Performance and energy loss mechanism of bifacial photovoltaic modules at a solar carport system

As photovoltaic modules are increasingly used in renewable energy systems, ensuring energy efficiency by reducing the levelized cost of electricity has become the focus of current research. Herein, the 1‐year performances of both monofacial and bifacial photovoltaic modules were monitored, compared, and analyzed at a solar carport system in the Korean peninsula. The environmental parameters during the four seasons were investigated for both systems under ambient conditions. Irradiance was determined as the primary influencing parameter in the system. The energy yield of the bifacial module system was 3.08% higher than that of the monofacial system owing to rear‐side absorption during the study period. The loss mechanism for this lower yield was determined by investigating the irradiance effect. It was attributed to low bifaciality due to cell properties (optical and electrical properties, and the sorting effect) and module properties (shading effects from the junction box cable and frame design, and glass grid effect), design of the carport system, and the albedo effect. These analyses revealed that the photovoltaic energy yield can be increased further by reducing these loss parameters at the carport system. This study can contribute to achieving higher energy yield from photovoltaic systems during field applications.


INTRODUCTION
The photovoltaic (PV) industry has continuously focused on developing energy-efficient modules by reducing the levelized cost of electricity (LCOE) through various research and development objectives, such as improving product performance (cell efficiency, module power, and performance), reducing the cost of PV products (reduced consumption of wafers, metal paste, gas, and chemicals, and increased productivities), and increasing the energy yield. Presently, passivated emitter and rear cell (PERC) and tunnel oxide passivated contact (TOPCon) solar modules are the leading cell structures with the highest cell efficiencies and a market share of approximately 85% in 2021. The stabilized cell efficiency for PERC and TOPCon (n-type) cells in industrial applications is approximately 23% and 24%, respectively, in 2022. 1 Additionally, several studies have been conducted on the development and large-scale application of improved products, such as heterojunction back-contact and Perovskite-based tandem solar cells. 2 Larger wafer sizes such as M10 (182 × 182 mm 2 ) or M12 (210 × 210 mm 2 ) have already been used in the industry since 2021 and the proportion of M10 and M12 Cz-mono wafers in the market has reached over 50% in 2022. 1,3 The larger wafer size is not related to cell efficiency, but it can contribute to increased cell or module power. Moreover, in terms of product performance, the bifacial PV product is a suitable candidate. Despite their same sizes, the bifacial module generates more power than the monofacial module, which is attributed to the front and rear irradiance in the field. The market share of the bifacial cell is over 60% in 2022, which is estimated to increase to over 80% in 2032. 1,[4][5][6] The cost reduction methods applied to the solar cell or module are also necessary to achieve a lower LCOE. A thinner wafer is currently used in the cell and module; for example, p-type wafers such as 160 µm (M6), 165-168 µm (M10 and M12) have been used in the modules manufactured in 2022, and the thicknesses are expected to be 140-145 µm in 2032. 1 Thus, for the same ingot length, more wafer quantities can be produced, which can contribute to cost reduction. Metal pastes are the second most expensive materials used in the cell product. The amount of silver paste required for the front and rear sides of an M10 bifacial cell in 2022 is approximately 200 mg/cell and is estimated to decrease to approximately 125 mg/cell in 2032. 1 The energy yield is another key parameter that can directly affect the LCOE. Several factors can contribute to an increase in energy yield from PV cells or modules. These include cell design for improved low-light behavior and low power temperature coefficient, improvement in module power efficiency, such as through the use of a bifacial module structure that has excellent cell properties, and system design for increased power generation in the module.
To the best of our knowledge, the 1-year performances of monofacial and bifacial modules with PERC have been compared and analyzed at a carport system in the Korean peninsula for the first time in this study. Additionally, the loss mechanism associated with the energy yield from the PV bifacial module was analyzed based on cell, module, and system design, as well as the albedo effect. Finally, future avenues for improving PV energy efficiency are suggested. This study can contribute to achieving higher energy yield and lower LCOE in PV systems during industrial applications.  Figure 1. The design and operation process for the front side of the cells is the same, including wafer dimensions, diffusion process, antireflection coating (ARC) layer process, metal lining, and metal pastes used. However, the rear sides are different. In the monofacial cell, the rear side is entirely covered by metal pastes (silver and aluminum), whereas in the bifacial cell, a part of the rear side is exposed to facilitate light absorption. Thus, the metal lining at the rear side has the busbar-finger and local contact structure similar to the front side, which is adversely affected by a higher series resistance and leads to lower cell efficiency with a lower fill factor (FF) than that in the monofacial cell.
Moreover, the dielectric structure on the rear side is different ( Figure 2). Both structures have p-type wafers with Al 2 O 3 , SiN, SiO x N y , and SiN dielectric layers and the same optical constant values, but the thickness of each layer is different ( Table 1). The role of Al 2 O 3 is for electrical passivation on the surface of the p-type silicon wafer. The other dielectric layers have the role for optical performance improvement which is to enhance the reflection of the infrared (IR) wavelength between the p-type silicon wafer and those dielectric layers. In addition to these roles, the bifacial cell should minimize the reflection on the rear side for the scattered or reflected light from albedo. Thus the thickness of the bifacial cell is thinner than those of the monofacial cell.

