Performance evaluation and comparative study of three 52-kW PV plants in India: a case study

Introduction

Solar energy has grown to be among the most popular sources of clean energy in recent years across many industries, and numerous studies are being conducted to improve its application and benefits. A continent like Asia has a higher potential for power generation from solar energy, as depicted in Figure 1(a). In this continent, developing country like India has vast potential, and their demand also increases with a population of nearly 140 crores. This power demand must be met through renewable power sources due to fossil fuel depletion. Since the country has higher global solar radiation, depicted in Figure 1(b), the country has set a target of 300 GW for solar energy by 2030. However, the country has already reached its installed capacity of 63 GW in March 2023, and the total installed capacity of renewable sources is depicted in Figure 1(c). Hence it is necessary to concentrate on this to reach the set target. Recently, many researchers have been concentrating on research based on several configurations in which solar photovoltaic (PV) systems can be installed, grid-connected PV and standalone PV, which may be designated as off-grid systems.¹ However, the installation capacity of both differs significantly as, through the years, it has been observed that a grid-connected PV system is much more developed than an off-grid system. There is another topology which is known as a Hybrid solar energy system. It can charge the system from the grid and solar PV directly, but these are expensive and are not usually preferred. Several studies are being conducted to improve its application in daily life and determine how its potential applications might be broadened. It has been observed that solar thermal collectors are utilized to turn energy into heat while also generating electricity with panels.² Some problems include variations in output energy due to changes in irradiance level.³ As we know, PV modules are made from silicon cells, thus limiting their efficiency to significantly less. Therefore, it is essential to increase their efficiency so that more people can be ready to invest in this.⁴ Solar insolation determines the sustainability and dependability of PV-based power generation systems; hence optimization is crucial to satisfy load demand.⁵ These factors are crucial during system installation because the performance is also influenced by the environment, location, and plant varieties.⁶

Figure 1. (a) Global solar radiation (GHI) of Asia; (b) Global solar radiation (GHI) of India.²⁴ Courtesy: Solargis maps; (c) Renewable energy (RE) share in India.

According to the IRENA report, even though many technologies exist, most governments concentrate on solar and wind, and major investments were made in PV and wind technology.⁷ The government has recently initiated a series of efforts to deploy roof panels over various offices and organizations, which can assist in addressing the situation.^8,9 Sometimes, PV degradation may also lead to variation in series, shunt resistances, and decreased output power.¹⁰ It can also diminish the impact of greenhouse gases brought by fossil fuels. PV energy is more affordable than any other sustainable energy source, and research has shown that it is incredibly profitable in rural regions.^11,12 At every stage of the global solar PV supply chain, China is currently by far the largest supplier; with a manufacturing capacity for PV modules of around 340 GW/year, it has more than twice the installed PV modules worldwide, with a manufacturing capacity utilization rate for solar components of between 40 and 50% in 2021. China directly supplies all markets, except North America, where import taxes on Chinese solar PV components have been imposed. However, Chinese businesses have been actively investing in production capacity in Southeast Asia to supply the area and export to the United States.¹³ Many studies suggest the appropriate areas for implementing PV systems, but they could be more extensive. The utilized parameters and sites discussed in this research are identified as an outcome of the literature review, and their applicability is noted in Table 1.

One of the first R analysis types thoroughly explored and applied realistically in many situations is linear regression. Creating a mathematical model that may be used to forecast one variable, known as the dependent variable, can be characterized using another variable, the independent variable. The degree of the linear relationship between two variables is measured using correlation analysis. This is because models with a linear dependence on their unknown parameters are more readily fitting than models with non-linear dependence because it is simpler to identify the data samples of the resulting estimators.²⁴

On the other hand, a significant amount of data must be managed, so the regression model is useful.²⁵ It is frequently used to predict time-series and regression models using conventional estimate approaches, which involve consideration of the predictor variables, the target variable, and their relationship.²⁶ This study compares the two models’ abilities to accurately forecast PV module performance: linear and non-linear regression models. A logarithmic linearized equivalent model serves as the mathematical representation of the non-linear model. In this paper, the site which has been selected is based on SRM Institute of Science and Technology in Kattankulathur, Chennai City, in Tamil Nadu, India. Many studies were conducted in our literature to investigate the behavior during one year, from January 2020 to December 2020. Performance parameters like global horizontal irradiation, energy yield, and capacity factor have been calculated. The power plants are installed on the rooftop of the Mechanical C Block, Civil Engineering Block, and Science & Humanities Block of SRM Institute of Science and Technology (SRMIST), Kattankulathur, 603203. In the paper, the description of installed PV systems and site details are discussed.

