双面组件跟踪系统的技术经济性表现-cell press -joule

Article Global Techno-Economic Performance of Bifacial and Tracking Photovoltaic Systems This work performs a comprehensive techno-economic analysis worldwide for photovoltaic systems using a combination of bifacial modules and single- and dual-axis trackers. We find that single-axis trackers with bifacial modules achieve the lowest LCOE in the majority of locations (16% reduction on average). Yield is Carlos D. Rodrı´guez-Gallegos, Haohui Liu, Oktoviano Gandhi,.,LiLi,ThomasReindl, Ian Marius Peters carlos.rodriguez@nus.edu.sg HIGHLIGHTS Techno-economic comparison of combinations of bifacial and tracking PV systems Worldwide assessment of yield and LCOE Bifacial modules with single-axis trackers achieve lowest LCOE in most locations Bifacial modules with single- and dual-axis trackers boost yield by 35% and 40% ll Rodrı´guez-Gallegos et al., Joule 4,1–28 July 15, 2020 ª 2020 Elsevier Inc. https://doi.org/10.1016/j.joule.2020.05.005 boosted by 35% by using bifacial modules with single-axis trackers and by 40% in combination with dual-axis trackers. Systems 1 Jai Prakash Singh, 1 3 Li Li, 4 Thomas Reindl, 1 ance reaching both the front and rear surfaces of the modules for the different system designs (validated based on data from real most commonly installed new generation source. 2 The majority of current PV in- stallations employ monofacial crystalline silicon PV modules 3,4 with a fixed-tilt sys- Context i.e.,willthesesystemshavealowerLCOEthanstandardmonofacialfixed- tilt installations? Hence, in this paper, we explore the cost effectivenessof monofacialand bifacialPV farms with fixed-tilt, 1T and 2T installations worldwide. A sketch of the analyzed mounting structures is presented in Figure 1, also indicating the irradiance reaching both the front and rear sides. As the first step, we developed a performance model for the different mounting approaches using satellite data (Methodology). We cali- brated this model against field-measured data from fixed-tilt and tracking installa- tions and showed agreement with an overall deviation below 3% (Validation of Irra- diance Calculation Methodology). We then extended this study globally, based on the parameters provided in Global Techno-Economic Performance Calculation, andinvestigatedthetechnicalandeconomicperformanceofthedifferentmounting configurations (Results). Finally, we explored the sensitivity of LCOE in a variety of locationsusingaMonteCarloandregionsensitivityapproach(Results).Thisanalysis shows what changes in LCOE should be expected for variations related to weather, module, and cost parameters. A general discussion on the previous results is pre- sented in Discussion, and the paper is concluded in Conclusions. Methodology Calculation of Global/Diffuse Horizontal and Direct Normal Irradiance Daily average (onevalue perday)global horizontal irradiance (GHI)isobtained from NASA’s Clouds and the Earth’s Radiant Energy System (CERES). 32 Due to the influ- enceofthesun’spositionontheoverallirradiancecollectedatthemodulesurface,it is necessary to estimate the irradiance values with a higher resolution. This is ll achieved by employing an approach similar to the one presented in Sun et al. 29 The hourly GHI is first estimated by applying the clear sky model proposed by Haur- witz. 33,34 Inanefforttoreduceestimationerrors,thehourlyGHIvaluesaremultiplied by a constant so that their average is equal to the daily average GHI provided by Erlangen-Nu¨rnberg for Renewable Energy, 91058 Erlangen, Germany 7 Lead Contact *Correspondence: carlos.rodriguez@nus.edu.sg https://doi.org/10.1016/j.joule.2020.05.005 2 Joule 4, 1–28, July 15, 2020 Table 1. Literature Review on the Energy Gain of Different Designs Location Module Elevation (m)/ Number of Modules Forming the System Albedo/Bifaciality Tilt( C14 )/Facing G (%) Comments Amsterdam, The Netherlands a,10 0.5/1 0.2(0.5)/0.923 90/east 10.4(29.5) monofacial module: optimum tilt angle/south Doha, Qatar a,10 0.5/1 0.2(0.5)/0.923 90/east C05.6(17.2) Amsterdam, The Netherlands a,11 ?