| Structures of the monofacial and bifacial modules
As shown in Figure 3, the module structures are different for both cases. 5,7 The monofacial module has the traditional structure implying that an antireflection glass sheet with a thickness of 3.2 mm was used to protect the front side of the module from environmental factors, such as hail or snow. Two ethylene vinyl acetate (EVA) films were used for cell protection with a thickness of 400 µm each. A white-colored EVA was used as the rear film to improve the encapsulation effect and achieve a higher cell-to-module (CTM) conversion ratio. Moreover, the back sheet has a PE/PET/PE structure with a thickness of approximately 350 μm. As shown in Figure 3A, the irradiance from the rear side reflects at the white film allowing only the front irradiance to contribute to the module's power.
The bifacial module had a different bill of materials (BOMs) from the monofacial module. Similar to the monofacial module, it consisted of one glass on the front side; however, it had another glass on the rear side to function as the back sheet. The thickness of each glass sheet in the bifacial module is 2.0 mm, which is thinner than that of the monofacial module to avoid heavy weight and high cost. Two transparent and thicker EVA films (500 μm) were used to prevent cell breakage during the lamination process owing to the use of two glass sheets. The rear EVA film is also transparent in the bifacial module to maintain a lower CTM than that in the monofacial module, which uses a white rear EVA film.

| Carport system structures with the monofacial and bifacial modules
Fifteen of these modules were installed in a carport system in the central Korean peninsula. The carport system was at a height of 2.51-2.97 m, which can be considered the optimized height for energy yield because the tilt angle of the module was 5°to the north; thus, the height range of the studied carport system is in better condition than that in a previous study. 8 Three cars can be parked under this carport module system. The bifacial and monofacial modules were installed in the same region to ensure the same ambient conditions. The albedo is not always constant for this system due to the presence of asphalt (in the area with no parked cars) or differently colored cars in the parking area, as shown in Figure 4A. Even during winter, the cars could be covered with snow. The reflection spectrum is different for the various albedo, which can affect the power generated from the rear-side irradiance. 9 This condition (nonuniform rearside irradiance) can produce different current values, which can lead to current mismatch, low bifaciality, and then, finally lower energy yield in the system. According to other F I G U R E 2 Structures of the rear dielectric layers of the monofacial and bifacial cells.   research, the power loss caused by the mismatch of solar cells is around 41% at the power distribution. Thus, this current mismatch is one of the critical loss items. 10

| Product performance and weather information
For 1 year (four seasons: spring, summer, fall, and winter), the energy yield, irradiance, as well as the module, and ambient temperatures were monitored. This carport system is located under the climate conditions of the Korean peninsula. 11 Additionally, the electrical parameters for the two modules were compared under the different climate conditions during the study period. Linear regression analysis was conducted between the different measured parameters.

