0 Голоса «за»0 Голоса «против»

Просмотров: 11366 стр.Oct 10, 2013

© Attribution Non-Commercial (BY-NC)

PDF, TXT или читайте онлайн в Scribd

Attribution Non-Commercial (BY-NC)

Просмотров: 113

Attribution Non-Commercial (BY-NC)

- Solar Photovoltaic Incentive Program
- WindFarm Ecnomics
- Design Guidelines for 30 100 KW Systems
- BIPV Training Presentation 7-10-08
- India Solar Sponsor Map Final Web 1
- OK MSc thesis: chapter 7 References
- State-of-Massachusetts-Incentive-Area-Commonwealth-Solar-II-Rebates---
- Future Direction of Solar PV Industry
- Renewable Intro Slides
- Solar-220
- Thin Film Photovoltaics (PV) - Advanced Technologies and Global Market (2010 - 2015)
- JacobRachel.ts.Step.by
- Italian Nursery School_ Conceptual Design Analysis _ Sustainability Workshop
- Intro
- b1
- 009-045
- Lecture 1
- Alonso Refs
- Scorecard Final Version
- 1-s2.0-S0960148108002619-main

Вы находитесь на странице: 1из 66

Prepared By: Shelton Honda Alex Lechner Sharath Raju Ivica Tolich

Authors / Disclaimer

San Jose State University (SJSU) engineering student and project advisor authored this report and provide the information to assist the industry with the assessment of existing PV systems that may or may not be performing to the expectation of the system owner. The objective of the project has been to investigate and evaluate PV system performance metrics and associated calculational methods in current use, and based on this evaluation, identify cost-effective assessment methods for short-term and long-term assessment periods. The information in the report can be used for further investigation and analysis of data to expand industry understanding of system performance assessment techniques. The project and content of this report are technical in nature and do not consider financial, legal, or ownership aspects of PV systems. It is expected that results of performance assessments may typically be used as input to operation and maintenance decisions. The content, conclusions and recommendations expressed in this report are those of the engineering student authors and advisor and are based on information gathered during project. San Jose State University (SJSU) is not responsible for any aspects of the project or report and is not liable for the use or misuse of the information provided herein.

ii

Acknowledgements

The project was performed with ongoing technical assistance and advice from SolarTech Performance Committee members including Joe Cunningham of Centrosolar, Laks Sampath of Neozyte, and Willard MacDonald of Solmetric. They provided valuable input from the solar industry perspective. SJSU engineering students contributed to various aspects of the project during the Spring and Fall 2011 semesters including testing, data collecting, analysis, method review, and preparation of content for the project report. Students that provided input to the authors included David Twining, Michael Holzkamp, and Dennie Her. The support and encouragement from SolarTech staff and committee members has been appreciated. Notably, support from David McFeely, and encouragement from Mike Balma and Tim Keating have helped with completion of the project. Also, it is appreciated that SolarTech gave SJSU engineering students an opportunity to participate in this project which contributed to the SolarTech organizations goals and simultaneously provided a learning opportunity to SJSU engineering students.

iii

Table Of Contents

Acknowledgements .... iv Table of Contents 1 Executive Summary 3 1.0 Introduction .,.. 4 1.1 Background 1.2 Purpose of Performance Assessment 2.0 PV Performance Assessment Methods . 7 2.1. Performance Metrics 2.1.1. Review of Current Industry Performance Metrics Literature Survey 2.1.2. Summary of Effective Performance Metrics 2.1.2.1. Short-term Assessment Power Performance Index 2.1.2.2. Long-term Assessment Performance Ratio 2.1.2.3. Long-Term Assessment Performance Ratio, Compensated 2.1.2.4. Long-term Assessment Energy Performance Index 2.1.3. Performance Metric Calculation methods 2.1.4. Uncertainty and Significance Tests 2.2. Numerical Models of Expected Performance 2.2.1. Review of PV Module and System Modeling Methods 2.2.2. Tabulated Comparison of Models and Methods 2.3. Measurement Methods of Actual Performance 2.3.1. AC Measurements 2.3.2. DC Measurements 2.3.3. Comparison of Direct Measurements to Inverter and Monitoring System Measurements 2.4. Performance Shortfall Investigation 2.5. Financial Model to Evaluate Cost/Benefit of Maintenance 3.0 Application of Above Methods to Existing Systems... 23 3.1 400 kW Local System Power Performance Index (PPI) Method 3.2 191kW Live Site System Performance Ratio (PR) 3.3 600 kW Local System Energy Performance Index (EPI) Method 4.0 Recommendations and Conclusion ... 41 4.1 Recommended Performance Assessment Methods 4.1.1 Simple approach with greatest uncertainty but ease of use 4.1.2 Moderate approach with moderate uncertainty 4.1.3 Accurate approach with least uncertainty requiring greater level of effort 4.2 Project Conclusions 1

5.0 6.0

References..... 42 Appendix 6.1 Performance Assessment Flow Chart 6.2 Method to calculate plane of array irradiance based on GHI 6.3 Method to calculate PV cell temperature based on ambient temperature 6.4 Derate Factor determination, assumptions 6.5 Sample of Excel Spreadsheets to calculate Performance Ratio (PR) 6.6 Sample of Excel Spreadsheet to calculate Energy Performance Index (EPI)

Executive Summary

Owners of existing photovoltaic (PV) solar energy systems are typically interested in the system short-term and long-term performance as input to operation and maintenance decisions. Performance metrics and methods to calculate the metrics are used by the industry for various purposes and it is difficult for the owner to know which metric is appropriate for which purpose. The objective of this project was to summarize metrics in use, determine the level of effort to calculate the metrics, review of the purpose of the metrics, and recommend which metric is appropriate for which purpose. Literature review and discussions with industry experts let to focusing on the following four metrics: Power Performance Index (PPI) of actual instantaneous kW AC power output divided by expected instantaneous kW AC power output. Performance Ratio (PR) of final yield divided by reference yield over an assessment period. Performance Ratio with final yield corrected for cell temperature (CPR) over an assessment period. Energy Performance Index (EPI) of actual kWh AC energy divided by expected kWh AC energy as determined from an accepted PV model, such as SAM, using actual climate data input to the model over the assessment period.

Data for use in the project were obtained from three systems having DC ratings of 191kW, 400kW, and 600kW. Data were obtained by direct power measurements, on-line live site, and access to monitoring system data, respectively. The three primary purposes for performance assessments of existing systems and the associated recommended metrics are listed below. A guideline summary is provided in Table 2.2: Monitoring of a specific PV system to identify degraded performance and need for maintenance based on condition. Use EPI metric and trend EPI for the specific system. Commissioning or assessment after major maintenance. Use PPI and EPI metrics. Determination of specific industry parameters, such as Yield or Performance Ratio, to allow comparison of systems in different geographic locations for design validation or investment decisions. Use PR, CPR and/or EPI depending on the level of effort and level of uncertainty.

Additional work is recommended to develop specific procedures for each of the four metrics summarized above and for making Excel spreadsheets available for general use. Additional long-term data should be analyzed to further investigate the ability of metrics to meet the stated purposes, and how to integrate metrics with currently available industry products such as monitoring systems and IV curve tracers.

1.0 Introduction

1.1 Background

Photovoltaic solar power systems provide clean renewable energy typically with high capital costs and low operating costs. Recent operating system data has identified that 22% of the PV strings produce less than 90% of the maximum output string [Ref. 3]. There is an industry need for a guideline which can be used to measure short-term power (kW) and long-term energy (kWh) performance after the system has been commissioned to aid in operation and maintenance decisions and for comparison of systems. Values for performance metrics, such as final yield, daily and monthly performance ratio, and AC DC yields, are often stated in reports, plots provided, and shown occasionally with uncertainty bars. The plots illustrate typical seasonal variation and the influence of weather and derating conditions, and are interesting but they are not useful to make near-term decisions. Relatively large uncertainty, which is inherent in metrics for short assessment periods, is not pointed out. Attempts have been made to reduce variation but work is still needed to make these metrics more useful. The purpose of this project has been to summarize existing performance metrics commonly used in the industry and to outline effective assessment metrics and associated calculation methods considering factors such as level-of-effort to perform the assessment (cost) and the value of the assessment (benefit). Consideration was given to the size of the system, availability of data, relative uncertainty inherent in the assessment, short-term or long-term assessment periods, known degradation mechanisms, and relevance to the industry need. The project included research, summarizing and comparing methods, evaluating effectiveness, and generally evaluating performance assessment methods currently used during and after commissioning. The objective of the project was to define methods which would enable quick identification of actual performance that deviates from expected performance. The Performance Assessment guideline is applicable to and may be used on any size system, however is targeted for the PV market segment of 100kW and larger systems which use technologies where the module characteristics are provided by the manufacturer and for which performance models are available. The guideline focused on systems with fixed flat panel PV modules.

1.2

The purpose and need for performance assessment can best be summarized from an owners perspective by the questions that are often asked: 1. How is my system, or portion of my system, performing currently in comparison to how I expect it to perform at this point in its life? 2. How is my system performing for both the short-term and long-term in comparison to how it is capable of performing with its given design and site location? 3. How is my system performing over an assessment period in comparison to other, similar systems in similar climates to help make operating and maintenance decisions? 4. Did I get what I paid for in terms of cost of energy? 4

5. 6. 7. 8. 9.

How is my system performing compared to the last assessment period (trending)? What is the annual kWh AC energy production per installed KW DC (Yield) What is my cost of energy in terms of intial and maintenance costs relative to kWh energy production over the life of the system? When do I need to do maintenance, such as cleaning, based on performance? I am aware that an individual module is damaged based on inspection, but the overall system performance is not noticeably affected, how do I do a more accurate performance assessment to determine if other modules are damaged?

The purpose of the assessment affects which metric is appropriate to use. For example, the commonly used metric of Performance Ratio (PR), as defined by IEC61724 and NREL, is appropriate to trend a specific system or compare systems in similar geographic locations. If PR is used to evaluate a system in San Francisco, CA, compared to a similar system in Daggett, CA, incorrect conclusions would be reached. Specifically, using PVWATTS Version 1, a 100kW system in San Francisco with latitude tilt has a calculated PR of 0.73 with an output of 145,000 kWh/year, while a 100kW system in Daggett with latitude tilt has a PR of 0.69 with an output of 171,000 kWh/year. Even with a lower PR, the Daggett system has higher output and therefore higher performance. One purpose of the performance assessment is to detect changes in system performance; usually decreases in performance, to allow the system owner to investigate and potentially perform cost effective maintenance. This can be done best on a relative scale of trend where the specific performance of the system is compared to itself. If the design related derate factors or calculation methods are not exactly correct, this approach reduces sensitivity of the metric value to the assumptions used. Another purpose is to provide a performance parameter to compare systems of potentially different designs and locations on an absolute scale. In this case it is necessary to use multiple derate factors or alternatively an accepted PV model to calculate expected performance with actual weather data over the assessment period. The actual performance divided by expected performance would then determine a Performance Index (PI), in this case an Energy Performance Index (EPI). The performance assessment method and associated metrics used must also consider the cost and benefit of the assessment. Generally, the degree of accuracy and depth analysis correlates to the level-of-effort and cost. The level-of-effort related to the amount of data and analysis in the performance assessment should be comparable to the expected benefits gained. Small systems could use simplified methods, with correspondingly greater uncertainty (10% to 20%) but still having the capability of detecting equipment failures and degrading conditions. Larger systems typically have comprehensive monitoring systems and analysis algorithms with correspondingly lower uncertainty (2% to 5%) to detect subtle performance changes. It should be emphasized that system performance is different than system value. The performance of a system is related to actual AC energy output relative to its capability as designed and operated; whereas, value of a system is the system cost relative to the annual AC energy output. Also, performance is different than reliability. Failures affect reliability which in turn affects performance. Figure 1.2 shows the relative types of assessment and the applications. 5

