System Optimization and Primary Electrical Design of 50 MW Agrivoltaic Power Station: A Case Study in China

Abstract

Agrivoltaic technology holds great significance for promoting the collaborative development of new energy industries and modern agriculture. A systematic optimization design and preliminary electrical scheme for a 50 MW agrivoltaic power station in Shaanxi Province, China, were studied in this work. A combination of checkerboard and long-row layouts was adopted, considering the influence of the shading rate on agricultural production and photovoltaic power generation. The checkerboard pattern features the highest system efficiency, the smallest irradiance loss, and a slight lead in power generation, with a moderate shading rate, when compared to the other patterns. The expected energy gain from the bifacial modules’ rear side in this specific setup is 7.6%. These layouts ensure the power generation efficiency of the photovoltaic power station, while minimizing the shading impact of shading on crop growth, thereby achieving efficient comprehensive utilization of agricultural greenhouses and solar power generation. The primary electrical system was designed, including the main wiring design, main transformer selection, and type selection of major electrical equipment. The research results provide a practical reference for the large-scale application of agrivoltaic power stations, which is beneficial to promoting the high-quality development of modern agriculture.

1. Introduction

Addressing climate change remains one of the most pressing challenges for humanity in the 21st century. Photovoltaic (PV) power generation provides clean renewable energy, and its large-scale application can effectively reduce greenhouse gas emissions, thereby facilitating the global decarbonization process. Therefore, expanding the installed solar power capacity has become a core initiative to achieve economic decarbonization [1]. Projections indicate that the global installed PV power capacity will reach 6000–8000 GW by 2050, with China accounting for 60% of the new additions [2]. However, the large-scale development of the PV industry has intensified land use conflicts [3]. Satellite monitoring data reveals that most non-residential PV systems worldwide are constructed on cultivated land [4]. By 2020, PV power stations had occupied 911 km² of arable land [5]. Agrivoltaic technology provides a comprehensive land use solution to balance food security and energy supply through the organic integration of PV power generation and agricultural production on the same land [6].

At present, agrivoltaic systems can be divided into three types according to the categories of agricultural land: agrivoltaic systems combined with cultivated land, livestock farms, and solar greenhouses [7]. When PV systems are integrated with cultivated land, the installation height and layout of PV panels may restrict the passage of large agricultural machinery (such as harvesters and seeders), thereby hindering mechanized agricultural operations [8]. In PV systems combined with farms, livestock manure, hair, and dust can easily adhere to the surfaces of PV panels, reducing power generation efficiency [9]. Moreover, the maintenance of PV panels requires access to farm areas, which may interfere with animal activities. In contrast, solar greenhouse PV systems are less constrained by compatibility with the natural environment and conflicts in production management, making them the preferred model for current agrivoltaic technology.

Research on solar greenhouse PV systems has involved crops such as tomatoes, lettuce [7], sweet peppers [10], grapes, onions [11], berries [12], and potatoes [13]. Yano et al. [14] found that the use of semitransparent PV panels has minimal impact on the growth of horticultural crops, only slightly reducing plant biomass yield while accelerating the apical growth mechanism of light-loving plants. Chatterjee et al. [15] explored a cost-effective combination of poly (3-hexylthiophene) (P3HT) donors and a non-fullerene acceptor based on fluorinatednaphthobisthiadiazole (FNTz-FA) for applications in green light wavelength-selective organic solar cells. These cells convert solar energy in the green light spectrum into electrical energy while utilizing the transmitted blue and red light to grow crops. Torrente et al. [16] proposed a new model to simulate the distribution and uniformity of radiation within agricultural greenhouses with PV panels installed on the roof, which are critical factors in PV greenhouse design. The model enables scientists and engineers to design agricultural power-generating greenhouses. Boccalatte et al. [17] designed a high-efficiency micro-PV greenhouse with near-zero energy consumption, establishing a dynamic simulation model of a real greenhouse to predict its performance and explore beneficial energy-saving control strategies. Petrakis et al. [18] investigated the impact of semitransparent PV units on global horizontal irradiance and photosynthetically active radiation. Jamil et al. [19] studied the impact of varying PV transparency on strawberry yield and growth with Cd-Te PV modules, and evaluated the potential of agrivoltaics as a sustainable solution for strawberry production in northern climates. Hermelink et al. [20] provided the yield response curves of individual berry crops to increasing shade, also distinguishing between different radiation intensity environments, which can help in selecting optimal crop and panel density combinations for different locations. Zhang et al. [21] presented spectral-splitting concentrator agrivoltaics (SCAPV), which can effectively photosynthetically harvest excess light energy for photovoltaic power without compromising crop productivity. This system transmits a selected light spectrum for plant growth while reflecting the remaining spectrum for electricity generation. Jamil et al. [22] compared strawberry agrivoltaics using two different types of solar PV modules—uniform illumination provided from semi-transparent thin-film cadmium telluride (Cd-Te) and non-uniform illumination from semitransparent crystalline silicon (C-Si) including rows of solar cells and transparent solar-grade glass. Results showed that non-uniform illumination from C-Si PV modules significantly increased fresh weight by 18% compared to controls while lowering soil temperature and increasing humidity.