| Energy loss analysis for the carport system
The differences in the energy yield (kWh/kWp) from the two modules were analyzed and the possible reasons for this variation are further discussed based on the design of the cells and modules, electrical and optical properties of the cells and modules used in the studied carport system, light-blocking or shading effect, and the albedo effect.

| Performance of the monofacial and bifacial cell structures
The energy yields from the two cell structures were compared. The bifacial cell efficiency for the measurement at the front side of the cell at the standard test conditions (STCs) was lower by approximately 0.1% (absolute value) than that of the monofacial cell with 12 busbar structures. This is because of the lower FF due to higher series resistance from the rear metal design, as shown in Figure 1B. This difference in cell efficiency depends on the number of busbars in the cell, amount of paste consumed, metal design, and wafer size. For example, 12 busbar cells have a smaller difference in energy efficiency than six busbar cells. The mean values of the cell efficiencies from the STC were found to be F I G U R E 4 Carport system with two types of modules (monofacial and bifacial modules). Fifteen modules can make the space for (A) three cars, and the height is 2.51-2.97 m. (B) The module tilt is in the north direction. This system can be exposed to three sources of albedo: asphalt, car color, and snow (January, February, and December).
22.31% at the front and 17.14% at the rear side. Thus, the bifaciality (ratio of the cell efficiencies of the rear and front sides ×100) for each cell was 76.8%. At the cell level, this bifaciality depends on two factors: (1) ARC on the rear side and (2) amount of shading from the rear metal.
For the ARC at the rear side ( Figure 2), the optimized design for the thickness of the dielectric layer in the bifacial cell varied from that of the monofacial cell. This bifacial structure should meet two conditions: (a) it maximizes the reflection for the IR wavelength at the interface between the p-type silicon wafer and dielectric layers and (b) functions as the antireflection layer for irradiance (all wavelengths) from the rear side. The metallization design including the finger's width and quantity and busbar quantity cause shading, which can directly affect the bifaciality. This has been discussed further in Section 3.6.4. Thus, the short circuit current (I sc ) primarily decreases from 9.811 A (measured and mean values at the front side) to 7.592 A (measured and mean values at the rear side) and the I sc bifaciality was 77.4%, as listed in Table 2.

| Performance of the monofacial and bifacial module structures
Each of the 15 modules in the monofacial and bifacial structures has an average measured power of 422.4 and 414.9 W, respectively (Table 3). Owing to the different BOM in the bifacial structure (such as the transparent rear EVA), the CTM power conversion ratio of the bifacial module was lower than that of the monofacial module, which decreases the generated power by 1.78%. The average bifaciality (which is determined as the power ratio of the rear to the front power values) of the bifacial module was 69.2%, which was lower than the value (76.8%) at the cell level. This has been discussed further in Section 3.6.

| Measured data during the four seasons
The measured energy yield, temperature, and irradiance are presented in Figure 5. These data were the daily averaged values for the irradiance and ambient temperature. The highest irradiance was 273 W/m 2 in June, and the highest temperature was 29.7°C in August.
In contrast, the lowest irradiance and ambient temperature were 103.4 W/m 2 in December and 8.2°C in January, respectively. Thus, a wide variance in temperature and irradiance (21.5°C and 169.6 W/m 2 ) was observed between summer and winter. Moreover, a rainy period following a typhoon occurred during summer so that the irradiance began to decrease in July, but high ambient temperatures with humid conditions were maintained. During winter, snow occurred from December to January, which also leads to low irradiance with lower ambient temperature.

