Comparison of Solar Panel Models for Grid Integrations Studies E. M. da S. Brito, A. F. Cupertino, L. P. Carlette, D. O. Filho, H. A. Pereira Gerência de Especialistas em Sistemas Elétricos de Potência Universidade Federal de Viçosa, Viçosa, Brasil erick.brito@ufv.br, allan.cupertino@yahoo.com.br, luan.carlette@ufv.br, delly@ufv.br, heverton.pereira@ufv.br Abstract— Photovoltaic systems are highly dependent on climatic conditions in which they are submitted. The incident solar irradiance and temperature are the main factors impacting on the power generated by a solar panel. This paper presents three different models of a solar panel and compare, through simulations, their accuracies and efficiencies, and also shows the advantages and applications of each model. Simulations for each model connected to the grid were also made through a controlled inverter. This inverter keeps the voltage at the terminals of the panel in a constant value equal to its maximum power point, provided by the manufacturer. In the end, it is possible to see that two of the presented models have almost the same behavior while the third one has some discrepancy. I. P. F. Ribeiro Technische Universiteit Eindhoven Department of Electrical Engineering, 5600 MB Eindhoven, The Netherlands pfribeiro@ieee.org Photovoltaic panels have an intermediary behave between a current and a voltage source. Moreover, variations in the incident solar irradiance and temperature have a great impact on the generated power as illustrated in Figure 1. The modeling of photovoltaic panels can be considered a multi physical modeling because it considers the temperature at the panel, the radiation incident on it, and its electric power generation. In this domain, it becomes necessary to make the study of each subset, considering their effects in cascade that influence on each other [3], [4]. INTRODUCTION The technological progress has changed the lifestyle of the modern human and the energy demand prompted motivates new researches for renewable energy sources. Among these are solar and wind power systems [1]. It has been happened a significant growth in the installed power on solar photovoltaic systems. In 2011 the value was of 69.68 GWp, a growth of 76 % in relation to the previous year. More than 95% of this power is generated by grid-connected systems [2]. (a) Although its strong growth, renewable sources does not produce a significant amount of energy compared to the world demand. This is due to lack of tax incentives, making these sources less competitive in the market. Many countries do not have specific legislation for photovoltaic systems connected to the grid, for instance. Nevertheless, expectations and forecasts for the future indicate another reality. Even though the sun represents an almost unlimited source of clean energy, an obstacle to develop the photovoltaic (PV) panels’ technology is low efficiency, compared to hydroelectric and nuclear plants. The energy production is further reduced on cloudy days or in shaded situations and it is even worse in the night time when there is no generation. The authors would like to thank to CNPQ, FAPEMIG and CAPES for their assistance and financial support. (b) Figure 1. Effect of incident irradiance (a) and temperature (b) in I x V curves of a solar panel. Some parameters used in the modeling are informed by manufactures and they are in TABLE I. TABLE I. MAIN PARAMETERS OF A SOLAR PANEL Symbol Maximum Power (W) Maximum Power Voltage (V) Maximum power current (A) Short circuit current (A) Temperature coefficient of (A/K) Temperature coefficient of (V/K) II. METHODOLOGY A. Model 1 – Equivalent Circuit There is in literature a simplified model of PV used in some works [5], shown in Figure 2. The resistances in series and in parallel represent the voltage drop when the charge migrates from the electrical contacts and the reverse leakage current of the diode, respectively. is the open circuit voltage and is a constant continuous current source. The parameters of the model are calculated according to the electrical characteristics of the photovoltaic panel. This model has some divergences compared to the characteristic curves provided by