Methodology

How we calculate.
What we don't.

Model version: 1.2  ·  Last reviewed: May 2026
No silent revisions — every change is logged below

This page explains exactly how preFeasibility works — the formulas, the assumptions, the data sources, and the boundaries of what the platform can and cannot tell you. We have written it for two kinds of readers: the practitioner who wants to verify the methodology before trusting a number, and the non-technical user who wants to understand what they're looking at before making a decision with it.

Both are legitimate. Both deserve a straight answer.

Feasibility classes — what they actually mean
The industry standard for classifying the rigour of an energy analysis

The renewable energy and infrastructure industry uses a classification system — commonly referred to as Class 1 through Class 5 — to describe how detailed, site-specific, and financially defensible an analysis is. You will see these referenced in financing term sheets, due diligence reports, lender requirements, and internal approval gates.

The classification is not standardised by a single body. Different institutions use slightly different terminology — some use "Order of Magnitude," "Feasibility Study," "Bankable," and so on. The table below reflects the most widely used framework across development finance, independent power producers, and investment committees.

Class
Also called
What it means
Typical use
Class 1
Bankable / Investment-grade
Independently verified, site-specific, audit-ready. Every assumption sourced and defensible to a lender or equity investor.
Completed resource assessment · Measured wind/solar data · Full EPC pricing · Independent engineer sign-off
Financial close
Class 2
Detailed feasibility
Site-specific study with significant measured data. Suitable for board approval, detailed term sheets, and advanced development spend.
12+ months wind data · Preliminary EPC quotes · Preliminary grid study · Environmental scoping
Board approval
Class 3
Pre-feasibility
Credible preliminary analysis using a mix of modelled and measured data. Sufficient to support early development decisions and internal investment approval.
Modelled resource data (GlobalWindAtlas / PVGIS) · Budget CAPEX estimates · Regional OPEX benchmarks · Assumed WACC
← We operate here
Class 4
Indicative / Screening
Fast, high-level analysis used to screen a site or market before committing development resources. Results carry meaningful uncertainty but are useful for go/no-go decisions.
Regional default inputs · Published benchmark CAPEX/OPEX · Atlas-based capacity factors
← And here
Class 5
Order of magnitude
Back-of-envelope. Useful only for deciding whether a concept is worth any further attention at all. Not suitable for any decision involving real money or real commitments.
Technology type · Very rough location · Nothing else
Napkin math

Class 3 vs Class 4 in practice: The difference is mainly input quality, not methodology. The same LCOE formula is used at both levels. A Class 4 result uses regional defaults — the system fills in typical CAPEX, OPEX, and a modelled capacity factor for your region. A Class 3 result uses your own project-specific inputs for CAPEX, OPEX, and WACC, combined with a location-derived capacity factor. The closer your inputs are to your actual project, the closer the output is to Class 3. preFeasibility is designed to help you move from Class 4 to Class 3 as your project matures, without changing tools.

Where preFeasibility sits — and where it stops
Honest about what the tool is for
Read this before you run anything

preFeasibility produces Class 3–4 pre-feasibility outputs. This means: credible enough to inform an early-stage development decision, rigorous enough to walk into an internal meeting with, not sufficient on its own to commit capital, seek project finance, or sign a land lease.

The platform is deliberately positioned here. We are not trying to replace a bankable feasibility study. We are trying to make the decision to commission one — or to screen out a site before you spend that money — faster, cheaper, and more defensible than the alternative.

The platform offers two products at this class level. The Site Screener is a Class 4 screening tool — an 8-axis go/no-go verdict designed to answer "Should I even look at this site?" in seconds. The Full Prefeasibility Study (Wind / Solar / Hybrid) operates at Class 3–4 — producing LCOE, IRR, NPV, and AEP with site-specific inputs. Both share the same data spine (GWA 3.0, PVGIS-SARAH3, Natura 2000, OSM, GEM, ENTSO-E, IRENA) and the same methodology for resource estimation and financial modelling. The screener filters; the full study quantifies.