| Module and ambient temperatures during the four seasons
The position of the temperature sensor at each type was shown in Figure 6A. Each sensor is located at the cell position of the rear side in each module. Each data point indicates the monthly average value of the daily average T A B L E 2 Comparison of measured electrical data for the STC at the front and rear sides of the same bifacial cells. temperature. All data including the module's electrical parameters such as current, voltage, and power are collected from 5:00 to 21:00 for 1 day. The graph in Figure 6B shows the temperature difference between the monofacial and bifacial modules and the ambient temperature. Both module types exhibit similar temperatures. Thus, the additional rear glass did not affect the module temperature difference. During summer, the temperature difference was less than 1°C. Furthermore, during winter, a similar or slightly higher temperature (approximately 1°C) was observed. Thus, the temperature effect for both types of modules is not significant. The winter temperature is around 8°C due to the data collecting time (5:00-21:00). It means that the midnight data are not included as mentioned above. And the difference between the ambient temperature and those modules is less than 5°C because the temperature is the average value for the collecting time and the direction of the carport is north, it means the irradiance amount to the system is less than the south direction case.

| Comparison of the electrical parameters
The monthly trend summary and linear regression analysis between the normalized current and monthly average irradiance for each month were shown in Figure 7A,B. The normalized current implies the ratio of the measured current in the system to the measured and mean I sc values of 15 modules at STC. The higher irradiance leads to the higher normalized current. In June, the peak normalized current values were the highest, at 28.4% on the monofacial module and 29.9% on the bifacial module at the highest irradiance (273.04 W/ m 2 ). The rainy and typhoon season started in July, and under these conditions, the irradiation and normalized current started to decrease. Both the bifacial and monofacial modules (yellow and blue lines in Figure 7B) show linear regression analysis between the normalized current and average irradiance. The R 2 values are extremely high (0.967 and 0.966 for the bifacial and monofacial modules). The slope of the bifacial module (0.105) was larger than that of the monofacial module (0.101) due to rear-side absorption. The major parameter is the irradiance for the normalized current. The ratio of the normalized current of the bifacial module to the monofacial module ranged from 102.0% to 105.4% from February to November. However, higher ratios of 108.1% and 107.4% were obtained in January and December, respectively, during winter because of the snow season. This snow can decrease the absorption from the front side and increase the albedo effect, which leads to increased absorption from the rear side.
Generally, the voltage in the module depends on the module temperature. As the temperature of the module or solar cell increases, the module voltage decreases due to the silicon bandgap narrowing effect (−2.2 meV/K). 12 The measured temperature coefficient for the open circuit voltage (V oc ) on these two module types was −0.27 (%/K). Although the rear structure of the cell and the BOM of the module are different, the temperature coefficient for the V oc of the cell and module was the same. 12 As shown in Figure 6B, the temperature difference between June (summer) and January (winter) of the bifacial module is 23.9°C which decreased the V oc by 6.4% according to a simple calculation. The temperature difference (23.6°C) was also similar to that of the monofacial module. On the basis of the temperature coefficient of V oc and this temperature difference, the normalized voltage was expected to be lower in June due to the highest temperature. However, the carport F I G U R E 5 Comparison of the daily averaged irradiance and ambient temperature during the four seasons in the study period. system investigated herein shows the field data, which are simultaneously affected by temperature and irradiance. The normalized voltage indicates the ratio of the measured voltage in the system to the summation of the V oc values for 15 modules at STC, which is similar to the normalized current and energy yield calculations. The monthly trend between the normalized module voltage and average temperature is shown in Figure 8A. The normalized voltage increased during the first half of the year, although the temperature increased. This is contrary to the general understanding of the temperature coefficient of V oc . 12 This trend may be attributed to the increased dependence on irradiance which can increase the system current and consequently increase the voltage in the field. The results presented in Figure 8B support this conclusion. The temperatures for the bifacial module are similar in June (32.5°C) and August (32.2°C). During the rainy and typhoon season, the irradiance begins to decrease in July, which leads to different irradiance levels in June (273 W/m 2 ) and August (181 W/m 2 ). Under this ambient condition (same temperature and different irradiance), Figure 8B exhibits two trend lines: one is for the first half of the year, including July and the other is for the second half (except July). Each trend had a similar slope but different y-intercept values, which was due to the different amounts of irradiance. The slopes had positive values for both modules. This implies that the normalized voltage increased despite the higher temperature. This set of field data also represents that irradiance is the dominant parameter.
The monthly energy yield for the irradiance in each month was shown in Figure 9. This characteristic is highly similar to the results presented in Figure 7A. Irradiance is the major parameter that affects the energy yield and the current. High R 2 values (0.962 and 0.958 at the bifacial and monofacial modules, respectively) also support the linear regression analysis results for both modules. During the entire year, the bifacial module exhibited higher energy yield, which is due to the higher current (as shown in Figure 7). Under this type of climate conditions, irradiance is the major factor causing the high energy yield. From February to November, the yield ratio of the bifacial module to the monofacial module increased from 100.7% to 104.6%. During winter, the ratio was higher at 106.6% (January) and 108.5% (December), as shown in Figure 9B. The normalized current ratios also exhibited similar trends, at 108.1% (January) and 107.4% (December) in Figure 7B. During winter, especially in January and December, considerable snowing was observed on the carport system. Because the albedo of snow is high, it can contribute to increased rear-side absorption and blockage of front-side absorption. Thus, the bifacial module has high current and energy yield ratios.
The energy yields for the monofacial and bifacial modules were 1093.1 and 1126.8 kWh/kWp, respectively, during the 1-year study period. As reported in several studies, the bifacial module exhibits a higher energy yield (increase of 33.7 kWh/kWp), which implies a gain of 3.08%. 8,13,14 2.6 | Loss analysis for energy yield on the carport system The energy yield of the bifacial module at the carport system in this study exhibits a lower gain (3.08%) than that reported in other studies. 8,13,14 This lower value is due to the following reasons: (1) Lower bifaciality in the cell or module due to (A) nonuniformity in the color of the rear side at the bifacial cell (optical appearance), (B) nonuniformity of each cell's electrical performance by electrical luminescence (EL) images, (C) rear metal design and process capability of the cell, (D) module frame design and metrology, (E) encapsulation effect caused by the grid design at the rear glass. (2) Light blocking or shading caused by objects in the carport system due to