/7 rows 0.3/0.8 optimum orientation for fixed-tilt systems 5 and 6 and 10 and 4 and 8 and 12 and 10 and 5 and 5 and 25) N/A Davis, USA b,22 ~2.5/3 rows 0.16/0.75 N/A 5 bifacial-1T versus monofacial-1T La Higuera, Chile b,23 ~2/575kWp desert sand/~0.85 N/A 12.8 bifacial-1T versus monofacial-1T San Felipe, Chile a,24 0.5/72 0.12(0.7)/0.85 30/north ~7(~34) N/A South Korea b,25 ~1/6 0.06(21)/~0.82 30/south 5.25(14.47) N/A Hokkaido, Japan b,26 ~1.5/12 grass/0.959 35/south 8.6 N/A ~1.5/12 snow/0.959 35/south 23.0 N/A ~1.5/12 scallops’shells/0.959 35/south 23.3 N/A N/A ~1.5/12 snow/0.959 35/south 23.8 N/A Kasese, Uganda a,15 ?/1 0.2(0.5)/? ?/? 16.47(43.77) N/A ?/1 0.2(0.5)/? ?/? 14.71(17.93) monofacial-1T versus monofacial fixed-tilt ?/1 0.2(0.5)/? ?/? 1.53(21.93) bifacial fixed-tilt versus monofacial-1T ?/1 0.2(0.5)/? ?/? 40.1(62.2) bifacial-1T versus monofacial fixed-tilt Unless otherwise indicated in the column ‘‘Comments,’’ G (gain) corresponds to the extra energy produced from a bifacial-fixed design with respect to its mono- facialcounterpart,both atthesameorientation (indicatedinthecolumn‘‘Tilt( C14 )/facing’’).Thesymbol‘‘?’’isappliedwhenthedataofinterestwerenotprovided in the literature. a Results obtained from simulations. b Results obtained from real systems. 4 Joule 4, 1–28, July 15, 2020 ll Please cite this article in press as: Rodrı´guez-Gallegos et al., Global Techno-Economic Performance of Bifacial and Tracking Photovoltaic Sys- tems, Joule (2020), https://doi.org/10.1016/j.joule.2020.05.005 Article model 38 and anisotropic model 5 when calculating I f and I r , respectively) contribu- tions, together with the reflection losses. 39,40 Once I f and I r areobtained,thepower production of a PV system can be defined as: P PV = P PV;f $ðI f +I r $bÞ$f 1 1000 $½1 + g$ðT c C025ÞC138$h inv $ð1C0b 0 C0y$b 1 Þ$ð1C0lÞ (Equation 1) whereP PV;f [W p ]isthe overallpowerproductionof the installedmodulesunderstan- dard test conditions (STC) whenlight only reaches the module front side,b [%]is the bifaciality factor (equal to zero when dealing with monofacial modules), and f 1 rep- resents the solar spectral distribution influence. 41,42 The temperature influence on the module performance is represented by the second term from Equation 1,where g[%/ C14 C]isthepowertemperaturecoefficientandT c [ C14 C]isthecelltemperature. 43,44 Thevariableh inv [%]representstheinverteraverageweightedefficiency.Inaddition, b 0 [%] is the initial PV degradation, while b 1 [%/year] is the yearly degradation rate (where y is the analyzed year). 45–47 The other losses presented in a PV installation (e.g., shading, ohmic wiring losses, module mismatch, etc.) are combined and rep- resented by the l [%] parameter. Figure 1. Sketch on the Analyzed Mounting Structures (A) Fixed-tilt. (B) Horizontal single axis tracker (HSAT). (C) Tilted single axis tracker (TSAT). (D) Dual-axis tracker. Joule 4, 1–28, July 15, 2020 5 Please cite this article in press as: Rodrı´guez-Gallegos et al., Global Techno-Economic Performance of Bifacial and Tracking Photovoltaic Sys- tems, Joule (2020), https://doi.org/10.1016/j.joule.2020.05.005 Article WithEquation1,theoverallenergyproducedbyaPVsystemduringaparticularyear y,E ðyÞ PV [Wh],canthenbecalculated.Adetailedexplanationonthepowerandenergy estimation can be found in Rodrı´guez-Gallegos et al. 30 Estimation of the System Cost The overall cost for the PV system during its lifetime (l S )isdefinedas: C PV = C Bank;int +C Bank;amor +C own +C war +C insu +C OM (Equation 2) wheretheinitialinvestmentcost,whichincludesthecostrelatedtotheacquisitionof solar panels and inverters as well as their installation cost factor (c ins ), is divided into the first three terms: cost of bank interest (C Bank;int ), cost of bank amortization (C Bank;amor ), and cost of ownership (C own ). To cover for the initial investment of a PV farm, it is usual to obtain a bank loan to pay for part of this investment (resulting in the bank interest and amortization costs), while the owner of the PV farm is ex- pectedtopayfortherest(resultinginthecostofownership).Inaddition,theinverter warranty cost (C war ) is incurred every few years, while insurance cost (C insu )andop- erations and maintenance (Oint , C Bank;amor , C war , C insu ,andC OM . Further details of each of thetermscanbefoundinRodrı´guez-Gallegos et al. 30 Estimation of LCOE TheLCOE[USD cents/kWh],is taken asthe parameter toevaluatethecost effective- ness of the analyzed PV installations. This is defined as: LCOE = C PV P l S y =1 E ðyÞ PV ð1+DRÞ y (Equation 3) The numerator of Equation 3 is the total cost, whereas the denominator is the energy generated by the PV system throughout its lifetime. Notice that the cost aspect of the yearly energy generation is also influenced by the discount rate, DR [%]. 48 Limitations of the Presented Approach There are several technical and economic aspects that are beyond the scope of this study. These include: No consideration of module row-row shading (and therefore, no backtracking algorithm is applied); we have assumed that the module rows were properly spaced so that these shading losses were marginal. In addition, potential shading due to the mounting structure (which includes the torque tube for trackers) and uneven irradiance at the module rear side are neglected. The influence that the row-row spacing has on other cost factors, such as site preparation works, wiring, fencing, etc., has also been neglected in this work. We do not consider governmental policies;thesecanhaveabiginfluenceontheLCOEandtherefore,determinewhether itiscosteffectivetoinstallaPVsysteminaparticularlocation.Wedonotconsidersoil- ing losses in detail; soiling losses are assumed to be a part of the loss parameter l and anydependenceonlocationormoduletiltangleisnotconsidered.Wedonotconsider transportation costs, which are expected to have a particular influence for locations with difficult access, e.g., deserts. We also do not consider the land cost. ll Most of these limitations affect the different considered system architectures in a similar way and should not affect our conclusions or performance rankings severely. If the land cost becomes significant, then the LCOE improvement from bifacial and 6 Joule 4, 1–28, July 15, 2020 ll Please cite this article in press as: Rodrı´guez-Gallegos et al., Global Techno-Economic Performance of Bifacial and Tracking Photovoltaic Sys- tems, Joule (2020), https://doi.org/10.1016/j.joule.2020.05.005 tracking technologies are expected to be diminished, as these technologies are likely to have a higher space requirement than the fixed-tilt ones. Validation of Irradiance Calculation Methodology Model validation is performed by simulating front and rear irradiance against measured values from experimental setups provided by three institutes. The measured irradiance values presented in this section were obtained from high qual- itysensors,whichhaveanon-linearityaccuracybelow0.3%,e.g.,pyranometersKipp and Zonen SMP10, SMP11, and CMP11. The first analyzed data set was provided by Sandia National Laboratories and covered a time range from 17 September 2017–21 November 2017, 30 March 2018–2 May 