PV Perf Assessment Report Draft Rev C

7/27/2019 PV Perf Assessment Report Draft Rev C

1/66

SJSU SOLAR PROJECT - 007January 2012; Rev. C

SOLAR PV SYSTEM PERFORMANCE ASSESSMENT GUIDELINE

For

SOLARTECH

Dated:

January 2012

Prepared By:

Shelton Honda

Alex Lechner

Sharath Raju

Ivica Tolich

Reviewed By:J. S. Mokri, Advisor

Solar Energy Project Team

Mechanical and Aerospace Engineering

San Jose, California


2/66


ii

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 systemsthat 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 thereport 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 theengineering 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.


3/66


iii

Acknowledgements

The project was performed with ongoing technical assistance and advice from SolarTech

Performance Committee members including Joe Cunningham of Centrosolar, Laks Sampath ofNeozyte, 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 preparationof 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 toparticipate in this project which contributed to the SolarTech organizations goals and

simultaneously provided a learning opportunity to SJSU engineering students.


4/66


1

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 SystemMeasurements

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


5/66


2

5.0 References..... 42

6.0 Appendix

6.1 Performance Assessment Flow Chart6.2 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)


6/66


3

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 tocalculate 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 byexpected instantaneous kW AC power output.

Performance Ratio (PR) of final yield divided by reference yield over an assessmentperiod.

Performance Ratio with final yield corrected for cell temperature (CPR) over anassessment period.

Energy Performance Index (EPI) of actual kWh AC energy divided by expected kWh ACenergy as determined from an accepted PV model, such as SAM, using actual climatedata 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 formaintenance 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, toallow 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. Additionallong-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.


7/66


4

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 withuncertainty bars. The plots illustrate typical seasonal variation and the influence of weather andderating 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 notpointed 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 ofactual 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 whichperformance models are available. The guideline focused on systems with fixed flat panel PV

modules.

1.2 Purpose of Performance Assessment

The purpose and need for performance assessment can best be summarized from an ownersperspective 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?


8/66


5

5. How is my system performing compared to the last assessment period (trending)?

6. What is the annual kWh AC energy production per installed KW DC (Yield)

7. What is my cost of energy in terms of intial and maintenance costs relative to kWh energyproduction over the life of the system?

8. When do I need to do maintenance, such as cleaning, based on performance?

9. 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 100kWsystem 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 thereforehigher 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 orcalculation 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 withactual 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 thelevel-of-effort and cost. The level-of-effort related to the amount of data and analysis in theperformance assessment should be comparable to the expected benefits gained. Small systems

could use simplified methods, with correspondingly greater uncertainty (10% to 20%) but stillhaving 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.


9/66

1

Fig. 1.2 - Performance Assessment Map showing

applicability of guideline covered by this report

System Size: Small Medium Large200kW >10 MW

Asset Class: Residential Commercial Large Commercial, PPA

LevelofEffort:

Minimum

Moderate

Maximum

Uncertainty:

10%t

o20%

5%t

o10%

2%t

o5% Proprietary

Algorithm

Inverter kWh meter,Utility billing, PPI,PR w/o adjustments

ProprietaryAlgorithm

ProprietaryAlgorithm

PPI, PR withtemperaturecompensationfactor, EPI






10/66


1

2.0 PV Performance Assessment Methods

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 ofthis 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 calculatingexpected 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 forsome 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 systemefficiency 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 systemdesign, 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 orenergy.

Actual output normalized divided by actual input normalized. An example of this metricis Performance Ratio and is used regularly to compare systems, however, may result inincorrect 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 consideredinstantaneous power output when the measurement is taken and denoted by kW (power). Long-


11/66


2

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 canbe 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 reachinappropriate 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, thereforeuncertainties are lower with Energy Performance Index.

Initial review of industry practice found various performance metrics as shown in Table 2.1.