manufacturers. The temperature is not a parameter. Consequently the PV curve does not change when variations on temperature happen. = = − − = Figure 2. Equivalent circuit of a photovoltaic panel. − 1# − = % + ∆'( Where , and are calculated by (1), (2) and (3): ! + The variable is calculated by (5): Open circuit voltage (V) = − Parameter B. Model 2 – Mathematical Model This model is based on equations of photovoltaic panels. Through equations it was possible to simulate the solar panel. The equation of the current in the solar panel is: (1) (2) (3) (4) (5) Where is the current in the nominal conditions, calculated by equation (6); ∆' = ' − ') (T is the solar panel temperature and T+ is the nominal solar panel temperature); e are the values of incident solar irradiance and the reference irradiance (W/m²), respectively. The variable 0 is the temperature coefficient of the short circuit current (A/K). = + (6) The reverse leakage current in the diode, I is: = + ∆' %3 678 ∆9( : 2 45 3 ; ! −1 (7) is the nominal short-circuit current, is the nominal open circuit voltage and is the coefficient of the open circuit voltage (V/K). The variable <is the ideality constant of the diode, contained in the range1 ≤ < ≤ 1.5. is strongly dependent on temperature and equation (7) is an alternative way to express this dependency showing a linear variation in open circuit voltage due to . This equation also simplifies the model canceling the errors around the open circuit voltage points and consequently in other points of the IxV curve. Finally, @ is calculated by: @ = A' (8) Where A is the Boltzmann’s constant, ' is the temperature of the panel (K) and is the electron charge. In [6] and [7] it is proposed an algorithm for adjusting and . The method is based on the fact that there is an only pair { , } in which the maximum power calculated by the IV model E is equal to the maximum experimental power from the datasheet . Using E = in equation (4), it will be obtained [6]: = % + ( − F EG EG # ! − 1HH − (9) The interactive process is shown in Figure 3. The initial values of and are [6]: EJ = 0 − P I EJ = − − (10) case, and calculated by (12 12) and (13), where default value. ' L QRS = N QRS = L,) # ') L,) = ^_`a exp b + ∆' exp W C. Model 3 – Solar Cell Multi Physical Model The multi-physics physics modeling represents the influence of various phenomena in which a real system is subjected. There is in Matlab 7.10.0/Simulink®, in Simscape library, a multi physical model of a solar cell. This model is defined by 16 parameters including some temperature coefficients and the reverse leakage current in both diodes, for example. This model also contains two diodes which better represent the non nonlinear characteristic of the cell. Moreover, it is possible do observe the change in the series and nd parallel resistances with temperature. The equation of the current in the solar panel is: = − L 2 −N 2 ,5MM ! ,5MM ! − 1; ;− − 1; − + , OO , OO (11) , OO and , OO are the series and parallel resistance of each cell, respectively. is calculated by (5), ( L and N are the reverse leakage current in each diode that are equal, in this = 3, as a cde 1 1 − #f <A ') ' :< g − 1 @ ' 9h\L , OO QRS = , OO # ') Figure 3. Algorithm of the method used to adjust the I x V model[6]. 9Z[\L (12) (13) (14) ' 9hiL , OO QRS = , OO # ') (15) Y, OO , OO = 2 ; Y, OO (16) 'X1 and '1 are the exponent that represent the influence of temperature on resistance values values. In this work, and are constant, independent of temperature, so 'X1 and '1are zero. The resistance values of the panel were estimated by an alternative ative method based on [6]. The resistances for each cell can be obtained by: Y, OO , OO = 2 ; Y, OO (17) Where Y, OO and Y, OO OO are the number of solar cells in series and in parallel of the panel, panel respectively. Comparing equations (7) ( and (13) it is possible to see the main difference between Model 2 and 3. In Model 2 there is one more coefficient for temperature influence, . D. Connecting the Models to the Grid Figure 4 shows a schematic of the grid connected photovoltaic system through a controlled control inverter. The control of the inverter keeps the voltage at the panel’s terminal