Existing research on PV greenhouses has primarily focused on improving crop yield and PV conversion efficiency, with less attention given to the integrated design of PV and electrical systems. This study presents the comprehensive design of a 50 MW agrivoltaic power station in Shaanxi Province, China. PV modules are installed on the greenhouse roof. The design encompasses both the PV power generation system (including simulation) and the primary electrical system for step-up and grid connection. Crucially, the PV system design carefully considers the impact of the shading rate on crop production. A variety of optimized PV module layout schemes are employed to minimize shading while ensuring power generation efficiency. The grid connection design encompasses the main circuit design (including N − 1 contingency analysis), short-circuit current calculation, and key electrical equipment selection. The design of this project addresses a gap in the integrated design of PV and electrical systems, which can serve as a valuable reference for similar engineering projects.

2. Project Overview

This study proposes the construction of a 50 MW centralized agrivoltaic power plant in Hengshan County, Yulin City, Shaanxi Province. The PV panels will be installed on the roofs of solar greenhouses, covering a total construction area of 30,000 square meters. The station is designed to operate in grid-connected mode. The power plant is planned to connect to the grid via a 110 kV outlet with a double feeder.

Hengshan District is located at approximately 37.95° N and 109.26° E, deep in the inland areas of Northwest China. The terrain is dominated by plains and hills, with relatively gentle undulations. Influenced by a temperate continental climate, it enjoys plenty of sunny days throughout the year, with few cloudy ones and minimal cloud cover blocking sunlight. Such climatic conditions are conducive to enhancing the power generation efficiency of photovoltaic modules.

As a major agricultural area in Shaanxi, Hengshan is suitable for large-scale deployment of PV systems due to its abundant solar radiation and relatively flat terrain. Additionally, its solid agricultural foundation provides favorable external conditions for agricultural cultivation in photovoltaic greenhouses. According to the 2024 monthly meteorological data (

Table 1. Monthly meteorological data of Hengshan District, Yulin City, Shaanxi Province, China.

	Total Horizontal Radiation	Horizontal Scattered Radiation	Temperature	Wind Velocity	Atmospheric Turbidity Factor	Relative Humidity
January	81.2 kWh/m2/mth	28.6 kWh/m2/mth	−8.4 °C	2.11 m/s	3.904	50.0%
February	102.8 kWh/m2/mth	33.8 kWh/m2/mth	−3.2 °C	2.49 m/s	5.184	43.3%
March	139.8 kWh/m2/mth	56.9 kWh/m2/mth	4.8 °C	2.90 m/s	5.925	32.3%
April	170.6 kWh/m2/mth	72.0 kWh/m2/mth	11.9 °C	3.19 m/s	6.374	31.3%
May	180.6 kWh/m2/mth	89.5 kWh/m2/mth	18.0 °C	2.99 m/s	6.259	33.2%
June	151.7 kWh/m2/mth	83.6 kWh/m2/mth	21.9 °C	2.70 m/s	6.352	42.5%
July	171.7 kWh/m2/mth	89.1 kWh/m2/mth	24.1 °C	2.69 m/s	5.749	54.1%
August	160.6 kWh/m2/mth	81.2 kWh/m2/mth	21.9 °C	2.40 m/s	6.006	61.1%
September	126.7 kWh/m2/mth	58.6 kWh/m2/mth	16.3 °C	2.39 m/s	5.496	62.4%
October	117.4 kWh/m2/mth	44.7 kWh/m2/mth	10.1 °C	2.39 m/s	5.270	54.8%
November	86.4 kWh/m2/mth	27.7 kWh/m2/mth	1.6 °C	2.40 m/s	4.107	47.8%
December	78.8 kWh/m2/mth	22.0 kWh/m2/mth	−6.4 °C	2.30 m/s	3.625	49.1%
Year	1568.3 kWh/m2/yr	687.7 kWh/m2/yr	9.4 °C	2.6 m/s	5.354	46.8%

), the annual average temperature in this area is about 9.4 °C, and the annual total solar radiation on the horizontal surface reaches 1568.3 kWh/m² with a peak sunshine duration of 4.3 h, making Hengshan a region with rich solar energy resources. This not only facilitates the construction of agrivoltaic power stations and agricultural production but also lays a foundation for building a new green and low-carbon agricultural model in the region.

The design consists of two main parts: the PV power generation system and the grid-connected step-up primary electrical system. The PV system design involves the layout of PV arrays as well as the selection of DC combiner boxes and inverters among other components. The grid-connected step-up primary electrical system design includes the main electrical wiring configuration, main transformer selection, short-circuit current calculation, and selection of key equipment (e.g., busbars, current transformers, voltage transformers, circuit interrupters, disconnector switches).

3. PV System Design

3.1. PV Module Selection

The selection of PV modules shall be comprehensively evaluated based on technical indicators including conversion efficiency, production cost, weather resistance, and impact on crop growth. As demonstrated in [11], double-glass modules (especially the semi-transparent, amorphous, silicon glass-based type) are better suited to meeting the lighting requirements of greenhouse crops due to their high light transmittance and low sensitivity to shadow obstruction. Compared with conventional polycrystalline silicon modules, they can significantly optimize the growth environment of crops in greenhouses, which was adopted for this project. The LR5-72 HBD 550 M bifacial double-glass module with a rated power of 550 Wp. The specific technical parameters are detailed in

Table 2. Main parameters of LR5-72 HBD 550 M PV module.

Name	Parameters
Peak Power	550 Wp
MPP Voltage (Vmpp)	41.95 V
MPP Current (Impp)	13.12 A
Open-Circuit Voltage (Voc)	49.8 V
Short-Circuit Current (Isc)	13.99 A
Voc Temperature Coefficient	−0.265%/°C
Peak Power Temperature Coefficient	−0.34%/°C
Module Efficiency	21.3%
Operating Temperature Range	−40~85 °C
Dimensions (L × W)	1.134 mm × 2.278 mm

, and a total of 90,909 pieces of this module type need to be configured for the project.

3.2. Inverter and DC Combiner Box Selection

It is proposed to adopt centralized converters for this project, in view of its large-scale system capacity of over 20 MW and the requirements for high-efficiency, economical, and stable operations of the system. Fourteen units (one unit on standby) of SG3125-HV-20 model centralized inverters from Sungrow, with a capacity ratio of 1.23, meet the power generation demand and were selected. The inverter’s parameters are shown in

Table 3. Main parameters of SG3125-HV-20 inverter.

Name	Parameter
Rated Output Power	3125 kW
MPPT Voltage Range	875–1300 V
Maximum Input Voltage	1500 V
Maximum Output Power	3593 kW
Peak Conversion Efficiency	99%
Maximum DC Current	4178 A
Maximum AC Output Current	3458 A
AC Output Voltage	600 V
Grid Frequency	50/60 Hz
Power Factor	0.95 leading to 0.95 lagging
Dimensions (L × W × H)	2.438 mm × 2.591 mm × 2.991 mm

.

Given the large number of connections between photovoltaic modules and inverters, DC combiner boxes were selected to reduce wiring complexity, enhance system reliability, and facilitate subsequent operation and maintenance. The KBT-PVX16 DC combiner box was selected, and detailed parameters are shown in

Table 4. Main parameters of KBT-PVX16 DC combiner box.

Name	Parameter
Single-Channel Input Voltage Range	DC 24~1500 V
Maximum Single-Channel Input Current	20 A
Number of Input Channels	16
Maximum Output Current	250 A
Rated Output Voltage	DC 1500 V
Operating Temperature Range	−25~+70 °C
Protection Class	IP 65
Dimensions (L × W × H)	700 mm × 550 mm × 220 mm

. A total of 232 units (with 5% as backup) of 16-in-1 combiner boxes are planned along the pre-reserved channels on the ground of each greenhouse to ensure stable operation of the PV system and efficient power collection.

3.3. PV String Design and Array Layout

Combining the above parameters of photovoltaic modules and inverters, the number of series-connected PV modules is calculated according to Formulas (1) and (2) in Standard GB 50797-2012 [23].

$$N\le {\displaystyle \frac{V_{\mathrm{d}\mathrm{c}\mathrm{m}\mathrm{a}\mathrm{x}}}{V_{\mathrm{o}\mathrm{c}}\times \left[1+\left(t-25\right)\times K_{\mathrm{v}}\right] }},$$

(1)

$${\displaystyle \frac{V_{\mathrm{m}\mathrm{p}\mathrm{p}\mathrm{t}\mathrm{m}\mathrm{i}\mathrm{n}}}{V_{\mathrm{p}\mathrm{m}}\times [1+\left(t^{\prime }-25\right)\times {K^{\prime }}_{\mathrm{V}}]}}\le N\le {\displaystyle \frac{V_{\mathrm{m}\mathrm{p}\mathrm{p}\mathrm{t}\mathrm{m}\mathrm{a}\mathrm{x}}}{V_{\mathrm{p}\mathrm{m}}\times [1+\left(t-25\right)\times {K^{\prime }}_{\mathrm{V}}]}}$$

(2)

where Kv and K′v are the PV module temperature coefficient of the open-circuit voltage and operating voltage, respectively. N is the number of series-connected PV modules (N is an integer), and t and t’ are the extreme low and high temperatures under the PV module’s operating conditions (°C), respectively. V_dcmax is the maximum DC input voltage permitted by the inverter (V). V_mpptmax and V_mpptmin are the maximum and minimum MPPT voltage of the inverter (V), respectively. V_oc and V_pm are the open-circuit voltage and operating voltage of the PV module (V), respectively.

The annual maximum and minimum temperatures at the project site are 24.1 °C and −8.4 °C, respectively. According to the above formulas, the calculated range for the number of series-connected PV modules N was 21 ≤ N ≤ 28. A total of 27 modules were selected for each string, and the rated power of each string was 14,850 W. Based on the total installed capacity of the PV power station and the rated power of each string, the number of PV strings required for the project was approximately 3367.

Shading is the most critical issue in greenhouse PV systems. PV panels block sunlight from reaching crops in the greenhouse, creating partial shadows that may affect crop growth [24]. Therefore, the setting of shading rates should be adjusted according to different crop types. When the shading rate was 30% in a previous study, there was no significant impact on lettuce yield, and the specific leaf area was increased [25]. However, at a 50% shading rate, lettuce yield decreased by 21–42%, with a significant reduction in leaf count [26]. When the shading rate was 25.9%, strawberry yield increased by 16.73% (equivalent to 447.9 g per plant) [27]. It is evident that light-loving crops such as vegetables and fruits require a shading rate below 30% [28]. Unlike light-loving crops, shade-tolerant crops like mushrooms and yellow sweet potatoes generally require a shading rate above 50% [28]. At a 17.97% shading rate, the growth and development of yellow sweet potatoes were inhibited, with dry weight reduced by 21.6% [29]. When the shading rate reached 75%, mushroom yield increased by 102% [30].

