NC-METH-001 v1.1 — Part B: The Pipeline¶

The Canonical 10-Stage Flow from Estate Boundary to Financial-Model Handoff¶

12 May 2026 · Internal · Author: AI-assisted, M. Forni review

B.0 Pipeline overview¶

The IEAT Solar methodology converts an industrial estate — a geographic boundary on a map containing buildings, parking lots, internal roads, canals, and tenants — into an investment-grade financial model: 34.59 MWp / $25.04M EPC / 12.2% sponsor IRR. The conversion happens through a 10-stage analytical pipeline.

Stage	Name	Output
B.1	Estate boundary	GIS polygon + total area (rai / km²)
B.2	Surface enumeration	Per-typology candidate surfaces (rooftop, ground, parking, road, canal)
B.3	Capacity sizing	Per-segment kWp DC
B.4	Yield	Per-segment P50/P90 kWh/kWp/yr
B.5	CAPEX	Per-segment EPC cost
B.6	Offtaker attribution	IEAT-direct vs tenant-attributed per segment
B.7	Confidence scoring	HIGH / MEDIUM / LOW / ASPIRATIONAL per segment
B.8	Tier assignment	Tier-1 / Tier-2 / Tier-3 for tenant-attributed segments
B.9	Proof status	VERIFIED / CONFIRMED / PENDING per segment
B.10	Lot packaging	Deployable lots for EPC procurement

Stages execute roughly sequentially but with feedback loops. Confidence scoring (B.7) and proof status (B.9) update throughout Phase 0 work as field verification arrives. Lot packaging (B.10) may iterate back to surface enumeration if EPC procurement constraints force re-grouping.

The pipeline outputs feed Part D (cross-cutting modules: yield modeling, CAPEX build, carbon, tariff/PPA, BESS economics) which in turn feed the financial model (NC-FM-LC-001 v1.0 for LC; analogous for other estates).

The descriptions below capture the pipeline as it exists in current artefacts (LC v1.0 register, LC IC paper, LC financial model, BP at-risk register, IC companion deck spec), flagged with [IMPLICIT] where a rule isn't written down but is being applied, [INFERRED] where I'm reconstructing a rule from observed behaviour, and [INCONSISTENT] where artefacts contradict each other.

B.1 Stage 1 — Estate boundary¶

Purpose: define the geographic extent within which all subsequent analysis happens.

Inputs¶

IEAT cadastral data (estate map; legal concession boundary)
OSM (OpenStreetMap) administrative polygons
IEAT lease records identifying which parcels are inside vs outside the concession

Decision rules¶

OSM polygon as primary source. The estate boundary is the OSM way (or union of ways) that matches the IEAT cadastral. Where OSM and IEAT cadastral disagree, OSM is preferred for geometric operations but IEAT data is authoritative for legal status.
Concession-inside / polygon-outside flags. Where IEAT operationally controls a parcel that falls outside the OSM polygon (e.g., the Delta Plant 9 case at Bangpoo: inside the concession but outside the OSM-mapped polygon), the segment is flagged but not automatically included. Inclusion requires IEAT confirmation.
Total area in rai is the methodology unit (1 rai = 1,600 m²; 1 km² = 625 rai). The memory's locked values for 13 estates (e.g., LC 3,498 rai; BP ~5,500 rai with OSM LOCKED at 5,851 rai = 9.36 km²) are canonical.
Sub-areas excluded by default: water bodies (unless designated for T6W canopy), security perimeters, fire lanes, government easements unless IEAT confirms availability for solar deployment.

Outputs¶

GIS polygon (GeoJSON or shapefile)
Total area in rai and km²
List of inside-concession / outside-polygon flagged parcels for IEAT review

Data sources¶

IEAT cadastral (per-estate Land Department records)
OSM ways (verified at lock date; LC ways and BP ways 223661193+1280772014 are locked per memory)
GISTDA satellite imagery (Sentinel-2 base + Pleiades high-resolution where commissioned)
LDD (Land Development Department) soil maps for ground typologies

Known weaknesses¶

OSM data quality varies by estate. Older estates (LC, BP, Lat Krabang) have higher-quality OSM. Newer or smaller estates (Smart Park, Phichit) may have less complete polygons.
IEAT cadastral updates lag. Land swaps, easements, and concession amendments may not appear in either OSM or IEAT cadastral for months.
Coastal estates (LC, MTP, MTP Port, Songkhla) have port boundaries that change with reclamation and dredging operations.
[INFERRED] The "1 rai = 1,600 m²" conversion is industry-standard but not stated in any current artefact; canonised here.
[IMPLICIT] The decision rule that OSM trumps IEAT cadastral for geometry but IEAT trumps OSM for legal status is operational practice; canonised here.

