High-Reflectivity Ground Covers for Energy Yield Enhancement in Single-Axis Tracked Bifacial Photovoltaic Systems: Field Evidence from Brazil

1. Introduction

Photovoltaic (PV) generation has become one of the main options for expanding electricity supply with low operational greenhouse gas emissions and the lowest levelized cost of energy. Its growth has been enabled by rapid reductions in PV module cost, improvements in efficiency, and manufacturing scale, which have progressively increased the competitiveness of solar electricity in different markets [1]. In countries such as Brazil, where abundant solar resource led to rapid growth in installed PV capacity [2,3], further performance improvements are particularly relevant because even modest percentage gains can translate into large cumulative energy gains over the lifetime of utility-scale PV assets.

Among the most important changes in PV plant design over the last decade has been the consolidation of bifacial modules and single-axis trackers [4,5,6]. Bifacial modules generate electricity from irradiance incident on both front and rear surfaces, while trackers increase the collection of direct irradiance by following the apparent motion of the sun. The combination is especially attractive for utility-scale plants because it can improve energy production per installed watt and reduce the levelized cost of electricity, with bifacial single-axis tracking systems shown to achieve the lowest levelized cost of electricity over more than 90% of the investigated land area in a global techno-economic assessment [7].

The relevance of rear-side irradiance is expected to increase further as the PV industry transitions from p-type PERC products to higher-efficiency n-type technologies, especially TOPCon and heterojunction modules [8]; as bifacial tandem perovskite-silicon PV evolves from pre-commercial to large-scale deployment, additional performance aspects related to subcell current mismatch will bring more complexity to the handling of the effects albedo. Technology roadmaps indicate that these module families tend to have higher bifaciality factors and lower temperature coefficients than conventional PERC modules [9]. As module bifaciality increases, the rear-side response becomes more effective in converting additional reflected irradiance into electrical output, thereby enhancing the incremental gain provided by reflective covers.

In conventional PV power plants, the ground surface is usually selected or modified according to civil works, vegetation management, access, erosion control, and drainage. In large-scale bifacial PV power plants, however, deliberately increasing ground albedo can increase rear-side irradiance and potentially improve annual energy yield. Artificial high-reflectivity covers such as polymeric films, white membranes, and geomembranes are therefore gaining attention as an approach to enhance bifacial performance without increasing the installed DC capacity or the occupied area. This paper uses the term high-reflectivity ground covers to refer to these artificial reflective surfaces.

The potential benefit of this strategy is conceptually simple but technically complex. Increasing albedo does not automatically lead to a proportional increase in delivered energy. Rear irradiance depends on the view factors between modules, sky, ground, and neighboring rows; on tracker position and backtracking; on the direct and diffuse components of irradiance; and on the spatial distribution of light on the rear side of the module. Additional rear-side irradiance will also influence mismatch, inverter clipping, module temperature, soiling, and maintenance practices. Therefore, field evidence is essential to understand the net gain that can be achieved under realistic operation.

This article reports experimental results from two annual monitoring cycles at the Fotovoltaica/UFSC pilot plant in Florianópolis, southern Brazil (27.4°S, 48.4°W), evaluating the effects of different ground-cover albedos on both irradiance and PV performance. Irradiance effects were characterized at a dedicated albedo measurement station, while the corresponding energy-yield response was measured directly in bifacial PV arrays mounted on single-axis trackers. Preliminary results from the first monitoring cycle, focused on two reflective films, were previously reported in Braga et al. [10] and Azevedo et al. [11]. The present article builds on those preliminary studies by integrating the first-cycle findings with a second annual monitoring cycle, including optimized reflective covers, and by providing a broader assessment of albedo, spectral response, energy yield, soiling, thermal behavior, electrical limitations, and practical installation aspects. The objective is to quantify how different reflective ground covers modified albedo, rear-to-front irradiation ratio, and energy yield in bifacial PV systems with single-axis tracking, while also documenting the observed influence of soiling, temperature, and installation methods. The manuscript is deliberately focused on the field results and on the reflective-cover literature that supports their interpretation.

2. Literature Background: Reflective Ground Covers for Bifacial PV

The energy gain of a bifacial PV system can be interpreted at different levels. The optical gain is associated with the additional irradiance reaching the rear surface. The module-level gain includes the bifaciality factor and conversion response of the rear side. The system-level gain includes thermal behavior, mismatch, DC and AC losses, inverter clipping, and the characteristics of the power conversion system [6,12]. This distinction is central for reflective ground covers because their primary action is optical, but the result of interest for plant owners is the net electrical energy delivered by the system.

