A bankable energy yield assessment (EYA) chains around twenty documented loss factors and a formal uncertainty analysis to turn a site’s solar resource into a probabilistic, lender-grade 25-year generation forecast. This methodology explains each step — from the resource cascade and the loss register to P50, P75 and P90 — and why the financing structure decides which probability of exceedance a lender needs.
Why the yield number decides the deal
For a renewable energy project, revenue, debt sizing and the debt-service-cover ratio (DSCR) all flow from one number: predicted generation. An over-optimistic yield inflates the financial model and puts debt service at risk in a poor year; a defensible, independently reviewed yield is the foundation of bankability.
That is why lenders expect the EYA to be produced — or independently reviewed — by a specialist, and why the assessment must be transparent about every assumption and every loss it applies.
Inputs and modelling tools
A bankable EYA starts from long-term solar resource data (satellite-derived or ground-measured), the detailed site layout, the module datasheet (as a PAN file), the chosen inverter model (as an OND file), and cable, transformer and transmission-line loss calculations.
These are run through an hourly — or sub-hourly — time-step simulation in industry-standard PV modelling software, so that plant behaviour is captured across the full range of irradiance and temperature the site actually experiences, rather than a single annual average.
From sun to array: the resource cascade
The starting point is global horizontal irradiance (GHI). This is transposed to the plane of the modules using a transposition factor that accounts for the split between direct-beam and diffuse light and for ground-reflected (albedo) gains — the latter being more significant on tilted and tracked arrays. Getting this plane-of-array irradiance right is the single largest driver of the final yield.
The loss chain, line by line
A credible EYA quantifies each documented loss rather than applying a blanket de-rate. The chain typically includes: horizon and far shading; near and inter-row shading (where a small shadow can cause a disproportionate electrical loss); the incidence-angle modifier; low-irradiance losses; module temperature losses; soiling; module quality versus nameplate tolerance; light-induced and first-year degradation; module and string mismatch; DC ohmic (cable) losses; inverter efficiency de-rating; AC ohmic losses; inverter-duty and power-transformer losses; medium-voltage cable and transmission-line losses; auxiliary consumption; plant and grid unavailability; and any clipping or unused energy caused by a grid export limit.
Two of these are worth calling out. Soiling is highly site-specific — driven by rainfall, dust and the cleaning regime — and is usually kept under about 3–4%, lower at high tilt where rain self-cleans the modules. And module temperature is one of the largest single loss lines in hot climates, because crystalline-silicon modules lose roughly 0.5% of output for every degree Celsius above 25°C.
Bifacial gains
For bifacial modules, the rear-side irradiance gain is modelled as its own line — net of the additional rear mismatch it introduces — because it depends heavily on mounting height, ground albedo and row spacing, and should never be assumed as a flat percentage uplift.
The headline outputs
The assessment reports first-year P50 energy (MWh/yr), specific yield (kWh/kWp), the plant load / capacity utilisation factor (PLF/CUF), and the Performance Ratio (PR) — the ratio of actual to theoretically possible output. Each should be stated with the basis on which it is calculated (for example, PR referenced to plane-of-array irradiance, not GHI), so the numbers can be compared like-for-like against operating data later.
Uncertainty and probability of exceedance (P50, P75, P90)
A yield figure without an uncertainty band is not bankable. The EYA combines an uncertainty stack — typically a couple of percent for the modelling software, a few percent for a well-maintained met-station pyranometer, and appreciably more (often 7.5% or higher) for satellite-derived resource — into a combined uncertainty that can approach 10%, dominated by the resource data itself.
That uncertainty is what converts a single P50 estimate into P75 and P90 — the yields that will be exceeded with 75% and 90% probability. A one-year P90 matches near-term debt service, while a ten-year P90 reflects decade-scale resource variation; the financing structure dictates which the lender requires. Characterising inter-annual variability (typically of the order of 5%) generally needs around ten years of resource data, and the uncertainty narrows over the asset’s life as year-to-year variability averages out.
The 25-year view
Finally, an annual degradation rate — commonly around 0.4–0.5% per year, taken from the module warranty — is applied across the debt term and the 25–30-year asset life, so the model reflects declining output over time rather than a flat first-year number held constant.
What makes an EYA bankable
Independence is the first test: SgurrEnergy’s assessments are free from EPC, OEM and equipment-supply interests, so the yield is not shaped by anyone selling into the project. Beyond that, a bankable EYA uses site-specific soiling and cleaning assumptions, models full near and far shading, quantifies its uncertainty explicitly, and includes a design-tolerance sanity check on the headline PR — giving lenders a number they can lend against.