Further, the simulation of the grid and the calculated results are shown through tables and graphs. Towards the end of the paper, the economic factor and environmental impacts are discussed. Global solar radiation (GHI) of Asia and global solar radiation (GHI) of India are illustrated in Figure 1. These global solar radiation maps are downloaded from Solargis, where several collections of solar resource maps are available for research purposes.²⁷ Various case studies on the PV performance are accomplished by researchers by considering certain locations,^28,29 based on techno-economic assessment,^30,31 feasibility studies,^32,33 comparative studies,³⁴ optimizing the performance,³⁵ solar insolation studies and estimations^36–38 and several other factors.^39–46

Methods

This section discusses the methodology followed to make this case study. Initially, the linear regression model is obtained using solar radiation, temperature, and wind speed data from NREL (National Renewable Energy Laboratory). Then, regression equations are obtained from this prediction model for further study. This statistical analysis will give the correlation among the control factors and its significance. The results from the prediction model will be compared with the running 52-kW plant installed in the institution. Finally, a complete description of all three 52-kW grid-connected PV systems is presented with the photographs, satellite map, and the specification of BoS (Balance of Solar PV system).

After that, a comparative study is performed on all three 52-kW grid-connected PV systems with respect to energy yield, performance ratio, capacity utilization factor, C_O2, and diesel saved. The procedure followed for this comparative analysis is presented in detail in a separate section. This 52-kW plant is simulated in PVsyst V7.1.8 simulation software and the results obtained are compared with the real-time data for 2020. This comparison will help us observe the performance of all three 52-kW PV systems. Finally, the inferences from the study are observed and listed for the conclusion. The flowchart of the methodology followed for the study is shown in Figure 2.

Figure 2. Methodology followed for the study.

Description of three 52-kW grid-connected PV system

The major components of the grid-connected PV system are a solar array, inverter with maximum power point tracking (MPPT), AC and DC disconnect, and other protective and connective equipment to the grid. It is more effective than a standalone PV system because it eliminates the losses incurred in energy storage. Another significant advantage of the grid-connected system is the eradication of the problem incurred due to the presence of batteries, i.e., cost and replacement. The general schematic diagram of all three 52-kW PV systems is represented in Figure 3.

Figure 3. 52-kW Solar PV plant.

Site location

All three 52-kW solar power plants are located at SRMIST with latitude and longitude of 12.8231° N, 80.0442° E, and elevation above the sea level of 51 m. Since the generated PV power significantly depends on the sun’s position and its radiation intensity, the institute studied solar radiation for one year and opted for these three locations on the campus.

Figure 4(a)-(c) presents the description of all the sites taken for the study. In addition, satellite map images and photographs of the Mechanical ‘C’ block, Science and Humanities, and Chemical Engineering block are depicted in Figure 4(a), (b), and (c), respectively.

Figure 4. Site description (a) Mechanical 'C' block; (b) Science and Humanities; (c) Chemical Engineering block.

Inverter specification

A 50 kVA inverter converts the DC power to AC power. The range of DC and AC voltage of the inverter are 200-1000 V and 320-480 V, respectively. The inverter’s efficiency is 98.60%, and the total input current is 100 A. The total harmonic distortion is less than 3%, with a 45-55 Hz frequency range. It has an inbuilt disconnect switch.

To observe the DC and AC voltage, current, and power of the plant, a few graphs are presented in Figures 5 and 6. Figure 5(a)-(f) presents the AC and DC voltage and current of all the 52-kW power plants. This observation is drawn from June 15^th, 2020. Similarly, the AC power output of three solar PV plants was observed on October 22^nd, 2020; these graphs are presented in Figure 6. Figure 3(a)-(c) presents the AC power output of the 52-kW plant of the Faculty of Science and Humanities, Chemical Engineering lab, and mechanical ‘C’ block, respectively.

Figure 5. Voltage and current on June 15^th, 2020 (a) Faculty of science and humanities: AC (b) Faculty of science and humanities: DC (c) Chemical Engineering lab: AC (d) Chemical Engineering lab: DC (e) Mechanical ‘C’ block: DC (f) Mechanical ‘C’ block: DC.

Figure 6. AC power on 22^nd October 2020 (a) Faculty of science and humanities; (b) Chemical Engineering lab; (c) Mechanical ‘C’ block.

Regression analysis for the selected site

The chosen site data was gathered on an hourly basis. The selected location receives 9 hours of solar radiation every day on average. For prediction, the AC hourly produced energy, direct beam and diffused radiation, ambient temperature, and wind velocity of the chosen site were considered. The regression model for the creation of AC energy outputs was developed using average hourly data at the location SRMIST, Kattankulathur, Tamil Nadu, India. From the prediction model, the regression equation was derived. Equation ((1), which has a linear relationship with beam radiation, diffused radiation, temperature, and wind speed, is used to estimate AC power from PV panels.

A regression model is a statistical method for determining the connection between the control variables. It is essential to check for residual plots before developing a regression equation to ensure linear regression. The statistical analysis and generation of the regression model for the system under consideration were carried out using Minitab software version 16.2.1.