A reasonable panel layout can minimize shading effects and maintain a uniform light distribution [31]. For instance, at a 10% shading rate, when PV panels were arranged in a linear pattern, tomato ripening was slightly delayed, with reduced fruit quality and diameter [32]. However, when the panels were arranged in a checkerboard pattern, tomato plants grew taller, and yields increased significantly [33]. Therefore, different from conventional long-row arrangements, a combination of checkerboard and long-row layouts was adopted for the PV array in this project to meet the shading requirements of various crops, given the large project scale and diverse crop types. The plans of this project include the installation of PV modules on the top of greenhouse sheds. Each greenhouse has an arched roof, with a north–south width of 8 m and an east–west length of 60 m. A fixed adjustable bracket system was adopted for the PV panels, with adjustment of the tilt angle twice a year: it is set at 20° from April to September, and adjusted to 40° from October to March of the following year. All modules are fixed facing due south. The arrangements are as follows:

• In greenhouses requiring a shading rate below 30%, the PV modules are arranged in a quasi-checkerboard pattern (Figure 1a): spaced along the east–west direction with 3 rows in the north–south direction and a 1.139 m gap between adjacent rows.
• In greenhouses with a shading rate between 30% and 50%, the checkerboard layout (Figure 1b) was adopted: east–west spacing with 4 rows in the north–south direction.
• In greenhouses requiring a shading rate above 50%, long-row parallel arrangements (Figure 1c) are used, with a 1.139 m gap between adjacent rows.

3.4. PVsyst Simulation and Results Analysis

The PVsyst software Version 8.0.5 was used to simulate and predict the power generation of the PV system based on the meteorological data of the project site, PV module parameters, and system configuration. The corresponding meteorological data are derived from Meteonorm 8.0. Given that the greenhouse roof is transparent, part of the radiation will be reflected onto the back of the PV modules, so albedo was designed in the PVsyst simulation. Based on the literature [34], the albedo was set to 40%. A three-dimensional shadow scenario was defined in the simulation, and the shadow coefficient was corrected according to the linear influence model.

Yulin City is located in the arid and water-scarce area of Western China. Strong wind and sand lead to dust accumulation on the surfaces of solar panels, which blocks light absorption and directly reduces the photoelectric conversion efficiency. The actual power generation efficiency of panels dropped from 23% to 25% to approximately 17% to 18% according to the measured data from a 20 MW PV power station in Yulin, Shaanxi Province, China. This was due to the failure to thoroughly solve the problem of panel cleaning. Therefore, the soiling loss was assumed to be 6% in this project. To simplify the design, local climate data were directly used for temperature parameters, and the effect of humidity was not considered in this simulation for the time being.

3.4.1. Analysis of Power Generation Gain of Bifacial Modules

Taking the arrangement of all PV modules in long rows in the PV power station as an example, the power generation gain of bifacial modules is analyzed. The power generation gain can be calculated with Formula (3).

$$G_{cg}={\displaystyle \frac{{M_{BPV}-M}_{npv}}{M_{npv}}}\times 100\%,$$

(3)

where $G_{cg}$ is the power generation gain; $M_{BPV}$ and $M_{npv}$ are the power generation of bifacial modules and conventional modules, respectively.

Figure 2 exhibits the power generation gain at different installation heights of bifacial modules (the height from the module edge to the vault of the greenhouse). When the installation height is 0.5 m, the power generation gain of bifacial modules is insignificant, due to severe light shading on the back of the modules. When the installation height is less than 1.5 m, the power generation gain increases rapidly with the rise in installation height. When the installation height exceeds 1.5 m, the growth rate of the power generation gain slows down. As the height increases, ground-reflected light attenuates and the proportion of scattered light decreases, leading to a significant reduction in the increase in backside irradiance. The modules in this project are installed on the greenhouse roof. The higher the installation height of the modules, the greater the wind load. Considering safety factors, the module installation height is set to 1 m, resulting in a power generation gain of 7.6%.

3.4.2. PV Panel Arrangement Pattern Analysis

Table 5. Performance indicators of different PV panel arrangement patterns.

	Checkerboard Pattern	Quasi-Checkerboard Pattern	Long-Row Pattern
Shading rate	29%	39%	58%
Performance ratio	84.0%	84.5%	84.3%
Irradiance loss	3.6%	2.9%	3.1%
Total system power generation	79,141 MWh/yr	79,623 MWh/yr	79,495 MWh/yr
Annual unit power generation	1583 kWh/kWp/yr	1592 kWh/kWp/yr	1590 kWh/kWp/yr
Unit power generation	4.34 kWh/kWp/day	4.36 kWh/kWp/day	4.36 kWh/kWp/day

presents the performance indicators of different PV panel arrangement patterns. Among the three patterns, the checkerboard pattern features the highest system efficiency, the smallest irradiance loss, and a slight lead in power generation, with a moderate shading rate (39%), making it suitable for greenhouses requiring a shading rate between 30% and 50%. The quasi-checkerboard pattern has the lowest shading rate (29%), but its lower system efficiency and larger irradiance loss result in slightly lower power generation, making it suitable for scenarios with poor light conditions or low shading rate requirements. The long-row pattern has the highest shading rate (58%). However, it achieves a unit power generation comparable to that of the checkerboard arrangement through balanced system efficiency and irradiance loss. This arrangement saves more horizontal land space. Due to its higher shading rate, it is more suitable for shade-tolerant plants such as mushrooms that require a higher shading rate.