Fig. 1.2 - Performance Assessment Map showing applicability of guideline covered by this report

Cost Effective Approaches to Performance Assessment

Maximum

2% to 5%

Proprietary Algorithm

Proprietary Algorithm

Proprietary Algorithm

2.1 Performance Metrics

Section 2.1.1 and Table 2.1 below summarize performance metrics commonly used in the industry based on a literature survey and discussions with industry experts. Section 2.1.2 and Table 2.2 summarize performance metrics found to be effective and practical to apply. Section 2.1.3 and Table 2.3 summarize methods to calculate performance metrics based on the results of this project. 2.1.1 Review of Current Industry Performance Metrics Literature Survey The review of currently used performance metrics included metrics from NREL, Sandia, IEC, equipment suppliers, and other organizations. It was found that there is variation among how the metrics are calculated, how they are used, nomenclature, and few industry standards to provide guidance. Some methods appropriately use a ratio of actual performance divided by expected performance, called Performance Index (PI). Some methods have established acceptance criteria which define the minimum output and are used primarily during commissioning. Methods for calculating expected performance included as-build system component ratings, irradiance, irradiation, ambient temperature, module temperature, and typical condition dependent derate factors. Methods for measuring actual AC and DC output power and energy typically used direct measurements, revenue grade watt-hour meters, inverter display, and/or online live-site data. The calculated performance metrics were then typically compared to industry average values for specific technology, established criteria used for commissioning, trend monitoring, and assistance in troubleshooting. Actual system conditions were usually included using Derate Factors, however there was a lack of information on how to select appropriate Derate Factors for some of the conditions. Derate Factors have a large influence on the performance calculation and also introduce significant uncertainty in the calculations. In principle, performance assessment could be based on any of the following: Actual output divided by actual input. This metric is representative of overall system efficiency and a normal system would have a value on the order of 0.1, largely dependent on the module efficiency. No analytical PV model is needed in this case. This metric has limited use most likely due to the negative perception of a low value around 0.1. Actual output divided by expected output. This metric is largely dependent on the system design, quality of installation, and the accuracy of the PV model. A normal system would be on the order of 1.0. This metric is used and can be based on either power or energy. Actual output normalized divided by actual input normalized. An example of this metric is Performance Ratio and is used regularly to compare systems, however, may result in incorrect conclusions if the systems being compared are in different locations with different irradiance and temperature.

Performance metrics can first be divided into short-term and long-term assessment periods. Various degradation mechanisms and intermittent anomalies develop and occur over long-term periods so both periods are needed to complete an assessment. Short-term is considered instantaneous power output when the measurement is taken and denoted by kW (power). Long1

Term, such as weekly, monthly, or annually determines energy and yield, and is denoted by kWh (energy). Performance metrics can also be divided into absolute and relative values. An absolute value can be used to evaluate a system by comparing to industry-wide values resulting in a figure of merit of the system. A relative performance metric can be used to trend a specific system using trend plots of the metric and associated parameters. Both the absolute and relative metrics would provide input to troubleshooting of degraded systems. Measurement uncertainty and error analysis should be addressed and used to define a tolerance band so differences in actual to expected performance that are due to measurement uncertainty would not be used to reach inappropriate conclusions. Some metrics, Yield and Performance Ratio are independent of a PV model, whereas Performance Index is related to calculated expected performance and is therefore dependent upon an accurate PVmodel. Yield and Performance Ratio uncertainty due to temperature variation and measured values is greater than uncertainties in the PV model, therefore uncertainties are lower with Energy Performance Index. Initial review of industry practice found various performance metrics as shown in Table 2.1.

METRIC Yield Performance Ratio Performance Ratio Performance Ratio Specific Production Performance Ratio Performance Factor Performance Index Performance Index Output Power Ratio Output power Output power Specific Production Acceptance Ratio Inverter comparison String comparison Utility billing Performance Ratio, temp. comp. (CPR) Energy Performance Index (EPI) CALCULATION kWh / kWDC STC (kWh/ kWDC STC ) / (H/GSTC) kWh / (sunhours area efficiency) (EActual / EIdeal) * 100% EIdeal is temp. and irrad. compensated MWhAC / MWDC STC

(100 * Net production / total incident solar radiation) / rated PV module eff.

REFERENCE NREL/CP-520-37358 IEC61724 SMA SolarPro, Taylor & Williams SolarPro, Taylor & Williams NREL/TP-550-38603 Sutterlueti SolarPro, Sun Light & Power Townsend SolarPro, Sun Light & Power CEC Commissioning SRP Arizona Utility SolarPro, Taylor & Williams Literature Qualitative Qualitative Qualitative Proposed in this report Proposed in this report

Actual Power / (Rated power * irrad adj. * temp adj * degradation adj * soiling adj * BOS adj)

kWmeasured / kWpredicted kW > CF-6R-PV Table kW > 95% expected MWhAC / MWDC-STC kWactual / kWexpected kWh of multiple similar inverters Imp, Vmp of multiple parallel strings Monthly comparison (kWh/ kWDC *KTemp) / (H/ GSTC) kWh AC actual / SAM AC Expected using actual weather data 2

Yield The standard Yield metric is considered to be the bottom-line indication of how well a system is performing since the purpose of the system is to maximize energy output for a given system size; however, it does not account for weather conditions or design and can only be applied for a consistent assessment period (such as annually). Since Yield increases proportionally with hours of operation, insolation, and lower temperature a high yield due to unusually high insolation can be misleading and potentially even mask a case of a degrading system. A system with an unusually low insolation may be incorrectly judged to have low performance. If systems are being compared using Yield, the hours of operation, insolation, and cell temperature should be equivalent for a fair comparison. The basic Yield equation is shown below as equation 1: kWh Yield = kW 1

The value of a system ultimately comes down to annual AC energy output relative to system cost. Therefore, Yield is a measure of system value rather than performance. Performance Ratio Performance Ratio (PR), as defined by IEC61724 and NREL, is a metric commonly used, however one shortcoming in the basic PR is that normal temperature variation influences PR and is not included in the basic equation. Specifically, cases with low temperature and moderate irradiation (such as late winter) result in higher PR and cases with high temperature and moderate irradiation (such as late summer) result in lower PR. A normal system would have a decreasing PR trend in spring that is normal and could incorrectly be judged to have a degrading system. Hourly data also has variation from morning to afternoon that is difficult to interpret. The seasonal variation of PR can be illustrated using PVWATTS to calculate monthly AC kWh and monthly irradiation. A 100kW system with latitude tilt in Sacramento was arbitrarily selected and analyzed resulting in the plot shown in Figure 2.1. It would appear that the system performance was degrading February through June.

Seasonal Variation of PR

0.8 0.75 0.7 PR 0.65 0.6 0.55 0.5 0 2 4 6 Month 8 10 12

Also as discussed above in Section 1.2, PR is more appropriate to trend a specific system or to compare systems in similar geographic locations rather than used to compare performance of various systems. If PR is used to evaluate a system in San Francisco, CA, compared to a similar system in Daggett, CA, incorrect conclusions would be reached. Even with a lower PR, the Daggett system has higher output and therefore higher performance. One of the advantages of using PR is that the expected performance is not calculated, therefore, a PV computer model is not needed and the uncertainty introduced by the model is not a factor. Performance Ratio, Compensated Compensation for factors such as cell temperature, KTemp, can be applied to the basic PR to adjust the DC power rating from Standard Test Conditions (STC), however since temperature varies continuously with irradiance and weather, the correction must be performed at each time increment (such as, hourly) and the PR calculated hourly. PR . kWh kW K = kWh 1kW 3

Typical hourly data includes night hours when the energy production and irradiance are zero. Dividing by zero is undefined, therefore, PR should be only be calculated using the SUMIF function in Excel to sum kWhAC and kWhSun only when energy production is greater than zero. A daily PR would then be obtained using equation (3). Hourly PR values vary from zero to a maximum either before or after noon depending on conditions and are considered to be of little use for performance assessment. Averaging hourly PR to obtain daily PR is incorrect versus summation of the hourly kWhAC and kWhSun values for the day. 4

Because irradiance and temperature change continuously, it would be beneficial to use a time increment less than an hour, however for practicality an average hourly temperature is considered acceptable unless the assessment is for a large critical system. The 2004 King paper, Ref 1, suggests that hourly averages is acceptable for most assessments, although a Ransome and Funtan paper, Ref. 5, says hourly average under-predicts performance due to the thermal lag when irradiance increases. If additional compensation factors are of interest to be included, such as balance of system losses, angle of incidence, soiling, shading, long-term degradation, etc, it is more practical to include them in the Energy Performance Index (EPI) using an accepted PV model, such as SAM, to incorporate the compensation factors rather than complicating PR. Performance Index Performance Index (PI) was found to be usually related to the actual output of a system divided by the expected output. The expected output was calculated using an accepted PV Model, such as the NREL Solar Advisor Model (SAM), therefore, the accuracy and uncertainty of PI is dependent on the accuracy of the PV Model. Temperature Correction Methods to calculate a cell temperature from ambient or module backside temperature are discussed in the literature, and were compared as summarized in Appendix 6.3 and section 3.3. Various PR temperature compensation techniques to reduce seasonal (or daily in this case) variation were tried with results shown in the Fig 3.2.1 plot, and summarized below. Temperature Compensated PR used averaged hourly measured module temperature for hours with measureable AC output power, 25C for STC temperature, and the temperature coefficient from the module datasheet. This method had the lowest seasonal variation compared to other methods and required less level-of-effort to apply. Weighted Temperature Comp PR used power weighting which gives weight to hours having higher power (or hourly energy) output. Formula used was Average Daily Temp = Sum of (hourly temp hourly energy) / total daily energy. This method was proposed by Townsend in Ref. 3 was helpful in reducing the seasonal variation; however the variation was still present. The Sunpower method in Ref 4 uses an irradiance weighted TMY annual average cell temperature that is difficult to apply for less than an annual period and resulted in PR values that matched the Basic PR formula, therefore further analysis is planned. The method to compensate for temperature using the Sandia temperature model with a and b coefficients, or textbook model, or NOCT method are used to obtain the best estimate of cell temperature and are not necessarily the best to reduce PR seasonal variation. A temperature adjustment as proposed by Sunpower is most likely a better method, but at this time did not produce better results, therefore additional analysis is planned. 2.1.2 Summary of Effective Performance Metrics The industry has used various metrics, often with similar names but different calculation methods, or with different names and similar calculation methods. Some metrics and 5

calculations presented in technical papers are not effective for the purpose intended. As the industry has evolved, data has become more available, and analyses easier to perform; newer methods have been proposed and used. Based on evaluation of these various metrics, those that are considered to be appropriate for assessments are summarized in Table 2.2. In general, performance assessment is the process of measuring or monitoring actual performance and comparing to expected performance. Either the actual performance or the expected performance must be adjusted to account for the actual weather and derate factor conditions. One approach is to adjust the actual system kW AC output up to STC (e.g. apply a ratio of 1000 W/m2 / Gactual) and compare to the expected STC system output from PV model calculations, or the other approach is to adjust the STC output from PV model calculations down to the actual condition (e.g. apply a ratio of Gactual / 1000 W/m2 ). The second approach is appropriate and more commonly used by the industry. Performance Index (PI) is typically the direct ratio of actual divided by expected. Performance Ratio (PR), as defined by NREL and IEC, is a normalized version of output divided by input and an expected value is not directly included so it is not the typical output divided by input ratio. However, by including the DC STC rating and irradiation ratio, the PR ratio is actual output divided by rough calculation of expected output. If compensation factors are added to PR, such as temperature, balance of system losses, etc, it converts PR to a ratio with expected value in the denominator and is then similar to PI. Energy Performance Index (EPI) is a ratio of actual kWh AC divided by expected kWh AC using actual climate data over the assessment period as input to an accepted PV system model with all relevant derate parameters included. Therefore, EPI incorporates the most complete metric for performance assessment. In the paragraphs that follow, the four metrics which are considered to be appropriate for performance assessment are discussed.

