Characterization and Analysis of Photovoltaic Modules and the Solar Resource Based on In-Situ Measurements in Southern Norway Georgi Hristov Yordanov

Transcription

1 Characterization and Analysis of Photovoltaic Modules and the Solar Resource Based on In-Situ Measurements in Southern Norway Georgi Hristov Yordanov Supervisor: Prof. Ole-Morten Midtgård (NTNU) Co-supervisor: Prof. Lars Einar Norum (NTNU)

2 OUTLINE Aims and objectives Context and background Research questions Experimental setups Solar resource in Grimstad PV performance analysis and modeling Contributions to PV industry Scientific contributions 2

3 AIMS AND OBJECTIVES Investigate the PV potential in S. Norway Explain with device physics the observed performance differences among c-si PV To achieve this, one needs to: 1. Study the local solar resource 2. Measure, analyze and model PV performance 3. Measure, analyze and model I-V curves 4. Identify quantitative links between performance and I-V curve parameters 3

4 HIGH LATITUDE: N 4

5 COASTAL CONTEXT 5

6 SEA TO THE SOUTH-EAST 6

7 PRIOR STUDY 1 Olseth and Skartveit, The solar radiation climate of Norway, Solar Energy 37 (1986) 423; GHI 1050 kwh/m 2 /yr 7

8 PRIOR STUDY 2 Midtgård et al., A qualitative examination of performance and energy yield of photovoltaic modules in Southern Norway, Renewable Energy 35 (2010)

9 SOLAR RESOURCE IN 2005 Ibid. 10 % more than predicted by the EU PVGIS 9

10 η VS. IRRADIANCE Ibid. The effects from G and T not separated. 10

11 I-V CURVE RESOLUTION Ibid. Poor-resolution I-V; Every 20 min; 73 % up 11

12 12 PRIOR STUDY 3 Huld et al., A power-rating model for crystalline silicon PV modules, Solar Energy Mat. and Solar Cells 95 (2011) T MOD = 40 C

13 13 PV PERFORMANCE MODEL Relative efficiency: η REL = η(g,t )/η STC G = G/G 0, G 0 = 1000 W/m 2 ; T = T 25 C An empirical performance model was proposed in Huld et al., Mapping the performance of PV modules, effects of module type and data averaging, Solar Energy 84 (2010) 324: REL 1 k1 ln G' k 2 ln G' T' k 3 k 4 ln G' k 5 ln G' k 6T' k 3 the rel. temp. coeff. of P MAX at G = G 0

14 RESEARCH QUESTIONS Seasonal distribution of solar resource in 2011 vs and PVGIS? Maximal year-to-year variability? Maximal overirradiances? Burst durations? Physical cause? Effects of discrete sampling? Optimal azimuthal orientation of PV? Energy lost due to clouds? Accuracy «3 % by using I SC of many PV modules? How individual I-V curve parameters such as R S and n affect η REL at different G? How can a PV system builder recognize the bestperforming modules on the market? How can a PV manufacturer design cells which make best-performing modules? 14

15 THE NEW TEST SETUP Altitude 60 m a.s.l. Tilt angle = 39 ± 1 15

16 THE NEW TEST SETUP 2 16

17 THE NEW SOFTWARE 17

18 THE NEW I-V CURVES 18

19 MPP TRACKING 19

20 2 nd SETUP IN S. NORWAY Tilt angle = 60 ; Mutual shadowing 20

21 IRRADIANCE SENSORS SOLDATA 80SPC KIPP & ZONEN CMP 3 21

22 IRRADIATION IN kwh/m 2 15 % more than in PVGIS (long-term!) As in 2005, very sunny April, March and January 22

23 STATISTICS

24 YEAR-TO-YEAR VARIABILITY GHI data from nearby Landvik; 20.5 % Max. (y2y); σ = 5.5 % 24

25 EFFECTS OF DISCRETE SAMPLING 1.6 % extra uncertainty (annual) if sampling every 20 min 0.33 % for 1-min sampling (with a slow sensor) 25

26 CLOUD ENHANCEMENT 26

27 FORWARD SCATTERING 27

28 FORWARD SCATTERING 2 28

29 MIE PHASE FUNCTION Strongly anisotropic Depends on droplet size, wavelength, etc. Important to e.g. 3D gaming graphics programmers LOG SCALE! 49% 49% 1% 29

30 SENSOR RANGE MATTERS! 30

31 TYPICAL DURATIONS 31

32 LONGEST DURATIONS Total no. of events: 13,000 32

33 RESULTS FROM May: 1521 W/m 2 ; 10 June: 1528 W/m 2 33

34 PRIOR STUDIES Emck & Richter (2008): 1832 W/m 2 (equatorial Andes) T. Buseth (Elkem Solar AS, 2011): >1800 W/m 2 (Kenya) Hansen et al. (2010): GHI, >1500 W/m 2 (New Mexico) Luoma et al. (2012): Tilted, >1500 W/m 2 (California) Zehner et al. (2010, 2011): Attributed to reflection Parisi et al. (2004): Cloud-enhanced UV skin cancer?!! Overirradiances impose range requirements on sensors Calculation of UV doses and UV index should account for cloud enhancement!!! 34

35 BEWARE OF UV!!!!!!! 13,000 Danes develop skin cancer each year 35

36 CLOUD RESOURCE A hypothetical cloud-free year: 2130 kwh/m 2 ; 44 % lost due to clouds in the year

37 OPTIMAL PV AZIMUTH Averaged all daily irradiance profiles from 2011; Center of mass : 13:05 p.m W from S 37

38 PV PERFORMANCE: ANALYSIS AND MODELING Relative efficiencies at 25 C of 10 c-si modules; Fitted performance model coefficients k 1 through k 6 38

39 LINKING PERFORMANCE TO DEVICE PHYSICS 39 k 1 nn v R I S 0 S M, STC V M, STC I k2 RS 2 V M, STC M, STC Assuming 1-exponential I-V curve model, no shunts n ideality factor; N S no. of cells in series; v 0 =k B T 0 /q (thermal voltage at 25 C); R S series resistance; (V M,STC,I M,STC ) MPP at STC k 1 determines the slope of η REL (G,25 C) at G=G 0 and thus the behavior at intermediate irradiances k 2 determines the low-light performance; always < 0 Very good agreement between fitted and theoretical k 1

40 CONTRIBUTIONS TO PV INDUSTRY Quantitative and qualitative guidelines for design and selection of PV devices with better performance If modules with screen-printed c-si cells are chosen, PV system builders should generally go for 2 busbars, not 3 PV module makers have 2 new methods to monitor R S 40

41 A RECENT RECOGNITION By: Bosch Solar Energy AG, Germany Referred to: Yordanov et al. (2010), 25 th EUPVSEC 41

42 THE RESULT (AS OF 5 DEC) 42

43 SCIENTIFIC CONTRIBUTIONS 43 A methodology for in-situ testing of PV modules A very detailed analysis of the local solar resource Identification of peaks > 1500 W/m 2 in S. Norway Two new methods for evaluation of I-V parameters 1 improved and 1 novel differential technique An equation for Equivalent Cell Temperature (ECT) calculation from V OC for PV devices with variable ideality factors which are not covered in IEC Showed limits of applicability of classic I-V curve models k 1..k 6 for 8 c-si and 1 CIGS modules; equations for k 1, k 2

44 2 NEW METHODS TO EVALUATE R S, n AND I 0 44

45 IMPROVED AND NOVEL DIFFERENTIAL TECHNIQUES 45

46 MUCH BETTER IRRADIANCE ACCURACY Self-referenced irradiance from I SC of many new PV modules uncertainty 1 %! Corrections! 46

47 SOME CITATIONS OF MY PAPERS Verma et al., 38 th IEEE PVSC (2012) p C.W. Hansen, Sandia Report SAND Attivissimo et al., IEEE Trans. Instrum. Meas. (2012) p Stošović et al., Proc. Small Syst. Simul. Symp. (2012) p. 28 Nuotio and Kernahan, US Patent 8,239,149, 2012 Kernahan - US Patent 8,093,754, 2012 Polverini et al., Prog. Photovolt: Res. & Appl. 20 (2012) p. 650 A.K. Das, Solar Energy 86 (2011) p. 26 Lee et al., Int. J. Photoenergy 2012, 11 pp. Lamont and El Chaar, Renewable Energy 36 (2011) p Zimmermann and Edoff, IEEE J. Photovolt. 2 (2012) p

48 THANK YOU! 48