Electrical and Financial Impacts of Inverter Clipping on Oversized Bifacial Photovoltaic Systems

Abstract

This paper studies the impacts of inverter clipping on bifacial PV modules under different weather and ground reflectivity. A 5 kW bifacial array was connected to a 3.8 kW grid-tied inverter, a 10 kWh Li-ion battery, and an EV charger. A PV output calculation model was developed to compare the estimated output of the modules with the actual measurements to evaluate the relation between ground reflectivity and clipping loss. The results showed that clipping potentially occurs on sunny days in summer from 10:00 to 15:00 during the period with the highest solar irradiance. Three colors of ground cover were also examined to compare the performance of bifacial modules under different albedo reflective properties. The results indicated that the white ground in winter leads to the highest bifacial gain (13.1%) and daily DC efficiency (22.2%) due to the combination of high reflectivity with low solar angle giving maximum upward reflection of direct sunlight. This same combination shows a minimal advantage in summer due to the clipping. The proposed model is evaluated, demonstrating 98.2% agreement between modeled and actual data for all conditions. Furthermore, simulation models based on the actual system with different system sizes and ground reflectivities have been studied to evaluate the impacts of the clipping in terms of technical losses and financial returns. The analysis shows that a high reflective ground condition can provide the best financial benefit, and the clipping loss does not have a great effect on the finance of the project since the loss is less than 4% of the annual production even in an extreme case.

1. Introduction

Due to an advancement in silicon solar cell fabrication technology, the manufacturing cost of photovoltaics (PVs) has decreased significantly over decades, causing PVs to be the fastest-growing renewable energy resource. The majority of modules are installed in ground-mounted utility-scale arrays (10 MW). A variety of technological approaches have been introduced to boost the financial viability of these large systems [1]. Bifacial PV modules are seen as one of the most attractive technologies owing to their ability to generate power from both front and rear surfaces, leading to a total energy boost compared to conventional monofacial Si-based PV modules [2,3,4]. The additional power from the rear side of a bifacial module over a monofacial one leads to a parameter called the bifacial gain, which has been investigated in a number of studies [4,5,6]. As the back side of the module is typically facing the ground, whether fixed- or single-axis tracking, the ground reflectivity (known as albedo) plays a crucial role in determining the bifacial gain [7,8]. For the same irradiance conditions, a bifacial module provides more daily energy and hence has a lower levelized cost of energy (LCOE) than the monofacial PV type when the albedo is sufficient, promoting higher economic payback for the bifacial module [9,10]. However, the capacity to reflect sunlight onto various surfaces, including vegetation, fabrics, and snow, varies significantly, and the choice of the ideal ground condition could result in a bifacial gain exceeding 15% [2].

Declining module costs have led to the common practice of over-sizing the PV DC capacity relative to the inverter AC rating, called the DC/AC ratio or sizing factor, in order to obtain the highest revenues and reduce their payback period [11,12,13]. This potentially leads to an inverter “clipping”, where the DC power going to the inverter is reduced due to the limitation of inverter AC output capacity. As PV modules require specific standard test conditions (STCs) of temperature and irradiance to reach the rated maximum power [14,15] and those conditions occur just a few times a year in most locations, the module output is nearly always less than the rated power. It is not necessary to match the inverter rating with the maximum power of the array. Additionally, inverters perform better while working near their maximum output power and this can help to improve the annual energy production [16,17]. The oversized PV system with a typical DC/AC ratio <1.3 can result in inverter clipping losses in certain weather conditions due to excessive energy production and consequently conceal the benefits of bifacial gain in bifacial PV systems [18,19,20]. In typical monofacial PV systems, the clipping loss associated with DC/AC ratio of 1.3 can be up to ∼3% of the total PV production [21]. Since bifacial PV systems generate more DC power over the day due to the benefit of the rear side, clipping loss is expected to be greater than 3%, and will be much higher with high reflective grounds.