Performance Assessment Guideline See acronym list at end of table [Rev. B]

PURPOSE OF REQUIRED DATA ASSESSMENT Condition Directed Maintenance of Existing System: Determine if performance is generally as expected, general monitoring. Monthly actual kWhAC, and general recollection of weather. Metered hourly kWhAC, actual hourly weather data (GHI, ambient temp, wind, etc). Note: Need accurate PV model (e.g. SAM). Metered hourly kWhAC, actual hourly weather data (GHI, ambient temp, wind, etc) for both systems. Note: Need accurate PV model (e.g.SAM). Metered hourly kWhAC, actual hourly weather data (GHI, ambient temp, wind, etc) for both systems. Note: Need accurate PV model (e.g. SAM) to calc POA irradiation and module temp but not for expected output or derate factors. METRIC Trend of monthly inverter output values EQUATION METHOD 1. Obtain monthly kWh report 2. Recall general weather 3. Compare months by ratio or difference of kWhAC Level of Effort = Low, Uncertainty = High 1. Obtain hourly metered kWhAC for assessment period. 2. Sum hourly kWhAC, for each day using Excel SUMIF. 3. Calc DNI, DHI from GHI using NREL DISC. 4. Create climate file in SAM in TMY3 format. 5. Model system in SAM, estimate derate factors. 6. Calc hourly kWhAC output. 7. Sum hourly kWhAC for day using Excel SUMIF. 8. Calc EPI for day, plot daily trend. Level of Effort = Moderate, Uncertainty = Low 1. Obtain hourly metered kWhAC for both systems. 2. Calculate DNI, DHI from GHI use NREL DISC. 3. Create climate file in SAM in TMY3 format. 4. Model system in SAM, save hourly data in Excel. 5. Total hourly kWhAC and calculate daily EPI. 6. Repeat for other system. 7. Compare EPI values using Comparison Ratio (CR). Level of Effort = High, Uncertainty = Moderate 1. Obtain hourly metered kWhAC data. 2. Calculate hourly POA irradiation, using hourly GHI, DISC to calc DNI, DiffuseHI, input to 1kW system in SAM, output Data Table with Irradiance Total hourly (Beam+Diffuse) w/o soiling,shading, save to Excel file. 3. Calc hourly cell temp and compensation factor KTemp using power weighted (or other, to be determined) cell temp approach for TCell-TSTC. 4. Calc daily sum of hourly kWhAC divided by kWSTC times KTemp. 5. Calc daily sum of hourly kWhPOA from step 2. 6. Calc monthly PR by dividing step 4 value by step 5. Daily: Level of Effort = High, Uncertainty = High REF.

Determine if maintenance is needed, based on daily trend of the metric for the specific system accounting for weather and derate factors.

EPI = Actual kWhAC / Expected kWhAC from SAM using actual hourly weather data over the assessment period.

Determine if maintenance is needed, based on comparison of daily metric of specific system to a reference system of different design and/or location.

Determine if maintenance is needed, based on monthly trend of the metric of the specific system, accounting for weather only.

Comparison Ratio (CR) of daily EPI of specific system divided by daily EPI of reference system. Trend of monthly or annual values of temperature compensated Performance Ratio (CPR). Seasonal variation must be established before CPR metric is dependable.

CR=EPISyst1/EPISyst2, where, EPI = Actual kWhAC / Expected kWhAC from SAM using actual hourly weather data over the assessment period, needed for both systems.

SJSU SOLAR PROJECT - 007 January 2012; Rev. C Monthly: Level of Effort = High, Uncertainty =Moderate Commissioning of New System (Operating 0 to 6 months) Determine if system is initially working based on actual versus expected instantaneous output power, within an estimated uncertainty range. Metered daily kWhAC, plane of array irradiance (IPOA), module backside temp. Dont need accurate PV model. 1. Obtain hourly metered kWhAC data. 2. Obtain POA irradiance and module backside temp. 3. Obtain temp coefficient for power from module datasheet and calc temp compensation. 4. Estimate uncertainty of values, use prop of uncertainty (e.g. SRSS) to estimate uncertainty range of PPI. 5. Calc instantaneous PPI. Level of Effort = Low, Uncertainty = Moderate 1. Obtain hourly metered kWhAC for assessment period. 2. Sum hourly kWhAC, for each day using Excel SUMIF. 3. Calc DNI, DHI from GHI using NREL DISC. 4. Create climate file in SAM in TMY3 format. 5. Model system in SAM, estimate derate factors. 6. Calc hourly kWhAC output. 7. Sum hourly kWhAC for day using Excel SUMIF. 8. Calc EPI for day, plot daily trend. Level of Effort = Moderate, Uncertainty = Low

Determine if system is working based on actual versus expected output energy, within an estimated uncertainty range..

Metered hourly kWhAC, actual hourly weather data (GHI, ambient temp, wind, etc). Note: Need accurate PV model (e.g.SAM).

EPI = Actual kWhAC / Expected kWhAC from SAM using actual hourly weather data over the assessment period.

Determination of Industry Parameter: Determine annual Yield. Although Yield is not an effective performance assessment metric because irradiance and temperature are not considered, Yield is widely used. Metered total hourly or Comparison of system annual kWhAC, output. Yields must be on comparable time base (e.g. annual) since Yield increases in proportion to time. Yield is a metric of system value more so than performance. Determine Performance Metered hourly or daily kWhAC, and Ratio (PR) per IEC61724 without compensation measured plane of factors. Note: PR is not array irradiation dependent on PV model or (POA).

Yield

Yield=(kWhAC/kWDC STC )

1. Obtain hourly or assessment period metered kWhAC system output values. 2. Sum hourly kWhAC, for assessment period using Excel SUMIF function. 3. Determine nominal DC rating of array at STC. 4. Calc Yield using equation. Level of Effort = Low, Uncertainty = Low

1. Obtain hourly metered kWhAC data and kWhPOA. 2. Calc daily (or monthly) total of hourly kWhAC and kWhPOA using Excel SUMIFS function. 3. Calc daily (or monthly) PR from the totals using IEC61724 equation.

SJSU SOLAR PROJECT - 007 January 2012; Rev. C assumed derates factors. period. Level of Effort = Low, Uncertainty = Low 1. Obtain hourly metered kWhAC data. 2. Calculate hourly POA irradiation, using hourly GHI, DISC to calc DNI, DiffuseHI, input to 1kW system in SAM, output Data Table with Irradiance Total hourly (Beam+Diffuse) w/o soiling,shading, save to Excel file. 3. Calc daily (or monthly) total of hourly kWh and kWhPOA using Excel SUMIFS function. 4. Calc daily (or monthly) PR from the totals using IEC61724 equation. Level of Effort = Low, Uncertainty = Moderate 1. Obtain hourly metered kWhAC data. 2. Calculate hourly POA irradiation, using hourly GHI, DISC to calc DNI, DiffuseHI, input to 1kW system in SAM, output Data Table with Irradiance Total hourly (Beam+Diffuse) w/o soiling,shading, save to Excel file. 3. Calc hourly cell temp and compensation factor KTemp using power weighted cell temp approach for TCell-TSTC. 4. Calc daily sum of hourly kWhAC divided by kWDC STC times KTemp. 5. Calc daily sum of hourly kWhPOA from step 2 above. 6. Calc daily PR by dividing step 4 value by step 5 value. Level of Effort = High, Uncertainty = High

Metered hourly kWhAC, actual irradiance data (GHI) Use PV model (e.g. SAM) to calc POA irradiation.

Determine performance of a system by comparing PR values of multiple Metered hourly systems. Temperature kWhAC, actual hourly compensation is included weather data (GHI, to reduce variation due to ambient temp, wind, temperature, (CPR). etc) for both systems. Compensation for other Need accurate PV factors also possible, but model (e.g.SAM) to consider using CR of EPI calc POA irradiation method. Note: PR is not and module temp. dependent on PV model or assumed derates factors. Investment Potential Site Comparisons: Determine LCOE. LCOE is an effective performance assessment metric but is dependent on many assumptions, and therefore beyond the scope of this report. Determine if system meets performance guarantee, based on metric agreed upon in original system contract and is therefore beyond the scope of this report.

Comparison Ratio (CR) of CPR of specific system divided by CPR of reference system. Compensated Performance Ratio (CPR) for daily, monthly, annual period.

SJSU SOLAR PROJECT - 007 January 2012; Rev. C Acronyms: PI = Performance Index, ratio of actual divided by expected PPI = Power Performance Index, instantaneous actual power divided by expected power PR = Performance Index CPR = Temperature compensated Performance Ratio EPI = Energy Performance Index CR = Comparison Ratio of EPI or CPR actual divided by EPI or CPR expected. kWhAC = AC Energy at system output at utility meter kWhPOA = Insolation normal to the plane of the array kWDC STC = DC rating of array at standard test conditions (STC) POA = Plane of array, as related to incident irradiance normal to an array surface plane SAM = Solar Advisor Model, from NREL TMY3 = Typical Meteorological Year, third version GHI = Global Horizontal Irradiance (based on 1 hour period average) DNI = Direct Normal Irradiance, normal to beam component of irradiance DISC = Direct Insolation Solar Code, developed by NREL to calc DNI DiffuseHI = Diffuse irradiance incident on a horizontal surface KTemp = Temperature compensation factor based on (TCell-TSTC)

2.1.2.1 Short Term Assessment Power Performance Index: The Power Performance Index (PPI) is the instantaneous actual AC kW power output divided by the expected AC kW power output including derate factors. The calculation of expected output using rated DC STC power times adjustment factors is called the PKs method in this report. (Refer to Flow Chart in Appendix 7.1) If irradiance and temperature measurements are taken manually, it is important to measure and note how steady the values are so that the inverter actual power value is correlated to the actual irradiance and actual temperature values. Experience has shown that apparently clear sky conditions can result in significant variations of irradiance over a short time. 1. Visually inspect system - Determine as-built configuration, identify conditions affecting performance, estimate typical derate factors per PVWATTS description or similar documentation and combine to obtain derate K factor (KDerate). 2. Measure Plane of Array (POA) irradiance. If only horizontal data is available (GHI), convert to POA using NREL DISC spreadsheet to calculate DNI, DHI and use Isotropic model to convert to POA irradiance. The Isotropic model formula based on Duffie/Beckman Reference 1 is: 1 + cos 1 cos = + + 2 2 3. Calculate irradiance K factor, KIrrad, from: = 1000

4. Measure module backside temperature and add 3C as an estimate of cell temperature, per King 2004 paper. If backside temperature is not available, measure ambient temperature and calculate cell temperature using NOCT value on module datasheet using the following formula: = + 20 800

5. Calculate temperature K factor (KTemp) for temperature relative to STC using the following formula, where is the power temperature coefficient and is a negative number, such as typically - 0.005/C. = 1 + 25 6. Calculate expected AC output power (kW): 1

= 7. Measure actual AC output power (kW) or use inverter displayed value at a time which is correlated with the irradiance and module temperature measurements. 8. Calculate ratio of measured actual AC power to expected power, define values as Power Performance Index (PPI) =

9. Estimate uncertainty values for measured and calculated values (apply propagation of uncertainty method using square root sum squares of each uncertainty in %). 10. Evaluate PI. If PI = 1.0 uncertainty, short-term system performance is acceptable, proceed to Long Term Assessment.

