Model/datatypes
Khalim Conn-Kowlessar 5c2158e6c4 Slice S0380.171: CHP heat-fraction split for community heating cost
Closes the +£104 cost / +4.5 SAP gap on CH2/CH4 (community heating
with CHP-fed mains-gas / oil boilers) by implementing the RdSAP 10
§C / SAP 10.2 Appendix C (PDF p.58) default heat-fraction split:

  "If CHP (waste heat or geothermal treat as CHP):
   - fraction of heat from CHP = 0.35
   - CHP overall efficiency 75%
   - heat to power ratio = 2.0
   - boiler efficiency 80%"

Verified against the corpus block 9b lodgement: CH2 worksheet (303a)
= 0.3500 + (303b) = 0.6500 + (305) = 1.00 + (306) DLF = 1.45. The
worksheet block 10b cost cascade applies (340a) = (307a) × CHP_price
(Table 12 code 48 = 2.97 p/kWh) + (340b) = (307b) × boiler_price
(Table 12 codes 51-58 = 4.24 p/kWh) with (307a) = 0.35 × (307),
(307b) = 0.65 × (307).

Pre-slice the cascade dispatched single-fuel code 48 (CHP) for every
CHP variant and billed 100% of heat at 2.97 p/kWh, under-charging by
~£104/yr versus the worksheet's 35% × 2.97 + 65% × 4.24 = 3.7945
p/kWh blended rate.

Three layers wired:

1. Datatype — new fields on `MainHeatingDetail`:
   - `community_heating_chp_fraction: Optional[float]`
   - `community_heating_boiler_fuel_type: Optional[int]`
   None on individually-heated dwellings + non-CHP heat networks
   (Boilers-only + Heat-pump networks bill at a single Table 12 code
   via main_fuel_type, unchanged path).

2. Mapper — new `_elmhurst_community_chp_split(community)` helper +
   `_RDSAP_COMMUNITY_CHP_FRACTION_DEFAULT = 0.35` constant. When the
   §14.1 Community Heat Source is "Combined Heat and Power": returns
   (0.35, boiler_fuel_code) where boiler_fuel_code is resolved from
   the §14.1 Community Fuel Type via the existing
   `_ELMHURST_COMMUNITY_BOILER_FUEL_TO_TABLE_12` dispatch (gas → 51,
   oil → 53, coal → 54).

3. Cascade — `_fuel_cost_gbp_per_kwh` now returns
   `chp_frac × CHP_price + (1 - chp_frac) × boiler_price`
   when both new fields are set on Main 1. Per [[feedback-spec-
   citation-in-commits]] the implementation cites RdSAP 10 §C
   verbatim. Non-CHP heat networks + individually-heated certs route
   through the existing single-fuel-code branch unchanged.

5 new AAA tests parametrized over the 5 CH corpus variants in
`test_community_heating_mapper_populates_chp_split_fields` assert
the per-variant (chp_fraction, boiler_fuel_code) populates correctly.

Closures vs pre-S0380.171 residuals (heating-systems corpus block 11b):

  variant            ΔSAP      Δcost      status
  CH1  (Boilers/Gas) +0.5915   -£13.63    unchanged (no CHP split)
  CH2  (CHP/Gas)     +4.50→-0.0076  -£104→+£0.17   ✓ CLOSED
  CH3  (HP/Elec)     +0.5915   -£13.63    unchanged (no CHP split)
  CH4  (CHP/Oil)     +4.50→-0.0076  -£104→+£0.17   ✓ CLOSED
  CH6  (CHP/Coal)    -3.52→-8.03   +£81→+£185     REGRESSED

The CH6 regression is exposed (not caused) by the spec-correct split:
pre-slice CH6 sat at -3.52 SAP / +£81 by coincidence — the cascade's
CHP-only pricing (2.97 p/kWh) cancelled with cascade DLF=1.45
(Table 12c age G default) against the CH6 worksheet's lodged DLF=1.0.
Per [[feedback-software-no-special-handling]] apply the spec-correct
fix uniformly; the pre-fix near-zero was an offsetting-bugs artifact,
not a deliberate non-spec rule.

The CH6 worksheet (306) DLF=1.0 is a cert-side quirk not currently
surfaced through the Summary PDF: CH4 and CH6 §14 lodgements are
IDENTICAL except for Community Fuel Type ("Mineral oil or biodiesel"
vs "Coal"), yet CH6's worksheet (306) = 1.0000 while CH4's = 1.4500.
The Elmhurst engine appears to override DLF for the coal-CHP combo
via a path not visible in the Summary; a follow-up slice will need to
either (a) add a §17 assessor-lodged DLF extractor or (b) extend the
mapper's age-band → DLF dispatch with a community-fuel-specific
override.

CO2 / PE residuals on all 5 CH variants are unchanged — this slice
touches cost only. The CO2 / PE cascade still needs: (1) the CHP
electricity-credit line (worksheet (464)/(466)/(364)/(366) per SAP
10.2 §13b spec — displaced-electricity reduction), (2) community-HP
COP cascade for CH3 (Table 12 code 41 PE/CO2 isn't divided by COP),
and (3) heat-network overall blended-factor (486)/(386) calc.

Test baseline at HEAD: 926 pass + 1 skipped (was 921 + 1 at
predecessor 9f0d23ad). Pyright net-zero on affected files
(epc_property_data.py, mapper.py, cert_to_inputs.py,
test_heating_systems_corpus.py + elmhurst_site_notes.py): 65 → 65.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-06-04 15:40:25 +00:00
..
epc Slice S0380.171: CHP heat-fraction split for community heating cost 2026-06-04 15:40:25 +00:00
magicplan typing and renaming 🟪 2026-05-07 13:26:49 +00:00
__init__.py further breaking up code 2023-07-20 12:24:34 +01:00
datatypes.py further breaking up code 2023-07-20 12:24:34 +01:00
enums.py completed build of new demo portfolio - some fixed required still 2023-11-20 14:42:16 +00:00