Docstoc

PV-Catapult WP9 Performance Modelling

Document Sample
PV-Catapult WP9 Performance Modelling Powered By Docstoc
					        PV-Catapult WP9 Performance
                 Modelling
Ralph Gottschalg, Sheryl Williams, Wlodzimierz Kolodenny,
  Mariusz Prorok, Tadeusz Zdanowicz, Antoine Guerin de
  Montgareuil, Nico van der Borg, Teun Burgers, Hugo de
 Moor, Gabi Friesen, Dirk Stellbogen, Hans-Dieter Mohring




  CREST
                   Outline
• Overview of Aims and Objectives
• Approach
• Review of modelling methodologies evaluated in
  the RR
• Results of RR1
• Results of RR2
• Conclusions



CREST
                  Aim of this WP
• Plan and execute one modelling round robin (RR) test
• Evaluate State of the Art of European Photovoltaic Module
  Modelling
• Identify measurement accuracy of European measurement
  methodologies – limited to modules
• Demonstrate that a 5% accuracy is possible

Performance also added:
• Identify pitfalls for future RR
• Make available knowledge and infrastructure to
  Performance



CREST
                Partners

Modelling Partners:   Data and other input
  –   CEA               –   FhG-ISE
  –   CREST             –   JRC
  –   ECN               –   TTU
  –   SUPSI             –   TÜV
  –   WrUT              –   ZSW




CREST
                       Timeline
• Start of work November 2004
• Questionnaire of current methods sent out January 2005
• First RR completed June 2005
   – Re-Modelling of same site/module/year (ideal case)
   – Modelling same module at same site, different year
• Second RR completed January 2006
   – Predicting same site, different module
   – Predicting different module, different site
   – Predicting same module, different site
• Dissemination/discussion meeting for future work January
  2006
CREST
                         Models
   European Performance Type of Model
  Name of   Developed By
  Modelling
   Method
   Matrix      LEEE       Surface Fit

 MOTHERPV         CEA        Statistical

On-Line Yearly    ECN          Matrix
Yield Simulator
      SSE         CREST      Measure-
                          correlate-predict

CREST
              1st Round Robin

• How different are the different sites
• How accurate are modelling approaches in ideal
  or close to ideal conditions




CREST
                                         Sites for testing
                        80000       Annual irradiation as a function of irradiance
                                                  at different sites
  Irradiation (Wh/m²)




                        60000
                                                    Cadarache
                                                                                      Lugano

                        40000


                        20000

                                                                   Wroclaw
                           0
                                0        200      400       600      800       1000      1200
                                           Irradiance (W/m ²), 20 W/m ² intervals


CREST
Effect of Data Treatment Variation amongst
       the Different Modelling Groups
                                            Matrix   MOTHERPV    SSE
                              1.2
  Annual Irradation (rel to




                               1

                              0.8
           Avg.)




                              0.6

                              0.4

                              0.2

                               0
                                    cSi_1       cSi_2    cSi_3         CIS   aSi-2j

 CREST
Comparison of Energy Yield (measured vs
           modelled) Year 1
      Matrix      MOTHERPV       Back Temp.   SSE    On-Line Simulator
            6

            4

            2
  Error %




            0

            -2

            -4

            -6
                 cSi_1   cSi_2        cSi_3         CIS      aSi_2j

 CREST
Comparison of Energy Yield (measured vs
           modelled) Year 2
       Matrix     MOTHERPV       Back Temp.   SSE    On-line Simulator
             8
             7
             6
             5
  Error %




             4
             3
             2
             1
             0
            -1
                 cSi_1   cSi_2        cSi_3         CIS      aSi_2j

 CREST
       Lessons Learnt RR1
• ORGANIZATIONAL
   • Dataset format
   • Database Structure
   • Data filter and handling
   • Not as conclusive as wanted because groups
     dropped out (budget cuts)
   • Late start but on time to complete all tasks