2.1.2.2 Long Term Assessment Performance Ratio: Long-Term assessment is needed to identify system degradation due to intermittent faults, outof-service time (outages), unavailability, low light performance, angle of incidence effects, solar spectrum effects, light induced degradation, and other conditions that cannot be detected during the Short Term assessment period using methods described in Section 2.1.2.1. The basic PR calculation uses the standard yield equation in the numerator and the actual measured plane of array (POA) irradiation summed over the assessment period divided by standard irradiation in the denominator. The units work out to be hours divided by hours. The numerator is equivalent to the number of hours the system operated at the DC STC rating and the denominator is equivalent to the number of peak sunhours of irradiation. Both the measured irradiation and standard irradiance are in terms of meter2, and cancel directly. kWh kW PR = kWh /m 1kW/m 2

Both the numerator and denominator are summations of the measured increment data, such as hourly, over the assessment period. The assessment period can be daily, weekly, monthly, annually. Analysis of data required filtering to eliminate hours with zero irradiance since dividing by zero is undefined. The Excel filter function was used in various scenarios such as to include mid-day hours and for irradiance greater than a defined value, such as 600 kWh/m2. Effectively, this was a mid-day flash test. Excel spreadsheet data for hourly AC output and hourly irradiance were summed and averaged using the Excel SUMIFS and AVERAGEIFS formulas.

Detailed Description of Long-Term Performance Ratio (Appendix 6.1 Flow Chart). PR = (kWhAC/DCRated)/(kWhSun/1kW) 1. Install POA irradiance and module temperature datalogger, or obtain access to existing POA data, or use data from another local site adjusted from horizontal to POA and ambient temperature using method discussed in Section 2.1.2.4 using NREL DISC Excel spreadsheet. 2. Read inverter kWh total on inverter display at beginning of assessment period, or obtain access to existing monitoring data. 3. Read totals for irradiation from datalogger and kWh from inverter (or from monitored data) at end of assessment period, calculate differences to obtain actual kWh of irradiance and kWh of AC energy over the assessment period. For simpler approach for annual PR estimate, use PVWATTS annual irradiation value. Annual PVWATTS irradiation is typically less discrepant from actual than monthly PVWATTS irradiation values. 4. Calculate Performance Ratio (PR). Calculate the hourly PR based upon the IEC61724 formula, after first filtering through the measured data and removed all hourly data sets that did not have a measured plane of irradiance of 600 W/m2. With the remaining hourly data sets a PR is calculated. A daily PR is then calculated by averaging the hourly PRs from each day. Since a minimum of 600 W/m2 was used, there are days in which a PR was not calculated. 5. Compare PR value to typical industry values, or to similar systems in other locations, or to previous PR values of the same system to establish trend of performance depending on the purpose of the assessment. 6. Evaluate PR. If PR uncertainty is within Long-Term criteria, system performance is acceptable. Otherwise proceed to investigate performance shortfall of individual components (Flow Chart pages 2 and 3).

2.1.2.3 Long Term Assessment Performance Ratio, Compensated: The basic Performance Ratio (PR) is directly influenced by energy (kWh) output, which is directly influenced by irradiation (kWh/m2) in inversely influenced by module temperature. Since the basic PR equation accounts for irradiation, changes in irradiation will have little direct effect on PR, however, since changes in temperature are not accounted for, the basic PR will increase as temperature decreases. In order to use a metric which is more indicative of system condition rather than design or environmental conditions that are outside the control of the owner, compensation factors can be added to the basic PR equation. One method to include temperature compensation is to adjust the DC rating in the numerator using the power temperature coefficient provided on the module manufacturers data sheet relative to the STC temperature of 25C. Other methods used for hourly calculations weight the compensation factor by the power output for the hour (energy), or to use factors based on average annual ambient temperature. Other factors besides temperature also affect PR and are also outside the control of the owner, 3

such as design, shading, degradation, balance of system, and could be included as compensation factors, however the basis for calculating these factors for use with PR is not well understood. Therefore if compensation other than temperature is desired, it is more practical to calculate Long-Term Energy Performance Index (EPI) using actual irradiation and temperature in one of the accepted models, such as SAM, as described in Section 2.1.2.4 below. If the purpose of the assessment is only to evaluate a specific system, trend analysis using a temperature compensated PR is reasonable because it is not influenced by the accuracy and/or uncertainty of a PV model.

Detailed Description of Long-Term Performance Ratio with temperature compensation (Appendix 6.1 Flow Chart). CPR = [(kWhAC/(DCRated KTemp )]/(kWhSun/1kW) 1. The procedure is the same as PR above in Section 2.1.2.2 but with temperature compensation added. 2. Obtain average ambient temperature from logged or monitored data. 3. Calculate cell temperature using formula in Duffie/Beckman, Reference 4 4. Calculate annual average cell temperature during the times included in the filtered range. 5. Obtain the power temperature coefficient () from the module data sheet. 6. Calculate KTemp using [1 - (Cell Temp annual average cell temp)]. 7. Using hourly data, Excel SUMIFS function, and KTemp calculate the daily PR values. 8. Average the daily PR values for the assessment period, such as monthly or annual. 9. Maintain records to monitor trend of temperature compensated PR to evaluate possible system degradation. 10. Compare PR value to typical industry values, or to similar systems, or to previous PR values of the same system to judge performance and/or trend depending on the purpose of the assessment. 11. Evaluate PR uncertainty. If PR uncertainty is within criteria, system performance is acceptable. Otherwise proceed to investigate performance shortfall of individual components (Flow Chart pages 2 and 3). Literature reviewed from PV systems found that typical uncompensated PR ranges from 0.60 to 0.80. System should be able to perform at near PR of 0.8 when there is high DC to AC conversion efficiency Derate Factors, potentially totaling 20% to 25%, due to: Manufacture tolerance and mismatch 5% Aging (10% after 10 years of age) 5% for new system

PR Acceptance Criteria: PR > 0.80 system OK PR between 0.80 and 0.60 normal, consider maintenance PR < 0.60 Corrective action recommended

Detailed description of Temperature Correction Procedure When applying a temperature compensating factor we decided to take two different approaches. The first approach we used was by adding a temperature compensating factor related to the nominal operating cell temperature (NOCT). The second approach used was by adding a temperature compensating factor developed by Timothy Dierauf from SunPower. PR = [(kWhAC/(DCRated KTemp )]/(H/ GSTC) NOCT Approach KTemp,NOCT = 1 + (TCell - TCell,STC) KTemp,NOCT Temperature correction factor using the NOCT approach TCell Calculated operating cell temperature (C) TCell Calculated operating cell temperature under STC (25 C) TCell = Ta + (TNOCT Ta,NOCT)*(H/GNOCT)*(1-(/)) Ta Ambient temperature (C) TNOCT nominal operating cell temperature (NOCT) at NOCT test conditions Ta,NOCT Ambient temperature at NOCT test conditions (20 C) H Measured irradiance in the plane of array (W/m2) GNOCT Irradiance at NOCT test conditions (800 W/m2) c/() 0.083/0.9 can be used to estimate this value 1. The procedure is the same as calculating long-term PR but with a temperature correction. 2. Calculate the temperature correction factor through the equations above. SunPower Approach KTemp,SP = 1 + (TCell - TCell_sim_avg) KTemp,SP Temperature correction factor using the SunPower approach TCell Calculated operating cell temperature (C) TCell_sim_avg Cell temperature computed form annual average measured meteorological data (C) TCell = Ta + H*e(a + b*WS) + (H/GSTC)*3 Ta Ambient temperature (C) 5

H Measured irradiance in the plane of array (W/m2) a, b = Empirically determined coefficients establishing the rate at which module temperature drops as wind speed increases GSTC Irradiance at standard test conditions (1000 W/m2) TCell_sim_avg = ( GPOA_sim_j*Tcell_sim_j) / (GPOA_sim_j) GPOA_sim_j Simulated average cell temperature from typical weather year (C) TCell_sim_j Simulated cell operating temperature for each year (C) j each hour of the year (8,760 hours total) 1. The procedure is the same as calculating long-term PR but with a temperature correction. 2. Calculate the temperature correction factor through the equations above. 3. TCell_sim_avg must be calculated prior to running a filter that removes all hours that do not meet the minimum hourly irradiance of 600 W/m2. 4. When calculating the cell operating temperature, the values for a and b can be determined by referencing Photovoltaic Array Performance Model by D. L. King. Table 4 below contains the empirically determined coefficients based on module type. Table 4. Empirically determined coefficients calculated by Sandia National Labs Module Type Mount a b T (C) Glass/cell/glass Open rack -3.47 -0.0594 3 Glass/cell/glass Close roof mount -2.98 -0.0471 1 Glass/cell/polymer sheet Open rack -3.56 -0.0750 3 Glass/cell/polymer sheet Insulated back -2.81 -0.0455 0 Polymer/thin-film/steel Open rack -3.58 -0.1130 3 22X Linear Concentrator Tracker -3.23 -0.1300 13 PR Compensation for Seasonal Variation: The above temperature compensation is an attempt to reduce the seasonal variation of PR. Another approach that was tried was to calculate a ratio of two PRs. In the numerator, the basic monthly PR was calculated using actual kWh AC and actual irradiation, and in the denominator a monthly PR was calculated using TMY3 input to SAM for kWh AC and TMY3 insolation, as shown in Figure 2.3. In this case the TMY3 data had minimal seasonal variation of PR and therefore did not normalize the actual PR value to remove seasonal variation. This approach can be considered after accumulating multiple years of data at the specific site in place of the TMY3 data.

1.00 0.90 0.80

PR TMY3+SAM

PR and PR Ratio

PR Climate+Meter PR Ratio

2.1.2.4 Long-term Assessment Energy Performance Index (EPI): When compensation factors are added to the PR equation, the equation is equivalent to Performance Index of Actual Energy divided by Expected Energy for the assessment period. Note that the PR equation which includes compensation for temperature or other factors is identical to the equation for Energy Performance Index (EPI), based on the following algebra: kWh kW PR = kWh 1kW kWh kW K K K = kWh 1kW kWh

PR .