Site Screener — 8-axis methodology
How the traffic-light verdict is calculated for each axis

The Site Screener evaluates a wind or solar site across eight independent axes, each producing a GREEN / AMBER / RED traffic-light result. The axes are scored against published thresholds using publicly available datasets. The overall verdict — PASS, CAUTION, or KILL — is synthesised from the individual axis results.

The screener is a Class 4 screening tool. It is designed to answer one question: Should I spend any more time on this site? It is not a substitute for a Class 3 pre-feasibility study or a Class 1–2 bankable analysis.

Where a country has sufficient installed capacity data (≥ 20 projects in our reference dataset), the Resource axis uses country-relative thresholds (P40/P25 percentiles of the operating fleet). Where country data is insufficient, global or latitude-banded defaults are applied. This distinction is noted in the screener output for each run.

01
Resource

What it measures: The estimated capacity factor (CF) for the site, derived from Global Wind Atlas 3.0 (wind) or PVGIS-SARAH3 (solar).

Wind thresholds — country-relative (used when ≥ 20 operating projects in reference fleet):

  • GREEN: CF ≥ P40 of the country's operating fleet
  • AMBER: P25 ≤ CF < P40
  • RED: CF < P25

Wind thresholds — global fallback (when country cohort is unavailable):

  • GREEN: CF ≥ 30%
  • AMBER: 22% ≤ CF < 30%
  • RED: CF < 22%

Solar thresholds — latitude-banded (no country cohort available):

  • Tropical (|lat| ≤ 25°): GREEN ≥ 18%, AMBER 14–18%, RED < 14%
  • Subtropical (25–40°): GREEN ≥ 15%, AMBER 11–15%, RED < 11%
  • Temperate (> 40°): GREEN ≥ 12%, AMBER 9–12%, RED < 9%

Terrain complexity override (wind only): If Weibull k < 1.6 and mean wind speed > 8 m/s, a GREEN result is downgraded to AMBER — indicating that complex terrain introduces resource uncertainty not captured in the atlas data.

Known limitation: Global and latitude-banded thresholds may misclassify sites in regions where the local fleet operates at different CF ranges than global averages. For example, a solar CF of 13% is typical in Northern Europe but would be flagged AMBER under the temperate threshold (≥ 12%). Country-relative thresholds reduce this problem where data is available.

02
Grid

What it measures: Proximity to high-voltage transmission infrastructure and grid saturation from nearby renewable projects.

Distance thresholds (HV / EHV, ≥ 110 kV):

  • GREEN: ≤ 15 km to nearest HV line
  • AMBER: 15–25 km
  • RED: > 25 km

Voltage adjustments: Medium-voltage lines (33–110 kV) tighten thresholds to 10 km (AMBER) and 20 km (RED). Low-voltage lines (< 33 kV) always return RED — they are not suitable for utility-scale injection.

Grid saturation (GEM pipeline data, 20 km radius):

  • GREEN: Pipeline capacity ≤ 75% of estimated headroom (default 500 MW)
  • AMBER: 75–100%
  • RED: ≥ 100% of headroom

The combined Grid verdict is the worse of distance and saturation.

Data source: OpenStreetMap (transmission lines). This is not a substitute for a hosting capacity study with the TSO.

03
Land

What it measures: The percentage of the project area that is buildable after accounting for slope, settlements, land cover constraints, and aviation buffers.

Buildable footprint thresholds:

  • GREEN: ≥ 80% buildable
  • AMBER: 60–80%
  • RED: < 60%

Deductions from 100% (cumulative): steep slope (> 20°: −60%), moderate slope (15–20°: −40%), settlements within setback zone (wind: 0.5 km, solar: 0.2 km: −30%), high dwelling density (wind: > 5 within 500 m: −40%), forest cover (≥ 50%: −30%), built-up area (≥ 20%: −25%), wetland (≥ 15%: −10%), waterbody proximity (−20%). Mega-projects incur an additional −5% per unit over 500 MW (wind) or 300 MW (solar), capped at −20%.

Aviation override (structural):

  • Airport or military installation ≤ 3 km → RED (structural)
  • Airport or military installation ≤ 8 km → AMBER
  • Heliport ≤ 1 km → AMBER

GEM permitting override: If national permitting data indicates high cancellation rates (> 35%), very long timelines (≥ 84 months), or very high complexity — a GREEN result is downgraded to AMBER.