B.2 Stage 2 — Surface enumeration¶

Purpose: identify all candidate surfaces within the estate boundary where each typology could be deployed.

Inputs¶

Estate boundary polygon (from B.1)
Building footprints (Microsoft Building Footprints or equivalent satellite-derived)
Road network (OSM highway tags)
Canal network (OSM waterway tags + field verification for >15m wide canals)
Parking lot polygons (manual digitisation from satellite imagery or Google Street View)

Decision rules¶

Per typology:

T1 rooftop: building footprints filtered by (a) area > 500 m² (smallest viable rooftop); (b) roof type compatible (flat or low-slope only; pitched > 25° excluded); © building age < 30 years; (d) ownership: IEAT-owned admin or factory buildings, or tenant-owned with consent
T2 ground-mount: parcel polygons filtered by (a) unbuilt status per satellite; (b) easement-free per IEAT cadastral; © buffer-strip eligibility per setback rules; (d) area > 1,000 m²
T4A carport: parking lot polygons filtered by (a) parking width consistent with carport spans (10m, 15m, 20m standard); (b) clear soffit possible (no overhead obstructions); © lot size > 2,000 m² (smallest viable)
T4B-DC arterial canopy: internal road segments filtered by (a) carriageway width 6–9m; (b) length > 100m; © no overhead utilities; (d) median splits or single-side configuration possible
T6W wide canal canopy: canal segments filtered by (a) clear water width > 15m; (b) concrete-lined banks; © length > 200m; (d) bridge-free run of > 50m
BESS: not a surface — sited within or adjacent to PV lots, near LV/MV interconnect points

Cross-typology rules: - Each surface gets one typology assignment (no double-counting) - Where two typologies could apply (e.g., a building rooftop with adjacent parking), the analyst chooses the higher-yield typology by default - Aspirational reserve typologies (T1X / T2X / T6X) capture surfaces below confidence floor

Outputs¶

Per-typology list of candidate surfaces (polygons + identifiers like LC-T1-01, LC-T6W-03)
Aspirational reserve segments
Surfaces explicitly excluded with reason

Data sources¶

Microsoft Building Footprints (2024 release)
OSM highway and waterway tags
GISTDA / Sentinel-2 satellite imagery
Google Street View (canal width verification, parking layouts)
IEAT engineering drawings (where available)
Field walks (Phase 0B)

Known weaknesses¶

Building rooftop enumeration is the weakest stage. Microsoft Footprints catches most buildings but misses smaller structures, has perimeter accuracy issues, and provides no roof age or condition data. Phase 0B ROOF-001 structural assessments fill this gap segment-by-segment.
Parking lot enumeration requires manual digitisation; no public dataset captures parking lot polygons with sufficient accuracy. Time-consuming and error-prone.
Canal width verification: OSM tags rarely include canal width. Field verification or high-resolution satellite required.
Tenant building rooftops: even where building footprint is captured, ownership and consent require separate IEAT lease record query.
[INCONSISTENT] LC v1.0 segment register has 47 active segments; older artefacts have 39.1 MWp / 42 segments and 50 V3 segments. The 47 is canonical (per AUDIT-019 closure) but the older numbers persist in some derivative documents.
[INFERRED] The decision rule "choose higher-yield typology when two apply" is operational practice — canonised here.

B.3 Stage 3 — Capacity sizing¶

Purpose: convert each candidate surface into a peak DC capacity rating (kWp).

Inputs¶

Surface polygon and area (from B.2)
Typology-specific module density (W/m²) and layout efficiency (fraction of usable area)
Setback rules per typology

Decision rules¶

Module density (TOPCon 610Wp on 2.4 m²): - Nameplate density: 254 W/m² (610 / 2.4) - Effective deployable density varies by typology layout

Per-typology kWp/m² coefficients [INFERRED from LC v1.0 register cross-check]: - T1 rooftop: 130–160 W/m² (after 50–60% layout efficiency + setbacks) - T2 ground-mount: 100–130 W/m² (after row spacing for tilt + access) - T4A carport: 105–125 W/m² (single-axis fixed; depends on canopy span) - T4B-DC arterial: 110–135 W/m² (dual N/S orientation captures more sky) - T6W wide canal canopy: 110–130 W/m² (north-south orientation across canal)