| Lower bifaciality value
This carport system consisted of 15 bifacial modules whose bifaciality distribution is shown in Figure 10A. The maximum, average, and minimum bifaciality values were 70.3%, 69.2%, and 68.1%, respectively. According to the International Technology Roadmap for Photovoltaic (ITRPV), the bifaciality in the PERC module used in the PV industry is approximately 70% in 2022. Thus, this average value (69.2%) is slightly lower than the average value in the PV industry, 1 with a delta value (difference between the highest and lowest values) of 2.2%. Considering the PV industry value (70%), some modules have module power losses; for example, for the lowest bifaciality case (68.1%), this module has a power loss of 7.6 W at the rear-side STC. This loss mechanism has been reported in previous studies. 15,16 This module power loss occurred due to the wide I sc distribution, which caused the current mismatch. In this study, all parameters at the cell, module, and system levels were studied to determine the cause of the current mismatch in the bifacial product. Moreover, the lower bifaciality in terms of the optical and electrical properties is discussed qualitatively and quantitatively in Sections 3.6.2 and 3.6.3, respectively.
2.6.2 | Nonuniformity in the color of the rear side at the bifacial cell (optical appearance) The images of the rear side of the bifacial module are shown in Figure 11 for the cell appearance itself and the cell appearance of the module. Thus, owing to the rear glass and rear EVA film, the color of the image of the cells in the module was darker. In both these images, some cells had a different color, such as dark or light blue. These colors are due to the variation in the thickness of the dielectric layers. As shown in Figure 2, the total thickness of the dielectric layers ((a) + (b) + (c) + (d)) was 90 nm, which was selected as the target thickness. Some cells have lower thicknesses, such as 80 or 50 nm, which causes the difference in the optical appearance. When the module power (for both the monofacial and bifacial modules) was measured at the front side, this color difference at the rear side was not significant because this optical variation does not affect the reflection of the IR wavelength at the interface between the silicon wafer and the rear dielectric structure. This finding implies that this type of optical variation has no effect on the module power performance. Thus, this type of color difference can be neglected when measuring the front module power of monofacial or bifacial modules in the PV industry. However, when measuring the module power at the rear side or in the field, this color difference is crucial. The rear side in the bifacial module is the absorbing side for the scattered or reflected light from the albedo. Therefore, all wavelengths can be affected by this color variation in the ARC layer, which leads to high variations in the I sc and current mismatch among the cells in the module, consequently exhibiting lower bifaciality. This explanation is supported by a similar study, which demonstrates that wavelengths below 1000 nm can be affected by this type of color difference in the reflection spectrum in other research. 17 In the PV industry, the cell efficiencies are not measured for all the cells at the rear side or they are solely measured as the sample base. This can cause the aforementioned color variation in the rear side and decrease bifaciality in the module due to the current mismatch that can cause a decrease in the module power.