PR . = EPI =

kW

kWh K K K 1kW 7

This equation is of the form of the Power Performance Index (PPI) presented earlier, however in this case it is in terms of energy and is EPI. Acceptable models (e.g. SAM) inherently include compensation factors as part of the model. It is necessary to input actual weather data in a climate file. In the case of SAM, actual hourly data for GHI, DNI, DHI, dry-bulb temperature, and wind speed can be incorporated into TMY3 format file and read by SAM. Other parameters included in the TMY3 file, such as dew-point, relative humidity, pressure, and albedo can be assumed to be acceptable from the original TMY3 file for the specific location. The rate of change of the compensation factors affects the time frame over which the summation is performed. The procedure to calculate EPI is listed below: Detailed Description of Long-Term Performance Index Procedure (Appendix 6.1 Flow Chart) EPI = Actual energy output / Expected energy output When calculating the expected energy output we used System Advisor Model (SAM) 2011 created by National Renewable Energy Laboratory (NREL). In order for SAM to create an expected energy output, it draws data points from a selected Typical Meteorological Year (TMY3) data set. In SAM, there is a function that allows you to create a TMY3 file specific for your specific PV system. In order to do so SAM requires a base TMY3 file and various data points measured from your PV system. Procedure: 1. Download an available TMY3 file to use as your base TMY3 file. We chose the San Jose International Airport since it was the nearest available TMY3 data. When downloading the file save it as a CSV file and do not open this file in Excel. SAM will automatically replace the values in this file using the SAMs Create a TMY3 File function. 2. Click and open the function Create a TMY3 File 3. Enter in the following data regarding the sites name and location. a. Site code b. Station name c. Station state d. Station time zone (GMT) e. Station latitude (DD) f. Station longitude (DD) g. Station elevation (m) 4. SAM also requires one year of hourly data for the following sections. From the data gathered by Deck Monitoring, we were able to obtain information for each section except DNI and DHI. a. Global horizontal irradiance, GHI (W/m2) 8

b. Direct normal irradiance, DNI (W/m2) c. Diffuse horizontal irradiance, DHI (W/m2) d. Dry-bulb temperature (C) e. Dew point (C) f. Relative Humidity (%) g. Pressure (mbar) h. Wind Speed (m/s) i. Albedo (unitless) 5. In order to calculate the Direct Normal Irradiance (DNI), we used the Direct Insolation Solar Code (DISC) model developed by Dr. E. Maxwell of the National Renewable Energy Laboratory, available on-line. In order to calculate an estimated DNI for each hour, DISC required the latitude, longitude, time zone, pressure, and global horizontal irradiance. Once DISC calculated DNI we replaced any negative calculated DNI with 0. 6. The Diffuse Horizontal Irradiance (DHI) was also calculated through DISC. Using the DNI, and z zenith angle, calculated by DISC, and the relationship between GHI, the direct horizontal irradiance (dHI), and DHI, we were able to calculate DHI. We also replaced any negative DHI with 0. = Where: = cos

7. Copy the hourly data for each subset in step 4 from your data set and paste it into the Create a TMY3 Function. 8. Using the Create a TMY3 Function, create a TMY3 file for your PV system. 9. Use SAM to calculate an expected hourly generation (kWhAC) based on the created TMY3 data file. 10. Calculate an hourly performance index using the measured energy generated and expected energy generation from SAM. 11. Apply a filter removing all hours where less than 1200 kWh were generated. This number can vary dependent on your criteria. A plot of EPI is provided below from the 600kW PV system applying the above method. It shows a potential performance problem in late summer that could be investigated, such as soiling. The EPI value should be investigated more to determine if some portion is due to normal seasonal variation although cell temperature has been compensated directly in the SAM PV model.

1.00 0.95 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 0 1 2 3 4 5 Month 6 7 8 9 10 11 12 Meter PI PI

2.1.3 Performance Metric Calculation Methods The method used is dependent upon the purpose of the assessment and must consider the cost and benefit of the assessment. Generally, the level of accuracy and depth correlates to the level of effort and cost.

10

Level of Effort Minimal Minimal Method Utility bill compared to previous months billing, on-line inverter monitoring if available. Instantaneous Power Performance Index (PPI) of actual AC power output divided by expected AC power output including derate factors. PExpected = PDC STC KIrrad KTemp KDerate This method is termed the PKs method in this report. Plane of array irradiance and cell temperature are needed. PActual is obtained by direct measurement or inverter display. Initial PPI and long-term Performance Ratio (PR) of specific system trend of Actual Yield to Reference Yield. PR = (kWhAC/DCRated)/(kWhSun/1kW) Initial PPI and long-term compensated Performance Ratio (CPR) for factors that are outside the control of the system owner, such as temperature to compare different systems. CPR = [(kWhAC/(DCRated KTemp )]/(kWhSun/1kW) If other compensation factors are of interest to be included, use EPI approach (below). Long-term Energy Performance Index (EPI) of actual AC energy output to calculated output from accepted program (e.g. SAM) with input of actual weather conditions (irradiance, ambient temp, wind) and derate factors over the assessment period of day, week, month or year. Proprietary algorithms Frequency Monthly Periodic

Moderate

Periodic

Moderate

Periodic, depending on level of monitored data available Periodic, depending on level of monitored data available Continuously

Extensive

Extensive

2.1.4 Uncertainty and Significance Tests When calculating the expected output for a PV system each component affects the expected value in different magnitudes. Table 7 displays the percentage for each factor and how it can contribute to the uncertainty of the expected calculation. Table 7. Effect on monthly energy production calculations (IEC=PVPS T2-07:2008)

11

One area that can be improved is the manner in which data is collected or monitored. Currently the data been provided and is an average of data during a one hour time period. In doing so the hourly averaging of data underestimates the actual energy production during high irradiance conditions. This occurs due to averaging the fluctuations of irradiance over an hour. With large fluctuations the power generated will adjust quickly, however the module operating temperature will adjust slowly and remain at a lower temperature.

2.2

2.2.1 Review of PV Module and System Modeling Methods The expected performance is calculated using measured irradiance, cell temperature, and estimated derate factors and a mathematical model of the modules and system components. Commonly used industry models using typical meteorological conditions are intended for system design and prediction of long-term future performance but are not appropriate for performance assessment where actual irradiance and temperature conditions are needed. Actual weather data can be formatted into TMY3 format by SAM and used for performance assessment over the assessment period. Several methods available to calculate expected performance are tabulated below: Table 2.1 - CURRENT MODELS OF EXPECTED PERFORMANCE MODEL 5-Parameter METHOD Determine equivalent circuit values adjusted for actual irradiance and temperature. Used to calculate IV curve and MPP. Use Sandia module tested data coefficients and effective irradiance Sunhours area efficiency Determine coefficients from standard module data, adjust for actual irradiance and temperature Use irradiance K factor on current and temperature K factor on voltage, and typical derate factors from PVWatts NREL developed model that generates an expected energy output based on TMY3 Estimates direct normal irradiance (DNI) based upon latitude, longitude, and global horizontal irradiance (GHI) Uses a polynomial regression model in order to predict power. Takes into account irradiance, ambient temperature, and wind 12 REFERENCE CEC, DeSoto paper

Sandia Empirical Efficiency Characteristic Parameters Correction factors, denoted PKs method System Advisor Model Direct Insolation Solar Code (DISC) model PVUSA Rating Method

Sandia, D. King 2004 paper SMA Rausehenbach eqns. Fitzpatrick paper, North Carolina Solar Center SolarPro, Sun Light & Power article SAM developed by NREL Dr. E. Maxwell of the National Renewable Energy Laboratory IEA-PVPS T2-07:2008

speed. A practical and readily available computer program, which incorporates the first three models, is Solar Advisor Model (SAM) available at NREL.gov website. Background on the models is summarized below. More complete information is available in the literature and references. Sandia/King Model The Sandia PV model is of the PV module and based on empirical curve fit equations with data obtained from specific outdoor tests. Rauschenbach Model The model [Ref. 4] ] develops I-V I V curve equations for a module, or array, based on Voc, Isc, Vmp and Imp, referred to as the characteristic parameters. The characteristic parameters vary with temperature. Temperature coefficients are used to calculate the value of characteristic parameters at operating temperature. Voc and Vmp are independent of insolation. Isc and Imp are directly proportional to irradiance. The values of Isc and Imp at a given irradiance are obtained by scaling the reference value by a factor equal to the ratio of operating irradiance to the standard irradiance. The model is accurate with less than 5 % error, when compared with experimental curves. The model has been tested on crystalline silicon, multi multi-crystalline crystalline and amorphous silicon on modules, and arrays.

5-parameter model A solar cell (or module or array of modules) can be modeled [ [Ref. 2] using the equivalent circuit shown below. The equivalent circuit consists of a diode in parallel with a lightlight dependent current source IL. The losses in the cell are represented by a parallel resistance Rsh, and a series resistance Rs.

The I-V V relationship for the above equivalent circuit is given by the equation below: =

The 5-parameters parameters viz. IL, Io, Rs, Rsh and a, a modified ideality factor determine the shape of 13

the I-V curve. In general, the 5-parameters vary with effective irradiance, cell temperature, and the incidence angle. They can be determined from the information presented in the modules datasheet. DeSoto [2] shows how to calculate the 5 parameters at reference conditions, and at operating conditions. Therefore, it is possible to predict the power output at any operating conditions. 2.2.2 Tabulated Comparison of Models and Methods The open literature has many comparisons of PV computer models that are currently accepted by the industry. Due to the availability of SAM and the brief comparison to an independent computer model, EPBB, based on PVWATTS, as reported in the Markley Ref 9 SolarTech paper, SAM is considered to be an acceptable model for use in the EPI method when actual weather data is input. It is worth pointing out the Rauschenbach Model PV model, because of its ease of use for calculating IV characteristics and programming into Excel models.

2.3

2.3.1 AC Measurements Per industry practice. 2.3.2 DC Measurements Per industry practice. 2.3.3 Comparison of Direct Measurements to Inverter and Monitoring System Measurements The actual AC and DC performance is measured using conventional voltage, current, and power devices. The inverter output display and remote monitoring capability (if available) is also used after validation using direct measurements. With some inverters, it may be difficult to calculate input DC power and output AC power to estimate the inverter efficiency. It was found that a Crest Factor was needed on the DC current to account for non-sinuoidal waveform since the digital multimeter calculated the RMS value assuming sinusoidal waveform. The actual and expected values are based on measurements and assumptions, all of which introduce uncertainty ranges and directly affect results. Statistical analysis of data can be used to determine validity of results consideration propagation of measurement and assumption errors. Uncertainty in derate factors can be reduced by inspection and calculation (such as DC wiring losses) specific to a design. The actual real-time irradiance and temperature data can be obtained from solaranywhere.com, or CIMIS (California Irrigation Management Information System), monitored systems with live14

site public data access, such as from FatSpaniel (PowerOne) on-line live sites, or potentially satellite data. Research should be undertaken to find other sources of real time weather and irradiance data for use in calculating long-term Performance Ratio.

2.4

Performance Shortfall Investigation - If either or both the PI or PR do not meet criteria, the array strings, inverter, and/or balance of system (BOS) components should be investigated for root cause of the shortfall (Flow Chart pages 2 and 3). String level IV curves can effectively be determined using Solmetric PVA600 or similar product. The following procedure was used on the 400kW Local System using direct measurements. 1. Measure POA irradiance and module temperature. 2. Estimate Derate Factors from visual inspections, supplier data, and calculations. 3. Calculate expected instantaneous performance of one array string for measured irradiance and temperature using Sandia or 5-Parameter model to determine Isc, Voc, Imp, Vmp, FF. 4. Disconnect strings (open fuses at combiner box) from inverter and measure Isc (make and use test switch box with 600 VDC rated switch to short string) and Voc. 5. Connect individual string to inverter and measure Imp and Vmp. 6. Compare the measured to expected values with uncertainty estimates (uncertainty bars) and plot per Figure 2.1. 7. The above steps can be performed using an industry IV curve tracer analyzer, if available. 8. If range of uncertainty bars overlaps Actual / Expected Ratio = 1.0, string is acceptable, proceed to investigation of inverter and/or BOS. If the string, inverter, or BOS shortfall is identified, proceed with troubleshooting to investigate root cause. Refer to the troubleshooting matrix (Flow Chart page 4) for possible causes of performance shortfall.

15

Figure 2.1 String Comparison Data Used on Representative Strings of the 400kW System

1.250 1.200 1.150 1.100 1.050 1.000 0.950 0.900 0.850 0.800 Isc Ratio Voc Ratio Imp Ratio Vmp Ratio FF Ratio Power Ratio Ratio of Actual/Expected

An economic analysis is needed to determine the cost and benefit of possible maintenance options. a) b) c) Calculate Present Worth (PW) of cost of maintenance on an expected schedule for a minimum of a 1 year period or through the maintenance interval for major maintenance. Calculate PW of savings expected from increased system output (kWh) for the 1 year period. If PW for maintenance is less than PW of savings, it is cost justified to performance cleaning or maintenance.

The above is based on an assumption that a typical cleaning schedule repeats annually. Major maintenance on an interval longer than annually would require calculating PW for the longer interval to be included in the financial analysis.