Per-typology setback rules: - T1 rooftop: 1.0 m perimeter setback; 0.5 m from rooftop equipment; 2.0 m from access hatches - T2 ground: row spacing 2–3 module-heights (fence-to-fence); 5 m from estate boundary - T4A carport: 0.5 m from pedestrian zones; full coverage of parking bays; 3.0 m soffit clearance - T4B-DC: 0.5 m from carriageway edges; bridge clearance preserved - T6W: 0.5 m from canal bank edge; pedestrian access maintained at intervals

Outputs¶

Per-segment kWp DC nameplate capacity
Sum: estate active envelope kWp (LC: 34.59 MWp across 47 segments)
Per-typology subtotals (LC: T1 2.15, T2 4.35, T4A 11.35, T4B-DC 1.83, T6W 14.91 MWp)

Data sources¶

TOPCon module datasheets (610Wp standard; per memory edit 16)
Typology kWp/m² coefficients from LC v1.0 register + segment register's per-segment data
Setback rules from IEAT engineering standards [IMPLICIT — need explicit canonisation in Annex H]

Known weaknesses¶

The kWp/m² coefficients are LC-derived. Whether they transfer cleanly to BP, MTP, or Lat Krabang depends on each estate's specific geometric constraints. WP2 engineering review should validate.
Module efficiency assumption is static. As 700Wp+ modules become available (2027–28), the methodology should refresh.
[INCONSISTENT] The per-segment kWp in the LC v1.0 register comes from a mix of (a) calculated from surface area × density × efficiency, and (b) hand-adjusted by Marc based on site visit observations. The two are not separately tracked in the register.
[INFERRED] The bifacial uplift treatment (parking ~5%, road ~5%, water ~1–2%, gravel ~7% per memory) is methodology-specified but not applied per-typology in the model's yield (1,380 is a single LC number). Resolution pending AUDIT-041 in WP4.

B.4 Stage 4 — Yield¶

Purpose: assign P50 and P90 annual specific yield (kWh/kWp/yr) to each segment.

Inputs¶

Per-segment kWp (from B.3)
Estate location (latitude, longitude, climate zone)
Per-segment shading and orientation [INCONSISTENT — currently estate-level not per-segment]
Typology bifacial characteristics

Decision rules¶

Estate-canonical P50 yields (per RES-001 methodology canonical, currently inconsistent with LC model):

Estate / Zone	P50 (methodology RES-001)	Notes
Bangpoo (BP)	1,499 kWh/kWp/yr	Bangkok-adjacent
Laem Chabang (LC, EEC)	1,485 (methodology) / 1,380 (model)	Convention reconciliation pending WP1
Northern (Lampang, Lamphun, Phichit)	1,292–1,372	Cooler climate
Saraburi	1,420	Central plains
Southern (Songkhla)	1,520	High irradiance

P90 = P50 × 0.90 (industry-standard lender convention) Degradation = 1.0%/yr (TOPCon standard; Y1 nameplate; Y25 = nameplate × 0.99^25 = 77.8%)

Per-typology adjustments [INFERRED, not applied in LC v1.0 model]: - Bifacial uplift: parking ~5%, road ~5%, water ~1–2%, gravel ~7% - Roof tilt vs ground tilt: -3% to -5% for low-tilt rooftops in tropics - Canal humidity / biofouling: -3% to -5% over 25 years (T6W only)

Outputs¶

Per-segment P50 kWh/kWp/yr (currently a single estate-level number per LC model)
Per-segment P90 kWh/kWp/yr (P50 × 0.90)
Year-by-year degraded yield trajectory (Y1 through Y25)
Total estate P50 generation in GWh/yr (LC: 34.59 × 1,380 = 47.7 GWh/yr)

Data sources¶

PVGIS (free, global, lower-resolution than commercial)
Solcast Time Series API + Bankable PV Yield study ($5–7K, WP1 procurement)
Solargis (commissioned study; optional cross-check)
TGO (Thai Greenhouse Gas Management Organization) for grid emission factor
Existing solar systems at estate for ground-truth validation (where available)

Known weaknesses¶

Yield convention reconciliation is the largest open methodology gap. AUDIT-001 in R3 register: LC model 1,380 vs methodology RES-001 1,485 = 7.6% revenue gap = ±220 bps IRR depending on resolution.
Per-segment yield is currently estate-level, not segment-level. Shading, orientation, and tilt vary across segments but yield is applied uniformly. AUDIT-017 in R3 (ACTIVE-METHODOLOGY); per-segment modeling via Solcast + DSM is WP1 scope.
Bifacial uplift not typology-specific in model; methodology has typology framework but it's not flowed through. AUDIT-041; WP4 closure.
Soiling losses assumed in P50 but not explicitly disaggregated. Annual cleaning frequency assumption (4× monthly per Y1 model) drives the implicit soiling allowance.
Y1 partial-year convention [INCONSISTENT] — LC IC paper treats COD Q4 2027 as Y1 = 2027 first revenue year; model treats Y1 = first full revenue year (potentially 2028). Per A.4.6, canonical decision needed in v1.1.1.

