Analysis of impact of distributed generation in a distribution grid by the use of photovoltaic generators

Transcription

1 Analysis of impact of distributed generation in a distribution grid by the use of photovoltaic generators M. F. da Silveira¹, J. B. Dias¹ and J. V. C. dos Santos² ¹ Graduate Program in Mechanical Engineering, UNISINOS, Av. Unisinos, 95. São Leopoldo-RS, Brazil. ² Graduate Program in Electrical Engineering, UNISINOS, Av. Unisinos, 95. São Leopoldo-RS, Brazil. Alternative sources of energy and distributed generation (DG) are current subjects, especially in Brazil where the regulations had just changed and now the possibility of DG and compensation of energy can be considered. The use of solar energy has been increased along the last years, thus as a heat source or as photovoltaic (PV) generation, based mainly on the low environmental impact of this energy source. This analysis aims to study the impact of DG in the electrical energy distribution system caused by the use of grid-connected PV power systems. The study is done in steady state condition, in a specific feeder, considering typical curves proposed for sunny and cloudy days. It had been considered a penetration of DG of 2%, in two different hourly times for each kind of day, using sunny and cloudy typical curves. The power generated had been normalized related to the power consumed in different points of the feeder. The impacts observed in voltage levels of the distribution grid are not relevant. However, the DG did contribute to slightly raise some voltage levels that are already among the critical zone, in some of the sets analysed. Keywords: Distributed Generation; Grid connected photovoltaic power system 1. Introduction Distributed generation (DG) is a technology raising its participation in the market. The main characteristic is the decentralization of generation allowing the use of small energy sources along the distribution or transmission system. To reduce the necessity of expand the transmission systems and to reduce the dependency in the big power plants are the major advantages of DG as the small power sources are usually located close to customers. Photovoltaic (PV) systems will contribute to DG as they are small in size, can be installed in a distributed way and interconnected to the grid, also PV is getting popular due to the environmental appeal they have, to the costs related that are lowering and the freeness of the primary source. A typical application is the PV installed in the roof of a house and the extra power generated would be released in the grid. In order to release the extra power on the grid some conditions has to be fulfilled, as no voltage difference (in module and angle) between the customer and the grid and same frequency must be guaranteed [1]. The inverter is responsible to guarantee these conditions before the interconnection of the PV system to the grid [2] and the meter is responsible to allow the compensation among the energy consumed and released to/from the grid. In Brazil, this compensation is possible since 212 when the legislation has changed allowing small customers to produce electrical energy and to compensate the overrun in a matter of debit and credit of energy [3]. To analyse the impact in the grid by the use of PV system in a DG configuration is the aim of this study. Specially to analyse the impact in a steady state at the voltage level in a specific feeder by the insertion of PV generators in a distributed manner considering the situations of sunny and cloudy days through the use of typical curves. 2. Methodology The methodology used in this essay is divided in six steps. The first was the research, especially in the state of art regarding PV in DG. The lines of [4], [5] and [6,7] were adopted, in a matter of use a data base of irradiance to analyse the impact of generation, to analyse the impact in the voltage levels in steady state only and to consider a DG penetration of 2%. Other impacts as short-circuit levels or protection systems are not seen in this study but can be found at [6,8]. The second step was to define and to treat a data base of irradiance and air temperature, allowing the third and fourth steps, which are to define the behaviour of the following kind of days: sunny, cloudy and partial cloudy, and to define the typical cuves of PV generation for those sorts of days. Once the typical curves are established the fifth step take over, to validate the curves proposed in a PV generation analysis. After this validation the analysy of impact in voltage levels finally could be done considering the study in consuption values of general customers in a specific feeder. FORMATEX 213 3