in a Figure 4. Simulated grid-connected photovoltaic system. fixed value equal to the maximum power point (provided by the manufacturer) and it injects the generated power in the grid with the power factor close to the unity. The controlled variables were written in direct and quadrature axis. The control loops of the inverter are shown in Figure 5, there is the reactive power loop that controls the power factor and a loop for regulate the DC bus voltage. The current control loops use proportional controllers and the external loops use proportional-integral controllers [8]. The controller’s gains were adjusted by the poles allocation method. A Synchronous Reference Frame – PLL (SRF-PLL) circuit is used as synchronism technique to connect the system to the grid. This structure estimates the grid voltage angle for the control of the inverter. To reduce the harmonics generated by IGBT’s switching a LCL filter was used. The design of this component is in [9]. TABLE II. KYOCERA SM-48KSM PARAMETERS FOR 1000 W/M² AND 25 ºC. Parameter TABLE III. Value 48l 18.6 2.59p 22.1 2.89p −0.070/ 1.66rp/ *STC: AM1.5 spectrum KYOCERA SM-48KSM PARAMETERS FOR 800 W/M² AND 47 ºC. Parameter Value 34l 17.1 2.02p 20.5 2.26p −0.070/ 1.66rp/ *STC: AM1.5 spectrum III. Figure 5. Control loops of the inverter. RESULTS In order to validate the three mentioned models, it was used the Kyocera panel, model SM-48KSM whose parameters are shown in TABLE II. The curves for each model were collected for different situations, varying the climate parameters such as temperature and radiation in order to verify the efficiency and accuracy of each one. Using the data shown in TABLE II, TABLE III and the estimation method from [6], the models were simulated for the nominal operating conditions. The final values for the resistance for Model 2 are shown in (18) and for Model 3 it is shown in (19). The different values are justified because the reverse leakage current of each model is calculated in a different way, as can be seen in (7) and (12). = 0.186200Ω s = 108.602693Ω P 3 (18) = 0.090600Ω s = 92.939664Ω P 2.5 (19) TABLE IV and TABLE V show the relative errors between the parameters obtained by the graphics and the provided by the manufacturer, for 1000 W/m² and 25°C. Some manufactures also indicate the panels’ data for other values of radiation and the most common is 800 W/m² and 47ºC. Thus, simulations were made to these new values and the results are shown in Figure 7. TABLE IV. ERRORS (%) ON THE PARAMETERS OF EACH MODEL, COMPARED WITH MANUFACTURES DATA - THE SITUATION OF 1000 W/M² AND 25 ºC. 1.5 1 Model 1 Model 2 Model 3 0.5 0 0 5 10 15 Voltage(V) 20 25 20 25 (a) 55 50 Model 1 Model 2 Model 3 45 40 35 Power(W) Analyzing Figure 6, Model 1 follows a solar panel behavior, but not its accuracy. It presented a different shape of the curve, more linear, compared to the other ones, showing a wide divergence from reality. Furthermore, it was observed that for different values of irradiance the open circuit voltage in Model 1 was constant. Despite being the less accurate, Model 1 has the fastest simulation and a small number of equations. Model 3’s simulation is slower than Model 2 and it has a satisfactory performance when compared to the Model 1, since its waveform was more accurate and its open circuit voltage decreased with a reduction of incident radiation. Besides, Model 3 presents the characteristics of each solar cell, what is an advantage. Current(A) 2 30 25 20 15 10 Model 1 Model 2 Model 3 4.2 0 0 ∆Pm(%) 3.6 0.3 0.5 ∆Voc (%) 1.9 0 0.4 ∆Isc (%) 4.2 0 0.8 ∆Vmp (%) 0.4 0 0.8 TABLE V. ERRORS (%) ON THE PARAMETERS OF EACH MODEL, COMPARED WITH MANUFACTURES DATA – THE SITUATION OF 800 W/M² AND 47 ºC. Models Model 1 Model 2 Model 3 18.6 0.6 0.5 ∆Pm(%) 11.7 0.6 0.8 ∆Voc (%) 0.4 3.5 3.1 ∆Isc (%) 17.8 0.5 1.3 ∆Vmp (%) 5 ∆Imp (%) 1.0 1.0 1.5 0 0 5 10 15 Voltage(V) (b) Figure 6. Curves I x V (a) and P x V (b) of the three panels for an irradiance of 1000 W/m² and temperature 25 ºC. 3 ∆Imp (%) 2.5 2 Current(A) Models 1.5 Model 1 Model 2 Model 3 1 0.5 0 0 5 10 15 Voltage(V) (a) 20 25 3.5 50 Model 1 Model 2 Model 3 2.5 Current(A) Power(W) 40 3 30 2 1.5 20 1 10 0 Model 2 Model 3 0.5 0 0 5 10 