B.5 Stage 5 — CAPEX¶

Purpose: calculate per-segment EPC cost and roll up to total project cost.

Inputs¶

Per-segment kWp (from B.3)
Typology-specific unit cost $/kWp
Total project cost markup factors

Decision rules¶

Per-typology unit costs (LC v1.0 register values):

Typology	$/kWp (LC)	$/kWp (methodology canonical)	Status
T1 rooftop	$412	$645	LC-specific lower (zero-foundation; simple installation)
T2 ground-mount	$813	$820	Aligned
T4A carport	$742	$950	LC-specific lower (3.0 m soffit; standard parking, not low-clearance industrial)
T4B-DC arterial	$700	$1,350	LC-specific lower (single-mobilisation; per-segment $700 vs methodology BP-specific T4B-1 $1,350)
T6W wide canal	$732	$1,150 (interpolated T6)	LC-specific lower (fixed-portal-frame on concrete-lined canals)

The differences between LC and methodology canonical are documented per AUDIT-040 as LC-specific cost drivers. WP4 produces the LC-specific cost driver annex (Annex H).

Total project cost build (per LC model Construction tab): - EPC × 1.08 = EPC + Development & soft costs (8% of EPC) - + Financing fees (2% of debt amount, calculated) - + Contingency (5% of EPC) - + Interest During Construction (IDC; calculated from debt drawdown schedule) - = Total project cost (LC: $25.04M EPC → $29.16M total)

Outputs¶

Per-segment EPC cost
Per-typology subtotals
Estate EPC total
Total project cost with build-up
Sources/uses table for the capital structure

Data sources¶

LC v1.0 segment register (per-segment unit costs)
Methodology canonical unit costs per typology (NC-MEM-COST-001 if exists; otherwise per Annex H)
EPC market study (NC-PR-LC-001 SOW-05 outputs)
Vendor quotes (BNEF, Solar Mart, supplier-direct)
Historical project costs (NewCo pipeline records)

Known weaknesses¶

LC unit costs are 14–46% below methodology canonical for carport/arterial/canal typologies. Whether this is genuine LC-specific structural advantage or systematic understatement is unresolved. WP2 engineering review under Phase 0 will validate.
Bottom-up cost stack is not formalised. Phase 3B desk bottom-up was retired (R3 register) as unvalidated. Annex L is deferred to v1.2 pending engineering review.
C4 corrosion premium [INFERRED — not visibly applied] — coastal estates (LC, MTP, MTP Port, Songkhla) should carry a BOS premium for marine-grade equipment. Whether this is in LC v1.0 unit costs is unverified; AUDIT-002 closure pending WP4.
Methodology canonical vs LC numbers — the gap (especially T4B-DC and T6W) is largest at LC and may reflect either LC-favourable conditions or memo drift. Documenting LC-specific cost drivers (AUDIT-040) is the correct response.
[IMPLICIT] The 8% / 2% / 5% / IDC markup convention is in the LC model but not standardised across estates in the methodology — should be canonised in Part D.

B.6 Stage 6 — Offtaker attribution¶

Purpose: assign each segment as IEAT-direct (offtake to IEAT) or tenant-attributed (offtake to a specific tenant).

Inputs¶

Per-segment location and typology (from B.2, B.3)
IEAT lease records (which parcels are tenant-leased vs IEAT-retained)
Tenant inventory (NewCo lc_tenants_v3.csv for LC; equivalent for other estates)
Building footprints + tenant location mapping

Decision rules¶

Attribution per typology:

