Bifacial gain and albedo: how much extra energy do bifacial modules produce?

In field test arrays, mostly with PERC modules, bifacial modules on trackers over natural grass added about 5% to 9% to energy; small arrays overstate this. Ground albedo (the share of sunlight the ground reflects, 0.15 to 0.26 for grass), the bifaciality factor and row geometry decide the result. Snow and white covers can push the gain past 10%.

What is the bifaciality factor, and how is it measured?

The bifaciality factor (φ) is the ratio of a module's rear-side performance to its front-side performance, each measured at standard test conditions (STC). IEC TS 60904-1-2 defines three versions: for short-circuit current, open-circuit voltage and maximum power (see the published preview).

The IEA PVPS Task 13 report describes the test. Each face is lit in turn at 1,000 W/m² and 25 °C (air mass 1.5) with the other face behind an opaque, non-reflective background. A second measurement gives the power gain per unit of rear irradiance (BiFi), using an equivalent front irradiance:

G_E = 1,000 W/m² + φ × G_r

In the 2024 second edition, φ here is the short-circuit-current factor; the 2019 edition used the lower of the current and power factors. Amendment 1 followed on 16 March 2026.

Bifaciality depends mainly on the cell design. The table compares passivated emitter and rear cell (PERC), tunnel oxide passivated contact (TOPCon) and heterojunction (HJT) modules, using the VDMA's International Technology Roadmap for Photovoltaic (ITRPV, 17th edition, Figure 62), printed datasheet values and an institute presentation.

Datasheet tolerances run from ±5% to ±10%. TÜV Rheinland (reported by IEA PVPS) found ±2% to ±5% variation within one module family. It also found that a 0.15 difference between module types changed the outdoor module performance ratio by less than 2%, which fits the formula below.

How is bifacial gain defined, and what is the formula?

Bifacial gain (BG) is the extra energy a bifacial system produces compared with a monofacial system of the same orientation and size. The IEA PVPS report defines it at three levels.

BG_optical = G_rear / G_front
BG_module  = φ × G_rear / G_front
BG_system  = (E_bifacial - E_monofacial) / E_monofacial

G is irradiance in W/m² on the rear and front faces, and E is annual energy in MWh. System gain falls below module gain because rear-side current raises ohmic losses and inverter clipping, the two main nonlinear effects the report names.

NREL modelled a rear-to-front irradiance ratio of 8% to 9% in Albuquerque and 9.5% to 11% in Seattle for a tracker over ground with an albedo of 0.25. Seattle is higher because of more diffuse light. TÜV Rheinland found that in optimised installations the ratio is around 50% of the albedo, so grass at 0.20 points to about 10%. The table applies the module formula to three example ratios.

These are module-level results before shading, mismatch, ohmic and clipping losses. The ratios are example inputs, not site data.

Which site factors decide bifacial gain?

Ground albedo and module bifaciality are generally the two most important factors, according to the IEA PVPS report. The other factors set how much reflected light reaches the rear face.

Trackers and fixed tilt also differ in total energy yield; the trackers vs fixed tilt explainer covers that comparison. The IEA report adds a warning for evaluators: small test sites show higher gains than utility-scale plants, because of less self-shading and more edge brightening.

What are typical albedo values for solar sites?

Grass reflects about 0.15 to 0.26 of the sunlight that falls on it. Snow and white covers reflect far more.

Albedo is not fixed: it changes with wetness, dryness and the greenness of vegetation. It also depends on the sensor. NREL compared a reference cell with a pyranometer and found a spectral effect over green grass about 10% larger than over drier summer grass.

What bifacial gain on grass do field studies measure?

The annual and seasonal figures below put gains over natural ground at about 5% to 9%, mostly for PERC modules. Small test arrays tend to overstate gain; white covers and snow give more.

Three points follow. First, annual grass results sit below 10%, while white covers mostly sit above it. NREL's global tracker model gives a typical 9%, but it assumes a bifaciality of 100% and no extra losses, which its authors say may be too optimistic.