| Nonuniformity of each cell's electrical performance by EL images
The electrical properties of the cells are another factor contributing to the reduced bifaciality, as shown in Figure 11C. The presented EL image is that of a low bifacial module. This image reveals relatively dark regions in the two cells (red box), which cause FF loss in measured power at the rear module. 18 This type of dark EL areas indicates electrical defects and can lead to lower electrical performance. The dark EL areas evidently caused increased power loss at the rear side of the module. Thus, to verify the occurrence of this phenomenon, some modules were intentionally designed in this study to account for the dark EL effect on the bifacial module and its bifaciality value. There are seven groups of cell efficiencies, as shown in Figure 12A. Each cell group has 150 cells and the average cell efficiency for this quantity was determined. A higher cell efficiency (22.46%) is associated with a higher bifaciality (74.4%), whereas a lower cell efficiency (21.21%) is related to a lower bifaciality (only 69.4%). In this analysis, two types of modules were designed. One module (Case A), used as the reference module, only consisted of the cell group (22.46%) composed of six strings. The other module (Case B), used as the test module, consisted of the main cell group (22.46%) composed of five strings and a lower cell group (21.21%) composed of only one string (green box) as shown in Figure 12B.
At the front side of these two modules, the measured power values were 424.1 and 421.4 W for Cases A and B, respectively. The module power delta was only 2.7 W although the lower-efficiency cells were present in only one string (green) of the test module, as shown in Table 4. However, during power measurement at the rear side, these dark EL cells considerably affected the module power and especially the FF value, as shown in F I G U R E 12 (A) Summary of the cell efficiency and bifaciality, (B) comparison of module electrical luminescence (EL) image and bifaciality for cells with (Case B) and without (Case A) dark EL areas, (C) comparison of the I-V curves for cells with (Case B) and without (Case A) dark EL areas with module power measured at the rear side, and (D) power loss simulation for the current mismatch with the two-diode model. Table 4. The decrease in module power and FF for Case B were 33.9 W and 9.9%, respectively. Figure 12C shows a comparison of the current-voltage (I-V) curve for both cases. The I sc and V oc did not vary considerably but a difference was observed at FF, which is generally due to the current mismatch. 10,[19][20][21] Notably, I sc or V oc was not affected by current mismatch, but the FF (especially I mpp ) decreased considerably, as indicated by the I-V curve in Figure 12C. The simulation result also supported this experiment result in Figure 12D. This simulation is based on the interconnection of the individual cells with characteristic curves. The characteristics are generated with the two-diode model. 22,23 Six cases were simulated from the current matched case to 20% difference for short circuit amount at one string. 17.5% difference (green color line) can cause low bifaciality with 62.9% which is matched with experiment data (Case B) in Figure 12B,C.
Thus, the modules with lower bifaciality, as shown in Figure 10A, exhibited nonhomogeneity in the rear cell colors ( Figure 10A,B) or in the dark EL cells ( Figure 11C). The delta value (2.2%) of the bifaciality in Figure 10A can be explained by the findings of a previous study 16 and the results of the EL analysis. The best case module (bifaciality of 70.3%) in Figure 10A consists of these cells (Case A in Figure 12). Some PV module manufacturers use the cell mixing method to control cell efficiency distribution and use the rear side as the sampling base to allow for some lower-efficiency cells in a module that can have lower bifaciality. Although this type of cell mixing does not lead to any power loss in the monofacial modules, it can adversely affect the bifaciality of the bifacial module, as observed in the analysis results in Figure 12 and Table 4. 2.6.4 | Rear metal design and process capability of the cell The rear metal design and print quality of the bifacial cells are other crucial parameters that can cause lower bifaciality, as shown in Figure 13. The design of the bifacial product has main two objectives. One is to improve the energy efficiency for both the front and rear sides. The other objective is to improve the soldering of the product with the wires. Accordingly, various rearside designs were analyzed, as shown in Figure 13A.
To improve the extent of soldering between the rear silver paste and the wires in the module, the rear silver paste area should be larger. However, this not only causes high silver consumption (high cost) but also increases the metal shading area for the rear-side irradiance. Three types of designs for the rear side were selected ( Figure 13A), each of which shows the benefits of this tradeoff between soldering, cell efficiencies, and the cost of silver. Furthermore, the printing quality affects these parameters. The increase in the width of the rear finger also causes shading, and finally leads to lower bifaciality ( Figure 13B). The linear regression analysis between the shading at the rear side and bifaciality shows a high R 2 value of 0.93 ( Figure 13C). A 1% increase in shading can cause a 1% decrease in bifaciality. As based on this trend line and the shading effect, the lower shading rate (which implies less rear paste consumption and silver paste area) results in higher bifaciality. 24,25 2.6.5 | Module frame design and metrology A major BOM of the PV module is the frame, which protects the module panels from wind and snow. One module consists of four frames (two long and two short frames). The frame design is another parameter influencing the bifaciality, as shown in Figure 14A. By reducing the width of the frame from types A to B at the short frame, the bifaciality can be increased from 68.9% to 70.8% (by 1.9%). This is attributed to the reduced shading effect achieved by decreasing the aluminum frame width. 26,27 PV module manufacturers have attempted to modify the frame design to reduce the manufacturing cost, increase the CTM ratio, improve module efficiency, and contribute to mechanical strength. 26 The rear-side design of the frame must be optimized for this bifaciality range.
During module power measurement, the junction cable is another factor that reduces the bifaciality due to the shading effect. The cross section of the cable was T A B L E 4 Comparison of the electrical data for the module with (Case B) and without (Case A) dark electrical luminescence (EL) areas in the cells. 4 mm 2 , which can increase the shading effect, as shown in Figure 14B. The red I-V curve shows the shading effect from the shading area due to the cables (yellow part) in the box. In contrast, the blue I-V curve has no cable shading and the delta of the bifaciality was 4.8%, which is relatively lower value compared with other research (7.5%) due to the different product. 10 This shading effect can occur not only during metrology but also during module installment in the system, as shown in Figure 15 (yellow circles).
2.6.6 | Encapsulation effect caused by grid design at the rear glass In this system, the front and rear glasses are transparent, as shown in Figure 14C. If the rear glass consists of the white grid as in Figure 14D, the power can increase by approximately 3% through the encapsulation effect at both the rear and front sides. The results presented in Table 3 also confirm this gain in power. Thus, this can contribute to the energy yield in the bifacial module system (not the bifaciality value).

| Light blocking or shading caused by objects in the carport system
As shown in Figure 15, some objects can create a shading effect on the bifacial module system and affect the bifaciality. This carport system has several bars (red circles) of a specific type, which contribute to major losses in the bifaciality of this system. Considering the cable shading effect (−4.8%) or the frame design effect (−1.9%), this type of system bars (red circles) increases the shading and finally results in significant current mismatch and lower bifaciality in the module system. Other PV systems, such as the utility or vertical systems, do not consist of this type of shading objects in the field, but this type of lightblocking objects must be accounted for in the carport system, to reduce the shading effect. 7,27-29 In the carport system considered in this study, the temperature sensor and cable are located at two positions each month (green circles). An unknown object (blue circle) was also observed to obstruct the irradiance. These types of light-blocking objects can also cause shading loss and current mismatch, lower bifaciality, F I G U R E 13 (A) Comparison of three metal designs among the rear side of the bifacial cells, (B) comparison of the metal printing quality between the normal and bad prints, and (C) linear regression analysis between the metal shading rate and bifaciality. and consequently lower energy yield. Other research also reported this kind of loss mechanism on the flat rooftop of an industrial building, which is also a good comparison with this research analysis on the carport. 30 2.8 | Nonuniformity in the irradiance from the different albedo conditions Unlike the utility system, the carport system is affected by different albedo conditions, as shown in   Figure 4A. This figure shows the 1-day parking status, as revealed by the red-andblack, black, and white cars under the bifacial modules. Each car has different reflection properties, which affect the bifaciality at each module and total energy yield. In the case of the carport system, reflection of incident light also occurs from the asphalt, as shown in Figure 4B. Moreover, the presence of snow in January and during winter can also cause the albedo effect, which further alters the reflection. This is also shown in Figure 9B. The energy yield ratios for the monofacial and bifacial modules are 106.6% in January and 108.5% in December, which are substantially higher values than those obtained during other seasons (100.7%-104.6%). During these months, snowing occurs, which can increase shading on the front side and reflection from the snowcovered surface. Similar results were also reported about the bifacial gain from the effect of albedo. 31 Thus, the energy yield of the bifacial module may be higher during winter. In the future, these data for the winter season will be continuously monitored and analyzed to verify the accuracy of this study's findings.