16

3.1 400kW Local System Power Performance Index (PPI)

The Power Performance Index (PPI) method was applied to an existing PV system and Excel spreadsheets used to analyze data and summarize results and are attached below.

17

PV SYSTEM FIELD PERFORMANCE ASSESSMENT:

Customer/Location: Accurate Solar Assessment Date: 6/21/2011 SJSU Team Lead: Ivica

SYSTEM DESCRIPTION:

System Configuration (per inverter): Number of modules/string = Number of strings = Module tilt angle (degrees) = Module azimuth (degrees) = Total Rated DC Power at STC (W) = 11 3 60 176 5445 Module nameplate data and rated parameters at STC (1000 W/m2, 25C, AM1.5): Mfg = Schott Voc (volts) = 4301 Model = SAPC 165 Isc (amps) = 5.46 Power = 165 Vmp (volts) = 34.6 Matl Type= mc-Si Imp (amps) = 4.77 [Product of module power x number modules x number strings] VoltageTempCoef.= CurrentTempCoef.= PowerTempCoef.= * * -0.005

Estimated module related derate factors: Module mismatch = 0.97 Module Soiling = 0.93 Mfg Tolerance = 0.92 Shading = 1 System Derate Factor = 0.738 Estimated string related derate factors: DC Voltage drop = 0.99 MPPT factor = 0.98 Estimated system related derate factors: Inverter efficiency = 0.96 AC voltage drop = 0.99 Age(years)= [Product of above 8 factors, refer to PVWATTS for explanation.] 6 0.965

1.Visually inspect system - Determine as-built configuration, identify conditions affecting performance, estimate derate factors per PVWATTS docs. 2.Measure Plane of Array (POA) irradiance, module backside temperature. 3.Calculate adjustment factors (K factors) for irradiance and temperature relative to STC. 4.Calculate expected AC output power (kW), P = PSTC x KIRR x KT x KDF 5.Measure actual AC output power (kW) 6.Calculate ratio of Measured kW / Expected kW, define as Power Performance Index (PPI)

PV SYSTEM FIELD PERFORMANCE ASSESSMENT:

Customer/Location: Accurate Solar Assessment Date: 6/21/2011 SJSU Team Lead: Ivica

INVERTER DESCRIPTION:

Mfg Model Pwr Rating Input Voltage Rating

System measurements to be taken at AC disconnect switch or at inverter output connection box. Irrad-iance (POA) (W/m 807 794 803

2)

Discrep. Meas. AC Meas. AC Meas. AC Meas. AC Measured Voltage line- Voltage lineCurrent line Current neut vs Calc. line neut (%) (volts) (volts) (amps) (amps ) 214.8 17 17 214.8 17.1 17.1 214.8 17 17

* Factor applied to convert from 208 VAC to calculate power output from one leg of 3-phase system.

PKs METHOD

String 1 String 2 String 3 Derate Factor (KS) Derate Factor (KS) Derate Factor (KS) 0.73847793 0.73847793 0.73847793 Irradiance Factor (KI) Irradiance Factor (KI) Irradiance Factor (KI) 0.807 0.794 0.803 Temp. Factor (KT) Temp. Factor (KT) Temp. Factor (KT) 0.805 0.81 0.805

Quantity of STC watts modules per module (watts) 33 165 33 165 33 165

Derate Irradiance Total STC watts Factor (KS) Factor (KI) (watts) 5445 0.73847793 0.807 5445 0.73847793 0.794 5445 0.73847793 0.803

Measured Power Perf watts ac Index (PPI) (watts) 2108.25224 0.807 2120.65373 0.820 2108.25224 0.811

3.2

In order to better understand the relationships that affect PR, data from the Arizona Game and Fish live site 191 kW PV system were analyzed. The live site data was reported hourly and includes system AC output energy, irradiance, module temperature, ambient temperature, and wind speed. Data was only available in one month blocks due to administrative limits. Analysis and plots of data are shown below in Figures 3.2.1, through 3.2.6. A summary of key observations from the analysis include:

Daily output (kWh AC) is linearly proportional to daily input (Irradiation) [Fig. 3.2.3] Hourly uncompensated PR is typically low at noon and lower in the afternoon (3:00PM) than in the morning. However in this data, the afternoon was higher. [Fig. 3.2.6] Daily uncompensated PR is typically higher in late winter and lower in late summer. Uncompensated PR is highest when temperature and irradiation are both low. Uncompensated PR is relatively high when temperature is low and irradiation is high. PR is influenced more by temperature than by irradiance. Change of irradiance directly changes kWh AC so PR ratio is approximately constant. Temperature is an independent effect and not included in basic PR. Plot of Daily PR shows that temperature compensation on DC rating is too severe. Need to develop less sensitive correction method or use Energy Performance Index method instead. Module temperature and daily irradiation are correlated, so difficult to separate the individual effects. When temperature decreases PR increases, so temperature compensation has the potential of reducing variation in PR due to normal seasonal temperature changes.

In addition to Arizona Game and Fish live site, San Jose Tech Museum data was available and basic PR calculated as summarized in table below. Standard Total Solar Nameplate Test Irradiance in Performance Total Useful Capacity Conditions Plane-ofRatio Output Wh Wh Irradiance G Array H PR=Yr/Yf W/m2 Wh/m2

PV System

TestInterval

02/21/1103/24/11 03/31/1105/01/11

191100

1000

253118

29960000

0.6194

185000

1000

150900

20407900

0.7310

Computed PR correlates to published PV System installation PR, ranging from 0.6 to 0.8. 1

Figure 3.2.1 Temperature Compensated PR Trends From 191kW Live Site Data

Daily PR

0.75 60 50 40 0.65 30 0.60 20 0.55 10 0 35 45 Assessment Period (Days) 55 65

Basic PR Temp Comp PR Weighted Temp Comp PR Sunpower Method Measured Cell Temp Total Insolation (kWh)

0.50

Various PR temperature compensation techniques to reduce seasonal (or daily in this case) variation were tried. Additional information on PR and CPR is given in section 2.1.1, and 3.3. Temperature Compensated PR used averaged hourly measured module temperature for hours with measureable AC output power, 25C for STC temperature, and the temperature coefficient from the module datasheet. This method had slightly lower seasonal variation than other methods. Weighted Temperature Comp PR used power weighting which gives weight to hours having higher power (or hourly energy) output. Formula used was Average Daily Temp = Sum of (hourly temp hourly energy) / total daily energy. This method was proposed in Ref. 3. The Sunpower method in Ref 4 uses an irradiance weighted TMY annual average cell temperature that is difficult to apply for less than an annual period and resulted in PR values that matched the Basic PR formula, therefore further analysis is planned. 1

Figure 3.2.2 Hourly Basic PR, 7AM to 5PM, To Remove Insignificant Night Values

Hourly PR Trend

Assessment Period 2/13/2011 + 14 days

1 Hourly PR 0.8 0.6 0.4 0.2 0 0 20 40 60 80 100 120 140 160 180 Assessment Period, Hours

Hourly PR values filtered to eliminate hours before 7 AM and after 5 PM which produced division by zero errors in Excel. It was determined that filtering should be in terms of power output when kWh AC > 0 rather than time.

1500 AC Output, kWh 1000 500 0 0 2 4 6 8 10 12 Insolation, kWh/M^2 y = 116.58x + 17.766 R = 0.9842 Daily Totals Linear (Daily Totals)

0.8 0.7 Basic PR Ratio 0.6 0.5 0.4 0.3 0.2 0.1 0 0 2 4 6 8 10 12 Daily Insolation, kWh/M^2 Basic PR Linear (Basic PR)

0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 20 40 60 Daily Temp, C Basic PR Ratio

Notice that PR slope is greater due to temperature than irradiance from Figure 2.4.

0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 24 48 72 96 120 144 168 192 216 240 Series1

3.3

System Description The solar photovoltaic (PV) system that was used for this study is a 595 kW system installed as a covered parking lot. The system is comprised of Sharp NU-U235F1 solar panels and three SatCon inverters. For this analysis we used hourly data measured by Deck Monitoring between 12/1/2010 and 11/30/2011. For simplicity purposes we also only focused on the data gathered by the 500 kW SatCon PVS500 inverter. PV module: Sharp 235W NU-U235F1 Type of cell: Monocrystalline silicon Module efficiency = 14.4% Temperature Coefficient (Pmax) = -0.00485/C Inverter: SatCon PVS500 480V Time Zone: -8 Altitude: 16 m Azimuth: 93 (S = 0) Tilt: 4

Figure 2. 600kW PV System Performance Assessment Review Based on Table 1, size of the system, and using a moderate effort level, it would be appropriate to look at long-term performance ratio with and without temperature correction factors. For this project we evaluated long-term performance ratio, long-term performance ratio with temperature correction, and long-term performance index. In our case we wanted to calculate PR and PI over a daily interval. 1. Long-Term Performance Ratio (PR, Flow Chart page 1). Standard NREL Performance Ratio (PR), and IEC61724 formula of Actual Yield to Reference Yield. Actual energy

output (kWh AC) normalized to DC rating divided by actual irradiation input normalized to rated irradiance. PR = (kWhAC/DCRated)/(H/GSTC) 2. Long-Term Performance Ratio (PR) of Actual Yield with temperature correction to Reference Yield. PR = [(kWhAC/(DCRated KTemp )]/(H/ GSTC) This is the IEC61724 formula of final yield over reference yield with the addition of the temperature correction in the final yield term. This term is useful for maintenance decisions and should be relatively constant over the year since it compensates for monthly variation of irradiance and temperature. Furthermore, the purpose of adding a temperature correction is to help remove the season bias that is inherent within the standard PR. Typically, the standard PR formula will produce larger values during winter season and lower PR values during the summer. 3. Long-Term Performance Index (PI). Actual Energy Output / Expected Energy Output. Actual energy output (kWh AC) divided by expected energy output (kWh AC). In this case we would use the System Advisor Model (SAM) developed by NREL in order to generate an expected output. If all conditions are modeled and the derate factors could be estimated properly, a fully generating system will have a PI = 1.0. Calculating the long-term performance index of the system is expected to provide a more accurate description of the PV system. Through the use of SAM, the expected energy output will account for more factors than just temperature correction. Although this method may be more accurate, it is also expected to require a higher level of effort. 1. Long-Term Performance Ratio Procedure Long-Term PR can be measured as the actual yield to the reference yield. PR = (kWhAC/DCRated)/(H/GSTC) 1. Prior to calculating the hourly PR based upon the IEC61724 formula, we first filtered through the measured data and removed all hourly data sets that did not have a measured plane of irradiance of 600 W/m2. Additional analysis is planned to filter using kWh AC greater than zero using the SUMIF function in Excel. 2. With the remaining hourly data sets a PR was calculated. 3. A daily PR was then calculated by averaging the hourly PRs from each day. Since we used a minimum of 600 W/m2 there are days in which a PR was not calculated. 2. Long-Term Performance Ratio with Temperature Correction Procedure When applying a temperature compensating factor we decided to take two different approaches. The first approach we used was by adding a temperature correction factor related to the nominal operating cell temperature (NOCT). The second approach used was by adding a temperature compensating factor developed by Timothy Dierauf from SunPower. 2

PR = [(kWhAC/(DCRated KTemp )]/(H/ GSTC) NOCT Approach KTemp,NOCT = 1 + (TCell - TCell,STC) KTemp,NOCT Temperature correction factor using the NOCT approach TCell Calculated operating cell temperature (C) TCell Calculated operating cell temperature under STC (25 C) TCell = Ta + (TNOCT Ta,NOCT)*(H/GNOCT)*(1-(/)) Ta Ambient temperature (C) TNOCT nominal operating cell temperature (NOCT) at NOCT test conditions Ta,NOCT Ambient temperature at NOCT test conditions (20 C) H Measured irradiance in the plane of array (W/m2) GNOCT Irradiance at NOCT test conditions (800 W/m2) c/() 0.083/0.9 can be used to estimate this value 3. The procedure is the same as calculating long-term PR but with a temperature correction. 4. Calculate the temperature correction factor through the equations above. Power Weighted Average (SunPower Approach) KTemp,SP = 1 + (TCell - TCell_sim_avg) KTemp,SP Temperature correction factor using the SunPower approach TCell Calculated operating cell temperature (C) TCell_sim_avg Cell temperature computed form annual average measured meteorological data (C) TCell = Ta + H*e(a + b*WS) + (H/GSTC)*3 Ta Ambient temperature (C) H Measured irradiance in the plane of array (W/m2) a, b = Empirically determined coefficients establishing the rate at which module temperature drops as wind speed increases GSTC Irradiance at standard test conditions (1000 W/m2) TCell_sim_avg = ( GPOA_sim_j*Tcell_sim_j) / (GPOA_sim_j) GPOA_sim_j Simulated average cell temperature from typical weather year (C) TCell_sim_j Simulated cell operating temperature for each year (C) j each hour of the year (8,760 hours total) 5. The procedure is the same as calculating long-term PR but with a temperature correction. 6. Calculate the temperature correction factor through the equations above. 7. TCell_sim_avg must be calculated prior to running a filter that removes all hours that do not meet the minimum hourly irradiance of 600 W/m2. 8. When calculating the cell operating temperature, the values for a and b can be determined by referencing Photovoltaic Array Performance Model by D. L. King. Table 4 below contains the empirically determined coefficients based on module type.