Rear-side irradiance is generally more spatially non-uniform than front-side irradiance because it is largely composed of sky-diffuse and ground-reflected contributions, whose distribution depends strongly on view factors, tracker position, torque-tube shading, row-to-row shading, module height, ground coverage ratio, and the spatial distribution and optical properties of the reflective surface. This non-uniformity can produce electrical mismatch between cells or substrings and reduce the fraction of the optical gain that is converted into useful power [13,14,15]. For this reason, measurements and simulations of bifacial systems often require more detailed geometric descriptions than those used for conventional monofacial systems [16,17,18,19].

Experimental evidence has shown that increasing ground reflectance can substantially improve bifacial PV performance. Chudinzow et al. simulated and assessed bifacial power-plant yield over different ground conditions and reported the strong dependence of bifacial yield on surface reflectivity [20]. Riedel-Lyngskaer et al. evaluated bifacial systems in a representative pilot plant in Denmark and found that highly reflective ground materials increased yield for both fixed-tilt and single-axis tracking configurations [21]. These results also emphasized that the economic value of the additional energy depends on the cost and durability of the ground modification.

More recently, Lewis et al. combined field measurements, ray-tracing simulations, and economic analysis to evaluate the size and position of artificial reflectors in single-axis tracked bifacial PV systems [22]. Their results showed that reflector geometry is a critical design variable and that electrical losses, particularly clipping, can reduce the value of additional irradiance. This finding is relevant for retrofits in existing plants, where the inverter loading ratio and DC string design were not necessarily selected to accommodate the extra rear-side current produced by high-albedo covers.

In parallel with the peer-reviewed literature, technical press reports indicate growing commercial interest in reflective membranes and geomembranes for bifacial PV applications. Articles published by industry news outlets have reported emerging commercial interest in reflective geomembranes and related ground-cover materials for bifacial PV projects, including manufacturer- or developer-provided estimates of yield gains, LCOE reductions, additional CAPEX, and payback periods [23,24,25]. Reported field applications indicate a wide range of yield gains, from approximately 5% to 20%, reflecting differences in system configuration, tracker geometry, baseline albedo, cover layout, and local operating conditions. However, the same reports also highlight uncertainties related to incomplete cost accounting and potential annual reflectivity losses caused by weathering and dust accumulation. More recently, a specialized PV industry news outlet reported the development of one of the reflective membranes evaluated in the present study by a Brazilian manufacturer, including its application in a bifacial PV plant in the Minas Gerais state region, with a reported 7% gain measured at the inverter output [26]. These technical reports do not replace independently validated scientific studies, but they show that reflective covers are already being considered in commercial, demonstration and showcase projects and reinforce the need for field-based assessments of energy gain, durability, soiling, and practical feasibility.

Besides broadband albedo, the spectral distribution of the reflected irradiance can influence PV performance. Different surfaces with similar broadband albedo may reflect differently across the spectral response range of silicon modules. Riedel-Lyngskaer et al. and Tonita et al. showed that spectral albedo can affect bifacial PV system output measurements and system model predictions, particularly when reference sensors and PV modules respond differently to the reflected spectrum [27,28]. In the Brazilian context, Braga et al. further demonstrated that local spectral irradiance distributions can deviate substantially from standard reference spectra, reinforcing the need to consider spectral effects in PV performance assessments [29,30]. Therefore, field characterization of artificial ground covers should, whenever possible, include both broadband albedo and PV-weighted measurements.

Operational exposure is another key issue. The reflectivity of artificial covers may decrease because of soiling, weathering, surface degradation, biological growth, or damage during maintenance. The effect can be stronger in locations with dust sources, nearby construction, dry periods, or insufficient natural cleaning. Studies of bifacial PV on reflective surfaces have noted that maintaining the optical properties of the ground material is essential for the long-term value of the solution [21,28]. Thermal effects must also be considered because additional reflected irradiance absorbed by the modules can increase operating temperature and partially offset energy gains [31,32].

Finally, utility-scale deployment requires practical feasibility beyond optical performance. Reflective covers must remain stable under wind, allow access for operation and maintenance, avoid creating new soiling traps, and be compatible with drainage and environmental requirements. Large PV plants can alter runoff and surface conditions [33], and additional impermeable or low-permeability covers may further affect hydrology and erosion. The present work therefore evaluates not only measured yield gains, but also the practical issues observed during the field campaign.

3. Materials and Methods

3.1. Pilot Plant and Albedo Characterization

The experimental campaign was carried out at the Solar Energy Research Laboratory Fotovoltaica/UFSC PV pilot plant in Florianópolis, Brazil (27.4°S, 48.4°W www.fotovoltaica.ufsc.br). The site includes bifacial PV systems mounted on horizontal single-axis trackers (SATs) and a dedicated albedo station adjacent to the tracker field, as shown in Figure 1. The experimental design compared systems installed over high-reflectivity ground covers with a reference system installed over gray gravel, whose measured albedo is within the 20–30% range typically associated with common ground conditions in Brazilian utility-scale PV plants. The comparison was performed under the same local meteorological conditions and using the measured energy output of each tracked PV system.