3.4.3. Final Simulation Results

Given the large scale of this project and the diversity of crop species, the design needs to meet the growth requirements of different crops. It is assumed that each of the three arrangement patterns accounts for one-third. The simulation results show that the annual energy conversion of the PV array is 79,796 MWh, and the final power generation after various losses is 84,004 MWh. The performance ratio (PR) is 84.65%. The irradiation, illumination, and temperature differences in the PV system are uncontrollable, and long-term operation may cause performance degradation of inverters and PV modules. The main energy losses of this system are pollution loss (6.0%), shield irradiance loss (2.7%), PV energy loss due to temperature (1.9%), module and string mismatch loss (2.1%), and PV loss caused by irradiation intensity (1.0%).

The Datang Huayin 100 MW Agrivoltaic Power Generation Project (Weinan City, Shaanxi Province, China) has an AC-side capacity of 100 MW, with an annual power generation of 150.3653 million kWh and an annual utilization hour of 1251.4 h, equivalent to an AC-side annual unit power generation of 1503.65 kWh/kWp/yr and a PR of 87.15%. The 100 MW Agrivoltaic Photovoltaic Power Generation Project in Baota District (Yan’an City, Shaanxi Province, China) has an AC-side installed capacity of 100 MW, with an average annual on-grid power of 158.5059 million kWh and an annual utilization hour of 1317.49 h, equivalent to an AC-side annual unit power generation of 1585.06 kWh/kWp/yr and a PR of 85.55%. The AC-side annual unit power generation of this project is 1596 kWh/kWp/yr, with a PR of 84.6%. Through comparison with the measured data of similar projects, this design is proven to be feasible.

Assuming the same equipment configuration (photovoltaic modules, inverters, etc.), a vertical bifacial agrivoltaic system was adopted with PV modules arranged in long rows. The simulation results show that the system generates 52,405 MWh/yr, with an annual unit power generation of 1048 kWh/kWp/yr and a daily unit power generation of 2.87 kWh/kWp/day, along with a system efficiency of 72.1%. All the above indicators are lower than the design indicators of this project.

In addition, to reduce shading of PV modules by the greenhouse in the vertical bifacial agrivoltaic system, the modules need to be installed above the greenhouse roof (4.5 m). When combined with the module height (9.112 m), the total height at the top of the modules reaches approximately 11 m. Excessively high modules will lead to excessive wind loads given the strong wind and sand in the project construction area, posing certain safety risks. Therefore, the system designed in this project is superior to the vertical bifacial agrivoltaic system.

4. Electrical Primary System Design

4.1. Main Wiring Design

The total installed capacity of the designed agrivoltaic power generation system was 50 MW. According to calculations, the system consists of 90,909 bifacial double-glass PV modules with a rated power of 550 Wp each and 14 centralized inverters rated at 3125 kW. Considering the project’s requirements for a straightforward configuration with minimal equipment and reduced investment costs, a non-segmented single-bus topology was selected. This configuration is particularly suitable for applications with a limited number of outgoing circuits, due to its simplicity in operation and maintenance.

The non-segmented single-bus configuration minimizes the use of bus couplers (e.g., circuit interrupters and disconnectors), reducing both the initial investment and operational complexity. Although the single-bus structure poses a single-point-of-failure risk, reliability is enhanced through the following measures: redundant design of inverters and combiner boxes (N + 1 configuration), comprehensive protection systems for the main transformer (differential, overcurrent, and overvoltage protection), remote monitoring and rapid fault diagnosis capabilities for critical equipment.

The power station is unable to transmit 100% of the power in the event of a short circuit, open circuit, or an external grid-side fault occurring on the 110 kV outgoing cable/overhead line. To ensure system safety, a double-circuit configuration (Line 1 and Line 2) is employed for the 110 kV outgoing line. During normal operation, Line 1 is in service, while Line 2 remains on cold standby, with the circuit interrupters (DL2) on Line 2 open and the disconnector switch (GS2) closed, allowing for immediate switching when needed. In case of a fault on Line 1, the power supply automatically switches to Line 2 (switching time < 200 ms).

The high-voltage side of the main transformer outgoing line via double-winding bushings taps into the two circuits, and an overcurrent intertrip device is included to prevent fault escalation. To enhance the system’s ride-through capability, an SVG (Static Var Generator) is configured on the high-voltage bus side, with a capacity of ±20 Mvar, a response time of ≤5 ms, and a reactance rate of 5%. A grounding resistor is installed at the neutral point of the main transformer.