T1 rooftop:
IEAT-owned admin building → IEAT-direct
IEAT-owned factory building with single tenant → either IEAT-direct (IEAT collects ESA, passes savings to tenant via reduced rent) or tenant-direct (tenant signs ESA directly with NewCo); methodology defaults to IEAT-direct unless specific tenant is named for direct ESA
Tenant-owned building → tenant-direct ESA required
T2 ground-mount:
IEAT-retained buffer parcels → IEAT-direct (LC: 6 of 7 T2 segments)
Tenant-adjacent parcels with shared use → tenant-attributed (LC: 1 T2 segment, LC-T2-07 Michelin)
T4A carport:
All IEAT parking infrastructure → IEAT-direct (LC: all 7 T4A segments)
Tenant-private parking → tenant-direct (typically not part of canonical T4A; goes to T4A-tenant subtype)
T4B-DC arterial canopy:
IEAT internal road network → IEAT-direct (LC: 8 IEAT-side segments)
Tenant-adjacent arterial → tenant-attributed (LC: 8 tenant-side segments — varies by side of road)
T6W wide canal canopy:
Adjacent to tenant operations → tenant-attributed (LC: all 5 T6W segments tenant-attributed; 3 to Thai Summit Group)
BESS:
Sited at IEAT load centre → IEAT-direct
Sited at tenant load centre → tenant-attributed (for demand charge management revenue)

Per-tenant credit tier (assigned at B.8, not B.6): - B.6 produces "tenant-attributed" with a named anchor or anchor cluster - B.8 then classifies that anchor as Tier-1 / Tier-2 / Tier-3

Outputs¶

Per-segment attribution: IEAT-direct or {tenant name(s)}
Estate-level split: % CAPEX IEAT-direct, % CAPEX tenant-attributed
LC: 26 segments IEAT-direct, 21 tenant-attributed; CAPEX 52% IEAT-direct, 48% tenant-attributed

Data sources¶

IEAT lease records (per-estate)
NewCo tenant inventory (LC: lc_tenants_v3.csv; 128 verified tenants)
GPlaces / Google Maps Places API (multi-source tenant verification)
OSM POI tags
Field walks (Phase 0B for boundary cases)

Known weaknesses¶

Per-tenant credit modeling is not in the canonical pipeline yet — AUDIT-027 in R3 register. The methodology says "tenant-attributed to X" but doesn't grade X's credit. Annex G (per-tenant credit framework) is WP4 scope.
Concentration risk — when multiple segments attribute to the same group (LC: 3 segments to Thai Summit Group; LC-T6W-01/02/04), the structure is exposed to single-counterparty refusal. The LC model handles this in the Tenant_Consent_Sensitivity tab (Thai Summit cluster out → 1.04× DSCR / 11.4% IRR / BREACH).
Boundary cases (parking shared between two tenants, factory rooftop with subletted floors) are resolved per-segment by analyst judgment — should be documented per case.
LC2 Power Plant is a documented edge case: power-producer offtaker that doesn't fit conventional self-consumption ESA. Methodology should formalise the "power-producer ESA" sub-class (AUDIT-043 model-handled; methodology documentation deferred).
[INFERRED] The default rule "IEAT-direct unless tenant named for direct ESA" is operational practice for tenants on IEAT-owned buildings; canonised here.

B.7 Stage 7 — Confidence scoring¶

Purpose: assign each segment a confidence score reflecting the strength of evidence that the segment is real, deployable, and as-characterised.

Inputs¶

Sources confirming the segment (count and quality)
Field verification status
Marc-verified or analyst-verified status

Decision rules¶

Confidence score = 0.0 to 1.0 continuous, with band labels:

HIGH (≥ 0.80): multi-source verified. Two or more of: (a) field walk; (b) multi-source satellite (GPlaces multi-source corroboration); © IEAT engineering drawing; (d) tenant LOI for tenant-attributed
MEDIUM (0.40–0.79): single-source verified. One of: (a) field walk only; (b) single-source satellite; © Marc-verified visually; (d) IEAT lease record only
LOW (0.20–0.39): single-source unverified. Inferred from satellite imagery without ground-truth
ASPIRATIONAL (< 0.20): below floor. Held in reserve typology (T1X / T2X / T6X)

Active envelope inclusion floor: 0.20 [canonised in Part A § A.4.10; previously [IMPLICIT]]

Per LC v1.0 register (47 active segments): - HIGH: T1 (12 segs, Annex A field-verified), T2 (7 segs, Annex A field-verified), T4A (7 segs, Annex A field-verified) - MEDIUM: T4B-DC (16 segs; 4 OSM-confirmed pairs + 12 inferred), T6W (5 segs; 1 OSM + 3 Marc-verified + 1 inferred) - Aspirational: T1X (0.5 MWp), T2X (1.5 MWp), T6X (1.5 MWp) — 3.5 MWp / $2.53M held in reserve below 0.20 floor