Table 4. Empirically determined coefficients calculated by Sandia National Labs Module Type Mount a b T (C) Glass/cell/glass Open rack -3.47 -0.0594 3 Glass/cell/glass Close roof mount -2.98 -0.0471 1 Glass/cell/polymer sheet Open rack -3.56 -0.0750 3 Glass/cell/polymer sheet Insulated back -2.81 -0.0455 0 Polymer/thin-film/steel Open rack -3.58 -0.1130 3 22X Linear Concentrator Tracker -3.23 -0.1300 13 Results As seen in Figures 3, 4, and 5, you can see the plotted daily PR for each method over a one year time span. Remember, the purpose of the temperature correction was to reduce variance caused by the annual seasons. With this in mind, Table 5 shows that the SunPower temperature correction factor allowed us to reduce the amount of deviation, variation, and range of our calculated daily PRs. However although this may be true, the data set that provided the best linear fit was the standard PR method. During this evaluation it was discovered that in order to calculate the daily PR using hourly data, there are many steps and calculations that were needed, requiring a higher level of effort than originally planned. This can be prevented by creating a software program that will automatically calculate the performance ratio dependent on the inputted data set and the various parameters set. However, if one is going to put in that level of effort, it would be smarter to use an existing model and calculate the long term Energy Performance Index (EPI) of the system. Table 5. Various PR methods and statistical description PR Method Standard, IEC61724 NOCT Correction SunPower Correction Mean 0.773 0.867 0.757 Standard Deviation 0.046 0.041 0.034 Standard Deviation 6.0% 4.8% 4.5% Sample Variance 0.0021 0.0017 0.0011 Range 0.3234 0.2971 0.2604

Standard PR

1 0.95 0.9 0.85 0.8 0.75 0.7 0.65 0.6 0 50 100 150 200

250

300

350

NOCT Corrected PR

1 0.95 0.9 0.85 0.8 0.75 0.7 0.65 0.6 0 50 100 150 200 250

300

350

Figure 4. Daily PR over one year using the NOCT temperature correction factor

SunPower Corrected PR

1 0.95 0.9 0.85 0.8 0.75 0.7 0.65 0.6 0 50 100 150 200 250 300 350 y = -6E-05x + 0.768 R = 0.0278

Figure 5. Daily PR over one year using the SunPower temperature correction factor 3. Long-Term Energy Performance Index (EPI) When calculating the daily EPI a minimum energy generated criteria was applied. In this case we applied a minimum of 1200 kWh per day. A statistical analysis was done on the filtered data set. As shown in Table 6, the average daily performance index was 0.9830, indicating that the system is running at an optimal level. Through this analysis, it was discovered that calculating the EPI of the PV system was much simpler and easier than calculating the PR or CPR. This was primarily due to the already existing tools and models that we could use in order to calculate the necessary data for the EPI. Table 6. Daily performance index statistics Daily PI Mean Standard Deviation Standard Deviation Range Minimum Maximum 0.9830 0.0365 3.71% 0.2482 0.8697 1.1179

Daily EPI

1.2 1.1 1 0.9 0.8 0.7 0.6 0 30 60 90 120 150 180 210 240 270 300 330 360 390 y = -0.0001x + 1.0039 R = 0.0899

Figure 7. Daily Energy Performance Index (EPI) over one year The slope of the EPI curve shows a gradual degradation of approximately 4% (from 1.0 at time zero and 0.96 at day 360). This is greater than the more typical 0.5% to 1% per year. Data Set Accuracy When calculating the expected output for a PV system each component affects the expected value in different magnitudes. Table 7 displays the percentage for each factor and how it can contribute to the uncertainty of the expected calculation. Table 7. Effect on monthly energy production calculations (IEC=PVPS T2-07:2008)

One area that can be improved is the manner in which data is collected or monitored. Currently the data been provided and is an average of data during a one hour time period. In doing so the hourly averaging of data underestimates the actual energy production during high irradiance conditions. This occurs due to averaging the fluctuations of irradiance over an hour. With large fluctuations the power generated will adjust quickly, however the module operating temperature will adjust slowly and remain at a lower temperature. In Figure 8, the measured plane of array irradiance at various irradiance levels can vary 7

dependent on the time interval in which the data is averaged. If you look at the values for hourly average and 15 second average at irradiances of 800 W/m2 or greater, the shorter time interval shows that there are larger amounts of energy available. Due to these discrepancies, a performance assessment could only be more accurate with data provided at shorter time intervals.

Figure 8. Plane of array insolation vs. irradiance using various averaged time periods (Ransome)

4.1 Recommended Performance Assessment Method

Literature review and discussions with industry experts let to focusing on the following four metrics: Power Performance Index (PPI) of actual instantaneous kW AC power output divided by expected instantaneous kW AC power output. Performance Ratio (PR) of final yield divided by reference yield over an assessment period. Performance Ratio with final yield corrected for cell temperature (CPR) over an assessment period. Energy Performance Index (EPI) of actual kWh AC energy divided by expected kWh AC energy as determined from an accepted PV model, such as SAM, using actual climate data input to the model over the assessment period.

The three primary purposes for performance assessments of existing systems and the associated recommended metrics are listed below. A guideline summary is provided in Table 2.2: Monitoring of a specific PV system to identify degraded performance and need for maintenance based on condition. Use EPI metric and trend EPI for the specific system. Commissioning or assessment after major maintenance. Use PPI and EPI metrics. Determination of specific industry parameters, such as Yield or Performance Ratio, to allow comparison of systems in different geographic locations for design validation or investment decisions. Use PR, CPR and/or EPI depending on the level of effort and level of uncertainty.

Additional work is recommended to develop specific procedures for each of the four metrics summarized above and for making Excel spreadsheets available for general use. Additional long-term data should be analyzed to further investigate the ability of metrics to meet the stated purposes, and how to integrate metrics with currently available industry products such as monitoring systems and IV curve tracers.

4.2

Project Conclusions

Owners of existing photovoltaic (PV) solar energy systems are typically interested in the system short-term and long-term performance as input to operation and maintenance decisions. A guideline summary of the recommended metrics and methods is provided in Table 2.2:

5.0 References

1. King, D.L., Boyson, W.E., Kratochvil, J.A., 2004. Photovoltaic Array Performance Model, Sandia Report No. SAND2004-3535 available from US Department of Commerce, National Technical Information Service, 5285 Port Royal Rd, Springfield, VA 22161. 2. Marion, Adelstein, et.al., Performance Parameters for Grid-Connected PV Systems. Feb 2005, NREL/CP-520-37358, Conference Paper 3. Townsend, Whitaker, et.al., A New Performance Index for PV System Analysis, 1994, IEEE CH3365-4/94/000-1036 from WCPEC Conference 1994, Hawaii. 4. Sunpower, Performance Metrics, ASES Solar 2011 Conference, May 20, 2011, presentation slides 5. Ransome and Funtan paper. 6. Duffie, J.A., Beckman, W.A., Solar Engineering of Thermal Processes, 3rd Ed, John Wiley & Sons Inc., New York 7. W. De Soto, S. A. Klein, and W. A. Beckman. "Improvement And Validation Of A Model For Photovoltaic Array Performance." Technical paper available on internet: http://www.physics.arizona.edu/~cronin/Solar/References/PV%20system%20modeling/Impr ovement%20and%20validation%20of%20model.pdf 8. Fitzpatrick, Shawn. "A Method For Predicting Pv Module And Array Performance At Other Than Standard Reporting Conditions." North Carolina Solar Center. <http://webtest.ncsc.ncsu.edu/media/research_papers/Method_for_Predict_PV.pdf>. 9. Markley, Chris, Solar PV Performance Calculators, SolarTech publication 7/12/2010.

10

6.0 Appendix

6.1 6.2 6.3 6.4 6.5 6.6 Performance Assessment Flow Chart Method to calculate plane of array irradiance based on GHI Method to calculate PV cell temperature based on ambient temperature Derate Factor determination, assumptions Sample of Excel Spreadsheets to calculate Performance Ratio (PR) Sample of Excel Spreadsheet to calculate Energy Performance Index (EPI)

11

PV PERFORMANCE ASSESSMENT GUIDELINE FLOW CHART Fig 6.1 [additional cleanup work needed] Short Term Assessment - Power Performance Index (PPI):

Measure POA Irradiance* Measure Amb, Mod. Temps.* Inspect AsBuilt System Measure Inverter AC Output V, I* Note Inverter Display Output Power*

If PI = 1.0 (uA/B)*, OK

Install** POA Irrad. and Temp. logger Note Inverter Display Initial Energy, kWh Calc PR with or w/out Temp Compensation Compare PR to Industry and/or Trend

assess. period, H Determine Industry Typical PR Values

Go to Page 3

Go to Page 4

*Apply propagation of uncertainty.

Go to Page 4

|/ | +

If [PRAct PRInd] < Equiv. OK. Repeat periodically for PR Trend of PR vs. time, irrad, temp.

** POA Irradiance options: 1) Use existing installed sensor if available, or 2) Use nearby live site, calculate POA using NREL DISC Excel, or 3) Install POA sensor *** After assessment period of 1-month, and/or 3-month, and/or 1year, plot weekly trend of PR and PR vs Irrad, and PR vs Temp

If PI = 1.0 (uA/B)*, OK

Measure POA beam and diffuse irrad.* 2 Estimate Soiling (SF) Measure ambient, and module backside temps.* Repeat above at 3 difference irradiance values if practical. Disconnect strings (open fuses) from inverter, measure Voc, Isc (use Test Switch Box [TSB] for measuring Isc) Measure Vop (Vmp), Iop (Imp) of each string at Combiner Box, while individually connected to Inverter (open fuses w/ TSB) Calculate uncertainty of measured and expected values.

Using Sandia and/or 5-Para Models, calc Expected FF, Voc, Isc, Imp, Vmp, of string. Include string derate factors.

Calc Actual/Expected ratios for Vop/Vmp, Iop/Imp, Voc/Voc, Isc/Isc, FF/FF, for each string and plot with uncertainty bands.