Footprint defaults: Wind: 40 ha/MW. Solar: 1.5 ha/MW.

04
Environmental

What it measures: Proximity of the site boundary to designated protected areas.

Thresholds:

  • GREEN: > 2 km from nearest protected area
  • AMBER: ≤ 2 km buffer zone (mitigable — may require environmental impact assessment)
  • RED (structural): Overlap with protected area boundary

Bird flyway override (wind only): Sites within one of eight major global flyway corridors are flagged AMBER if they would otherwise be GREEN. This is a precautionary flag — not a KILL — but indicates that an ornithological impact assessment should be commissioned early.

Data sources: EU sites — Natura 2000 (EEA, authoritative). Non-EU — OpenStreetMap Overpass query. Query radius: 5 km.

05
Market

What it measures: The overall market viability of renewable energy investment in the project's country, adjusted for price cannibalisation risk and grid curtailment.

Composite viability score (0–10 scale, pre-computed from country-level indicators including policy strength, investment attractiveness, and market scale):

  • GREEN: Composite score ≥ 6.0
  • AMBER: 4.0–6.0
  • RED: < 4.0

Capture-rate overlay (price cannibalisation):

As renewable penetration increases, the correlation between high-generation hours and low wholesale prices tends to increase — reducing the effective revenue per MWh. The capture rate (actual revenue-weighted price / baseload price) quantifies this effect. Sourced from ENTSO-E (EU), AEMO (Australia), EIA (US), and CEN (Chile).

  • RED (structural): Capture rate ≤ 60%
  • AMBER downgrade: Capture rate 60–75%
  • Normal range: 75–90%
  • Premium: ≥ 90%

Curtailment overlay: Sourced from ENTSO-E (EU) and equivalent national TSO publications. Not available for all markets.

  • RED (structural): Curtailment ≥ 15%
  • AMBER downgrade: 7–15%
  • Normal: 3–7%
  • Low: < 3%

The combined Market verdict is the worst of the base composite score, capture-rate overlay, and curtailment overlay.

06
Sponsor / Financing

What it measures: The cost and risk of financing a project in the given country, using WACC as the primary metric.

Thresholds (project WACC midpoint):

  • GREEN: WACC mid < 10% AND FX risk ≠ High AND Offtaker risk ≠ High
  • AMBER: WACC mid 10–15%, OR either FX or Offtaker risk = High
  • RED: WACC mid > 15%

Data sources: Damodaran country risk premium, OECD country risk classification, internal country financing database.

07
Social

What it measures: Country-level social and political risk using the INFORM Risk Index, with a settlement-awareness check for the specific site.

Thresholds:

  • GREEN: INFORM Risk ≤ 3.4 AND nearby settlements identified
  • AMBER: INFORM ≤ 3.4 but no settlement data available, OR INFORM 3.5–6.5
  • RED: INFORM Risk > 6.5

The settlement check acts as a confidence modifier: if the INFORM score is low but we cannot confirm nearby population centres (which may indicate data gaps rather than absence of social risk), the result is downgraded to AMBER.

Data sources: INFORM Risk Index 2024 (EU JRC), OpenStreetMap settlement data.

08
Logistics

What it measures: Physical accessibility for equipment delivery and installation, scored across three sub-components.

Sub-component thresholds:

Sub-componentGREENAMBERRED
Port distance≤ 100 km100–250 km> 250 km
Road distance≤ 20 km20–50 km> 50 km
CAPEX benchmark≤ $2,000/kW$2,000–3,000/kW> $3,000/kW

The combined Logistics verdict is the worst of the three sub-components. If both port and road data are unavailable, the axis defaults to AMBER (mitigable).

Data sources: OpenStreetMap Overpass (ports within 300 km, roads within 100 km), IRENA country CAPEX benchmarks.