Outputs¶

Per-segment confidence score (0.0–1.0)
Per-segment confidence band (HIGH / MEDIUM / LOW / ASPIRATIONAL)
Estate confidence profile (counts and CAPEX by band)
Replacement reserve set (aspirational typologies)

Data sources¶

Each segment carries a source column in the register (e.g., "GPLACES", "GPLACES (multi-source)", "OSM+GPLACES", "EXISTING_VERIFIED+GPLACES_MERGED", "EXISTING_VERIFIED", "MARC_VERIFIED")
Phase 0B field walk outputs (per gate: ROOF-001, RTK GPS, bathymetric, field walk)

Known weaknesses¶

Confidence scoring is partly subjective. A "Marc-verified visually" segment can be promoted to MEDIUM or even HIGH if Marc was confident; the methodology doesn't formalise the visual-verification standard.
Multi-source aggregation rule is implicit. The scoring algorithm "count distinct sources + apply quality weights" is operational practice — should be formalised in Annex.
Confidence updates during Phase 0 — as ROOF-001 / RTK / bathymetric / field walks come back, scores update. The methodology needs to specify how the update happens (monotonic confirmation only? downward revision possible?).
[INFERRED] The 0.20 floor is the de facto cut for active envelope inclusion; canonised in Part A.
[INCONSISTENT] Some Marc-verified segments are scored HIGH (multi-source) in some artefacts and MEDIUM in others; need a single source of truth (the segment register).

B.8 Stage 8 — Tier assignment¶

Purpose: classify the offtaker (IEAT or tenant) by credit and ESA-readiness tier.

Inputs¶

Per-segment offtaker attribution (from B.6)
Tenant credit profile (where tenant-attributed)
ESA-readiness signals (LOI status, prior engagement, group corporate health)

Decision rules¶

For IEAT-direct segments: tier is N/A (IEAT itself is the offtaker; methodology defaults to "IEAT investment-grade" with no further sub-classification)

For tenant-attributed segments:

Tier-1 anchor: HIGH-confidence segment adjacency + investment-grade group parent OR public company OR wholly-owned subsidiary of investment-grade multinational + multi-source verified (GPlaces multi-source + named in IEAT records)
Tier-2: MEDIUM-confidence with tenant directly identified but credit profile not validated as IG (e.g., Thai SME with no public credit rating); may need Phase 0B credit due diligence
Tier-3: LOW-confidence; tenant inferred from segment geometry; not suitable for ESA execution; would be replaced from aspirational reserve

Per LC v1.0 register (12 Tier-1 anchors):

Tenant	Group / parent	Adjacent segments
Michelin Siam Company	Michelin (FR)	LC-T2-04, LC-T2-07
MMTh Engine Co.	Mitsubishi Motors (JP)	LC-T4B-01
Thai Summit Harness PCL	Thai Summit Group (TH)	LC-T6W-01
Multibax Public Co.	(TH, public)	LC-T1 / T4A
Astemo Chonburi Manufacturing	Hitachi Astemo (JP)	LC-T4B-03
Siam Compressor Industry	Mitsubishi Electric (JP)	LC-T4B-05
Summit Auto Body Factory	Thai Summit Group (TH)	LC-T6W-02
Kimball Electronics (Thailand)	Kimball (US)	LC-T4B-07
Summit Laemchabang Auto Body	Thai Summit Group (TH)	LC-T6W-02
Techno-Metal	(TH)	LC-T4B-09
Sankyu Laem Chabang (Thailand)	Sankyu (JP)	LC-T1-10
Laem Chabang 2 Power Plant	(TH, power producer)	LC-T6W-03 (special case)

Outputs¶

Per-segment tier assignment (Tier-1, Tier-2, Tier-3, or N/A for IEAT-direct)
Tenant-cluster identification (e.g., Thai Summit Group cluster covering 3 segments at LC)
Anchor ESA prospect list (Tier-1 only) for Phase 0B LOI campaign

Data sources¶

GPlaces multi-source corroboration
Group parent identification (DBD Thailand business registry; S&P Capital IQ for international)
IEAT lease records (tenant ground-truth)
Public company financial filings (where applicable)
DBD Thailand financials (for Thai-incorporated tenants)

Known weaknesses¶

Tier-1 designation doesn't quantify credit risk. "Investment-grade group" is necessary but not sufficient for low default risk — local subsidiary may be undercapitalised relative to parent. Per-tenant credit framework (Annex G) addresses this in WP4.
Concentration is tier-blind. Three Tier-1 segments all attributing to Thai Summit Group is documented in B.6 but the tier scoring treats them equivalently to three independent Tier-1s.
Group / parent identification lags. Thai-incorporated companies may not always have clear group attribution in public records.
[INFERRED] The "named in IEAT records" requirement for Tier-1 is operational practice; canonised here.

