop, same class
as the PV bug — tracked follow-up), the rest dormant or awaiting FE tables.
Each is now explicit and justified.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
sap_energy_source.photovoltaic_arrays has no table, so every array is
dropped on save — worth ~12 SAP points on an electrically-heated dwelling
(persist != score). Inject two ordered arrays onto a PV-free fixture.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
is_exposed_floor / is_above_partially_heated_space have no
epc_floor_dimension column, so a True flag round-trips back to the False
default and silently flips the floor's heat-loss path (persist != score).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
A Landlord heating-system override was applied as a sparse patch, so the
replaced system's fields bled through. A storage flat reclassified as a gas
combi (property 728513) kept mains_gas=False, heating category 7, the 2401
storage charge control, a Dual meter and an electric-immersion cylinder — an
incoherent record that gated out the gas-boiler-upgrade Measure and made the
heating Generator read the dwelling as off-gas (offering HHRSH storage).
Extend the ADR-0035 drag-along to gas boilers (Table 4b 102/104/120): the
overlay now sets the whole coherent companion set — mains_gas, gas main fuel,
heating category 2, fanned flue, full modern controls (2106), a single-rate
meter, and hot water from the main system with the cylinder set from the boiler
type (combi → none, regular/CPSU → cylinder). The main_fuel overlay also flips
mains_gas=True for a "mains gas" fuel. Non-off-peak archetypes now drag an
explicit Single meter so a system switch never leaves a stale Dual.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The heating-donor display synthesis reads donor.epc.main_heating, which has no
dataclass default — so a partial object.__new__ EpcPropertyData must set it.
test_validation's _comparable builder didn't, failing the two leave-one-out
scorer tests in CI (the full epc_prediction suite wasn't run pre-push).
main_heating_controls / sap_ventilation default to None via class attributes.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Three corrections found by re-running property 742003 end-to-end:
- roofSegmentStats are POSITIONAL — real responses omit the segmentIndex field
the fixture happened to carry; key the centre/area lookup by array position.
- Base the cap on ground_floor_area (the footprint the roof covers), not the
greatest per-storey area; roof_area is the fallback.
- Clamp the basis by total_floor_area: predicted EPCs borrow the structural
template's geometry (742003: a 118.62 m² MAIN ground floor) decoupled from
the predicted 55 m² (ADR-0029), so without the clamp the cap reads the
template's larger footprint.
Result: 742003 plan A/92.4 (16 kWp) -> C/74.4 (6.4 kWp). 29 solar tests +
orchestration threading + products green.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
select_conservative_configs must accept the dwelling's roof area and cap panels
to its usable roof (ADR-0038) — bounding a 55m² dwelling to ~16 panels under
Google footprint conflation, while staying a no-op on correctly-matched homes.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The Dwelling-Roof Cap (ADR-0038) sizes by usable roof area and ranks segments
by distance from the dwelling, so the projection must carry each panel's
footprint and each segment's centre + area (from roofSegmentStats).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Prediction never synthesises ventilation — it keeps the size-template's
sap_ventilation, so a predicted dwelling in an MEV/MVHR neighbourhood is scored
+ displayed as natural (predicted property 721167 follow-up). Mode the
mechanical_ventilation_kind across the cohort like glazing.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
_apply_heating_donor copies the donor's calc sap_heating but leaves the
display rows (main_heating, main_heating_controls) on the structural template
— incoherent, and 'Heating Control: Unknown' when the template lodged no
control (predicted property 721167, ADR-0029 follow-up).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
When refetch_epc=False and no stored lodged EPC exists, the handler no longer
falls back to a live EPC API call — it treats the property as EPC-less and
hands it to the prediction path. This keeps REFETCH_EPC (lodged path) and
REPREDICT_EPC (prediction path) cleanly independent.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Full-SAP certs mapped property_type=None, so the hard cohort filter silently
excluded them as comparables. Correctly typing them admits real lodged EPCs as
donors — a ground-truth-method change (cf #1245). Net over the n=36 fixture: 16
components better, 4 worse, 6 unchanged; gains concentrated in the physical
characteristics full-SAP certs measure (window_count 3.83->1.69, building_parts,
total_window_area, floor_construction, construction_age_band, glazing, walls).
The 4 that fell are new-build-vs-old-stock service mismatch on 1-2 targets each
(heating/water fuel, cylinder insulation) + floor_area. Tighten 16, loosen 4.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Route the Google Solar client through the shared call_with_retry with
full jitter (de-synchronises the 32 concurrent containers per Google's
"avoid synchronised requests" guidance), honouring Retry-After, a 60s max
backoff (rides out the 600 QPM per-minute window), and 6 bounded retries.
429/5xx/transport errors are transient; other 4xx propagate immediately;
404-entity-not-found stays BuildingInsightsNotFoundError. On exhaustion a
TransientHttpError surfaces so the subtask fails and is re-triggered (no
silent degrade).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Even after batching the data writes, the handler still wrote to the DB per
property through the orchestrator's SubTask bookkeeping: create + start +
complete each self-committed, and _cascade re-listed every sibling and re-saved
the parent on every transition — ~5 writes per property plus an O(N^2) cascade.
- TaskOrchestrator.run_subtasks: create all children in one INSERT, run each
(failures isolated per child), then persist all terminal states in one bulk
save and cascade the parent once. Children go WAITING -> terminal; the
transient IN_PROGRESS row is never written.
- SubTaskRepository.create_many / save_many (bulk INSERT / bulk fetch + update).
- _cascade short-circuits when the Task is already FAILED (terminal) — skips the
sibling roll-up entirely.
- modelling_e2e handler fans out via run_subtasks instead of per-property
create_child_subtask + run_subtask.
Per N-property batch the SubTask bookkeeping drops from ~5N writes + an O(N^2)
cascade to ~2 writes + 1 cascade.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The handler fired ~2+2N read round-trips and N+N write transactions per
SQS batch, pinning RDS CPU under ~32 concurrent containers on pool_size=1.
Reads: merge the duplicate property query and add overrides_for_many /
SolarRepository.get_many so overrides, solar, and property rows each load
in one query (2+2N -> 3).
Writes: buffer each modelled property's persistence intent in memory
(_PropertyWrite) during the loop, then flush the whole batch in one
PostgresUnitOfWork with a single commit, and run the baseline orchestrator
once for all written ids (N+N -> 2 transactions). Per-property modelling
failures stay isolated in the loop; the batch write is all-or-nothing and
retried via SQS (saves are idempotent upserts).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
An Economy-7 storage dwelling now prices heating at the 0.20-day/0.80-
night blend through cert -> calculator -> bill, instead of raising
UnpricedFuel and aborting the modelling_e2e batch.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Surface the hot-water (Table 13 / HP-DHW), secondary (direct-acting),
main-2 and ALL_OTHER_USES High-Rate Fractions on CalculatorInputs from
the same Table 12a helpers the SAP cost path uses, so Bill Derivation's
day/night split matches the rating's exactly.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