Verdict synthesis — how the overall result is determined
PASS, CAUTION, or KILL

Each axis produces a traffic-light result (GREEN / AMBER / RED) and a tag (structural or mitigable). The overall verdict is synthesised from the individual axis results:

PASS: All axes are GREEN. No material concerns identified at the screening level. The site is a strong candidate for a full pre-feasibility study.
CAUTION: One or more axes are AMBER, and no axis is RED. The site shows potential but has identifiable risks. Each AMBER axis includes documented mitigation paths. A full pre-feasibility study should address these areas specifically.
KILL: One or more axes are RED. The site has a fundamental constraint that makes it unsuitable for development at this time. KILL verdicts are further classified:

Structural KILL: At least one RED axis is tagged structural — meaning the constraint is physical, regulatory, or market-level and cannot be mitigated through project redesign. Examples: overlap with a protected area, capture rate ≤ 60%, airport within 3 km, WACC > 15%.

Data-coverage exception: Axes where data coverage is rated INSUFFICIENT are excluded from the verdict calculation (treated as informational only) — unless they return RED, in which case the RED stands. This prevents missing data from inflating verdicts while still flagging confirmed risks.

Confidence levels — how much to trust each axis
HIGH, MEDIUM, or LOW — driven by data coverage

Every axis result includes a confidence rating (HIGH, MEDIUM, LOW) that reflects the quality and completeness of the underlying data, not the verdict itself. A RED result with LOW confidence still means "proceed with extreme caution" — but you should verify the data before acting on it.

Per-axis confidence is calculated as: base_score × coverage_multiplier. The base score reflects the inherent reliability of the data source (e.g., Global Wind Atlas = 0.70, OpenStreetMap = 0.60, INFORM = 0.80). The coverage multiplier reflects data availability: FULL = 1.0, PARTIAL = 0.75, INSUFFICIENT = 0.5.
Labels: score ≥ 0.75 → HIGH, ≥ 0.50 → MEDIUM, < 0.50 → LOW.
Overall confidence is a weighted average of all axes: Resource (30%), Grid (20%), Land (10%), Environmental (10%), Market (10%), Sponsor (10%), Social (5%), Logistics (5%). If any axis returns RED, the overall confidence is capped at the Resource axis confidence (max 0.80).
Pipeline intelligence — what GEM data tells you
Success rates, attrition, and deployment velocity from Global Energy Monitor

The screener report includes a pipeline intelligence panel sourced from the Global Energy Monitor (GEM) database. This is displayed for context only — it does not affect the verdict. It is computed client-side from GEM project data within the specified radius.

Local pipeline (100 km radius): Projects from GEM filtered by status (operating, construction, planned, announced, cancelled, shelved). Shows count and MW by status.
Success rate: operating.n / active.n × 100 where active = operating + construction + planned + announced.
Attrition rate: (cancelled + shelved).n / total.n × 100.
Local velocity: operating_mw / (year_max − year_min) for operating projects within 100 km, commissioned 2010–present (MW/yr deployed).
National velocity: total_operating_mw / (year_max − year_min) for the country, 2015–present.
Pipeline depth: pipeline_mw / national_velocity — the number of years of development backlog at the current deployment rate.
Known limitations of the Site Screener
What it doesn't catch — and what to verify separately
Global thresholds are not always regionally calibrated. The Resource axis uses country-relative thresholds where data permits (20+ operating projects in the reference fleet). For countries without sufficient data, global or latitude-banded defaults are used. This can produce false positives (flagging a normal European solar yield as AMBER because the temperate threshold is ≥ 12%) or false negatives (passing a site that would be sub-economic in a competitive local market). Country-relative thresholds are being expanded as IRENA capacity data coverage improves.
Search polygon vs. project polygon. The screener evaluates the area defined by the user's search polygon. This may not correspond exactly to the final project footprint. Land, environmental, and grid distances are measured from the polygon boundary — changes to the project boundary after screening may change the results.
Grid analysis uses OpenStreetMap only. Transmission line data from OSM does not include hosting capacity, load flow constraints, or TSO-specific connection queues. A site that passes the grid distance check may still face a multi-year connection wait or significant reinforcement costs. This is not a substitute for a grid connection study with the TSO.
Land cover breakdown is forthcoming. The Land axis currently produces a single buildable-percentage figure based on slope, settlement, and land cover deductions. A detailed breakdown (forest, agriculture, urban, wetland by area and percentage) is in development.
Capture-rate and curtailment data is not available for all markets. The Market axis overlays capture-rate cannibalisation and curtailment risk where data is available (ENTSO-E for EU, AEMO for Australia, EIA for US, CEN for Chile). For other markets, these overlays are skipped — meaning the Market verdict may be more favourable than reality in high-penetration markets without published capture-rate data.
Pipeline intelligence is informational only. GEM data reflects publicly tracked projects and may not include all developments, particularly those in early pre-permitting stages or markets with limited transparency. Pipeline velocity metrics should be treated as indicative, not authoritative.
The LCOE formula
Levelised Cost of Energy — what it is and how we calculate it