B.9 Stage 9 — Proof status¶

Purpose: track the verification state of each segment as it moves from desktop analysis through Phase 0 field work toward FID.

Inputs¶

Segment current verification artefacts (per source column in register)
Phase 0 work plan outputs (NC-WP-LC-001 v1.0 for LC)

Decision rules¶

Three proof states:

VERIFIED: GPS-bounded geometry + field walk completed OR multi-source satellite verification + tenant LOI signed (for tenant-attributed)
CONFIRMED: GPS-bounded geometry + single-source satellite verification + IEAT-direct OR Tier-1 anchor identified (with field walk pending)
PENDING: GPS-bounded geometry only; awaiting Phase 0B verification (ROOF-001 / RTK GPS / bathymetric survey / field walk)

Phase 0 verification gates per typology: - T1 rooftop: ROOF-001 structural + ACM (asbestos-containing material) screening - T2 ground: RTK GPS surveys for parcel boundary precision - T4B-DC: field walk for Sois C/D and other internal road segments - T6W: bathymetric survey for canal water depth and bank stability - BESS: site assessment for interconnect at PCC (point of common coupling)

Promotion rules: - PENDING → CONFIRMED on first successful verification artefact - CONFIRMED → VERIFIED on second corroborating artefact (typically field walk + LOI) - Downgrade is possible (VERIFIED → CONFIRMED → PENDING) if verification fails or LOI withdrawn

Outputs¶

Per-segment proof status (VERIFIED / CONFIRMED / PENDING)
Pre-FID gate burden (segments × verification gates × cost)
FID-readiness estate profile (% VERIFIED at planned FID date)

Data sources¶

Phase 0 work plan procurement SOWs (NC-PR-LC-001 v1.0: SOW-01 LOI, SOW-02 ROOF-001, SOW-03 PEA hosting capacity, SOW-04 EIA addendum, SOW-05 KTB IE)
Segment register source column (live document)

Known weaknesses¶

Proof status is not in the LC v1.0 segment register as a discrete column. The register has proof_status per row but the values predate this methodology canonisation. The methodology should formalise the three-state model going forward.
Promotion rules are implicit. Operational practice: a tenant LOI raises proof status; a failed ROOF-001 demotes it. Should be documented.
Downgrade tracking is weak. Once a segment is CONFIRMED, downgrade events (LOI withdrawn, structural assessment fails) need to flow back to the register reliably.
[INFERRED] The PENDING → CONFIRMED → VERIFIED progression is operational; canonised here.

B.10 Stage 10 — Lot packaging¶

Purpose: group segments into deployable lots for EPC procurement, financing, and commissioning sequencing.

Inputs¶

Per-segment kWp, CAPEX, offtaker, confidence (from B.3–B.7)
EPC procurement strategy (single-EPC-multi-lot vs multi-EPC-multi-lot)
Financing structure (debt sizing per lot; portfolio-level vs lot-level)
Commissioning sequence (which lot first; technical dependencies)

Decision rules¶

Lot grouping principles:

Typology aggregation where possible: same-typology segments grouped for EPC efficiency (one mobilisation cost, common procurement)
Geographic clustering: physically adjacent segments share infrastructure (cabling, inverter siting, switchgear) — lower BOS cost
Offtaker aggregation: IEAT-direct segments in one lot vs tenant-attributed in another (simplifies ESA negotiation)
Risk dispersion: avoid one lot concentrated in single Tier-1 anchor (mitigates counterparty risk)
Size targeting: lots typically 5–20 MWp; below 5 MWp loses EPC procurement efficiency; above 20 MWp triggers separate financing tranche

Lot configurations observed in practice: - BMA RFP package: 3 lots, non-geographic, by building type and SC uplift (per memory) - LC: implicitly single-lot or 2-lot at 34.59 MWp scale; per IC paper §1 the program is treated as one investment - BP: typically 2–3 lots (per memory: BP $102M / 77 MWp likely 3+ lots)

Outputs¶

Per-lot composition (segment list + total kWp + total CAPEX)
Per-lot offtaker mix
Per-lot commissioning sequence
Procurement strategy per lot (EPC RFQ scope)