For a low or moderate DC/AC ratio with monofacial modules, the boost in annual energy production exceeds clipping losses, which leads to a net decrease in LCOE [22]. Still, oversizing bifacial systems leads to more pronounced clipping losses due to the additional power generation from the rear side also being clipped out, thus negating part of their benefit [18]. The impact of inverter clipping is seasonally dependent [23,24] where the clipping loss is usually greatest around noon in summer when the solar irradiance reaches the highest intensity. This challenge has been addressed in a number of previous studies [20,25,26,27]. The inverter clipping not only results in energy loss but also affects the LCOE and inverter lifetime due to accumulative stress from operating most of the time at the saturated active power [19]. Most studies of clipping loss have been conducted via software simulations such as System Advisor Module (SAM) developed by the National Renewable Energy Laboratory (NREL), MATLAB Simulink, and PVSyst, where the solar irradiance, wind speed, module temperature, and DC/AC ratios can be modified to observe different behavior of the clipping and to optimize system size with minimal losses [20,28,29,30]. The optimal DC/AC ratio usually lies between 1.2 and 1.3 for the minimum losses [31,32,33].

While different studies addressed individual challenges with a combination of modified factors impacting the performance of bifacial PV, most of these studies are only based on computational modeling. Only a few studies were tested with the actual systems where the ambient temperature and shading effect were more realistic [34]. An evaluation under practical scenarios with the actual hardware is vital to help understand the limitations of the bifacial PV system implementation, especially in residential-scale systems. In this work, we study the inverter clipping behavior in a system that includes a 5 kW bifacial solar array tied to a 3.8 kW smart inverter and a 10 kWh DC-coupled battery to represent a residential scale PV+battery with generation and consumption data that have been collected for over two years. This system with a DC/AC ratio at 1.3 is comparable to a typical DC/AC ratio of both residential and commercial PV systems [35]; thus, the results in this paper can apply to most oversized PV systems with a DC/AC ratio in that range.

The rest of this paper is organized as follows: Section 2 presents an elaborate description of the information related to the experimental setup. Section 3 describes the methodology we followed for collecting data. Section 4 discusses our key findings along with the analysis of the clipping behavior. Finally, our concluding remarks are presented in Section 5.

2. Experimental Setup

2.1. System Configuration

The experiments were performed on a 5 kW south-facing bifacial PV testbed at the Institute of Energy Conversion (IEC) at the University of Delaware (39.7° N latitude). The PV system comprises twelve 405 W bifacial modules with a fixed 20° tilt angle accounting for the total array area of 24 m². Module-Level Power Electronic (MLPE) devices attached to each module provide the optimum power level and collect DC power output at the module level. A 3.8 kW grid-tied inverter is connected to the modules through the MLPE devices, providing a DC-to-AC ratio of 1.31. The oversizing of DC production is an ordinary practice in PV systems where the typical DC to AC ratio lies around 1.2 or higher, as discussed above. Note that the module rating of 405 W and system rating of 5 kW are for monofacial operation, as is common practice, and do not include any additional benefit from the rear irradiance since this is how bifacial modules are rated. The bifacial module industry has not yet established power rating protocols to account for rear illumination and bifacial gain due in part to the dependence on variables we studied here. In addition to the array and the inverter, the system is DC-coupled with a 10 kWh Li-ion battery; the system also includes a backup transfer panel and an electric vehicle (EV) charging station. Figure 1 shows the system configuration of the IEC bifacial system. We accessed data from the electronic components and sensors through the inverter’s user portal for our data analysis.

In addition to the MPLE devices that report the DC voltage, current, and power of each individual module, five irradiance sensors and three temperature sensors are installed on the plane-of-array and on representative modules to gather measurements from both the front and the rear sides. Figure 2 shows the locations of all sensors. All sensor data are recorded and stored in the inverter’s portal.

The irradiance sensors are Si PV cells with 0–10 V output and temperature compensation (model IMT Si-V-10TC, IMT Solar, New York, NY, USA). The temperature sensors are Pt1000-style resistive temperature detectors (RTDs) with 4–20 mA output (model IMT Tm-I-4090, IMT Solar, New York, NY, USA). All sensors are connected to the communication gateways (SE1000-CCG-G, SolarEdge Technologies, Milpitas, CA, USA), which interface with the split-phase inverter (SE3800H-US, SolarEdge Technologies, Milpitas, CA, USA). The module-level power optimizers are model Solar Edge P505. The irradiance sensor readings were validated by determining their offset and linearity using filters and known calibration standards but, we found no need to adjust their calibration. The module and ambient temperature sensors were evaluated by comparing their temperatures at night when they should all read the same to ambient thermal sensors and weather data. They were found to have small offsets (±2 °C) which were corrected using the gateway sensor calibration correction.

