Losses in photovoltaic systems: from incident irradiance to liquid AC

The Performance Ratio (PR) is the magnitude that synthesizes all the inefficiencies of a photovoltaic system: it is the ratio between the liquid AC energy actually delivered to the network and the energy that would be produced by an ideal system with the same area and actual irradiance, but without losses. A PR of 83% means that 17% of the incident solar energy was lost between the module and the output meter. For a bankable generation rating, each percentage point of these losses needs to be quantified, justified and traced to the normative literature.

Differently from resource analysis (NT.07), which is dominated by the uncertainty of GHI of the long term, the uncertainty of the estimate of losses in VF is dominated by three operational components: soiling (dirtyness), module degradation and availability. These are terms that the analyst can model with increasing precision — but that the lender knows that they tend to be underestimated in optimistic studies.

Classification of losses

The losses in FV systems are divided into four groups with different physical mechanisms and methodological treatment:

Optical and conversion losses: Cell temperature, incidence angle effect (IAM), spectral reflection. They are deterministic — calculable from climatic conditions and module specifications — and with relatively low uncertainty.

Field losses: It varies with the intensity of rain, the frequency of cleaning and the characteristics of the local dust.

Electrical losses: DC wiring resistance, investor efficiency (including clipping), transformer losses and AC wiring. They are relatively fixed and well modeled by the design specification.

Operational losses: System availability (failure, scheduled maintenance, disconnections), progressive degradation of modules over the life of the system. Degradation is the main source of uncertainty in the long-term P90 in projects with a 25-year contract.

Cell temperature: the largest component of loss

The efficiency of photovoltaic modules decreases as cell temperature rises, at a rate characterized by the power temperature coefficient (γ), typically −0.32% to −0.45%/°C for modern monocrystalline modules and −0.26% to −0.32% for high-efficiency bifacial modules. Cell temperature (T_cel) is calculated from ambient temperature (T_amb), irradiance (G) and NOCT (Nominal Operating Cell Temperature):

T_cel = T_amb + (NOCT - 20) × G / 800

For the Brazilian Northeast, the mean T_amb during solar hours is typically 32–38°C in dry period, resulting in T_cel of 55–70°C under full irradiance (1000 W/m2). With NOCT = 44°C and γ = −0.36%/°C, the temperature loss in relation to STC (25°C) is:

= 0.36% × (62 − 25) = 0.36% × 37 = 13.3% (instant) Average annual (weighted by irradiance): ~3.5%

Bifacial modules with open frame and higher ventilation speed present T_cel 3–6°C lower than modules with back-sheet, reducing annual temperature loss to ~2.5–3.0% — one of the technical arguments for their adoption in high-heat regions.

Soiling: variable with greater operational uncertainty

The deposition of particulate material (dust, biological material, poultry excrement) on the surface of the modules reduces the transmission of the glass cover and, consequently, the generation. In Brazil, the salting is strongly seasonal — inversely correlated with precipitation — and the interregional variation is the largest of any component of loss:

• Northeastern semiarid (PI, CE, RN, PB, PE, BA, GO): Dry period of 5–7 months with practically zero clean rain. Loss by soiling can reach 5–8% in peak months (August–October) without active cleaning. Yearly average without maintenance: 3.5–5.5%. With monthly cleaning: 1.5–2.5%.
• Central-West (GO, MT, MS, MG - Closed): – Dry May–September, burns increase particle load. Annual average: 2.0–3.5%.
• South and Southeast: More distributed precipitation, lower dust load. Annual average: 0.5–1.5%.

Module degradation: the long-term risk

The efficiency of photovoltaic modules decreases with exposure time — a phenomenon called long-term degradation (LTD). The two main mechanisms are:

• LID (Light-Induced Degradation): It occurs in the first hours of exposure to sunlight due to the formation of boron-oxygen complexes in Czochralski cells. The typical loss is 1–3% in the first week of operation, then stabilizing. New generation PERC and TOPCon modules have shrinking treatments that reduce LID to 0.5–1.0%.
• Progressive degradation (Letid and other mechanisms): After stabilization of the LID, efficiency drops to a rate of 0.3–0.7% per year, depending on technology, manufacturing quality and climatic conditions. The standard IEC 61215 specifies that the maximum loss in 25 years should be 20% (equivalent to 0.8%/linear year).First-line manufacturers offer 25-year warranty with a maximum of 20% drop.

The financial impact of the degradation rate is enormous in 25-year projects with long debt. The difference between a degradation rate of 0.30%/year and 0.65%/year results in ~7.7% less energy in year 25 — for a project of 300 GWh of P50 in year 1, this represents ~23 GWh of difference in the twenty-fifth year. Long-term generation P90 is therefore strongly determined by the uncertainty in the degradation rate adopted in the financial model.

Investor: clipping and efficiency

The loss by conversion in the investor combines two distinct contributions that are often confused:

• DC/AC conversion efficiencyThe ratio between AC power output and DC energy in the inverter input. Modern string and central investors have European efficiency (weighted by typical load distribution) of 97.5–98.5%. Typical conversion loss is 1.2–1.8%.
• ClippingIt occurs when the available DC power of the modules exceeds the nominal power of the inverter (DC/AC ratio > 1). The excess power is discarded. The clipping is intentional in the sizing of systems with high irradiance — a DC/AC ratio of 1.15–1.30 is common in Brazil to maximize the capacity factor in the peak time. The loss of clipping is typically 0.3–1.5% of the annual generation, depending on the adopted ratio and the irradiance profile.

In bankable reports, the two contributions should be reported separately. Clipping should be demonstrated with validated hour-to-hour simulation against the irradiance profile of the adopted database (NT.07) — not estimated by pocket rule. Investors with thermal derating (power reduction by high internal temperature) introduce a third contribution that many reports omit, especially relevant in hot climates with closed investor chambers.

Documentation for bankability

IEC 61724-1:2021 defines minimum requirements for FV system performance monitoring. For due diligence and bankable P90 purposes, the loss report shall include:

• Detailed loss table with each component, value in % of gross AEP, calculation methodology and normative reference;
• Soiling: regional validation-based data measured or estimated with explicit reference to source (SONDA/INPE station, peer-reviewed publication, own measurement);
• Degradation: central rate and uncertainty distribution, with reference to module technology, IEC 61215/IEC 61853 standards and manufacturer's warranty;
• Availability: probability of annual non-availability, including scheduled maintenance, electrical failures and network disconnections under contract;
• The sensitivity of PR and P90 to ±30% variation in the soiling, degradation and availability assumptions.