15 Voltage(V) 20 25 0 5 10 15 Voltage(V) 20 25 (b) Figure 7. Curves I x V (a) and P x V (b) of the three panels for a irradiance of 800 W/m² and temperature 47 ºC Besides irradiance another very influential climatic factor in the efficiency of photovoltaic panels is the temperature. In this context, the Model 1 is not able to simulate changes in temperature, showing a major limitation in its applications. In order to compare I x V and P x V curves of the Models 2 and 3 for different values of temperature, simulations were made for 25 ° C and 75 ° C. Figure 8. I x V curves for 25 ºC (a) and for 75 ºC (b) of the Models 2 and 3, with radiation of 1000 W/m². 50 45 Model 2 Model 3 40 35 Power(W) (b) It was noticed that both of the curves follow the same behavior, it is hard to see the two curves due to their proximity. Although the manufacturer does not provide data for different temperature values, it can be observed in Figure 8 that the temperature variation causes a slight increase in the shortcircuit current and it reduces the open circuit voltage. 30 25 20 15 10 5 0 0 5 10 15 Voltage(V) 20 25 20 25 (a) 3 45 Model 2 Model 3 40 2.5 35 30 Power(W) Current(A) 2 1.5 1 25 20 15 Model 2 Model 3 0.5 10 5 0 0 5 10 15 Voltage(V) (a) 20 25 0 0 5 10 15 Voltage(V) (b) Figure 9. P x V curves for 25 ºC (a) and for 75 ºC (b) of the Models 2 and 3, with irradiance of 1000 W/m². As a comparison criterion, it was simulated panels connected to the grid. A controlled inverter maintains the output voltage of the panels in a constant value. Figure 10(a) and (b) shows the generated power and voltage in the panel. The slight difference in the injected power is justified by Figure 11, where the given by the manufacture is highlighted and it is possible to see that the for Models 2 and 3 is almost the same in this point, but in Model 1 the is a little higher. Injected Power(W) 50 40 20 Model 1 Model 2 Model 3 10 0 0 0.02 0.04 0.06 time(s) 0.08 20 10 0 Voltage(V) 20 25 Figure 11. P x V curves of the three panels for a irradiance of 1000 W/m² and temperature 25 ºC, highlighting the voltage in the maximum power point. Another analyzed characteristic of the model is the simulation time. Using a personal computer with an Intel CPU of 2.1GHz processor speed and 3GB of RAM, the time to simulate the grid-connected system, for each model, is shown in TABLE VI. It is possible to see that despite Model 2 and 3 are very similar, Model 3 takes much more time to simulate than Model 2. COMPARISON BETWEEN SIMULATION TIME FOR THE MODELS CONNECTED TO THE GRID Models Simulation time (s) Model 1 Model 2 Model 3 19.0 22.55 1140.35 CONCLUSIONS The advantages of high simplicity of the Model 1 and the study of individual cells in Model 3 should not be disregarded. Besides electrical and thermal characteristics, the simulation time is a critical factor since Model 3 might become impracticable using a lot of solar cells. There will be situations where each of these models will be very important in the development of works in PV studies and it is up to the researcher to decide which one fits better. 19 0.02 10 15 Voltage(V) The results show that the mathematical model (Model 2) is more accurate if compared to the equivalent circuit model (Model 1) and to the panel made by individual cells association (Model 3). It allows obtaining the results closer to the reality. Model 1 Model 2 Model 3 0 5 IV. 20 19.5 0 0.1 (a) 18.5 30 TABLE VI. 30 Model 1 Model 2 Model 3 40 Power(W) Figure 10(b) shows the smooth operation of the inverter to keep the output voltage of the panel at a constant value. This fact was evidenced for the three analyzed models. Note in Figure 10(a) that in steady state the power in all the Models is close to the maximum power provided by the manufacturer. The only difference in Model 1 is the overshoot due to the different dynamics behavior that comes from the linearization of the model. 