2.2. Ground Reflectivity

To study the influence of reflectivity and albedo in different ground conditions, tarps with different colors were used to cover the ground. The colors of tarps used in this work are white and black, to represent a wide range of ground reflectivity, while the normal ground condition is gravel. The reflections of white and black tarps are measured as approximately 70% and 5% in the wavelength range of 400–1200 nm. Our setup allows the ground height to be adjusted by mounting the tarps on a platform, yet all findings in this paper were obtained from the same ground height setting in order to focus only on the impact of different albedo.

3. Experimental and Analytical Methods

3.1. Data Acquisition

The relevant data, including the daily energy, module temperature, front and back insolation, and AC inverter production, were collected starting in July 2021 when the bifacial system was initially installed. Data were exported in an Excel format with a 15-min resolution. Several tarp conditions were explored in order to compare the benefit of bifaciality along with different weather conditions. We generalized all the data into two seasons, namely summer (from May to October) and winter (from November to April), for simplification. Similarly, we separated the sky conditions into two categories, sunny and cloudy, determined by the solar insolation at the front surface. While sunny cloudless days are days with a front insolation between 3.5 and 7 kWh/m², cloudy days are represented by days with a front insolation lower than 3.5 kWh/m². These insolation values were obtained by integrating the irradiance (kW/m²) from the two plane-of-array (POA) front sensors over the entire day.

3.2. Expected PV Output Power Calculation

The DC power generation was calculated with and without inverter clipping to compare the actual power output. The front and rear irradiance data from the sensors attached to the modules were used in the calculation of the expected power output. The formula for the ideal DC power at the Standard Test Condition (STC) for the bifacial array without losses is shown in (1). This formula agrees well with the model developed by Janssen et al. [36].

$$\mathrm{Daily}{\mathrm{P}}_{\mathrm{calc}}\left(\mathrm{kW}\right)=(0.192\times \mathrm{FI}(\mathrm{kW}/{\mathrm{m}}^2)+0.137\times \mathrm{RI}(\mathrm{kW}/{\mathrm{m}}^2\left)\right)\times 24{(\mathrm{m}}^2)$$

(1)

where Daily P_calc is the calculated bifacial power output from the array without temperature correction; FI represents the average irradiance on the front side of the array from the two POA irradiance sensors; RI represents the average irradiance on the rear side of the array from the three rear irradiance sensors; 0.192 and 0.137 represent the front side efficiency and rear side efficiency of the modules obtained from the temperature corrected flash test measurements at STC; and 24 is the total area of the array surface.

The rear irradiance is due to diffused light from the sky and reflected light from the ground, as explained above. To analyze the effect of the bifacial component, the monofacial power was predicted using (1), yet without consideration of power generation from the rear side. The monofacial power output calculation formula is presented in (2).

$$\mathrm{Monofacial}{\mathrm{P}}_{\mathrm{calc}}\left(\mathrm{kW}\right)=0.192\times \mathrm{FI}(\mathrm{kW}/{\mathrm{m}}^2)\times 24{(\mathrm{m}}^2)$$

(2)

By eliminating the rear side term of (1), the calculated monofacial power output Monofacial P_calc at STC can be obtained, and the bifacial gain is calculated from the difference between bifacial power output and monofacial power output. Since the power calculations in (1) and (2) are performed at STCs, the temperature correction needs to be performed to correct the power production of each module at the actual temperature, as shown in (3).

$${\mathrm{P}}_{\mathrm{corrected}}\left(\mathrm{kW}\right)={\mathrm{P}}_{\mathrm{calc}}\left(\mathrm{kW}\right)\times [1-\mathrm{TC}(\%/°\mathrm{C})\times \left(25\right(°\mathrm{C})-{\mathrm{T}}_{\mathrm{avg}}(°\mathrm{C}))]$$

(3)

where P_corrected represents the temperature corrected power; P_calc is the power output calculated from (1); and (2) TC stands for the temperature coefficient for the modules, which the data sheet says is −0.36%/°C; T_avg represents the average module temperature from the sensors, and 25 (°C) is the cell temperature at STC.

In consideration of the inverter clipping effect, 4000 W_DC was statistically observed to be the peak DC power into the inverter and was converted to 3800 W_AC, which is the inverter-rated AC capacity. This threshold of 4000 W_DC was also defined by the inverter operating efficiency at 95%. A higher clipping point on the DC side is likely from the inverter conversion efficiency and wiring losses between the array and inverter. Therefore, if the calculated power exceeds 4000 W_DC, the expected output will be limited to 4000 W_DC due to the clipping effect. The calculation flowchart is demonstrated in Figure 3.

The clipping power was obtained from the difference between the Bifacial P_corrected and the measured DC power output from the bifacial array P_DC, as presented in (4).

$${\mathrm{P}}_{\mathrm{clipped}}\left(\mathrm{kW}\right)=\mathrm{Bifacial}{\mathrm{P}}_{\mathrm{corrected}}\left(\mathrm{kW}\right)-{\mathrm{P}}_{\mathrm{DC}}\left(\mathrm{kW}\right)$$

(4)

3.3. Energy Efficiency

In addition to the expected power output, an assessment metric was also developed to evaluate the effect of inverter clipping on the benefit of the bifacial array using daily DC energy efficiency. To calculate the DC energy efficiency, the formula in (5) was used.

$$\mathrm{Daily}\mathrm{DC}\mathrm{Energy}\mathrm{Efficiency}(\%)={\displaystyle \frac{\mathrm{Daily}{\mathrm{E}}_{\mathrm{DC}}\left(\mathrm{kWh}\right)}{\mathrm{Front}\mathrm{Insolation}(\mathrm{kWh}/{\mathrm{m}}^2)\times 24({\mathrm{m}}^2)}}$$

(5)

where Daily E_DC (kWh) is the measured total DC energy production from the array in a day and 24 is the total array surface (m²). It is important to note that in (5), the Daily E_DC is normalized by only the array’s total front insolation. In reality, the DC energy is generated from both the front and rear sides, so the DC energy efficiency should increase when the rear insolation increases because the additional energy from the rear contributes to the array output but is not included in the denominator. As a result, it includes the bifacial gain. All calculation methodologies presented in this paper are developed for real-time calculation from the measured field data. Applying the concept of Occam’s razor, this model is relatively straightforward but adequately sufficient to be used for the implementation of real-time energy dispatch controls in other system components such as battery and EV. The measured data used in all equations were collected from the sensors that have been properly calibrated to reduce uncertainty. As natural variations of sunlight, weather conditions, and temperature over the day are much larger than the variation from measurements, the uncertainty in the calculation method and data acquisition will not be discussed in this paper.

4. Results and Discussion

The inverter clipping behavior was investigated on sunny days in summer and winter. The overall finding is that the inverter clipping loss can be estimated by the front solar insolation. As the capacity of the inverter is restricted to 3.8 kW_AC, we can estimate that the inverter clipping loss would present in power production higher than 4 kW_DC, which is equivalent to 3.8 kW_AC after the DC-to-AC conversion loss. The average measured front irradiance values on cloudless sunny days are 973.2 W/m² and 773.6 W/m² in summer and winter, respectively. Based on the field observations, the front solar insolation can exceed 5.5 kWh/m² on cloudless sunny days and potentially deliver over 4 kW_DC during the peak sunlight period. This leads to severe clipping loss, resulting in a degradation of the system’s efficiency.

4.1. Analysis of the Bifacial Modules Performance

Two cloudless days with high solar insolation (4.5–6 kWh/m²) were selected here to highlight the impact of the clipping loss, which are August 13 (summer) and November 22 (winter). Their daily DC output power in the two seasons with white ground is presented in Figure 4. Overall, there is a significant seasonal impact on the clipping loss of bifacial modules. In summer, the expected DC output power obtained from (3) shows that the maximum hourly DC output could reach 5 kW at noon, yet the actual power output was limited to around 4 kW due to the inverter rating. The data indicated that the inverter clipping potentially occurs from 10:00 to 15:00 in summer, which accounts for 8.5% of daily energy loss on the selected day shown in Figure 4. In contrast, the calculated DC output in winter aligned almost perfectly with the actual output without inverter clipping since the power production did not exceed the inverter capacity. This occurs largely because the relatively shallow angle of the array does not harvest much sunlight in the winter when the solar angle is low. In comparison to the output power from a monofacial PV calculated using (2), the advantage of bifaciality is hidden due to the clipping loss despite the expected bifacial gain being over 15%. However, the bifacial gain on a cloudless day in winter can be as high as 14% since the peak production is typically below the inverter rating point, which avoids clipping. Note that the data presented in Figure 4 were collected with the white ground to represent the best scenario for obtaining the highest bifacial power output. The relatively steep decrease in DC power in the late afternoon is due to tree shading. Even though the shading slightly affected the front irradiance, it did not affect the rear irradiance from the diffused light, so the benefit of bifacial gain will not be impacted by the shading loss. Additionally, the MLPE provides module-level maximum power point tracking (MPPT) that helps reduce the mismatch loss due to shading. This adds an element of realism to the study as tree shading can also be experienced in residential-scale PV systems with extreme sun angles.

The effect of different ground reflectivity on the inverter clipping has also been explored over two years. For each season, two sunny days that have comparable front insolation were selected as listed in

Table 1. Data comparison of two different ground covers in summer (24 May and 19 August 2023) and in winter (21 November 2022 and 16 January 2023).

Parameters	Summer White	Summer Gravel	Winter White	Winter Gravel
Average front insolation (Wh/m2)	6550	6681	4444	4470
Average rear insolation (Wh/m2)	1210	782	1058	499
Calculated total energy from the front (Wh)	28,360	28,916	20,895	21,177
Calculated total energy (Wh)	32,131	31,349	24,438	22,863
Actual total DC output energy (Wh)	28,267	30,362	23,879	23,105
Calculated bifacial gain (%)	13.3	8.4	17.0	8.0
Actual bifacial gain (%)	1.7	5.0	14.3	9.1
Energy difference (Wh)	3863	987	559	242
Energy difference (%)	12.0	3.2	2.3	1.0

. Figure 5 shows the PV power output on two cloudless days in summer (Figure 5a) and winter (Figure 5b) with white and gravel ground conditions. The key finding is that the impact of clipping loss critically conceals the benefit of more reflective ground. With 70% reflectivity from the white ground tarp, the calculated daily output energy is over 32.1 kWh in summer and 24.4 kWh in winter, yet almost 12.0% of daily energy is wasted due to inverter clipping when having the white ground in summer, as well as 2.3% in winter. In comparison, the clipping loss with gravel is very small (less than 3.5%) due to lower rear insolation as the ground reflectivity is lower, resulting in lower overall PV output. Consequently, the white ground condition leads to a substantial drop in the actual bifacial gain from 14.3% in winter to 1.7% in summer due to the loss in excess power generation. This also affects the bifacial efficiency where the white ground will provide the benefit of boosted power in winter than in summer since the additional output from the rear can be converted through the inverter without a significant inverter clipping. A further quantitative comparison between white and gravel grounds can be found in Table 1.

In terms of the DC energy efficiency, we used one year’s worth of data with different weather, distributed front insolation, and different ground conditions to investigate the influence of clipping loss on the overall production of the bifacial array. The temperature correction was carried out using the measured module temperature to isolate the clipping effect from temperature losses. The result shown in Figure 6 indicates that the effect of clipping loss for the bifacial PV system with a DC/AC ratio <1.3 has a substantial influence on reducing the daily DC energy efficiency on sunny days, i.e., for daily front insolation > 3.5 kWh/m². There is a clear trend showing that the higher insolation at the front surface leads to lower DC energy efficiency. On a very sunny day with a clear sky condition, the energy efficiency can drop to as low as 17% due to the inverter clipping but can be as high as 26% on days with very low front insolation due to the relatively higher rear irradiance on those days.

With a limited amount of direct sunlight on cloudy days (front insolation < 3.5 kWh/m²), the reflected light from the ground and the diffuse light significantly contribute to energy production. High DC energy efficiency with an average of 22.6% can be achieved on cloudy days, while the average will be only 19.1% on sunny days. This value is close to the front efficiency of the bifacial array (19.2%), implying that the benefit of the additional rear irradiance is hidden on sunny days due to the clipping loss. A steeper trend line of data on sunny days indicates that the DC energy efficiency is more sensitive to the front insolation as direct sunlight plays an important role in energy production.

From the relatively greater impact of diffuse light on cloudy days, the DC efficiency decreases slightly with the front insolation as the rear insolation does not decrease as much with front insolation. Hence, high bifacial gain at low light conditions can be achieved [37,38]. This is due to our definition of DC energy efficiency as being normalized by the daily front insolation. However, as the daily front insolation increases beyond 3.5 kWh/m², the clipping losses become steadily larger, leading to an even greater decrease in DC efficiency.

In addition to the technical analysis, a cost–benefit assessment of oversized bifacial PV systems was performed in software simulation to estimate the cost-effectiveness of adding highly reflective ground components that increase the upfront cost but can be traded off with the electricity savings. A techno-economic analysis is discussed in Section 4.3.

4.2. Model of Daily Energy with Clipping

Figure 7 demonstrates the difference between the calculated and measured daily energy of the bifacial PV modules under different front insolation values. The calculated data were obtained from (3), and the difference between the calculated and actual data represents energy loss due to the clipping effect. As higher front insolation could lead to PV production greater than the inverter capacity, this results in larger clipping loss, which will be seen mostly on sunny days. Note that the data presented in Figure 7 do not include the clipping correction, so the modeled values are the expected DC power if the inverter does not clip.

The DC daily energy was modeled from (3) and (4) to take into account the temperature correction and potential clipping effect. A comparison of the modeled and measured results is presented in Figure 8. Because of the low solar irradiance on cloudy days, the PV production is unlikely to exceed the inverter rating, resulting in no clipping effect throughout the day. Therefore, the model shows a very good agreement with the actual data. As the front irradiance increases, more PV power is generated, which leads to a higher possibility of inverter clipping. As discussed in Section 4.1, the clipping is usually experienced from 10:00 to 15:00 on very sunny cloudless days in summer, and the effect becomes more severe with higher energy production.

The quantitative results of Figure 8 are presented in

Table 2. A quantitative result of calculating model assessment with different sky conditions and ground colors.

Weather Condition	Ground Colors	Calculated Energy (kWhDC)	Actual Energy (kWhDC)	Agreement (%)
All sky conditions	All colors	469.8	461.4	98.2
All sky conditions	White	219.9	213.7	97.2
All sky conditions	Gravel	249.9	247.8	99.1
Sunny days	All colors	395.5	386.7	97.8
Sunny days	White	181.5	175.1	96.5
Sunny days	Gravel	214.0	211.6	98.9
Cloudy days	All colors	74.3	74.8	99.4
Cloudy days	White	38.4	38.6	99.5
Cloudy days	Gravel	35.9	36.2	99.2

. The sums of the modeled and actual data in each condition were calculated, and the ratio of calculated to measured values was used to evaluate the agreement of the modeled data with the actual measurement. The results demonstrate 98.2% overall agreement, which is higher on cloudy days (99.4%) than on sunny days (97.8%). The agreement on cloudy days is within 1% due to no clipping loss in the actual outputs, validating the relatively simple model for bifacial array energy production calculated using (3) in the absence of clipping. Even though the clipping effect has already been taken into account in the model (Figure 3), the clipping loss still leads to a slight deviation. For a comparison of ground colors, white ground usually has a slightly lower agreement due to its higher energy production from the rear, causing a greater clipping effect, while the energy output with the gravel grounds is more predictable because the system has a lower possibility to experience clipping, resulting in deviations of 3.5% and 1.1% for white and gravel ground on sunny days, respectively. However, there is excellent overall agreement between the calculated data with the measurements for both white or gravel ground conditions on cloudy days (>99%). The data for black tarps in any weather conditions were excluded from the quantitative analysis in Table 2 due to a very limited sample size of black ground.

4.3. Evaluation of Electrical Losses and Financial Returns

To evaluate the impacts of the inverter clipping in terms of electrical losses and financial returns, we simulated a residential bifacial PV system model with different DC/AC ratios and ground reflectivities using SAM 2022.11.21 developed by NREL, Golden, CO, USA. The model was created based on our actual system described in Section 2. The installation cost is USD 3.35/W_DC and USD 3.5/W_DC for the monofacial and bifacial PV models, respectively [39,40]. A range of DC/AC ratio from 1.1 to 1.5 was simulated by changing PV capacity while keeping the inverter size fixed at 3.8 kW. Three albedo values, 0.1, 0.35, and 0.7, were selected to represent black, gravel, and white ground, respectively.

For the financial assessment of the bifacial PV systems, we applied the time-of-use (TOU) tariff structure from Baltimore Gas & Electric (BGE) [41], where the peak and off-peak periods are shown in Figure 9 and Figure 10. The electricity rates during peak demand are approximately 3.5 times higher than the off-peak periods. In addition to the rates shown in Figure 9 and Figure 10, there is a fixed charge of USD 9.32 applied monthly to the customer’s bill. Note that this TOU schedule excludes weekends and holidays, which are billed at the off-peak rates. Three economic parameters, levelized cost of energy (LCOE), net present value (NPV), and payback period (PB), were compared to analyze the financial benefits of oversized bifacial PV systems with different reflective grounds.

The key finding from the simulations is that, as expected, the high reflective ground (albedo = 0.7) leads to a larger clipping loss in an oversized bifacial PV system due to higher DC power production from the rear. As shown in Figure 11, with the DC/AC ratio of 1.3, the amount of energy loss from clipping can increase by 110% of the clipping loss in a monofacial PV system with the same DC/AC ratio. This amount of loss can be converted to USD 13 per year.

However, the financial measures show significant benefits of high reflective ground in an oversized system, which can outweigh the electrical losses. Figure 12 presents the financial parameters of the oversized bifacial PV system with a different albedo. While highly reflective ground leads to higher losses, it demonstrates a remarkable return by 15% higher NPV than the monofacial PV. Additionally, the bifacial PV with a highly reflective ground can expedite the payback time by 4% and lower the LCOE down to 7.52 cents/kWh. The sensitivity analysis of additional costs for a highly reflective ground was also performed to identify the break-even point where the installation of any grounds with an albedo of 0.7 will not be a cost-effective option anymore. Our results suggest that a high albedo ground that costs more than 30 USD/m² will not be financially viable since it offers the same NPV as a natural gravel ground. However, an average market white ground costs less than 5 USD/m², which is much lower than the yield of 30 USD/m².

Figure 13 and Figure 14 present the heatmaps of NPV and LCOE from 12 system configurations with different DC/AC ratios and ground albedo. Overall, the higher reflective grounds provide higher NPV and lower LCOE regardless of system sizes. Bifacial systems with the highest ground albedo of 0.7 achieve the best financial return among all ground colors with the same system size. It can provide up to 14% higher NPV and 4% cheaper LCOE than the monofacial system.

However, the larger-size systems (DC/AC = 1.5) will require a higher ground albedo than the smaller-size systems (DC/AC = 1.1 and 1.3) in order to see the benefits of bifaciality. This is because the capital cost and electrical losses of large PV systems are greater, leading to a longer payback time. The additional cost of bifacial arrays will reduce the financial return further. However, as this difference is shrinking each quarter as bifacial module manufacturing scales up, relatively lower albedos will be needed in the near future to achieve cost-effectiveness. Based on our simulations, a system with a DC/AC ratio = 1.5 will be financially feasible only when it has a ground albedo of 0.7 to achieve a better financial return compared to a monofacial system.

All simulation results are listed in

Table 3. A summary of results of 16 simulation cases.

