COMPARISON OF MEASURED AND ESTIMATED SOLAR RADIATION DATA: A CASE STUDY FOR ISTANBUL Şaban Pusat 1, Erdal Bağcı 2 1 Yildiz Technical University, Mechanical Engineering Department, Istanbul 2 Uzman Enerji Renewable Energy and Energy Efficiency Solutions, Istanbul Corresponding email: spusat@yildiz.edu.tr telephone: 0537-438-32-20 Abstract The main objective of this study is to review the horizontal global solar radiation estimation models and compare the results with measured data and national database for Istanbul. The mathematical models used are based on the sunshine duration data as common in the world. For comparisons, the most common statistical methods (MBE, MPE and RMSE) are used. We came up with the result that solar radiation and sunshine duration measurements in Turkey are very unreliable and inaccurate. Keywords: Solar energy; Global solar radiation; Sunshine duration; Istanbul; Turkey 1. INTRODUCTION In solar energy studies, solar radiation and its components for a specific location are the main necessities. The solar data should be sufficient, reliable and contemporary. Solar energy data measurement is a difficult process: measurement devices need to calibration and maintenance continuously. Due to the difficulties in radiation measurements, scientists are developed many mathematical models to estimate the solar energy data. Menges et al. studied the 50 models [1] and Bakirci reviewed 60 models [2] for solar radiation estimation. Radiation can also be estimated by satellite images. Turkey has a high potential on solar energy. The yearly average solar radiation is about 4.18 kwh/(m 2.day), and the daily average sunshine duration is about 7.50 h [3]. This potential is one of the best of Europe. Authority (EIE - General Directorate of Electrical Power Resources Survey and Development Administration) calculates that Turkey can produce 380 GWh electricity per year on 4600 m 2 area which has a potential approximately of 4.52 kwh/(m 2.day). In this study, we tried to evaluate the twelve solar radiation estimation models available in the literature for Istanbul. Models are based on different regression methods that use the sunshine duration data. Two different solar radiation and sunshine duration data are used for analysis: first one is ten years meteorological data taken from State Meteorological Affairs, known as DMI in Turkey, and second one is database of Solar Energy Potential Atlas, known as GEPA in Turkey. Statistical methods are used for comparisons of the results and selection of the best models. 1
2. DATA AND MODELS a) Data and Methods of Comparison In this study, the hourly measured irradiation and sunshine duration data on horizontal surface between 1997 and 2006 of Istanbul are used to calculate the contemporary average values. Basic data of the meteorological station are presented in Table 1. City Station Elevation (m) Latitude Longitude Average Temperature ( 0 C) a Average Wind Speed (m/s) b Average Total Radiation (cal/cm 2.day) a Istanbul Goztepe 33 40.97 N 29.8 E 14.2 2.8 307 5.9 a: 1971-2000 period b: 1961-1996 period Table 1: Basic information about meteorological station Average Sunshine Duration (h) a Monthly averages of two databases are presented in Table 2. As seen from the Table, there are high differences between two databases. These differences are not scope of this study and are examined in another study [4]. TYM a GEPA b orizontal global solar radiation Sunshine duration orizontal global solar radiation Sunshine duration kwh/m2.day h kwh/m2.day h JANUARY 1,42 2,18 2,00 3,46 FEBRUARY 2,28 3,70 2,57 4,43 MARC 3,46 5,00 4,20 5,32 APRIL 4,46 5,85 5,28 6,85 MAY 5,90 8,38 6,30 8,61 JUNE 6,49 10,05 6,79 10,51 JULY 6,49 11,06 6,79 11,17 AUGUST 5,38 9,44 6,07 10,14 SEPTEMBER 4,33 7,56 5,09 7,83 OCTOBER 2,88 5,23 3,74 5,22 NOVEMBER 1,77 3,29 2,37 3,85 DECEMBER 1,24 2,02 1,80 2,96 MEAN 3,84 6,15 4,42 6,70 a: Ten years database monthly average (TYM) b: Solar energy potential atlas database (GEPA) Table 2: Monthly average daily radiation and sunshine data The performances of the models were evaluated by the statistical error analysis methods: mean bias error (MBE), mean percentage error (MPE) and root mean square error (RMSE). These are the most common statistical methods in solar radiation estimation models comparisons [1, 5, 6, 7 and 8]. MBE, MPE and RMSE are formulas are given below: MBE = N i,c i,m i=1 (1) N 2