50 0.04 0.06 time(s) 0.08 0.1 (b) Figure 10. Power (a) and Voltage (b) in the panel connected to the grid. Obtaining a model of PV panels with high accuracy is vital for future works in the photovoltaic solar engineering area. These Models make it possible to develop works and techniques to improve the performance and applicability of PV systems in the context of global renewable energy production. Luan Peterle Carlette was born in Cachoeiro de Itapemirim, Brazil. He is student of Electrical Engineering at Federal University of Viçosa, Viçosa, Brazil. He works with Power Systems, especially with photovoltaic energy and control applied to converters. ACKNOWLEDGMENT The authors would like to thank to CNPQ, FAPEMIG and CAPES for their assistance and financial support. REFERENCES [1] GUIMARÃES, A. P. C. et al. Manual para Engenharia de Sistemas Fotovoltaicos. Edição Especial. ed. Rio de Janeiro: PRODEEM, v. I, 2004. Delly Oliveira Filho received his BSc in Electrical Engineering at the Federal University of Minas Gerais in 1979 and his MSc in Thermal Engineering at the same university in 1983. In 1995 received his PhD in Electrical Engineering at the McGill University in Montreal, Canada. He has been working as an Associate Professor of the Energy Area of the Agricultural Engineering Department of Federal University of Viçosa, Minas Gerais, Brazil, since [2] EPIA. Global Market Outlook for Photovoltaics until 2016. 1ª. ed. [S.l.]: EPIA, v. I, 2012. [3] MENNER, R. J. C. M. Multi-Physical Modeling and Experimental Verification of Respiratory System. University of Technology. [S.l.], p. 13. 2009. (Graduation Symposium). [4] MICHAEL POKORNY, Z. R. Multi-Physical Model of Gunn Diode. [S.l.]. (IEEE Paper). [5] MINEIRO, S. E. et al. Photovoltaic system for supply public illumination in electrical energy demand peak. The Applied Power Electronics Conference and Exposition, California, v. 3, p. 1501-1506, 2004. [6] VILLALVA, M. G.; GAZOLI, J. R.; FILHO, E. R. Comprehensive Approach to Modeling and Simulation of Photovoltaic Arrays. IEEE Transactions on Power Electronics, v. 24, n. 1, p. 1198-1208, Maio 2009. [7] VILLALVA, M. G.; GAZOLI, J. R.; FILHO, E. R. Modeling and circuitbased simulation of photovoltaic arrays. Brazilian Journal of Power Electronics, v. 14, n. 1, p. 35-45, 2009. [8] SUUL, J. A. et al. Turning of control loops for grid connected voltage source converters. PECon, Johor Baharu, Malaysia, p. 797-802, December 2008. [9] LISERRE, M.; BLAABJERG, L.; HANSEN, S. Design and control of an LCL-filter based three-phase active rectifier. IEEE Transactions on Industry Applications, v. 41, n. 5, p. 1281-1291, September 2001. BIOGRAPHIES Erick Matheus da Silva Brito was Born in Feira de Santana, Brazil. He is student of Electrical Engineering at Federal University of Viçosa, Viçosa, Brazil. He works with Power Systems, especially with photovoltaic energy. Allan Fagner Cupertino was born in Visconde do Rio Branco, Brazil. He is student of Electrical Engineering at Federal University of Viçosa, Viçosa, Brazil. Currently is integrant of GESEP, where develop works about power electronics applied in renewable energy systems. His research interests include solar photovoltaic, wind energy, control applied on power electronics and grid integration of dispersed generation. 1988. Paulo F. Ribeiro received B.S. degree in electrical engineering from the Federal Univesity of Pernambumco, Brazil, in 1975, and the Ph.D. degree in electrical engineering from the University of Manchester, Manchester, U.K., in 1985. Presently, he is a Professor at Calvin College, Grand Rapids, Michigan. In 2010 he was a visiting professor at the Eindhoven University of technology, The Netherlands. Dr. Rbeiro is active in the IEEE,. International Council on Large Electric Systems (CIGRE), and International Electrotechnical Comission (IEC) technical working groups Heverton Augusto Pereira was born in São Miguel do Anta, Brazil, in 1984. He received the B.S. degree in electrical engineering from the Federal University of Viçosa (UFV),Viçosa, Brazil, in 2007, the M.S. degree in electrical engineering from the Universidade Estadual de Campinas (UNICAMP), Campinas, Brazil, in 2009, and currently is Ph.D. student from the Federal University of Minas Gerais (UFMG), Belo Horizonte, Brazil. He is currently an Assistant Professor with the Department of Electric Engineering, Federal University of Viçosa, Brazil. His research interests are wind power, solar energy and power quality.