	DC/AC Ratio	Albedo	AC Inverter Efficiency Loss (%)	Inverter Clipping Loss (kWh/yr)	Annual AC Energy (kWh/yr)	Annual DC Energy (kWh/yr)	Energy Yield in Year 1 (kWh/kW)	Electricity Bill Savings (USD/yr)	Payback Period (yrs)	NPV (USD)	LCOE (cents/ kWh)
Monofacial	1.1	0.0	0.67	0	6007	6300	1466	684	14.8	2474	7.69
	1.3	0.0	0.89	64	7063	7471	1437	806	15.0	2827	7.82
	1.5	0.0	2.15	162	7310	7831	1274	835	17.0	2117	8.77
	1.7	0.0	2.31	176	7330	7866	1118	837	19.7	1072	9.99
Bifacial	1.1	0.1	0.01	1	6070	6366	1484	691	14.8	2499	7.69
		0.35	0.03	2	6181	6484	1511	704	14.5	2651	7.55
		0.7	0.10	6	6335	6648	1548	722	14.1	2861	7.36
	1.3	0.1	1.10	80	7134	7561	1453	814	15.0	2862	7.81
		0.35	1.36	101	7246	7700	1476	828	14.7	3026	7.68
		0.7	1.78	136	7397	7894	1507	846	14.4	3250	7.52
	1.5	0.1	2.74	212	7434	8013	1136	850	19.5	1142	9.94
		0.35	3.13	247	7538	8158	1552	863	19.2	1299	9.80
		0.7	3.50	281	7654	8314	1336	878	16.3	2560	8.44
	1.7	0.1	2.74	212	7434	8013	1136	850	19.5	1142	9.94
		0.35	3.13	247	7538	8158	1152	863	19.2	1299	9.80
		0.7	3.73	301	7678	8361	1173	881	18.7	1512	9.61

. Even though systems with a DC/AC ratio = 1.7 have the largest PV capacity than the smaller systems, the annual energy provided to the systems of systems with DC/AC = 1.7 is comparable to the system with DC/AC = 1.5. Overgeneration in an extremely high DC/AC ratio system causes a dramatic increase in an inverter’s operating voltage. However, the inverter cannot operate at voltage beyond a certain limit. Hence, a maximum power point tracker (MPPT) in the solar inverter will lower the operating voltage to and reduce energy DC output of the system. This is the reason that clipping loss in systems with DC/AC = 1.7 and 1.5 are comparable.

The evaluation of financial measures in an energy system needs to be considered deliberately as each financial parameter reflects different system values. For example, NPV determines the total financial return at the end of the project, while LCOE represents the price of electricity that the system generated, which will consider the system’s self-consumption. This is why a system with the highest NPV is not always the system with the lowest LCOE. A more detailed study for the DC/AC ratio and the financial analysis of bifacial clipping on NPV, LCOE, and PB has recently been published [42].

5. Conclusions

This paper discusses the impact of inverter clipping on a bifacial PV array performance in terms of DC power output, bifacial gain, and DC energy efficiency, as well as discusses the bifacial power calculating model with assessment. The key findings indicate that even though the modules generate higher DC output in summer than winter due to higher solar insolation, the expected benefit of bifacial energy collection is partially offset due to clipping losses. Thus, the standard practice of oversizing the DC/AC ratio minimizes the benefit of bifaciality. The clipping loss will be more severe on sunny days in summer with high front insolation. For our system, the clipping loss occurs between 10:00 and 15:00, when solar irradiance reaches its peak, but this will vary with the DC/AC ratio and albedo. The studies of different ground colors show that the white ground in winter leads to the highest bifacial gain (13.1%) and daily DC efficiency (22.2%) as expected, since there is maximum light reflected on the back surface and no clipping. While the energy production benefits of the high albedo ground cover are reduced due to clipping, the financial benefits in terms of payback time and net present value are significant.

These results quantify the possible benefits of increasing the bifacial gain. Yet the white ground tarp shows minimal advantage in summer due to the inverter clipping. This issue could be an obstacle for the system to optimize module performance, so the power calculation model and energy storage management strategy are introduced to help utilize the clipped PV power. The results of the relatively simple PV power calculation model developed here demonstrate the ability to estimate the daily DC energy of the bifacial modules with 98.2% overall agreement between the modeled and actual data. The estimation of the energy on cloudy days showed a higher agreement at 99.1% on average, followed by 96.7% agreement on average for data on sunny days. This PV power calculation model will be used as a tool for the implementation of the automated battery management strategy we are currently developing and will be further discussed in future publications. The battery management strategies would be able to offset the impact of clipping loss and recover some or all of the bifacial gain depending on battery dispatch modes and the system size.

The simulation results suggest that bifacial system with highly reflective white ground provides the best financial returns to the project, which can decrease the LCOE by up to 4% and increase NPV by 15% compared to a monofacial system. Even though the clipping loss in systems with higher ground reflection will be larger, the clipping loss contributes a minimal impact on the system finance and can be compensated by an increase in overall production.