Data sources¶

LC v1.0 register (current single-program treatment)
BMA RFP package (3-lot non-geographic precedent)
IEAT lease records (boundary cases)
EPC market intelligence (procurement appetite by lot size)

Known weaknesses¶

Lot packaging is the least formalised stage in the pipeline. Current practice is per-estate analyst judgment.
Lot boundaries influenced by EPC procurement strategy — may not align with technical or commercial groupings. Adds friction late in the pipeline.
Financing-driven lot constraints — debt sizing requires lots to be financially viable individually; this can fragment otherwise-clean technical groupings.
Lot-level vs portfolio-level financial modeling — the LC model treats the active envelope as one program; if 3 lots are needed for procurement, lot-level DSCR profiles may differ from program profile.
[INFERRED] The "5–20 MWp typical lot size" range is implicit from BMA + LC + BP precedent; canonised here.
[INCONSISTENT] LC sometimes appears as one lot, sometimes as multi-lot in different artefacts. Should be locked in v1.1.

B.11 Handoff to financial model¶

The pipeline output is a structured input file for the financial model (per LC: NC-FM-LC-001 v1.0 Inputs tab + Active_Envelope tab):

Input	Source stage	LC value
Active envelope MWp	B.3 (sum across segments)	34.59
Active CAPEX $M	B.5 (sum × FX)	25.04
Per-typology breakdown	B.2 + B.3 + B.5	T1/T2/T4A/T4B-DC/T6W per-typology rows
P50 yield kWh/kWp/yr	B.4	1,380 (model; methodology 1,485 pending WP1)
P90 yield	B.4 (P50 × 0.90)	1,242
Offtaker mix (% CAPEX)	B.6	52% IEAT / 48% tenant
Tier-1 anchor count	B.8	12
Aspirational reserve	B.7	5.28 MWp / $4.49M in T1X/T2X/T6X
Tenant cluster concentration	B.6 + B.8	Thai Summit cluster $5.50M / 22% of envelope
Pre-FID gate burden	B.9	ROOF-001 / RTK / bathymetric / field walk / LOI procurement

The model then applies Part D cross-cutting modules (yield → revenue × tariff; CAPEX → debt sizing × DSCR check; OPEX → per-typology build-up; carbon → if applicable; BESS → if applicable) to produce the returns metrics (IRR, DSCR, NPV, MOIC, payback, Y10 exit).

B.12 Acknowledged pipeline gaps¶

The Part B pipeline has these substantive gaps to address in WP1–WP5 and subsequent versions:

Per-segment yield modeling (B.4) — currently estate-canonical, not per-segment. Solcast + DSM via WP1.
LC-specific cost driver documentation (B.5) — methodology vs LC numbers diverge significantly for T4B-DC and T6W; engineering review and documentation via WP2 + WP4 (Annex H).
Per-tenant credit framework (B.6, B.8) — Tier-½/3 is necessary but not sufficient for credit risk grading. Annex G via WP4.
C4 corrosion premium (B.5) — verify presence in unit cost build-up; AUDIT-002 via WP4.
Confidence scoring algorithm (B.7) — currently implicit; formalise multi-source aggregation rule with quality weights.
Proof status promotion / demotion rules (B.9) — formalise the state-transition model.
Lot packaging (B.10) — least formalised stage; needs estate-deployment runbook treatment in Part E.
BESS pipeline integration (across all stages) — BESS is the 6^th typology but the pipeline above mentions it cursorily. WP4 expansion + dedicated Part C subsection.
Boundary-canal-canopy edge cases (B.1, B.2) — concession-inside / polygon-outside flags need IEAT review SLA documented.

These gaps are tracked in the audit register (R3 v0.3) and folded into the v1.1 work plan.

B.13 References¶

NC-IS-LC-001 LC Investment Segment Register v1.0 — per-segment data for all 47 active + aspirational segments
NC-IC-LC-001 IC Paper v1.1 — investment thesis with pipeline outputs
NC-FM-LC-001 Laem Chabang Financial Model v1.0 — financial model with pipeline inputs
NC-WP-LC-001 Phase 0 Work Plan v1.0 — verification gate procurement
NC-PR-LC-001 Procurement SOWs v1.0 — five SOWs for Phase 0
NC-MN-001-R3_audit_v0_3 — live audit register tracking pipeline gaps
Memory edits 1–30 (per memory_user_edits view 12 May 2026)
NC-METH-001 v1.1.0 Part A — framing document with conventions canonised

End of Part B v1.1.0.