LCOE (Levelised Cost of Energy) is the average cost per unit of electricity generated over the lifetime of a project, expressed in $/MWh. It is the standard metric for comparing the economics of different energy technologies and projects, independent of scale or market context.

Conceptually, it answers one question: at what minimum electricity price does this project break even over its lifetime? If the LCOE is $42/MWh and the available PPA is $55/MWh, there is headroom. If the LCOE is $68/MWh and the PPA is $55/MWh, there is not.

LCOE Formula
LCOE = (CAPEX_total + Σ OPEX_t / (1+r)^t) ───────────────────────────────── Σ AEP_t × D_t / (1+r)^t
CAPEX
Total capital expenditure ($/kW × capacity in kW). Incurred at t=0, not discounted.
OPEX_t
Annual operating expenditure in year t ($/kW/yr × capacity in kW). Held constant in real terms.
AEP_t
Annual Energy Production in year t (MWh). Year 1 AEP = Capacity × CF × 8,760 hours.
D_t
Degradation factor in year t = (1 − 0.005)^(t−1). Output declines 0.5% per year from year 2.
r
Discount rate (WACC). Applied to both cost and energy streams.
t
Year, running from 1 to project lifetime (fixed at 25 years in current model).

The summation runs over 25 years. Both the cost stream (OPEX) and the energy stream (AEP) are discounted by the same WACC — a standard approach that avoids the need to separate nominal and real cash flows and is consistent with IRENA and IEA methodology.

The degradation rate of 0.5% per year is a conservative industry standard applied to both wind and solar in the current model. For solar PV, this is consistent with crystalline silicon manufacturer warranties and falls within the typical 0.4–0.7%/yr range observed across utility-scale deployments. The specific yield base is derived from PVGIS-SARAH3 irradiance data per the installed panel technology.

Input parameters — what each one means
What you're telling the model when you move a slider
Capacity (MW)
Nameplate capacity of the project
The rated output of the wind farm or solar plant in megawatts. Used to scale CAPEX, OPEX, and energy production. Does not affect LCOE directly — LCOE is scale-independent — but affects absolute NPV and AEP outputs.
CAPEX ($/kW)
All-in capital cost per kilowatt
The total installed cost: turbines/panels, civil works, electrical infrastructure, grid connection, and development costs. Expressed per kilowatt of nameplate capacity. Benchmarks from IRENA and BNEF are used for comparison. Does not include financing costs (those flow through WACC).
OPEX ($/kW/yr)
Annual operating cost per kilowatt
Covers O&M contracts, insurance, land lease, asset management, and any ongoing regulatory fees. Held constant in real terms across the project life. Escalation is not modelled in v1.0 — see Limitations.
WACC (%)
Weighted Average Cost of Capital
The blended cost of your capital structure — reflecting both the cost of equity and the cost of debt, weighted by their proportions. This is the most sensitive input in the LCOE calculation. A 1% change in WACC typically moves LCOE by 3–5 $/MWh. If in doubt, use the regional benchmark WACC displayed alongside the slider.
Capacity Factor (%)
Average utilisation of the plant
The ratio of actual annual energy output to maximum possible output if running at full capacity 24/7. A 35% CF wind project generates 35% × 8,760 = 3,066 MWh per MW per year. Heavily location-dependent. Use the AEP estimator or GlobalWindAtlas to derive a defensible P50 for your site.
Target PPA ($/MWh)
The electricity price you're targeting
The Power Purchase Agreement price — what a buyer will pay per MWh. Used to calculate NPV and IRR. If LCOE > PPA, the project does not recover its costs. The gap between LCOE and PPA is the margin available for debt service and equity return above the cost of capital.
Fixed assumptions in v1.0
What we've locked in — and why