MPE = i,m i,c N i,m i=1 N x 100 (2) RMSE = ( i,c i,m ) 2 N i=1 (3) N where N is the total number of observations, i,m is the i th measured value and i,c is the i th calculated value. b) Basics of Estimation Models The simplest and most known model developed for estimation of monthly average daily global solar radiation on horizontal surface is the Angström-type regression model. After this model, dozens of models developed on this base. Today, the basic model is based on the fraction of the measured and clear day monthly average sunshine hours at the location in question [9]: 0 = a + b S S 0 (4) where is the monthly average daily global solar radiation, 0 is the monthly average daily extraterrestrial solar radiation, S is the monthly average daily bright sunshine hours, S 0 is the monthly average day length, and a and b are empirical coefficients. The monthly average extraterrestrial daily radiation on a horizontal surface, 0, can be calculated by the following equation [9]: 0 = 24 π I sc f [cos φ cos δ sin w s + π 180 w s sin φ sin δ] (5) where I sc is the solar constant (I sc=1353 W/m2), f is the eccentricity correction factor, φ is the latitude of the location, δ is the solar declination angle, w s is the mean monthly sunrise hour angle. The eccentricity correction factor, solar declination and sunrise hour angle can be computed by the following Equations (6)-(7)-(8), respectively [9]: 360 D f = 1 + 0.033 [cos ] 365 (6) δ = 23.45 sin [ 360 (284+D) ] (7) 365 w s = cos 1 ( tan φ tan δ) (8) where D is the number of the days of the year starting from the first of January. The clear day monthly average daily sunshine duration (the maximum possible sunshine duration) can be calculated by the following equation: S 0 = 2 15 w s (9) c) Model 1 Bakirci proposed a linear exponential model for Istanbul [7]: 0 = 0.7380 + 1.5454 S S 0 0.6294 exp S S 0 (10) d) Model 2 Bakirci also suggested the following exponential equation for Istanbul [7]: 3
= 0.6726 ( S ) 0.5283 0 S 0 (11) e) Model 3 Ulgen and epbasli proposed the following linear equation for Ankara, Istanbul and Izmir [6]: 0 = 0.2671 + 0.4754 S S 0 (12) f) Model 4 Ulgen and epbasli also suggested the following third order polynomial equation for Ankara, Istanbul and Izmir [10]: 0 = 0.2854 + 0.2591 S S 0 + 0.6171 ( S S 0 ) 2 0.4834 ( S S 0 ) 3 (13) g) Model 5 Tiris et al. obtained the following correlation from the experimental data measured in Gebze, Kocaeli [11]: 0 = 0.2262 + 0.418 S S 0 (14) h) Model 6 Yildiz and Oz proposed the following equation for Turkey [12]: 0 = 0.2038 + 0.9236 S S 0 0.3911 ( S S 0 ) 2 (15) i) Model 7 Tasdemiroglu and Server developed the following correlation for the six locations (Ankara, Antalya, Diyarbakir, Gebze, Izmir and Samsun) in Turkey [13]: 0 = 0.225 + 0.014 S S 0 + 0.001 ( S S 0 ) 2 (16) j) Model 8 Akinoglu and Ecevit proposed the following equation for Turkey [14]: 0 = 0.145 + 0.845 S S 0 0.280 ( S S 0 ) 2 (17) k) Model 9 Kilic and Ozturk have determined the regression coefficients a and b as a function of the solar declination angle, latitude and elevation (km) [15]: = a + b S 0 S 0 a = 0.103 + 0.017 Z + 0.198 cos(φ δ) a = 0.533 0.165 cos(φ δ) (18a) (18b) (18c) 4
l) Model 10 Glover and McCulloch suggested the following model that depends on the latitude [16]: 0 = 0.29 cos(φ) + 0.52 S S 0 (19) m) Model 11 Page proposed the following equation [17]: 0 = 0.23 + 0.48 S S 0 (20) n) Model 12 Gopinathan has derived following correlation [18]: 0 = 0.309 + 0.539 cos(φ) 0.0693 Z + 0.290 S S 0 + (1.527 1.027 cos(φ) + 0.0926 Z 0.359 S S 0 ) S S 0 (21) 3. RESULTS AND DISCUSSION Twelve mathematical models have used for estimation of the monthly average daily global solar radiation data. Validation of the models has been performed by using the basic statistical methods (MBE, MPE and RMSE). As mentioned in the previous section we have used two different databases. So we have two different global solar radiation and sunshine duration data. Therefore we have made four different evaluations: 1. By using TYM sunshine duration data, we estimated the solar radiation data and compare with a. TYM solar radiation data, b. GEPA solar radiation data. 2. By using GEPA sunshine duration data, we estimated the solar radiation data and compared with a. TYM solar radiation data, b. GEPA solar radiation data. According to the results, best models are defined as Model 10 for 1.a, Model 6 for 1.b, Model 1 for 2.a and again Model 1 for 2.b. Results are presented in Figure 1 and Table 3. 5