Second, gain does not stay constant. In NREL's BEST field the PERC group's gain fell from 6.6% in the first partial year to 2.9% in the fifth. The bifacial strings lost performance at about 0.94% a year for PERC and 1.46% for HJT, against 0.34% for monofacial. These are 2019-era modules in one field. The degradation calculator shows how such rates compound.

Third, models differ. In the IEA comparison, predicted gains varied by up to a factor of two between models. PVEL found mean bias errors between -1.8% and 1.3% in predicted bifacial gain for three software platforms against its measured array.

What this means for a grassland park in Hungary

For context, SolServices' Lumen Park Mezőtúr (66.3 MWp DC) uses bifacial modules on a sun-tracking system, according to the construction contractor Greencells and the company's project page. That sentence is context only: no gain figure in this article comes from SolServices data.

Seasonal albedo

Hungary lies between 45°45' and 48°35' north (HungaroMet). The Hungarian synthesis gives meadow albedo of 0.17 to 0.20 from April to October, inside the general grass range. Dry grass can reflect more: a California grassland measured about 0.15 when green and nearly 0.30 when dead. That site is not Central European, so it shows a direction, not a Hungarian value.

Using Merodio et al.'s coefficient, an albedo error of 0.05 moves annual yield by at most about 1.5%. The 0.03 spread in the Hungarian meadow values is well inside that margin, but a measured dry-season value for Hungary is missing.

Snow

The central Great Hungarian Plain has the fewest snow-cover days in the country, about 30 to 35 a year. Across the wider Plain, winters range from no snow cover to 80 to 100 days (Tar et al., 2025, citing Mersich et al., 2002).

Those days fall in months with little output. A PVGIS run for a fixed system at Szolnok (47.18°N, 20.19°E), with PVGIS's optimal angles and the default 14% loss, gives 1,253.72 kWh/kWp a year: 45 kWh/kWp in December and 146 in July. December to February deliver about 13.0% of annual energy. Snow lifts rear irradiance sharply, but a Hungarian year gets little of it. Sandia's modelling guide also notes that the snow gain assumes the array itself stays snow-free.

Vegetation management and rear-side gain

The ground cover is the reflector, so mowing, grazing and regrowth act on rear-side gain through albedo. Two points follow from the sources above.

• The size is bounded. Yield moves by at most 0.3% per 0.01 of albedo in the simulations of Merodio et al., so a cutting regime matters only through the albedo change it causes.
• Plan and measurement must match. Albedo measured before construction may not hold afterwards: Merodio et al. found that land-use change from construction creates an albedo uncertainty larger than any measurement-condition uncertainty.

This page found no measurement of how mowing height or regrowth changes albedo or output on a tracker plant in Central Europe. On-site albedo logs would close that gap.

What to check in a yield assessment

• The albedo basis: measured on site, by season, for the vegetation plan the plant will keep. Lu et al. found that a fixed albedo cannot represent snow events.
• The measured bifaciality factor and its tolerance.
• Row-end share, torque-tube shading, mismatch, ohmic and clipping losses, listed separately.
• The model and its validation against field data.

Effect on cost per MWh

Extra energy enters levelised cost of electricity (LCOE) through the yield input. The LCOE calculator can test a gain assumption; its results are estimates. The 2021 IEA report summarises a global analysis in which bifacial modules on single-axis trackers gave the lowest LCOE across 93% of the world's land area, and ITRPV 2026 reports that price premiums for bifacial modules are nearly gone. How to read the resulting yield figures is covered in capacity factor vs performance ratio.

Limits of this explanation

• No measured gain from a Hungarian park is used. The nearest measured analogues are in Denmark, Germany and the United States.
• Most field data come from small test arrays, many with PERC modules, gathered in 2019 to 2021 or earlier. No TOPCon field result is included; that column in the formula table is arithmetic.
• Risø covers only July to September 2019, its covers differ in GCR, and its tarp and gravel plots were narrow, so some rear light still came from the surrounding grass.
• ITRPV module averages were read from a chart, and datasheet values vary by product.
• Albedo ranges are published values. A site needs its own measurement.