2 3. Development The methodology development will be explained in the following sub-chapters. 3.1 Data base and day classification A twelve months data base of irradiance and air temperature is the base to this study. The data was acquired in Porto Alegre, Brazil, from July/24 to June/25 at UFRGS solar laboratory, the PV modules were positioned in an angle of 3 heading to the geographic north. The irradiance and air temperature had been acquired once per minute. The data base was treated so the days could be classified among sunny, cloudy or partial cloudy. First the days had been classified considering a qualitative criteria as suggested by [9], shown in Fig. 1, where the time zero means 6 a.m. and 78 minutes means 7 p.m Irradiance Irradiance a) b) Fig. 1 Sunny (a) and cloudy days (b), irradiance versus time. Some days couldn t be classified as sunny or cloudy as shown in Fig. 1 and they were classified as partial cloudy. Figure 2 represents a curve of irradiance versus time of March 17 th, 25 which was partial cloudy. 15 Irradiance Fig. 2 Irradiance curve of a partial cloudy day. The intermittence of this kind of days is notary as shown in Fig. 2 and it means that no standard can be predicted. Typical behaviour of each kind of days was sought in order to predict the behaviour of PV power generation curve in each case. Daily average irradiance H and daily reference average productivity Y R were calculated for every day of the data base according to Eq. 1 and 2. 1 H = Γ G( t) dt (1) where H represents the daily average irradiance, G(t) the irradiance acquired and Γ the period of one day. H Y R = (2) G ce 4 FORMATEX 213

3 where G ce is the irradiance of W/m² The Y R was calculated for every day of the data base, already classified among sunny, cloudy or partial cloudy day. The average Y R and the standard deviation for each kind of the day was calculated considering the whole data base and also the maximum and the minimum value in the data base was identified. The results are shown in Table 1. Table 1 Maximum, minimum and average Y R (h/day) and standard deviation for sunny, cloudy and partial cloudy days. Sunny Cloudy Partial Cloudy Y R Maximum Y R - Minimum Y R Average Standard deviation As expect, partial cloudy days have a huge variance in Y R in different date as shown in the high value of the standard deviation or in the big difference between the maximum and minimum value of Y R. In a quantitative analysis only of Y R the date of maximum value for partial cloudy could be classified as sunny and the date of minimum value could be classified as cloudy in a wrong way as if a qualitative analysis is done the intermittence of those days can be observed. The lack of standard in partial cloudy days can be seen in Fig. 3 where a dispersion analysis of Y R for all cloudy day in data base is shown. Fig. 3 Dispersion analysis of Y R for all cloudy day in data base. Due to the unpredictable behaviour, clear seen in the Y R dispersion shown in Fig. 3, the cloudy days had been disregarded in this study on the further steps. 3.2 Typical curves In order to reduce the quantity of information needed to calculate the energy generated in a PV power system, it s desired to have only Y R and the type of day to estimate the energy. Typical curves with the amount of energy in each hour were sought to achieve this considering the standards shown in Fig. 1 and the behaviour verified in section 3.1. Sunny days have a well defined standard with a smooth variation along the day. To define the typical curve of PV for a sunny day the irradiance acquired was integrated in periods of one hour for every hour of each day that forms the data base and an average of each hour was calculated, as shown in Table 2. FORMATEX 213 5

4 Table 2 Average of relative energy in each hour Sunny day. Time Energy - calculated (%) Energy - chose (%) 5 h - 5 h 59min.7 6 h - 6 h 59min.58 7 h - 7 h 59min h - 8 h 59min h - 9 h 59min h - 1 h 59min h - 11 h 59min h - 12 h 59min h - 13 h 59min h - 14 h 59min h - 15 h 59min h - 16 h 59min h - 17 h 59min h - 18 h 59min h - 19 h 59min.1 Total 1 1 Hours which zero PV energy were not shown in Table 2. The column Energy - chose (%) represents the values chosen to form the typical curve. These values are slightly different from the values calculated in each hour, considering an average of data base, however they were chosen to do a symmetrical behavior in the curve. Cloudy days do not have a well defined standard as observed in sunny days so an average of energy per hour would be pointless, thus a fixed rate of 11% was chosen for each hour between 8h and 16 h 59 min. Figure 4 shows the typical curve proposed for a sunny day and for a cloudy day. 3.3 Validation a) b) Fig. 4 Typical curves sunny day (a) and cloudy day (b). To validate the typical curves an algorithm in Matlab was adapted. The algorithm consider the irradiance (G) for every minute and the air temperature as inputs and give as output the DC and AC power (PPVC and PCA respectively). The ratio between PCA and PPVC represents the efficiency of the inverter, considering an empirical equation. The algorithm consider given characteristics for PV. Six different days was chosen in the data base to validate. The days that represents the higher and lower Y R of each category of days and the days closest to the average for each kind of day, and shown in Table 3. Table 3 Days used in validation. Sunny Cloudy Date Y R (h/day) Date Y R (h/day) Y R Maximum October 28 th, February 3 rd, Y R Minimum May 27 th, September 22 nd, Y R Average - Closest April 17 th, January 15 th, FORMATEX 213