In the aforementioned configuration (main transformer impedance of 12% + Line 1 current-limiting reactor with a reactance rate of 5% + Line 2 in cold standby), the periodic component of the short-circuit current on the 110 kV side bus is 6.51 kA, the impact short-circuit current value is 16.57 kA, and the maximum effective value of the total short-circuit current is 9.84 kA. All these current values are far below the limit values of conventional 110 kV equipment (25 kA for breaking/40 kA for closing), providing a sufficient safety margin. It can be seen that although the double-circuit line scheme increases the short-circuit current, the combination of the main transformer’s high impedance (12%), the 5% reactance rate of Line 1’s current-limiting reactor, and the cold standby operation of Line 2 can suppress the short-circuit current to a safe level while ensuring N − 1 reliability.

Therefore, the system architecture was structured as follows: 3367 strings of PV arrays were aggregated via 232 16-input-1-output combiner boxes into 14 inverters (1 unit on standby), which feed into a 600 V non-segmented single bus. The power was then stepped up to 110 kV through a main transformer and transmitted via two outgoing lines. The detailed configuration is illustrated in Figure 3. The single-bus design allows for future expansion by adding transformers and bus section interrupters.

4.2. Main Transformer Selection and Short-Circuit Current Calculation

The main transformer is tasked with stably stepping up the low-voltage electrical energy output by the inverter to a voltage level suitable for grid connection. Meanwhile, it reduces voltage fluctuations and distortions, transmits power smoothly to the grid, and ensures the normal operation of equipment in the grid. The main transformer selected was of the SFSZ11-63000/110 model, the parameters of which are shown in

Table 6. Main parameters of SFSZ11-63000/110 main transformer.

Name	Parameter
Rated Capacity	63 MVA
Rated Voltage	110 ± 8 × 1.25%/0.6 kV
Voltage Regulation Mode	On-Load Tap Changer (OLTC)
Cooling Method	Forced Oil Circulation with Air Cooling
Connection Group	YN, d11
Short-Circuit Impedance	12%

below.

The design capacity of this project was 50 MW, the main transformer capacity was 63 MVA, the transformer impedance U_K% = 12, and the reference capacity was taken as S_b = 100 MVA. The short-circuit current was simplified into an equivalent circuit diagram, as shown in Figure 4. The short-circuit current was calculated based on the equivalent circuit diagram, and the calculation results are shown in

Table 7. Parameters of short-circuit points.

Short-Circuit Point	D1	D2	D3
Effective Value of Periodic Component	1.64 kA	0.91 kA	6.51 kA
Impact Short-Circuit Current Value	4.17 kA	2.32 kA	16.57 kA
Maximum Effective Value of Short-Circuit Full Current	2.48 kA	1.37 kA	9.84 kA

.

4.3. Selection of Main Electrical Equipment

4.3.1. Busbars

Under normal operating conditions, the rated current of the equipment (I_al) for the 0.6 kV side should meet the condition that I_al ≥ I_max (the maximum continuous operating current); that is,

$$I_{\mathrm{a}\mathrm{l}}\ge I_{\mathrm{m}\mathrm{a}\mathrm{x}}=1.05\times {\displaystyle \frac{P}{\sqrt{3}UCOS\mathsf{\varphi }}}=1.05\times {\displaystyle \frac{3125}{\sqrt{3}\times 0.6\times 0.95}}=3323.56\mathrm{A},$$

(4)

When determining the busbar cross-sectional area, the economic current density method is adopted for calculation. The economic density value j is set to 1.65 A/mm², so the economic cross-section of the busbar is 2014.28 mm².

The aluminum busbar of model LMY-125 × 10 was selected. The specification of the aluminum busbar was double strips placed vertically with 125 mm × 10 mm. At the reference ambient temperature of 25 °C, when the maximum allowable temperature of the busbar was set to 70 °C (${\theta }_{\mathrm{a}\mathrm{l}}$), the skin effect coefficient Ks was calculated to be 1.45, and the long-term allowable current-carrying capacity I_N of the busbar reached 3426 A. In view of the actual ambient temperature of 24.1 °C ($\theta$), which deviated from the reference temperature, a temperature correction coefficient K needed to be introduced for corresponding correction as Formula (5).

$$K=\sqrt{{\displaystyle \frac{{\theta }_{\mathrm{a}\mathrm{l}}-\theta }{{\theta }_{\mathrm{a}\mathrm{l}}-25}}}=\sqrt{{\displaystyle \frac{70-24.1}{70-25}}}\approx 1.01,$$

(5)

Therefore, the long-term allowable current-carrying capacity is as follows:

$$I_{\mathrm{a}\mathrm{l}}=K\times I_{\mathrm{N}}=1.01\times 3426=3460.26\mathrm{A}>I_{\mathrm{m}\mathrm{a}\mathrm{x}},$$

(6)

Thus, the requirements for long-term heating conditions were satisfied.

Similarly, the aluminum busbar of model LMY-63 × 6.3 was selected for the 110 kV side, with a single strip placed horizontally in the specification of 63 mm × 6.3 mm. The economic cross-section of the busbar is 221.50 mm².