If ratios 1.0(uA/B)*. NOT OK. Investigate perf. discrepancy. Refer to Troubleshooting Matrix, identify possible causes, obtain more data to support or refute each possible cause.

Re-evaluate derate factors for DC losses, mismatch, mfg tolerance. Use thermography camera to check each module of string.

Calc (Vop/Voc)act/(Vmp/Voc)rated proportional to Rs, (Iop/Isc)act/(Imp/Isc)rated prop to Rp Measure soiling using IR gun for emissivity. Consider cleaning.

Measure bypass diode junct. box, temps for diode turned on.

Use collected data to assess condition of strings/modules If above inconclusive, use IV Curve Tracer, Plot IV Curve, Identify Imp,Vmp, FF. Compare to Model Calculated Values. Go to Page 4.

Measure Inverter DC and AC Power, kW* Check inverter MPP Tracking performance by monitoring range of Vmp and Imp variation

Calculate Efficiency Determine rated efficiency at operating load from inverter data sheet

Calculate and Compare Efficiency Difference, Use Equivalency Test With Uncertainty

Measure DC Home-Run Voltage Between + and -, + and green, and green at both ends Measure voltage drop across DC Disconnect switch and fuses

Visually inspect connections and switch contacts for discoloration due to overheating

Troubleshooting Matrix: Purpose: This table relates the string parameters which are used to calculate ratios to those conditions relationship may assist in identifying the cause of a ratio that is outside the range of 1.0 uncertainty. Calculated Values Description Related To Act Isc / Exp Isc of Actual Isc at measured Irrad and Temp / Irradiance and cell string Expected Isc as calculated using Sandia or conversion efficiency 5-Parameter model for measured Irrad and Temperature conditions. Act Vop / Exp Actual Vop at measured Irrad and Temp / Irradiance and cell Vmp of string Expected Vmp as calculated using Sandia conversion efficiency or 5-Parameter model for measured Irrad and Temperature conditions. Act Iop / Exp Imp Actual Iop at measured Irrad and Temp / Irradiance and cell of string Expected Imp as calculated using Sandia or conversion efficiency 5-Parameter model for measured Irrad and Temperature conditions. Act Voc / Exp Voc Actual Voc at measured Irrad and Temp / Cell temperature higher than of string Expected Voc as calculated using Sandia or module back surface 5-Parameter model for measured Irrad and temperture Temperature conditions. (Vop/Voc)/(Vmp/Voc) Actual operating point voltage / actual Voc Ratio is proportional to of string divided by expected max power voltage / amount of series resistance. expected Voc at measured Irrad and Temperature conditions. (Iop/Isc)/(Imp/Isc) of Actual operating point current / actual Isc Ratio is proportional to string divided by expected max power current / amount shunt resistance. expected Isc at measured Irrad and Temperature conditions. Actual FF / Actual FF using measured Vmp, Imp, Voc, String and module overall Expected FF of Isc at actual Irrad and Temp / Expected as quality, cell conversion string calculated using Sandia or 5-Parameter efficiency, optical model for measured Irrad and Temperature degradation, cell conditions. degradation, light-induced degradation, temperature induced degradation. 5

that affect the parameter. This Possible Causes Soiling, shading, or degraded cells. Model inaccuracy. Soiling, shading, degraded cells or

or

Delamination assembly

of

module

Electrical connections between modules, degrading cells Cell and module degradation allowing current paths in parallel with pn junctions Module mismatch, increased series resistance due to connections, or decreased shunt resistance (at low light), moisture diffusion into laminant, anti-reflective coating degradation.

SF

Estimate Soiling Fraction (SF) to adjust Reduced system output measured irradiance. Use NREL method to estimate SF. Measure SF using IR gun.

6.2

NREL has developed the DISC program to calculate DNI. Given GHI and DNI, diffuse horizontal irradiance can be calculated from the zenith angle and used as input to SAM to calculate POA irradiance as described in more detail in section 3.3 and Table 2.2.

6.3

When determining PV system performance, cell temperature is an important parameter of interest. Unfortunately PV monitoring equipment may not always provide cell temperature; temperature however, local weather station data are commonly available and can be easily provided via the web. There are multiple approaches available that can be used to determine cell temperature based on weather conditions. This Appendix shows the analysis of three different methods that were used and compared in order to determine which approach is most accurate by comparison of the calculated value to a measured value of the module backside on the 191kW system discussed previously in this report. report Simple Correction Method:

where TNOCT NOCT = 47C, Ta,NOCT = 20C, GNOCT = 800 W/m W/m2 Sandia Method:

Where WS = Wind Speed and constants a and b are provided by sandia model specific test data. data The additional 3C with the irradiance ratio should also be added to TC based on the Sandia model calculation. Text Book Method:

Equation provided by Solar Engineering of Thermal Processes 3rd Edition, , Reference 2. These equations were used on a PV system data where both the cell temperature and weather conditions were measured. Cell temperature was calculated using wind speed and ambient air temperature and compared to measured cell temperature. This was done for hourly data for a months duration. The standard deviation of of the difference between measured and calculated cell temperature for the sample set was then used to determine which approach is most accurate.

Table 1. Cell temperature method comparison Cell Temperature Method Simple Sandia Text Book Stand. Deviation of Temp Difference 4.86 8.04 3.48

Figure 1. Calculated Cell Temperature Comparison Of Four Consecutive Days Within Data Set

90 80 70 Temperature (C) 60 50 40 30 20 10 0 0 10 20 30 Hour 40 50 60 Measured Simple Text Book Sandia

The results from the analysis can be seen from Table 1 and Figure 1. The results show that the text book method appears to be more accurate. Table 1 lists the standard deviation of the difference between measured and calculated cell temperature for the entire sample set. By having a smaller deviation the difference between values is smaller through the sample set. Figure 1 visually demonstrates that the text book method is consistently closer in value to the measured. Independent calculations were performed to check the other students calculations as shown in Fig 2 and found to validate conclusions to support use of the textbook method, but it found that the Sandia method is also close to the measured values than concluded above. The study shows that there can be more than one approach in determining cell temperature based on weather conditions. It also suggests that the equation provided by the text book is more accurate than the simpler methods proposed. However, the study is still inconclusive due to the small sample size. Another important factor to note is the weather data used was not measured on site and was instead provided by a nearby weather station. Further studies in a more controlled environment will be necessary to further justify the use of one method over the other.

Figure 2. Calculated Cell Temperature Comparison Of Seven Consecutive Days Within Data Set

80 70 Temperature, C 60 50 40 30 20 10 0 1320 1340 1360 1380 1400 1420 1440 1460 1480 Ambient Temp Measured Cell Temp Sandia Calc Temp Energy Weighted Daily Avg Textbook Cell Temp

The textbook formula predicted cell temperatures that was closer to the measured cell temperature that the Sandia formula, however both are within the range of uncertainty and therefore recommended for use. The Energy Weighted Daily Average was used to calculate the temperature compensated Performance Ratio (CPR).

6.4

6.5

- Solar Photovoltaic Incentive ProgramЗагружено:Genability
- WindFarm EcnomicsЗагружено:Pedro Galvani
- Design Guidelines for 30 100 KW SystemsЗагружено:Maan E
- BIPV Training Presentation 7-10-08Загружено:savage68
- India Solar Sponsor Map Final Web 1Загружено:Ravi Singh Bisht
- State-of-Massachusetts-Incentive-Area-Commonwealth-Solar-II-Rebates---Загружено:Genability
- OK MSc thesis: chapter 7 ReferencesЗагружено:omerkorech
- Future Direction of Solar PV IndustryЗагружено:Dan688
- Solar-220Загружено:Banci Ngamen
- Renewable Intro SlidesЗагружено:chahat
- Thin Film Photovoltaics (PV) - Advanced Technologies and Global Market (2010 - 2015)Загружено:aarkstor
- JacobRachel.ts.Step.byЗагружено:saiddz4ever
- Italian Nursery School_ Conceptual Design Analysis _ Sustainability WorkshopЗагружено:Shashank Jain
- IntroЗагружено:Ramy Abd Elkhalek Shahin
- b1Загружено:James Terry
- 009-045Загружено:eng_waleed2008
- Lecture 1Загружено:KiKon Kwak
- Alonso RefsЗагружено:qaz_69
- Scorecard Final VersionЗагружено:Seetharaman Chandramani
- 1-s2.0-S0960148108002619-mainЗагружено:Zeensoni Yusof
- ENT ProjectЗагружено:saksham
- Renesola Installation ManualЗагружено:Tarun Bharadwaj
- final research projectЗагружено:api-341862798
- ForecastingЗагружено:Lee Ryan Bueno
- 1-s2.0-S1876610213000088-mainЗагружено:Rouf Arifin
- Effect of Nanoparticles Shape on the Efficiency OfЗагружено:keydeco
- Icweaa2018 Paper 15 2Загружено:Hamou Walt
- 100507324-Solar-Water-Pump-28th-June-2012.pptxЗагружено:Anonymous HRkuPF4gXJ
- Solar (Slide)Загружено:SOURABH MONGRA
- 1-s2.0-S0196890419300810-main.pdfЗагружено:Miguel Angel

- Combustion Plate 1Загружено:Marc Bitang
- Watertube_AlstomЗагружено:kappecito
- Synchronized 12-Pulse Generator (SimPowerSystems)Загружено:Pavan Kumar
- Electrical-Engineering-portal.com-The Role of Medium Voltage Switchgear in Power SystemЗагружено:Ricardo A Vergara
- Orgone Field Pulser II - Mobius-driven Bioenergy Generator.pdfЗагружено:김종빈
- Exelon 2020: A Low Carbon RoadmapЗагружено:jennabush
- NTPC Training_reportc gЗагружено:Anahita Aggarwal
- 14246_8120321944Загружено:طه محمد
- Commutation Scr PDFЗагружено:Hiram
- Icse Class-xth-physics Model Paper(1)Загружено:Upendra Shukla
- Mayer Joule ExpЗагружено:Anil Kumar Rout
- 05. Daily Energy Report Mei 2019Загружено:Riefal Dy Iskandar
- philips-2-by-2-led-commercial-panel-rc370b.pdfЗагружено:mehfooz
- Nuclear Chain Reactions _ Nuclear Fission _ Science _ AtomicarchiveЗагружено:gopinadh333
- I-DE-3010.96-1350-940-CO1-016_Rev.0Загружено:Ivan Ho
- {DD0E06E1-1E95-40F2-AF00-ACAAD23EC5C2}.pdfЗагружено:ScottWaldman
- 5054_s12_qp_11Загружено:Breezy Brown
- Gas Power Plant - Layout and Operation PPT (1) Electrical Seminar TopicsЗагружено:deepanshi2
- Development Ohmic Heating SystemЗагружено:achmatsarifudin1317
- 261218709-EN-12952-12-2003-E.pdfЗагружено:Domagoj Butumović
- Daikin VRV IV catalogueЗагружено:Myo Kyaw Kyaw
- StacoAVR2.pdfЗагружено:maung
- Notes on Everyday ScienceЗагружено:Atif Rehman
- ECX6332-ASS123&4-2015-2016 REVISEDЗагружено:Madusanka Weebedda
- Solar SolarЗагружено:ALDRICK JAYSON CADACIO
- ACE-G226LH_306LH_456LHЗагружено:Anonymous JcOZwo0Ox
- E_100Загружено:warmaster81
- Fuel and Waste -(NATURE)Загружено:Sindhuja Devineni
- An Ordinary TransformerЗагружено:sopan sa
- PCTVSG1808aprv.pdfЗагружено:mjat2006

## Гораздо больше, чем просто документы.

Откройте для себя все, что может предложить Scribd, включая книги и аудиокниги от крупных издательств.

Отменить можно в любой момент.