Global solar radiation (kwh/m 2.day) Solar Future 2010 Best for 1.a Best for 1.b Best for 2.a and 2.b Databases Model 10 Model 6 Model 1 TYM GEPA kwh/m 2.day JANUARY 1.301 1.521 1.521 1.425 2.000 FEBRUARY 2.156 2.577 2.308 2.282 2.570 MARC 3.258 3.892 3.314 3.457 4.200 APRIL 4.155 4.957 4.459 4.462 5.280 MAY 5.820 6.788 5.770 5.897 6.300 JUNE 6.550 7.460 6.368 6.485 6.790 JULY 7.268 8.057 6.737 6.486 6.790 AUGUST 5.981 6.771 5.815 5.381 6.070 SEPTEMBER 4.288 4.963 4.234 4.330 5.090 OCTOBER 2.841 3.377 2.801 2.881 3.740 NOVEMBER 1.734 2.070 1.822 1.768 2.370 DECEMBER 1.212 1.412 1.338 1.237 1.800 MEAN 3.880 4.487 3.874 3.841 4.417 Table 3: Comparison of the results 9 8 7 6 5 4 Best for 1.a Best for 1.b Best for 2.a and 2.b TYM GEPA 3 2 1 1 2 3 4 5 6 7 8 9 10 11 12 Month Figure 1: Comparisons of the results This study briefly says that there are no perfect measurements for solar energy. The results show that measurements cannot be used directly for any analysis. They can include wrong and missing data parts. The solar radiometry science has some developments to increase the measurement quality and reliability [19]. owever, today radiometric devices aren t reliable. Therefore researchers should notice the uncertainty level of measurements and make the energy calculations concerning the inaccuracies. The solar energy investments highly depend on the radiometric measurements. For this kind of energy investments, which are so expensive today, the feasibility studies should be done by professional persons. In this study we used the measured data from DMI station and GEPA database. Investors should have their own measurement station at the best location for the planned action, and the measured data also should be analyzed by specialists. 6
Acknowledgements The Turkish State Meteorological Service (DMI) is acknowledged for the supply of data. The authors are also grateful to Prof. Dr. Ismail EKMEKCI at Marmara University, Department of Mechanical Engineering, for his valuable comments and suggestions. References 1. Menges. O., Ertekin C. and Sonmete M.., Evaluation of global solar radiation models for Konya, Turkey, Energy Conversion and Management 47 (18 19) (2006), pp. 3149 3173. 2. Bakirci K., Models of Solar Radiation with ours of Bright Sunshine: A Review, Renewable and Sustainable Energy Reviews 13, (2009), pp. 2580 2588. 3. EIE, www.eie.gov.tr 4. Uzman Enerji, www.uzmanenerji.com 5. Ulgen K, epbasli A. Comparison of diffuse fraction of daily and monthly global radiation for Izmir, Turkey. Energy Source 2003; 25:637 49. 6. Ulgen K, epbasli A. Solar radiation models. Part 2: Comparison and developing new models. Energy Source 2004; 26:521 30. 7. Bakirci K., Correlations for estimation of daily global solar radiation with hours of bright sunshine in Turkey, Energy 34 (4) (2009), pp. 485 501. 8. Bulut., Büyükalaca O., Simple model for the generation of daily global solar-radiation data in Turkey, Appl Energy 84 (2007), pp. 477 491. 9. Duffie JA, Beckman WA. Solar engineering of thermal processes. New York: Wiley; 1991. 10. Ulgen K, epbasli A. Comparison of solar radiation correlations for Izmir, Turkey. International Journal of Energy Research 2002; 26:413 30. 11. Tiris M, Tiris C, Ture IE. Correlations of monthly-average daily global, diffuse and beam radiations with hours of bright sunshine in Gebze, Turkey. Energy Conversion and Management 1996; 37:1417 21. 12. Yildiz M, Oz S. Evaluation of the solar energy potential of Turkey. In: Proceedings of the 6th National Energy Congress; 1994. p. 250 60 [in Turkish]. 13. Tasdemiroglu E, Sever R. An improved correlation for estimating solar radiation from bright sunshine data for Turkey. Energy Conversion and Management 1991; 31(6):599 600. 14. Akinoglu BG, Ecevit A. A further comparison and discussion of sunshine based models to estimate global solar radiation. Solar Energy 1990; 15:865 72. 15. Dogniaux R, Lemoine M. Classification of radiation sites in terms of different indices of atmospheric transparency. Solar energy research and developmentin the European Community, Series F, vol. 2. Dordrecht, olland: Reidel; 1983. 16. Glover J, McGulloch JDG. The empirical relation between solar radiation and hours of sunshine. Quarterly Journal of the Royal Meterological Society 1958; 84:172 5. 17. Page JK. The estimation of monthly mean values of daily total short wave radiation on vertical and inclined surface from sunshine records for latitudes 40N 40S. Proceedings of UN Conference on New Sources of Energy 1961; 4(598):378 90. 7
18. Gopinathan KK. A general formula for computing the coefficients of the correlations connecting global solar radiation to sunshine duration. Solar Energy 1988; 41:499 502. 19. Gueymard C. A., Myers D. R., Modeling Solar Radiation at the Earth s Surface, Springer, 2008 8