The photovoltaic system measurements were carried out using the instrumentation installed on the tracker systems, including sensors for module temperature, rear-side irradiance, and front-side plane-of-array irradiance, as shown in Figure 2 and described in Table 3.

Table 1. Instrumentation used for photovoltaic system measurements at the single-axis tracker bifacial PV pilot plant. Selected sensors are shown in Figure 2.

Measurement	Instrument	Specification
Front- and Rear-Side Global POA Irradiance	EKO MS-80S	Classe A Pyranometer;
Front- and Rear-Side POA Irradiance	IMT Si-mV-85-PT100	Classe A Reference Cell
PV Module Temperature	PT1000	Classe A
Electrical Parameters	Huawei SUN2000	String Inverters
Meteorological Variables	Fotovoltaica/UFSC Weather Station	Various Sensors

Table 1. Instrumentation used for photovoltaic system measurements at the single-axis tracker bifacial PV pilot plant. Selected sensors are shown in Figure 2.

Measurement	Instrument	Specification
Front- and Rear-Side Global POA Irradiance	EKO MS-80S	Classe A Pyranometer;
Front- and Rear-Side POA Irradiance	IMT Si-mV-85-PT100	Classe A Reference Cell
PV Module Temperature	PT1000	Classe A
Electrical Parameters	Huawei SUN2000	String Inverters
Meteorological Variables	Fotovoltaica/UFSC Weather Station	Various Sensors

Two annual experimental cycles were considered. The first cycle evaluated a white film and a pearl-white film against the gray gravel reference over a 12-month period. The second cycle evaluated a black-and-pearl film and a white geomembrane, again using the gray gravel system as reference, also over a further 12 months period. During the experimental campaign, the PV modules installed in the pilot plant were replaced. The initial configuration used p-type monocrystalline silicon PERC bifacial modules, while the current configuration uses n-type monocrystalline silicon bifacial modules. The main characteristics of both module types are summarized in Table 2.

Before the reflective covers were evaluated, baseline periods were carried out for both module configurations. During these periods, black tarps were installed below all evaluated trackers to homogenize the ground reflectivity and impose a uniform low-albedo condition, as shown in Figure 3. The purpose of this procedure was to verify the relative consistency among the identical PV systems before attributing performance differences to the ground covers.

Albedo was characterized by broadband radiometric measurements and by silicon reference cells, complemented by spectral reflectance measurements. The instrumentation used at the dedicated albedo station is shown in Figure 4 and described in Table 3. Broadband albedo represents the ratio between reflected and incident solar irradiance integrated over the sensor response. Reference-cell albedo provides a PV-weighted estimate that is closer to the spectral response of crystalline silicon modules. Spectral measurements were used to compare the wavelength-dependent reflectance of natural surfaces and artificial covers within the relevant range for silicon PV conversion.

Table 3. Description of the instrumentation used at the albedo station of the Solar Energy Research Laboratory Fotovoltaica/UFSC in Florianópolis, Brazil (27.4°S, 48.4°W www.fotovoltaica.ufsc.br).

Measurement	Instrument	Specification	Time Resolution / Measurement Range	Application
Incident Global Irradiance	Kipp & Zonen SMP22	Class A Pyranometer	1 s	Broadband Albedo Calculation
Reflected Global Irradiance	Kipp & Zonen SMP22	Class A Pyranometer	1 s	Broadband Albedo Calculation
Incident Irradiance	IMT Si-mV-85-PT100	Class A Reference Cell	1 s	Silicon-Response Albedo
Reflected Irradiance	IMT Si-mV-85-PT100	Class A Reference Cell	1 s	Silicon-Response Albedo
Incident and Reflected Spectral Irradiance	EKO MS-720	Portable Spectroradiometer	350–1050 nm	Spectral Reflectance

Table 3. Description of the instrumentation used at the albedo station of the Solar Energy Research Laboratory Fotovoltaica/UFSC in Florianópolis, Brazil (27.4°S, 48.4°W www.fotovoltaica.ufsc.br).