Some parameters are fixed in the current version rather than user-adjustable. This is a deliberate choice: pre-feasibility analysis benefits from consistency, and allowing every parameter to vary freely can produce results that are technically calculated but practically meaningless. Here is every fixed assumption, its value, and the rationale.

Project lifetime: 25 years. Standard bankable project life for utility-scale wind and solar. Used by IRENA, IEA, Lazard, and most lenders as the default. In practice, many projects operate beyond 25 years with repowering — this is not modelled.
Annual degradation: 0.5% per year (wind and solar). Applied from year 2 onward. For wind, this is a conservative industry standard; actual degradation varies by turbine class and maintenance regime. For solar PV, 0.5%/yr is consistent with crystalline silicon manufacturer warranties and sits within the typical observed range of 0.4–0.7%/yr.
OPEX escalation: 0% real. OPEX is held constant in real terms. In practice, O&M costs may escalate slightly in later years (major component replacements, end-of-warranty transition). This is a known limitation — see below.
CAPEX timing: Year 0 (not discounted). The entire capital cost is assumed to be incurred at project start. Construction period financing costs are captured implicitly through the WACC rather than modelled as a separate construction drawdown schedule.
Tax and depreciation: Not modelled. Project IRR is computed on a pre-tax, pre-debt basis — consistent with how it is used in cross-country and cross-technology comparison. Equity IRR, which requires a full capital structure, debt sculpting, and tax treatment, is not in scope for pre-feasibility.
Curtailment: Not modelled. Grid curtailment — where output is physically limited by grid operator instruction — can materially reduce actual AEP in constrained markets. Regional curtailment rates are noted in the benchmark panel for markets where this is significant.
Inflation: Not modelled. All values are in real (constant) terms. Nominal LCOE — which would be higher — is not calculated.
IRR and NPV — what we calculate and what we don't
Project IRR is not equity IRR. This matters.

preFeasibility calculates Project IRR — the internal rate of return on total invested capital, before debt service. This is the discount rate at which the project's NPV equals zero, treating the entire CAPEX as an equity investment and the PPA revenue stream minus OPEX as the cash inflow.

This is not the same as Equity IRR, which is what most developers and investors ultimately care about. Equity IRR requires:

A defined capital structure (debt-to-equity ratio)
A debt interest rate and amortisation schedule
Debt service coverage ratio (DSCR) constraints
Tax treatment and depreciation schedule
Reserve accounts (DSRA, MRA)
Often: sculpted debt repayment tied to actual generation

Rule of thumb: For a typical utility-scale wind project with 70% debt at 6–8% interest, Equity IRR will be approximately 1.5–2.5× the Project IRR shown here — the leverage amplifies returns (and risk). A Project IRR of 9% might correspond to an Equity IRR of 14–18% depending on the capital structure. Use the Project IRR as a screening metric; commission a full financial model before making capital allocation decisions.

NPV Calculation
NPV = −CAPEX_total + Σ (Revenue_t − OPEX_t) / (1+r)^t
Revenue_t
PPA price × AEP_t × D_t. Annual revenue from selling electricity at the target PPA price.
r
WACC — same discount rate applied to revenue and cost streams.
Project IRR — Bisection Method
Find r* such that NPV(r*) = 0 Solved by bisection over 50 iterations. Convergence to <0.001% accuracy. Search range: −50% to +100%.
How country benchmarks work
What the benchmark comparison is telling you

Every LCOE result is compared against a country benchmark — the published median LCOE for that technology and market from our primary sources. The comparison shows you where your project sits in the distribution: whether your assumed inputs are conservative, typical, or aggressive relative to what has actually been built and financed in that market.

The benchmark is not a target. A project with an LCOE above the benchmark isn't necessarily unviable — it may reflect a lower-wind site, higher WACC due to market risk, or a smaller project scale. A project below the benchmark isn't automatically a winner. The benchmark is context, not verdict.

Benchmark data is updated annually following the publication of IRENA's Renewable Power Generation Costs report (typically released Q1 each year). The current dataset is from the most recently published edition. Publication year and methodology version are cited on every benchmark shown.

Benchmark source hierarchy
Where the numbers come from — in order of preference
Primary — IRENA Renewable Power Generation Costs (annual). The most comprehensive published dataset for global LCOE benchmarks. Country-level CAPEX, OPEX, and capacity factor data with full methodology. Our first source for any market with coverage.
Secondary — Lazard LCOE+ Analysis (annual). US-centric but with global context. Used to cross-check IRENA figures and for markets where IRENA coverage is thin. Methodology differences are noted where they affect the benchmark materially.
Tertiary — BloombergNEF (BNEF) New Energy Outlook. Used for forward cost trajectories and markets where IRENA and Lazard have limited data. BNEF figures are modelled rather than transaction-based — this distinction is flagged on any benchmark that relies on BNEF as the primary source.
Wind resource — GlobalWindAtlas 3.0 (DTU / World Bank). Publicly available, globally consistent wind speed and capacity factor data at 250m resolution. Used in the wind AEP estimator for P50 capacity factor derivation. The tool applies a loss stack (wake, availability, electrical, curtailment) on top of the raw atlas capacity factor.
Solar resource — PVGIS (EC JRC / SARAH3 dataset). EU Joint Research Centre’s publicly available irradiance API, covering Europe, Africa, Asia, and the Americas. Used in the solar AEP estimator for plane-of-array (POA) irradiation, specific yield (kWh/kWp/yr), and inter-annual variability (CV) to derive P90. Updated annually by JRC.
Market electricity prices — EMBER Global Electricity Review (annual). Used in the PPA structuring tool as a wholesale market reference. Transaction-based where available; modelled for markets with limited spot market data.

Every benchmark shown in the platform includes a source tag, publication year, and methodology note. If we are using a modelled figure rather than a transaction-based one, we say so. We do not blend sources without disclosure.

What we don't model — and why it matters
The honest list of where the platform stops

Every analytical tool has a boundary. Here is ours, stated plainly. These are not bugs or oversights — most are deliberate decisions to keep the tool appropriate for its class. But you need to know about them before you rely on an output.

01
Grid connection costs and constraints
Grid connection is one of the largest and most variable cost items in a renewable project — ranging from near-zero for a project close to an existing substation to $50–150/kW or more for a remote site requiring a new transmission line. We include grid connection in the CAPEX benchmark range but cannot model it project-specifically without knowing the connection point. If your site has a known connection cost, include it in your CAPEX input explicitly.
02
Curtailment
In markets with grid congestion or must-run generation, a wind or solar project may be instructed to curtail output — producing less than its technical capacity. Curtailment rates of 5–20% are not uncommon in constrained grids (parts of China, India, and some European markets). This directly reduces effective capacity factor and increases LCOE. We note regional curtailment risk in the benchmark panel but do not model a curtailment adjustment in the core calculation.
03
Construction period and drawdown schedule
We assume CAPEX is incurred at Year 0 as a lump sum. In reality, capital is drawn down over 12–30 months of construction, with interest during construction (IDC) accumulating before the first MWh is generated. For large projects, IDC can add 3–8% to effective CAPEX. We account for this implicitly through the WACC assumption — if your WACC correctly reflects your all-in cost of capital including financing, the LCOE will be approximately correct — but the NPV profile will not show the construction cash flows accurately.
04
OPEX escalation
OPEX is held constant in real terms. Older wind turbines and solar plants typically see increasing O&M costs as components age and warranty coverage expires — particularly in years 15–25. For projects where later-life costs are a concern, add a conservative buffer to your OPEX input (typically 10–20% above the early-life benchmark).
05
Taxes, incentives, and subsidies
Tax treatment varies significantly by jurisdiction — accelerated depreciation, production tax credits, import duties on equipment, withholding taxes on debt service. We do not model any of these. Our LCOE and IRR are pre-tax. In markets with material tax incentives (India's accelerated depreciation, US ITC/PTC, various European feed-in tariff regimes), the actual post-incentive economics may be substantially better than shown.
06
Decommissioning costs
End-of-life decommissioning — turbine removal, foundation removal, site restoration — is not modelled. For onshore wind, decommissioning costs are often partially or fully offset by scrap value of the steel. For offshore wind, decommissioning is a material liability. This platform covers onshore wind and solar only; decommissioning is excluded.
07
P90 and uncertainty quantification
We display a P50 capacity factor — the median expected output (50% probability of exceedance). Lenders and project finance structures typically require a P90 analysis (the output level with 90% probability of exceedance) to size debt. The AEP estimator produces an indicative P90 based on the statistical distribution of the GlobalWindAtlas data, but this is not a bankable P90 — that requires a measured wind data campaign and an independent resource assessment.
08
Merchant tail and price cannibalisation
The LCOE and NPV calculations assume a constant PPA price across the full 25-year project life. In reality, most projects have a contracted period (typically 10–20 years) followed by a merchant tail at spot market prices. As more renewables enter a market, the correlation between high-generation periods and low prices tends to increase — "price cannibalisation." The PPA structuring tool (in development) will model these structures explicitly.