• SCIENTIFIC
   • Temporal predict was successful, virtually all
     <6%
   • Different sites result in different accuracy
   • Surprisingly small problems for double junction
CREST
     Comparison of Methodologies
                   ZSW Dataset

    Base                       Predicted
   Modules                     Modules
    CIS_6                        CIS_16
                 Same Loc.
   CdTe_17                       CdTe_20



                             CIS_17 & CIS_20
            Different Loc.   CdTe_1 & CdTe_4
CREST
             2nd Round Robin
• How much does the module variability influence
  energy calculations
• How well can we cope with the variability in TF
  devices
• Are there site specific differences between the
  different data sets
• Test:
   – Same site different module
   – Same module different site
   – Different module, different site
CREST
            Results of 2nd RR (ZSW Dataset)

                 ECN     CREST     LEEE      CEA
            5
            0
 Error %




            -5
           -10
           -15
           -20
                 CIS 6    CIS 16    CIS 17     CIS 20


CREST
            Results of 2nd RR (ZSW Dataset)

                   ECN     CREST     LEEE   CEA

            12
            10
             8
  Error %




             6
             4
             2
             0
            -2
                 CdTe 17   CdTe 20    CdTe 1      CdTe 4


CREST
        Comparison of Methodologies
          PVCatapult Meas. RR Dataset

     Base                   Predicted Modules
    Modules Loc.1               SolarLab



     CEA            Loc.2         ECN




                                 CREST
            Loc.3
CREST
         Comparison of Methodologies
          PVCatapult Meas. RR Dataset
    Measureme       Time       Time of
    nt Location    Period       Year
                  /Weeks
        CEA           3        Spring

        SolarLab      4       Summer

          ECN         3        Late
                              Summer
        CREST      2(6days)   Autumn

CREST
      Results of 2nd RR( PVCatapult Dataset)
Measured at ECN
               LEE   ECN   CREST(NP)   CREST    CEA

              15
              10
     Error%




               5
               0
               -5
              -10
                     A60    KC60       US32    WS11007


 CREST
                  Lessons Learnt
• Module to module production effect dominate
• Modelling based on outdoor measurements needs
  careful cleaning
   – Overfiltering might result in statistically
     insignificant corrections which might introduce
     large errors
   – Not detecting/cleaning outliers will result in
     deviations also introducing large errors
• Models work well for c-Si
• Accuracy not as good as desirable for thin films

  CREST
                  Conclusions
• Main impact is the module-to-module variation
• Different environments are no problem if the
  module data is known accurately
• Error analysis of the measurement campaign is
  ongoing
• The work of WP9 provide a good foundation for
  further investigation of performance modelling




CREST
        Dissemination Efforts




CREST
                    Workshop

•   35 participants
•   Visitors from US & Japan
•   Presentation of results
•   Discussion of
     – Presentations in general
     – Further need for research



CREST
                     Publications
• EUPVSEC-20, Barcelona
   – Williams, S. R., et al. (2005): Evaluating the State of the
     Art of Photovoltaic Performance Modelling in Europe.
• WCPEC-3
   – Zdanowicz, T. et al (2006): Photovoltaic Performance
     Measurements in Europe: PV-Catapult Round Robin
     Tests
• EuroSun 2006
   – Strobel, M.B. et al (2006): Measurement Accuracy of
     Energy Yield of Photovoltaic Systems


CREST
                   Publications 2
• Dresden
  – Gottschalg et al (2006): Round Robin Comparison of
    European Outdoor Measurement Systems
  – Gottschalg et al (2006): Variability of Electrical
    Parameters Determined by Using Different Solar
    Simulation Systems for Different PV Module
    Technologies
  – Williams et al (2006): Variability of Electrical Parameters
    Determined by European Research Facilities for
    Different PV Module Technologies



CREST
      Other Dissemination Efforts

• Expert participation in IEA-PVPS Task 2
• Expert participation in IEA TC82-WG2
• Expert participation in CENELEC TC82-WG1

 WP 8/9 asks question, IP Performance will answer
  these




CREST

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:13
posted:8/2/2011
language:Polish
pages:26