Measurement	Instrument	Specification	Time Resolution / Measurement Range	Application
Incident Global Irradiance	Kipp & Zonen SMP22	Class A Pyranometer	1 s	Broadband Albedo Calculation
Reflected Global Irradiance	Kipp & Zonen SMP22	Class A Pyranometer	1 s	Broadband Albedo Calculation
Incident Irradiance	IMT Si-mV-85-PT100	Class A Reference Cell	1 s	Silicon-Response Albedo
Reflected Irradiance	IMT Si-mV-85-PT100	Class A Reference Cell	1 s	Silicon-Response Albedo
Incident and Reflected Spectral Irradiance	EKO MS-720	Portable Spectroradiometer	350–1050 nm	Spectral Reflectance

3.2. Characterization and Performance Indicators

For the characterization of the materials installed at the albedo station, the broadband albedo, ($\alpha$), was calculated as the ratio between the reflected irradiance measured by the downward-facing pyranometer ($G_{\mathrm{r}\mathrm{e}\mathrm{f}\mathrm{l}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{e}\mathrm{d}}$) and the incident global irradiance measured by the upward-facing pyranometer ($G_{\mathrm{i}\mathrm{n}\mathrm{c}\mathrm{i}\mathrm{d}\mathrm{e}\mathrm{n}\mathrm{t}}$), as shown in Equation (1). Similarly, the silicon-response-weighted albedo (${\alpha }_{\mathrm{S}\mathrm{i}}$) was obtained from the pair of silicon reference cells ($G_{\mathrm{r}\mathrm{e}\mathrm{f}\mathrm{l}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{e}\mathrm{d},\mathrm{S}\mathrm{i}}$ and $G_{\mathrm{i}\mathrm{n}\mathrm{c}\mathrm{i}\mathrm{d}\mathrm{e}\mathrm{n}\mathrm{t},\mathrm{S}\mathrm{i}}$), as shown in Equation (2). The spectral reflectance ($\alpha \left(\lambda \right)$) was defined as the wavelength-dependent ratio between the reflected and incident spectral irradiance, as shown in Equation (3).

$$\alpha ={\displaystyle \frac{G_{\mathrm{r}\mathrm{e}\mathrm{f}\mathrm{l}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{e}\mathrm{d}}}{G_{\mathrm{i}\mathrm{n}\mathrm{c}\mathrm{i}\mathrm{d}\mathrm{e}\mathrm{n}\mathrm{t}}}}$$

(1)

$${\alpha }_{\mathrm{S}\mathrm{i}}={\displaystyle \frac{G_{\mathrm{r}\mathrm{e}\mathrm{f}\mathrm{l}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{e}\mathrm{d},\mathrm{S}\mathrm{i}}}{G_{\mathrm{i}\mathrm{n}\mathrm{c}\mathrm{i}\mathrm{d}\mathrm{e}\mathrm{n}\mathrm{t},\mathrm{S}\mathrm{i}}}}$$

(2)

$$\alpha (\lambda )={\displaystyle \frac{E_{reflected}\left(\lambda \right)}{E_{incident}\left(\lambda \right)}}$$

(3)

For the PV systems, the main bifacial optical indicator was the rear-to-front irradiation ratio, ($R_{\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{r}/\mathrm{f}\mathrm{r}\mathrm{o}\mathrm{n}\mathrm{t}}$), defined as the ratio between the irradiation incident on the rear plane of array ($H_{rear}$) and the irradiation incident on the front plane of array ($H_{front}$) over the selected analysis period, as shown in Equation (4).

$$R_{\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{r}/\mathrm{f}\mathrm{r}\mathrm{o}\mathrm{n}\mathrm{t}}={\displaystyle \frac{H_{\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{r}}}{H_{\mathrm{f}\mathrm{r}\mathrm{o}\mathrm{n}\mathrm{t}}}}$$

(4)

Energy performance was evaluated using the specific yield of each tracker system, ($Y$), calculated as the ratio between the energy generated during the selected period ($E$) and the installed front-side nominal power ($P_{STC}$), as shown in Equation (5). The energy gain of each reflective-cover system ($G_{cover}$) relative to the gray gravel reference was calculated using the specific yield obtained under the same time window and filtering criteria ($Y_{cover}$and $Y_{ref}$), as per Equation (6).

$$Y={\displaystyle \frac{E}{P_{STC}}}$$

(5)

$$G_{cover}=\left({\displaystyle \frac{Y_{\mathrm{c}\mathrm{o}\mathrm{v}\mathrm{e}\mathrm{r}}}{Y_{\mathrm{r}\mathrm{e}\mathrm{f}}}}-1\right)$$

(6)

For the first cycle, the analysis included both monthly results and an annual synthesis. For the second cycle, monthly and daily results were evaluated to observe cleaning events and the stability of the gains. The analysis was restricted to the 9:00–15:00 time interval to reduce the influence of external and non-uniform shading on the comparison between systems. As a result, the absolute specific-yield values reported in this work should not be interpreted as full-day yields for the site, but as consistent comparative indicators for the evaluated systems.