Capture-rate handling: The Site Screener uses ENTSO-E capture-rate data (and equivalents for non-EU markets) to flag cannibalisation risk on the Market axis. The full prefeasibility LCOE/IRR engine does not yet adjust revenue projections for cannibalisation — the PPA price you input is applied as a flat rate. This gap will be addressed in the PPA structuring tool. For markets with significant cannibalisation (capture rate < 75%), treat the screener's Market flag as a signal to apply a haircut to the PPA price you use in the full study.

When to go beyond pre-feasibility
Signs that your project has outgrown this tool

preFeasibility is the right tool at the right stage — until it isn't. Here are the signals that your project needs more rigorous analysis:

You're preparing a term sheet or seeking indicative financing. Lenders will require a project-specific financial model, often independently reviewed. A preFeasibility output can be a useful starting point for that conversation but cannot substitute for it.
You're committing to a land option or development agreement. Any contractual commitment based on project economics should be backed by analysis that includes site-specific resource data, not modelled atlas data.
The grid connection situation is unusual. Remote sites, constrained grids, or markets with significant curtailment risk need a grid study before the LCOE means anything useful.
Tax and incentive structures are material to the decision. If production tax credits, accelerated depreciation, or local content requirements significantly affect project economics, model them explicitly.
You're presenting to an external investor or lender. Use our output as context, not as the primary deliverable. Commission a Class 1–2 study with independent engineer sign-off.
Model changelog
Every change, versioned and dated

We version every change to the calculation methodology, benchmark datasets, and fixed assumptions. There are no silent revisions. If a change affects previous outputs materially, we note it here and notify registered users by email.

May 2026
v1.2 — Site Screener methodology documented. 8-axis framework, verdict logic, capture-rate handling on Market axis, pipeline intelligence calculation, and known limitations formally documented. Site Screener repositioned as flagship product. "Where we sit" section updated to distinguish screener (Class 4) from full study (Class 3–4). Capture-rate cannibalisation handling clarified in limitations.
Apr 2026
v1.1 — Solar preFeasibility live. Full six-step solar workflow added: PVGIS-SARAH3 irradiance (EU JRC), site-specific specific yield, P90 from inter-annual CV, 25-year degradation at 0.5%/yr, LCOE/IRR/NPV with country benchmarks, OSM site access study, and WDPA environmental screening. Degradation assumption now documented for both wind and solar.
Jan 2026
v1.0 — Initial release. Wind LCOE calculator live. 25-year project life, 0.5% annual degradation, bisection-method IRR. Benchmark data from IRENA 2024, Lazard v17.0. GlobalWindAtlas 3.0 integration for AEP estimator.

Questions about the methodology? Spotted an error? Write to methodology@prefeasibility.com. We read every message and respond to technical questions directly — not with a support ticket.