Where ρ - AC energy output (kWh), α - beam irradiance (W/m²), β - diffuse irradiance (W/m²), γ - ambient temperature (C), and μ - wind speed (m/s).

A graphical method of analyzing residuals is essential to assess for a ‘good fit’ regression model the best fit of a set of data in a regression line. The residual plots reveal the degree of correlation between the variables and the projected results. The figure shows the residual plots of AC energy yield. As demonstrated in the figure, the points are spaced linearly straight. It indicates that the projected and actual values are more closely related, which is denoted as a normal probability plot.

Figure 7 shows a comparison between residual and anticipated values. Both appear to be the most similar to one another, so there is very little difference between them. The histogram plot of AC energy is shown in Figure 7. In the histogram graph, clear data regarding the residuals are shown. The figure demonstrates the residuals vs. trial run order. Both positive and negative residual values are present, indicating the existence of certain relationships. The models show promise for adequacy due to the thorough study of AC residual plots.

Figure 7. Residual plot of AC energy.

The R² (Coefficient Determination) value of the generated regression model is higher, indicating appropriate accuracy. For AC energy, the R² value achieved is 96.39 percent. The corrected R² (R_adj) value is 96.95 percent, indicating that the generated regression model is very significant. In addition, the R² (R_pred) value obtained is 93.54 percent. Figure 8 depicts the influence of irradiance and temperature on the AC energy produced. At maximum beam irradiance and temperature median, the maximum array energy production is seen in Figure 8.

Figure 8. Influence of (a) Beam Irradiance and Temperature, (b) Diffuse Irradiance and Temperature, (c) Beam Irradiance and Wind Speed, (d) Diffuse Irradiance and Wind Speed.

Increased temperature may result in a drop in production. The figure shows a high array output at the median of diffused radiation and temperature. It has been discovered that for the installed plant to produce more power, the temperature must be between low and high. The figure depicts the influence of irradiance and wind speed on the AC energy produced. The illustration depicts the effects of wind speed and beam irradiance on AC output shown in the image. For a high Voltage AC output, a full beam irradiance and a medium wind speed are required.

In the illustration, the median of the graph yields the highest output. As a result, it is found that maximum beam irradiance, medium dispersed radiation, temperature, and wind speed are the finest examples of high production yields.

Procedure taken for the analysis

These 52-kW plants are analyzed, and their performance is studied by dividing the study into three stages.

First stage: Retrieving data from the online (DelREMO) monitoring system of all three plants. The plant location and its structure are also studied thoroughly.

Second stage: The key metrics like yield ratio, performance ratio, and capacity utilization factor of the plants are analyzed and compared.

Third stage: Finally, the energy yield of the plants is compared with the result obtained from the modelling software PVsyst 7.1.8. The loss diagram of the plant is obtained and discussed.

There are specific performance parameters like reference yield, array yield, final yield, performance ratio and capacity utilization factor to determine the overall system’s performance. International Energy Agency has developed certain performance parameters for evaluating and analyzing the performance of grid-connected PV systems.⁹ DelREMO online monitoring system is shown in Figure 9. Comparison of all three 52-kW PV plant is illustrated in Figure 10. Performance parameters of the 52-kW power plant at Mechanical ‘C’ block, chemical engineering block, Faculty of Science and Humanities is listed in Tables 2, 3 and 4, respectively. A comparison of key highlights of 52-kW power plants at the institute is given in Table 5.

Figure 9. DelREMO online monitoring system (a) Home screen taken on June 7^th 2021; (b) AC energy yield display on August 1^st, 2020, for Mechanical 'C' block 52 kW plant; (c) Comparative dashboard of all the three 52 kW PV plant.

Figure 10. Comparison of all the three 52-kW PV plant (a) Performance ratio (PR); (b) Energy yield; (c) Average capacity utilization factor and PR.

Simulation using PVsyst

The maximum energy generated is in the month of March (6627 kWh), and the minimum is generated during July (4428kWh). The total energy produced during that year was 64606 kWh.

Normalized production

Figure 11 shows the L_c value recorded as 0.45kWh/kWp/day and the L_a value as 0.16kW/kWp/day. Similarly, Y_F is recorded as 4.55kWh/kWp/day.

Figure 11. Normalized production of 52-kW PV plant.

Loss diagram

The global horizontal irradiance is 1927 kWh/m²/y, as shown in Figure 12. The effective irradiation on the collector plane is 1842 kWh/m²/y. After the PV conversion, the nominal array energy is 5559 kWh. The efficiency of the PV array is 15.46% at STC, while the virtual energy is 4726 kWh. The energy at the output after having the inverter losses comes out to be 4524 kWh.

Figure 12. Loss diagram of 52-kW PV plant.

Conclusions

A performance study of three 52-kWgrid connected solar photovoltaic power plants installed on the Mechanical Engineering Block, Chemical Engineering Block, and Science & Humanities Block of rooftop SRMIST Kattankulathur was evaluated on an annual basis. The observations drawn from this case study are: