DNVGL

DNV GL © 2018 February 2018 Observations and trends from 4 years of extended-duration testing of PV modules Henry Hieslmair Ryan Desharnais Henry.Hieslmair@DNVGL.com Ryan.Desharnais @DNVGL.com SAFER, SMARTER, GREENER DNV GL © 2018 February 2018 1 DNV GL’s Solar Team has considerable experience in many solar aspects About DNV GL DNV GL is the world’s largest independent energy 33 PV Module Manufacturers; ≈ 2500 modules Tests Description TC600 Thermal Cycling between -40°C and 85°C for 600 cycles DH2000 Damp Heat: 85°C delamination, backsheet cracking, J-box issues, … DNV GL © 2018 February 2018 4 What can we learn about the tests themselves? DNV GL © 2018 February 2018 5 Correlation coefficient r ! = #$%&!((, *) , - , . DNV GL © 2018 February 2018 6 Correlations between the tests for Δ P max TC 600 - DH 2000 0.31 - HF 20 0.74 0.39 - UV 90 0.14 0.26 0.10 - DML+TC50 +HF10 0.64 0.10 0.63 0.03 - PID 0.24 0.9 0.22 0.11 0.26 - TC 600 DH 2000 HF 20 UV 90 DML+TC5 0+HF10 PID DNV GL © 2018 February 2018 7 Correlations of test duration for Δ P max TC 600 0.84 DH 3000 0.15* HF 20 0.81 UV 90 0.64 DML+TC50 +HF10 0.86 PID 192h 0.69 TC 200 DH 2000 HF 10 UV 45 DML+TC50 PID 96h DNV GL © 2018 February 2018 8 Correlations of test duration for Δ P max TC 600 0.84 DH 3000 0.15* HF 20 0.81 UV 90 0.64 DML+TC50 +HF10 0.86 PID 192h 0.69 TC 200 DH 2000 HF 10 UV 45 DML+TC50 PID 96h DNV GL © 2018 February 2018 9 Duration of the testing (various tests) • Eliminate failures 8% and the correlation improves. • Sign of catastrophic failures after 2000h 10 DNV GL © 2018 February 2018 How are the modules changing over 4 years. 11 DNV GL © 2018 February 2018 Modules efficiency over time About 0.75% abs /year Module efficiency = (Nameplate P) / (glass area) 12 DNV GL © 2018 February 2018 Module changes over time Item Change / year r Module Efficiency (%) Number of bus bars* +0.75% abs +0.6 BB 0.63 0.68 Ribbon width -0.20mm 0.67 Front Encapsulant Thickness Rear Encapsulant Thickness Frame thickness Glass thickness *excluding multi-wire +0.025mm +0.012mm No change No change 0.50 0.32 13 DNV GL © 2018 February 2018 Change in BOM over time +0.6 BB/year Excluding multi-wire: 12 wires -0.2mm/year 14 DNV GL © 2018 February 2018 Have the test results improved over 4 years? 15 DNV GL © 2018 February 2018 Test results over time for Δ P max Item Change in Δ P max /yr TC600 +0.8% abs DH2000 -0.4% abs HF20 +0.5% abs UV90 +0.1% abs DML+TC50+HF10 +0.6% abs PID +0.4% abs However, all correlations 0.28 16 DNV GL © 2018 February 2018 Changes in test results over time for Δ P max 17 DNV GL © 2018 February 2018 Do higher module efficiencies have better reliability? PERC? 18 DNV GL © 2018 February 2018 Test results vs module efficiency The difference between mono and The difference between mono and multi is not statistically significant. multi is statistically significant. (p=0.26) (p=0.005) Not enough PERC data Not enough PERC data 19 DNV GL © 2018 February 2018 Test results vs module efficiency 20 DNV GL © 2018 February 2018 Test results vs module efficiency 21 DNV GL © 2018 February 2018 How do BOM properties impact test results? 22 DNV GL © 2018 February 2018 Correlations of test duration Δ P max Module property Number of busbars Front Encapsulant Thickness Rear Encapsulant Thickness Total encapsulant thickness Glass Thickness Frame Thickness Ribbon Width Ribbon Thickness ? Test Result TC 600 DH2000 ? HF20 UV 90 kWh DML+TC50+HF10 PID 96 Hr 23 DNV GL © 2018 February 2018 0.29 0.10 0.16 Glass Thickness (mm) 0.17 -0.02 0.14 -0.05 0.11 -0.19 Frame Thickness (mm) 0.22 -0.10 0.32 0.19 0.05 -0.06 Ribbon Width (mm) -0.49 0.07 -0.32 0.10 -0.39 -0.30 Ribbon Thickness (mm) 0.29 0.05 0.10 -0.01 0.29 0.07 Module Efficiency (%) 0.22 -0.20 0.32 -0.08 0.32 0.18 Correlations between BOM and Test Δ P max 0.41 TC 600 DH2000 HF20 Date 0.28 -0.09 0.28 Nameplate Pmax (W) 0.26 -0.18 0.30 Number of busbars 0.48 -0.07 Front Encapsulant Thickness (mm) 0.35 -0.20 Rear Encapsulant Thickness (mm) 0.27 -0.06 Total Encapsulant thickness (mm) 0.32 -0.14 0.38 UV 90 kWh DML+TC50+HF10 PID 0.12 0.26 0.09 0.11 0.14 0.01 -0.01 0.44 0.32 0.21 0.18 0.19 0.29 0.05 0.10 0.34 0.37 Excluding multi-wire and PID 10% 24 DNV GL © 2018 February 2018 Encapsulant and Bus Bars vs TC 600 • Possible mechanism includes: – Mechanical stresses at solder pads and ribbons – Soldering stress incurred by soldering larger ribbons 25 DNV GL © 2018 February 2018 How do test results vary by manufacturer? 26 DNV GL © 2018 February 2018 Manufacturer 27 DNV GL © 2018 February 2018 Conclusions from Δ P max analysis • Observed correlations between: – the tests which involve thermal cycling – early test results and later test results, except DH • Over the last 4 years, modules have: – become more efficient – more bus bars – thinner ribbons – not improved in test results • We do not have enough PERC data to make reliability comparisons (yet) • Mono did better than multi on TC600 • Module manufacturer is still a primary determinant of test results 28 DNV GL © 2018 February 2018 Photos placed in horizontal position with even amount of white space between photos and header PVLifetimeProject–Challenges ofMeasuringPVModuleDegradation Joshua S Stein , BruceKing, andCharlesRobinson SandiaNationalLaboratories ChrisDeline andBillSekulic National RenewableEnergyLaboratory Sandia National Laboratories is a multimission laboratory managed and operated by National Technology and Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International, Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525.