5 Air temperature influence The first simulation aims to settle a standard air temperature instead of acquired data. To achieve it the following simulations took place: i) to analyse the difference between PV power generated using an hourly average air temperature instead of the temperature acquired every minute; ii) to analyse the difference between PV power generated using a daily average air temperature instead of the temperature acquired every minute, and iii) to analyse the difference between PV power generated using 25 C as air temperature instead of the temperature acquired every minute. All the comparison was based in PCA. The biggest difference observed in the six days used in this validation was 1.32% in the time of 12 h to 12 h 59 min of the May 27 th, 25, representing 47.5 W and happened in case ii), in this same point the difference observed in case iii) was smaller than 1%. Thus, the air temperature of 25 C was chose for further simulations Typical curve To validate the typical curves, the PV power generated in alternate current (PCA) was considered and the comparison between PCA founded using irradiance acquired every minute and PCA founded using Y R of each day in the typical curve. To make a better comparison the values of irradiance were integrated in hourly bases when the irradiance acquired was used. The results for sunny days are in Fig. 5, Fig. 6 and Fig. 7 and also in Table 4. Time zero means 8 a.m. and 54 minutes means 5 p.m Fig. 5 PV power generation curves, October 28 th, 24 Sunny day a) irradiance acquired every minute; b) average irradiance per hour and c) typical curve Fig. 6 PV power generation curves, April 17 th, 25 Sunny day a) irradiance acquired every minute; b) average irradiance per hour and c) typical curve Fig. 7 PV power generation curves, May 27 th, 25 Sunny day a) irradiance acquired every minute; b) average irradiance per hour and c) typical curve. FORMATEX 213 7

6 Table 4 Average of relative energy in each hour Sunny day. Date PCA (W) PCA (W) Irradiance acquired Typical curve Difference (%) October 28 th, April 17 th, May 27 th, The results for cloudy days are in Fig. 8, Fig. 9 and Fig. 1 and also in Table Fig. 8 PV power generation curves, February 3 rd, 25 Cloudy day a) irradiance acquired every minute; b) average irradiance per hour and c) typical curve Fig. 9 PV power generation curves, January 15 th, 25 Cloudy day a) irradiance acquired every minute; b) average irradiance per hour and c) typical curve Fig. 1 PV power generation curves, September 22 nd, 24 Cloudy day a) irradiance acquired every minute; b) average irradiance per hour and c) typical curve. Table 5 Average of relative energy in each hour Cloudy day. Date PCA (W) PCA (W) Irradiance acquired Typical curve Difference (%) February 3 rd, January 15 th, September 22 nd, The results shown in Tables 4 and 5 validate the typical curves proposed. Although some difference can be found, the amount of data needed to estimate the PV power is much smaller if the typical curve and Y R is used than if all irradiance data has to be used. 8 FORMATEX 213