Tracker operation was also considered because the rear-side contribution is sensitive to module orientation. Specific daily comparisons were made for sunny and cloudy conditions, with active tracking and with the tracker stopped, to evaluate how sky condition and tracker position affected $R_{rear/front}$ and the corresponding energy gain. In addition, inverter input-current profiles were inspected in selected high-irradiance periods to identify whether part of the optical gain could be limited by electrical constraints.

3.3. Soiling, Thermal and Installation Assessment

Soiling was evaluated through cleaning events and performance comparisons before and after cleaning. In the first 12-month cycle, construction activities adjacent to the pilot plant led to substantial deposition of particles on the reflective films, which provided a specific opportunity to quantify the loss associated with soiling under intense accumulation. In the second 12-month cycle, cleaning events were also documented, but the absence of nearby construction allowed a more representative assessment of normal exposure at the site.

The thermal behavior of the systems was assessed using irradiance-weighted module temperature. This metric gives greater weight to temperatures measured during high-irradiance periods, when energy production is more relevant. Estimated additional thermal losses were calculated from the measured temperature differences and the module temperature coefficient. Because the reported energy gains are based on measured energy generation, the thermal penalty is already embedded in the net energy results; the separate thermal analysis was used to interpret the physical mechanisms.

Practical installation aspects were documented during the campaign. Different fixation methods were tested for the reflective films, including metallic staples, raffia bags filled with sand, PVC tubes filled with sand, and trench anchoring. The assessment was qualitative but operationally important because anchoring, wind stability, durability, dirt accumulation, drainage, and maintainability can determine whether a high-albedo cover is feasible in a utility-scale plant.

4. Results and Discussions

4.1. Albedo and Spectral Characterization of the Materials

The measured albedo values confirmed that the artificial covers substantially increased the reflectivity of the ground surface relative to gray gravel. The gray gravel reference presented 24% reference-cell albedo and 25% broadband albedo. The white and pearlescent white films presented reference-cell albedo of 58% and broadband albedo of 53% and 54%, respectively. The black-and-pearl film reached 63% reference-cell albedo and 58% broadband albedo. The white geomembrane presented the highest reflectance, with 76% reference-cell albedo and 72% broadband albedo (Table 4).

The difference between broadband and reference-cell albedo highlights the importance of sensor spectral response. Reference cells weigh the reflected irradiance in a way that is closer to the response of silicon PV modules, while broadband pyranometers integrate a wider spectral range. For bifacial performance modeling, these two measurements should not be mixed without care. A model may either use PV-weighted albedo directly or apply a spectral correction, but applying both approaches simultaneously could double-count the same effect.

The spectral measurements showed that the artificial covers had higher reflectance than the natural surfaces over a large portion of the crystalline-silicon response range (Figure 5). The geomembrane and the reflective films maintained elevated reflectance across the visible and near-infrared intervals, while the natural reference surfaces were considerably less reflective. The increased noise observed near the lower and upper wavelength limits is attributed to the reduced sensitivity of the measurement equipment close to the boundaries of its operating range. The shape of the curves is relevant because the contribution of rear-side irradiance to photocurrent generation depends on both the intensity of the reflected radiation and the PV device’s quantum efficiency at each wavelength.

Measurements on representative clear and cloudy days indicated that the relative ranking of the materials’ reflectance was preserved under different sky conditions, although the absolute reflectance level varied between measurements (Figure 6). These differences should be interpreted cautiously because spectral acquisitions are instantaneous and sensitive to short-term irradiance variability, especially under overcast conditions, whereas daily albedo measurements integrate the surface response over a full day and across different incidence angles.

The temporal analysis showed that gray gravel maintained a relatively stable albedo over a long monitoring period (Figure 7a), with variations mainly associated with sky conditions and possibly influenced by rainfall events and changes in surface moisture. In contrast, the white film exhibited a gradual reduction in albedo during outdoor exposure (Figure 7b), with partial recovery after cleaning events. These results indicate that the optical durability of reflective covers should be assessed through long-term outdoor monitoring or accelerated degradation tests. Lewis et al., for example, evaluated the reflective geomembrane used in their study after accelerated ageing tests based on the IEC TS 62788-7-2 standard procedures and reported no significant loss of reflectance, suggesting that some materials retain their optical performance during their service life [22]. Nevertheless, field- and material-specific factors such as soiling, abrasion, biological growth, maintenance activities, and drainage conditions must still be considered when extrapolating accelerated-test results to long-term utility-scale operation.

The relationship between albedo and clear-sky index (k_c) presented in Figure 8 reinforces this conclusion. The gray gravel reference showed a clearer reduction in measured albedo under cloudier conditions (low k_c), while the artificial covers were less sensitive to the sky index (Figure 8). This behavior suggests that reflective covers may not only increase average albedo but also reduce the variability of rear-side irradiance contributions under changing sky conditions.

4.2. System-Level Reflective Cover Performance and Electrical Limitations

The annual energy-yield analysis was carried out separately for the two experimental cycles and then consolidated in Table 5. In both cycles, the systems installed over reflective covers were compared with the gray gravel reference operating under the same local meteorological conditions and data-filtering criteria. The main performance indicators were the rear-to-front irradiation ratio (R_rear/front, Equation (4)), and the annual energy gain relative to the reference system (Equation (6)).

The first annual cycle compared the white and pearl-white films with the gray gravel reference. As shown in Figure 9a, the module update described in Section 3.1, which was implemented to follow current industry trends, occurred during the first experimental cycle and was followed by a baseline period. This baseline period was used for system verification and was excluded from the annual energy-gain calculations presented in Table 5. Both events are identified in Figure 9a because they affected system operation and comparability during the corresponding months. Despite these operational interruptions, the reflective films consistently increased the rear-side irradiance contribution and produced positive energy gains relative to the reference surface. The cleaning event carried out in November 2024 was followed by an increase in measured gains, which may be associated with both the removal of accumulated soiling and seasonal variations in irradiance conditions. Soiling effects are further discussed in Section 4.3.

The second evaluation cycle assessed two improved ground covers with higher albedo, the black-and-pearl film and the white geomembrane, whose monthly energy gains are shown in Figure 9b. Cleaning events were carried out more frequently during this cycle than during the first. However, the results show limited correlation with these cleaning events, except for the October 2025 cleaning, after which a clearer increase was observed. To further investigate the short-term variability of the gains, daily energy gains are presented in Figure 10, with days classified according to sky condition: clear days in yellow, partially cloudy days in orange, and overcast days in red.

The results from Figure 10 show that energy gains tended to increase under cloudier conditions. This behavior indicates that, during sunnier days, especially in summer months (December to February) with higher solar resource, the performance of high-reflectivity covers was more frequently constrained by electrical limitations of the system, reducing the fraction of the additional rear-side irradiance converted into useful energy. In contrast, under cloudier conditions, the system operated farther from inverter limits, allowing a larger fraction of the optical gain to be converted into electrical output. This same effect is also evident in the monthly results of both evaluation cycles shown in Figure 9, where lower energy gains are consistently observed during summer months despite the higher solar resource available. No systematic response to the cleaning events was observed in the daily energy-gain series, suggesting that soiling had a limited and context-dependent effect during the second cycle; this aspect is discussed in more detail in Section 4.3.

The combined results from both evaluation cycles are summarized in Table 5. For the reflective films, increasing broadband albedo from 25% for gray gravel to 53–58% raised the rear-to-front irradiation ratio from 7.6% to approximately 20–21% and produced average annual energy gains of 8.8–9.5%. In the first cycle, the pearl-white and white films increased R_rear/front to 20.4% and 21.1%, with corresponding energy gains of 8.9% and 9.5%, respectively. In the second cycle, the black-and-pearl film increased R_rear/front to 19.8% and produced an energy gain of 8.8%, while the white geomembrane achieved the highest performance, with broadband albedo of 72%, R_rear/front of 23.1%, and an average annual energy gain of 11.3%. These results confirm previous findings that artificial increases in ground albedo can substantially enhance bifacial PV performance [21,22,23,24,25,26], while providing additional field evidence for engineered ground covers under Brazilian operating conditions.

The relationship between albedo and energy gain was therefore not linear. Although the geomembrane had substantially higher albedo than the black-and-pearl film, the additional energy gain was more modest than the albedo difference alone would suggest. This indicates that the net energy benefit is also influenced by system geometry, rear-side irradiance distribution, thermal behavior, and electrical conversion limits. This effect is also suggested by the results in Figure 9 and Figure 10, where lower gains were observed during summer months and clear days, when the systems operated more frequently close to inverter limits and part of the additional rear-side contribution was likely affected by clipping. Similar system-level limitations associated with the conversion of additional irradiance from reflective covers were also reported by Lewis et al. [22].

The reduction of energy gains during periods with higher solar resource suggests that clipping and other inverter-related constraints affected the conversion of the additional rear-side irradiance. To further investigate this effect, selected inverter input-current profiles during operation with the white geomembrane are shown in Figure 11. No current limitation is observed on a clear day (Figure 11a), and the additional rear-side irradiance could be converted into electrical energy without exceeding the inverter input-current limit. In contrast, during days with frequent cloud-edge effects (Figure 11b), the measured current repeatedly reached the current limit. This behavior is related to short-duration overirradiance events, which occur only under specific atmospheric conditions, particularly when cloud edges temporarily enhance the irradiance incident on the array. Such events have been reported in PV systems and can be especially relevant in high-irradiance climates and in the specific location of the study [34,35,36,37]. In the present case, the increased rear-side contribution from the high-reflectivity ground cover further amplified the current generated by the modules during these transient events. Therefore, in addition to conventional inverter power clipping, reflective covers may introduce or intensify input-current limitation when the electrical design is not able to accommodate the extra rear-side current. This effect indicates that part of the optical gain associated with reflective covers may not be converted into useful energy under specific high-irradiance and cloud-edge conditions.

These results indicate that systems with reflective ground covers may require different electrical and layout design choices from systems installed over natural or lower-albedo surfaces, including the optimization of tracker geometry, DC string configuration, inverter current limits, and inverter loading ratio. These aspects are particularly important for retrofitting existing plants, where the original electrical design may not be able to fully convert the additional rear-side irradiance into useful energy.

4.3. Soiling Effects

Soiling was one of the main operational factors observed during the field campaign. During the first cycle, construction activities east of the trackers created a substantial source of particle deposition. The spatial pattern shown in Figure 12, with less intense accumulation farther from the construction area, supports the relationship between the nearby works and the soiling observed on the reflective covers. Before cleaning, both the white and pearl-white films exhibited visible particle accumulation. Energy measurements before and after a cleaning event showed that the resulting soiling losses were measurable and depended on sky conditions. For the white film, the estimated loss was 2.1% on a sunny day and 3.4% on a cloudy day, while the corresponding values for the pearl-white film were 1.5% and 3.4%, respectively. Overall, the average soiling loss was estimated at 2.6%, corresponding to an average progression of approximately 0.65% per month during this construction-affected period (Table 6).

The second 12-month cycle allowed a complementary assessment under less disturbed conditions because there were no adjacent construction activities. The cleaning events of 10/06/2025, 08/10/2025, and 26/02/2026 were documented with aerial images of the black-and-pearl film and geomembrane before and after cleaning (Figure 13). The visual inspection confirmed that soiling still occurred, but the accumulation was less severe than during the first cycle.

The energy-gain analyses previously presented in Figure 9 and Figure 10 provide context for interpreting soiling effects during the second cycle. In particular, Figure 10 shows the daily energy gains, with cleaning events marked by vertical dashed lines. The gains fluctuated substantially from day to day, especially under cloudier conditions, partly because electrical limitations affected the conversion of the additional rear-side irradiance. This variability makes it difficult to isolate cleaning effects from meteorological conditions using daily energy data alone.

Therefore, a complementary analysis based on incident irradiation gain (Figure 14) was carried out to reduce the influence of system electrical limitations on the results. This analysis showed a weaker correlation with cloudiness and no systematic response to cleaning events, indicating that soiling had a limited impact on cover performance during the second experimental cycle. These findings suggest that the higher soiling losses observed in the first cycle were likely associated with specific exposure conditions, particularly construction activities near the pilot plant, while under more representative operating conditions the impact of dirt accumulation may be lower and more dependent on local environmental and maintenance factors.

Therefore, the results from the two cycles indicate that soiling should be treated as a relevant operational variable, but one that is strongly dependent on the deployment context. For utility-scale techno-economic analyses, it is important to consider sensitivity scenarios for different soiling-loss rates and cleaning frequencies, rather than directly extrapolating a single value obtained from the pilot plant. This approach makes it possible to represent both situations with higher particle deposition, such as construction periods, exposed soil, or intense vehicle traffic around the plant, and more stable operating conditions, in which the impact of soiling on the covers may be reduced.

4.4. Thermal Behavior

The thermal behavior of PV systems operating with reflective ground covers was evaluated separately because increasing the irradiance incident on bifacial modules can also increase operating temperature, which may partially offset the electrical benefit of the additional rear-side contribution.

In the first evaluation cycle, the irradiance-weighted average module temperature was 42.1 °C for the gray gravel reference and 44.7 °C for both the white film and pearl-white film systems. The average temperature increase was therefore approximately 2.6 °C, with monthly differences reaching up to 3.8 °C in summer. Using a module temperature coefficient of −0.3%/°C, the annual thermal loss was estimated at about 0.8%, and the maximum monthly effect was around 1.1%.

In the second cycle, which used optimized reflective covers based on the spectral response of silicon modules, the thermal penalty was smaller despite the higher rear-side contribution. The irradiance-weighted average temperature was 33.7 °C for the reference, 34.0 °C for the white geomembrane, and 34.9 °C for the black-and-pearl film. The corresponding temperature increases were 0.3 °C and 1.2 °C, leading to estimated additional thermal losses of approximately 0.1% and 0.4%, respectively.

The results from both experimental cycles indicate that the use of reflective covers produced only small increases in module operating temperature and even smaller temperature-related PV losses. These losses are much lower than the measured energy gains and therefore do not compromise the net benefit of the reflective covers in the evaluated configurations. Their relative importance may become even smaller in future applications, as modern PV cell and module technologies increasingly present lower temperature coefficients of power. It is important to highlight that, because the gains reported in the previous subsections are based on measured energy output, they already include the combined effects of rear-side irradiance, temperature, electrical losses, and real operating conditions. Therefore, the thermal analysis should be interpreted as a supporting diagnostic to quantify one of the mechanisms affecting net performance, rather than as an additional correction to the measured energy gains.

4.5. Practical Installation Aspects

The campaign also provided relevant information on installation and fixation methods. Four alternatives were tested for the reflective films: metallic staples, raffia bags filled with sand and PVC tubes filled with sand for anchoring, and trench anchoring (Figure 15). Each method had different implications for stability, soiling, maintenance, and ease of installation.

The methods based on staples, bags, and tubes presented limitations during field exposure. Metallic staples corroded, sand containers contributed to dirt spreading on the films, PVC tubes also became sources of local soiling, and insufficient fixation allowed film displacement during windy conditions (Figure 16). The trench-anchoring method, in which film edges were buried in the soil, provided better practical stability among the evaluated options. This result shows that utility-scale application depends not only on optical performance but also on robust civil and operational design.

5. Conclusions

This paper presented field results from a Brazilian pilot plant evaluating high-reflectivity ground covers in bifacial PV systems with single-axis tracking. The experimental evidence confirms that reflective covers can substantially increase broadband albedo, rear-to-front irradiation ratio, and measured energy yield relative to a natural reference surface with albedo representative of common ground conditions in Brazilian utility-scale PV plants.

The white and pearl-white films increased broadband albedo to 53–54%, raised the rear-to-front irradiation ratio from 7.6% to approximately 20–21%, and produced annual energy gains of 9.5% and 8.9%, respectively, compared with the gray gravel reference. The further optimized black-and-pearl film increased broadband albedo to 58%, rear-to-front irradiation ratio to 19.8%, and energy yield by 8.8%. The white geomembrane achieved the highest values, with 72% broadband albedo, a rear-to-front irradiation ratio of 23.1%, and an annual energy gain of 11.3%. These results confirm previous findings that artificial increases in ground albedo can enhance bifacial PV performance, while providing field evidence for engineered reflective covers under Brazilian operating conditions.

The results also show that the implementation potential of reflective covers cannot be assessed from albedo alone. The relationship between albedo and energy gain was not linear, indicating the influence of system geometry, rear-side irradiance distribution, thermal behavior, and electrical conversion limits. Inverter-related constraints were particularly relevant during high-irradiance periods, when part of the additional rear-side contribution was likely affected by clipping. In addition to conventional power clipping, current limitation was observed under specific cloud-edge overirradiance conditions, showing that reflective covers may amplify transient current peaks when the electrical design cannot accommodate the additional rear-side irradiance. Therefore, systems designed to operate with high-reflectivity ground covers may require different electrical design criteria and an optimized inverter loading ratio compared with systems installed over natural or lower-albedo surfaces. This aspect is especially important when retrofitting existing PV plants, where the original DC/AC ratio, string configuration, and inverter current limits were not necessarily selected to accommodate the additional rear-side current produced by reflective covers.

Operational aspects also affected the interpretation and long-term feasibility of the solution. Soiling losses were measurable during the first cycle, which was strongly influenced by construction activities near the pilot plant, but no systematic cleaning response was observed during the second cycle under more representative exposure conditions. This indicates that soiling should be treated as a context-dependent operational variable rather than extrapolated from a single value. Thermal losses were small compared with the measured gains and were already embedded in the net energy results. The optical stability of the covers, especially under long-term outdoor exposure, remains an important factor to be assessed through extended field monitoring or accelerated degradation testing. Practical aspects such as anchoring, wind stability, durability, drainage, maintainability, and compatibility with operation and maintenance routines are also decisive for utility-scale deployment.

Overall, high-reflectivity ground covers represent a promising strategy for enhancing the productivity of bifacial single-axis tracking systems. However, the results reported here are specific to the evaluated site, materials, tracker geometry, module technology, electrical configuration, and local operating conditions. The magnitude of the energy gain and the economic attractiveness of the solution are expected to vary with climate, solar resource, latitude, natural ground albedo, system geometry, electrical limits, soiling conditions, and maintenance practices. Therefore, application in utility-scale PV plants should be supported by integrated techno-economic analysis combining energy gains, cleaning and maintenance needs, anchoring, drainage, environmental compatibility, durability, and cost. The field results reported here provide an experimental basis for validating simulation models for bifacial PV systems operating under high rear-side irradiance conditions and for future techno-economic studies of reflective ground covers in Brazilian PV plants.