4.3.2. Circuit Interrupters

To meet the selection requirements for electrical equipment, the rated current of the circuit interrupter for the 0.6 kV side must be greater than the maximum continuous operating current I_max:

$$I_{\mathrm{a}\mathrm{l}}\ge I_{\mathrm{m}\mathrm{a}\mathrm{x}}=1.05\times {\displaystyle \frac{S_{\mathrm{N}}}{\sqrt{3}\times U_{\mathrm{N}}}}=1.05\times {\displaystyle \frac{3125}{\sqrt{3}\times 0.6}}=3157.38\mathrm{A},$$

(7)

After comprehensively considering that the rated voltage and rated current under normal operating conditions both meet the requirements, the DW45-4000 type circuit interrupter was selected. The detailed parameters are shown in

Table 8. Main parameters of circuit interrupters.

Name	0.6 kV Side	110 kV Side
Model	DW45-4000	LW36-126
Rated Voltage	0.69 kV	126 kV
Rated Current	4000 A	2000 A
Rated Breaking Current	31.5 kA	31.5 kA
Rated Making Current	100 kA	100 kA
Dynamic Stability Current	100 kA	100 kA
Thermal Stability Current (4 s)	31.5 kA	31.5 kA

. After verification, the dynamic stability current of the DW45-4000 type circuit interrupter (100 kA) was greater than the impact short-circuit current value (2.32 kA), and the thermal stability current (4 s) (31.5 kA) was greater than the maximum effective value of the full current (1.37 kA), thus meeting the dynamic and thermal stability requirements.

Similarly, the LW36-126 type circuit interrupter for the 110 kV side was selected, and other detailed parameters are shown in Table 8.

4.3.3. Disconnector Switches

The maximum continuous operating current I_max for the 0.6 kV side was 3157.38 A. The model selected for this project was the MGR3-4000, the specific parameters of which are shown in

Table 9. Main parameters of disconnector switches.

Name	0.6 kV Side	110 kV Side
Model	MGR3-4000	GW4-110D
Rated Voltage	0.69 kV	126 kV
Rated Current	4000 A	1250 A
Dynamic Stability Current	50 kA	100 kA
Thermal Stability Current (4 s)	50 kA	31.5 kA
Rated lightning impulse withstand voltage (peak value, to ground)	/	450 kV
Rated power frequency withstand voltage (dry test, 1 min)	/	185 kV
Rated power frequency withstand voltage (wet test, 1 min)	/	140 kV
Creepage distance of external insulation (outdoor)	/	240 cm

. After verification, the dynamic stability current of the MGR3-4000 type disconnector switches (50 kA) was greater than the impact short-circuit current value (2.32 kA), and the thermal stability current (4 s) (50 kA) was greater than the maximum effective value of the full current (1.37 kA), meeting the dynamic and thermal stability requirements.

Similarly, the model selected for the 110 kV side was the GW4-110D, and the specific parameters are shown in Table 9.

In the selection of disconnectors, the verification of the dielectric margin is a core step to ensure that the insulation performance of the equipment remains reliable under various voltage stresses. The rated voltage of the disconnectors on the 110 kV side is 110 kV. In accordance with DL/T 620-1997 [35], the standard value of the maximum operating voltage (U_max) for the 110 kV system is 126 kV.

The dielectric margin shall meet the requirement of $K={\displaystyle \frac{Systemovervoltagelevel}{Equipmentinsulationlevel}}\ge 1.25$, (as specified in DL/T 620-1997).

• Verification of lightning overvoltage: The lightning overvoltage in the 110 kV system is limited by the arrester. The lightning impulse residual voltage of the 110 kV zinc oxide arrester (under 10 kA) is approximately 260 kV (peak value). The lightning impulse withstand voltage of the GW4-110D is 450 kV. Thus the insulation margin K = 450/260 ≈ 1.73 ≥ 1.25, which meets the requirement.
• Verification of switching overvoltage: The switching overvoltage multiple of the 110 kV system (relative to the maximum operating voltage) is generally ≤ 2.0, that is, the peak value of switching overvoltage U_op = 2.0 × 126 × √2 ≈ 356 kV. The lightning impulse withstand voltage of 450 kV for the GW4-110D already covers the switching overvoltage (there is no need to separately specify the switching impulse withstand voltage), so the margin is sufficient.
• Verification of power frequency overvoltage: The system power frequency overvoltage (relative to the maximum operating voltage) is generally ≤ 1.3, that is, U_pf = 1.3 × 126 = 163.8 kV (effective value). The power frequency dry withstand voltage of the GW4-110D is 185 kV, and the insulation margin K = 185/163.8 ≈ 1.13. Although slightly lower, it meets the “short-time withstand” requirement. The duration of power frequency overvoltage is short, and 185 kV is much higher than the maximum operating voltage of 126 kV.

The core of the external insulation margin lies in the creepage distance meeting the requirement of a creepage ratio distance corresponding to the pollution level. The creepage ratio distance $\mathsf{\lambda }={\displaystyle \frac{Maximum operating voltage }{Creepage distance }}={\displaystyle \frac{U_m}{L}}$ (unit: cm/kV).

The Yulin area is dominated by coal and chemical industries, with relatively serious industrial pollution. Based on operational experience and GB/T 26218.2-2010 [36], it is tentatively classified as pollution class II. According to GB/T 16434, the required creepage ratio distance for 110 kV outdoor equipment is λ ≥ 1.6 cm/kV. Therefore, the minimum required creepage distance is L_min = 1.6 × 126 = 201.6 cm. The typical creepage distance of the GW4-110D is 240 cm, which is greater than 201.6 cm, meeting the requirements for pollution level II.