7 Special attention can be drawn to Table 5, the little contribution of PV power generated in a cloudy day is easily observed, with such a small amount generated in this sort of day its impact will not be considered in further analysis in this study. 3.4 Impact of a sunny day in the grid by the use of DG Once the typical curve is validated the impact on the grid can be done considering DG using PV power systems in a sunny day. The impact was analysed in the voltage levels in a specific feeder. The feeder analysed is based in real data, located in the city of Viamão / RS Brazil, close to the place where irradiance data was acquired. It has a length of 38 km, nominal voltage of 23 kv, maximum current of 35 A, which represents a three-phase power of 13.9 MVA. It has 542 distribution power transformers, 1158 bars, 5 voltage regulators and 3 capacitor banks. To consider the penetration of the DG in the system, the curves of consumption of each power transformer was changed, reducing the consumption in each time, which is the same as included power generation in a compensation scheme. Table 6 states the PV power considered in each hour, following the typical curve proposed and the 2% penetration of PV power systems. Power is given in per unit (pu). Table 6 PV power generated in each hour Sunny day. Hour Power generated (pu) 8 h - 8 h 59min.7 9 h - 9 h 59min.13 1 h - 1 h 59min h - 11 h 59min.2 12 h - 12 h 59min.2 13 h - 13 h 59min.2 14 h - 14 h 59min h - 15 h 59min h - 16 h 59min.7 Normal levels of voltage are considered between.93 pu and 1.5 pu. Three buses had been considered in the analysis, in two different hours, at 8 am and at noon. That is to consider one hour with a small power generation from PV power systems, which is early in the morning and one hour in the maximum generation in a day, which is twelve o clock. Figure 11 and figure 12 show the voltage level in buses, without DG and with DG Fig. 11 Voltage (pu) at different buses a) 8 am without DG; b) 8 am with DG. a) b) Fig. 12 Voltage (pu) at different buses a) 12 am without DG; b) 12 am with DG. The results are shown in Table 7. a) b) FORMATEX 213 9

8 Table 7 Voltage level (pu) at each bar. Bar 8 h - 8 h 59 min 12 h - 12 h 59 min Without DG With DG Without DG With DG Small contributions can be observed in values presented in Table 7. Special attention shall be drawn to bar 343 at 8 h to 8 h 59 min where the voltage was slightly under the normal zone without DG and it got to the adequate zone once DG was present. The same behaviour occurred at noon, with a short raise in voltage levels in all bars. 4. Conclusion The use of data base was useful to set the behaviour of sunny days and cloudy days and also to see the huge intermittence of partial cloudy days, where an quantitative analysis could make some wrong classifications. The typical curve proposed and the non dependency of air temperature were important tools to estimate the PV power generated with just two information: Y R and the type of the day. Besides, the impact in the voltage levels by the use of DG with grid-connected PV power systems are not relevant considering the penetration of 2% studied, with small raising in voltage levels which did not affect the normal operation zone. Acknowledgements The support by UNISINOS through Pe. Milton Valente Fund is gratefully acknowledged. References [1] Elgerd OI. Introdução à Teoria de Sistemas de Energia Elétrica. São Paulo; McGraw-Hill do Brasil, [2] Rüther R, Edifícios Solares Fotovoltaicos: O Potencial da Geração Solar Fotovoltaica Integrada a Edificações Urbanas e Interligada à Rede Elétrica Pública no Brasil. Florianópolis, UFSC, 24, 113 pages. [3] Normative resolution 482, dated April 17 th, 212. Agência Nacional de Energia Elétrica ANEEL. Available at Accessed July, 212. [4] Baumgartner FP; Achtnich T; Remund J; Gnos S; Nowak S. Steps towards integration of PV-electricity into the grid. Progress in photovoltaics: research and applications. Wiley Online Library (wileyonlinelibrary.com). DOI: 1.12, 21. [5] Vieira Júnior CM. Impacto da Geração Distribuída no Perfil de Tensão de Regime Permanente de Redes de Distribuição de Energia Elétrica. Laboratório de Sistemas de Energia Elétrica. 28. Available at Accessed August, 212. [6] Wang X, Gao J, Hu W, Shi Z, Tang B. Research of Effect on Distribution Network with Penetration of Photovoltaic System. 45 th International Universities' Power Engineering Conference UPEC 21, Wales, UK, 21. [7] GE Energy Consult. Intermittency analysis project: Appendix b impact of intermittent generation on operation of California power grid. California Energy Commission, 27. [8] Pizzali LFO. Desempenho de redes de distribuição com geradores distribuídos pages. Thesis (PhD in Electrical Engineering) Graduate Program in Electrical Engineering, Universidade Estadual Paulista Júlio de Mesquita Filho, Ilha Solteira, São Paulo, Brazil, 26. [9] Duffie JA, Beckman WA. Solar Engineering of Thermal Process. Third Edition. New Jersey. John Wiley & Sons, INC pages. 1 FORMATEX 213