| CONCLUSIONS
In this study, the performance of the bifacial module was compared and analyzed with the monofacial module for 1 year at a carport system in the Korean peninsula for the first time. The region is characterized by four seasons (spring, summer, fall, and winter), with extremely high temperatures and humid conditions during summer and cold weather and snow during winter. The analyzed monofacial and bifacial module types exhibited similar temperatures, and thus, similar product voltage. During each season, the major parameter affecting the current and energy yield is the irradiance, which shows strong linear correlations. When developing new PV products, many researchers and engineers have attempted to increase the V oc because it has a high CTM power conversion ratio, rather than the J sc and FF values. Moreover, considering irradiance during the four seasons under the same conditions, such as the Sunbelt region, this type of voltage temperature coefficient is a crucial parameter. Thus, many activities have been focused on improving the V oc in these modules. However, under these four-season weather conditions and this type of carport system design, the PV product with high I sc value can contribute to increased energy yield as discussed in F I G U R E 15 Low bifaciality owing to the shading caused by (1) mounting system (red circles), (2) junction cable (yellow circles), (3) temperature sensor and cable length (green circles), and (4) unknown object (blue circle). this paper. The gain of the bifacial module system was +3.08% (33.7 kWh/Wp/year) compared with that of the monofacial module system investigated in this study. This value is considerably lower than that reported in other studies. Therefore, the energy loss mechanism was analyzed in terms of the optical and electrical properties of the cell, rear metal design of the cell, module frame design, rear glass effect with white grid, light-blocking objects, and nonuniformity of the irradiance from the different albedo conditions. These analysis results revealed that increasing the I sc value and reducing the current mismatch can help increase the energy yield to a value higher than that obtained in the present study (+3.08%). This can contribute to the design of products with higher energy yield and lower LCOE in the PV industry. Furthermore, the current mismatch described in this study should also be considered when designing the next industrial product, such as the tandem solar cell. For future work, the system performance with high bifaciality cells and modules will be investigated and the loss mechanism will be quantitatively analyzed for precise breakdown for the PV industry. And more modules will be installed for statistics analysis.