According to GB/T 11022-2020 [37], when the altitude is higher than 1000 m, the external insulation strength decreases by 1% for every 100 m increase in altitude. The average altitude of Hengshan County is approximately 1000–1200 m, and the average value of 1100 m was selected. Therefore, altitude correction is required.

The withstand voltage correction factor K_h = 1/(1 − 1 × 1%) = 1.01.

The corrected required lightning impulse withstand voltage is 450/1.01 ≈ 445.5 kV, which is higher than the system lightning overvoltage of 260 kV, indicating a sufficient margin.

Therefore, under the conditions that the Yulin area is assumed to be of pollution level II and the altitude is ≤1500 m, the lightning impulse and power frequency withstand voltage margins of the GW4-110D disconnector meet the requirements, the external insulation creepage distance is suitable for pollution level II, and the model passes the insulation margin verification.

4.3.4. Current Transformers

The maximum continuous operating current I_max for the 0.6 kV side was 3157.38 A. The current transformer model selected was LDZJ-10-4000/5, and the specific parameters are shown in

Table 10. Main parameters of current transformers.

Name	0.6 kV Side	110 kV Side
Model	LDZJ-10-4000/5	LB7-110-500/5
Rated Voltage	0.69 kV	126 kV
Rated Primary Current	4000 A	500 A
Secondary Current	5 A	5 A
Accuracy Class	0.5/10P20	0.2/10P20
Dynamic Stability Current	100 kA	100 kA
Thermal Stability Current (4 s)	31.5 kA	31.5 kA

. After verification, the dynamic stability current of the LDZJ-10-4000/5 type current transformer (100 kA) was greater than the impact short-circuit current value (2.32 kA), and the thermal stability current (4 s) (31.5 kA) was greater than the maximum effective value of the full current (1.37 kA), meeting the dynamic and thermal stability checks.

Similarly, the current transformer model selected for the 110 kV side was LB7-110-500/5, and the specific parameters are shown in Table 10.

4.3.5. Voltage Transformers

When selecting a voltage transformer, the rated voltage must first be satisfied; that is, $U_{\mathrm{N}}\ge U_{\mathrm{N}\mathrm{s}}=10\mathrm{k}\mathrm{V}$. The model TYD110/√3-0.02H was selected, and the specific parameters are shown in

Table 11. Main parameters of TYD110/√3-0.02H voltage transformer.

Name	Parameters
Rated Voltage	126 kV
Rated Primary Current	500 A
Accuracy Class of Secondary Winding	0.2
Protection Winding	10P20
Dynamic Stability Current	100 kA
Thermal Stability Current (4 s)	31.5 kA

.

5. Conclusions

This study focused on the design of a 50 MW agrivoltaic power station in Shaanxi Province, China, which integrates PV power generation with agricultural production. An integrated model of PV modules and agricultural greenhouses was adopted, with PV modules installed on the sunny slope roof of the greenhouse at an inclination angle of 40°. Bifacial modules with high light transmittance were selected to facilitate the growth of greenhouse crops. To address the shading problem caused by PV modules, the modules were arranged in quasi-chessboard (coverage < 30%), chessboard (30% < coverage < 50%), and long-row (50% < coverage < 80%) patterns, according to the growth characteristics of different crops. Among these patterns, the checkerboard pattern features the highest system efficiency, the smallest irradiance loss, and a slight lead in power generation, with a moderate shading rate. The expected energy gain from the bifacial modules’ rear side in this specific setup is 7.6%. Centralized inverters and DC combiner boxes were configured. The PVsyst simulation results showed that the system power generation is 79,796 MWh/year, the annual specific power generation is 1585.06 kWh/kWp/yr, and the PR is 84.65%, which is in agreement with actual measurement data from similar projects, proving the feasibility of the design’s implementation.

The primary electrical design includes the completion of the main wiring design, main transformer selection, short-circuit current calculation, and selection of major electrical equipment. The single-busbar non-segmented mode was adopted, and the SFSZ11-63000/110 main transformer steps up the power output by the inverter to 110 kV for double-circuit connection to the power grid, meeting the power transmission requirements. This design rationally utilizes the roof space of greenhouses, meets grid-connected requirements, provides technical reference for the construction of agrivoltaic power systems, and has the potential to be widely adopted.

However, this study still has some shortcomings in terms of agricultural adaptability and systematic technical integration. Although various layout schemes take into account the differences in light requirements of different crops, they lack quantitative analysis of key parameters such as the light compensation and saturation points. Research on the spatial heterogeneity of the temperature, humidity and wind speed fields formed in different layout modes remains insufficient, making it impossible to accurately predict the microenvironmental characteristics of different crop growing areas. The design lacks a dynamic tracking system with a crop alignment algorithm, which limits the potential of synergistic efficiency improvement between agriculture and photovoltaics.

Future research should focus on synergistic efficiency improvement between agriculture and PV power generation. Key efforts should be made to break through the crop–light environment feedback mechanism, establish long-term microenvironment observation, develop bifacial module layout algorithms based on crop spectral response, expand long-term economic benefit evaluation based on life-cycle cost, and promote the qualitative leap of agrivoltaic systems from a “mechanical superposition” to “organic integration”.