2018, and 1 January 2019 – 9 March 2020 with a 1-min resolution for the following PV installations: (1) Fixed-tiltPVsystem:withtiltangleof35 C14 facingtheequator,row-to-rowpitch of4.9m,collectorwidthof2m,lengthoftherowsof21mandheightofmod- ule lower edge of 1.1 m. (2) Horizontal single-axis tracker system (HSAT): with east-west rotation composed of two tracker rows (referred as HSAT 1 and HSAT 2), row-to-row pitch of 6.1 m, collector width of 2 m, length of the rows of 25 m and tracker axis height of 1.7 m. These PV installations are located at Albuquerque, USA (latitude of 35.05 C14 ,longi- tude of C0106.54 C14 and elevation of 1,657 m). The data provided by Sandia National Laboratories include GHI, DNI, DHI, module orientation over time, height of rear irradiance sensors, and albedo. In addition, the sun position is estimated using the SPA algorithm introduced in Calculation of Global/Diffuse Horizontal and Direct Normal Irradiance. These data are then applied to compute the irradiance reaching thefrontandrearsidesofthemodulesforthesePVinstallations.Theresultsarethen compared with the measured irradiance falling on either module plane. Figure 2 shows the time series of the simulated and measured irradiance for three selected days with specific weather conditions (sunny, partly cloudy, and cloudy). Fig- ure3showstherelationbetweenthemeasuredandsimulatedirradiancevaluesforthe entire period considered. In general, we find decent agreement between measure- ment and simulation. A larger discrepancy occurs at the module rear side with a normalized root-mean-square error (NRSME) of approx. 5%–6%. The rear irradiance onlycontributesasmallfractionofthetotalirradiance.Hence,wefindsimilarNRMSEs of approx. 2%–3% for both front and front + rear. Figure 3 also provides the overall relative difference between the measured and simulated overall insolation (I d ). In addition to validating the bifacial irradiance calculation model, the accuracy of long-term irradiance calculation using satellite data (following the approach from Calculation of Global/Diffuse Horizontal and Direct Normal Irradiance)isalsovali- datedagainstfieldmeasurements.Here,weusedfieldmeasurementresultscollected by Harbin Institute of Technology (HIT) with a testbed located at Weihai, China (lati- tude of 37.53 C14 , longitude of 122.08 C14 , and elevation of ~10 m). More than one year of irradiance data are used. The following describe their analyzed PV systems: Article (1) Fixed-tiltPVsystem:withtiltangleof34 C14 facingtheequator,row-to-rowpitch of 10 m, collector width of 2 m, length of the rows of 36 m and height of mod- ule lower edge of 0.3 m. Only front side irradiance is recorded. Joule 4, 1–28, July 15, 2020 7 Please cite this article in press as: Rodrı´guez-Gallegos et al., Global Techno-Economic Performance of Bifacial and Tracking Photovoltaic Sys- tems, Joule (2020), https://doi.org/10.1016/j.joule.2020.05.005 Article ll (2) Tilted single-axis tracker system (TSAT): with 20 C14 tilt and east-west rotation, row-to-row pitch of 5.6 m, collector width of 1.7 m, length of the rows of 31 m and tracker axis height of 2.4 m. Front and rear side irradiance are re- corded. (3) HSAT: with east-west rotation, row-to-row pitch of 9.2 m, collector width of 2 m,lengthoftherowsof47mandheightofthetrackeraxisof2.2m.Onlyfront side irradiance is recorded. The monthly front and rear insolation received by the tracker systems is compared against the front side insolation of the fixed-tilt system by calculating the following ratios: (1) TSAT f+r /fixed f : Ratio between combined front and rear insolations collected bytheTSATsystemwithrespecttothefrontinsolationofthefixed-tiltsystem. (2) HSAT f /fixed f : Ratio between front insolation collected by