Table 2.1- Commonly Used Performance Metrics

METRIC CALCULATION REFERENCE

Yield kWh / kWDC STC NREL/CP-520-37358

Performance Ratio (kWh/ kWDC STC ) / (H/GSTC) IEC61724

Performance Ratio kWh / (sunhours area efficiency) SMA

Performance Ratio(EActual / EIdeal) * 100%

EIdeal is temp. and irrad. compensated

SolarPro, Taylor & Williams

Specific Production MWhAC / MWDC STC SolarPro, Taylor & Williams

Performance Ratio(100 * Net production / total incident

solar radiation) / rated PV module eff.NREL/TP-550-38603

Performance Factor ISC,G*RSC*FFR*ROC*VOC,T Sutterlueti

Performance Index kWmeasured / kWexpected SolarPro, Sun Light & Power

Performance IndexActual Power / (Rated power * irrad adj.

* temp adj * degradation adj * soiling adj

* BOS adj)Townsend

Output Power Ratio kWmeasured / kWpredicted SolarPro, Sun Light & Power

Output power kW > CF-6R-PV Table CEC Commissioning

Output power kW > 95% expected SRP Arizona UtilitySpecific Production MWhAC / MWDC-STC SolarPro, Taylor & Williams

Acceptance Ratio kWactual / kWexpected Literature

Inverter comparison kWh of multiple similar inverters Qualitative

String comparison Imp, Vmp of multiple parallel strings Qualitative

Utility billing Monthly comparison Qualitative

Performance Ratio,

temp. comp. (CPR)(kWh/ kWDC *KTemp) / (H/ GSTC) Proposed in this report

Energy Performance

Index (EPI)

kWh AC actual / SAM AC Expected

using actual weather dataProposed in this report


12/66


3

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 systemsize; 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 beequivalent for a fair comparison. The basic Yield equation is shown below as equation 1:

Yield= kWhkW 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.


13/66


4

Figure 2.1 Basic PR Seasonal Variation Without Temperature Correction

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, aPV 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 timeincrement (such as, hourly) and the PR calculated hourly.

PR . =

kWhkW KkWh1kW 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 littleuse for performance assessment. Averaging hourly PR to obtain daily PR is incorrect versus

summation of the hourly kWhAC and kWhSun values for the day.

0.5

0.55

0.6

0.65

0.7

0.75

0.8

0 2 4 6 8 10 12

PR

Month

Seasonal Variation of PR


14/66


5

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 isconsidered 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 andFuntan 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, suchas 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 coefficientfrom 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 inRef. 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 bcoefficients, 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


15/66


6

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; newermethods 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 theactual 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. PerformanceRatio (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, suchas 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.


16/66


17/66


18/66


19/66

4

Acronyms:

PI = Performance Index, ratio of actual divided by expected

PPI = Power Performance Index, instantaneous actual power divided by expected powerPR = 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)


20/66


1

2.1.2.1 Short Term Assessment Power Performance Index:

The Power Performance Index (PPI) is the instantaneous actual AC kW power output divided bythe 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 actualpower 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 + c o s2 + 1 c o s

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 + 256. Calculate expected AC output power (kW):


21/66


2

= 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, out-

of-service time (outages), unavailability, low light performance, angle of incidence effects, solarspectrum 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. Thenumerator is equivalent to the number of hours the system operated at the DC STC rating and thedenominator 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.

PR =kWhkW

kWh /m1kW/m 2

Both the numerator and denominator are summations of the measured increment data, such ashourly, 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/m 2. 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.


22/66


3

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 existingPOA 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 willincrease 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 adjustthe 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,


23/66


4

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 ofthe 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


24/66


5

Soiling (up to 20%) 5%

Inverter efficiency (95%+) 5%

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 atemperature 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



25/66


6


a, b = Empirically determined coefficients establishing the rate at which module temperaturedrops 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)



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 0Polymer/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 andtherefore 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.


26/66


7

Figure 2.3: Plot of ratio of actual system PR to typical TMY3 based PR

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:

PR =kWhkW kWh1kW

PR. =kWhkW K K K kWh1kW

PR. = E P I = kWh

kW kWh1kW K K K

0.50

0.60

0.70

0.80

0.90

1.00

0 1 2 3 4 5 6 7 8 9 10 11 12

PRandPRRatio

Month

PR Ratio To Reduce Seasonal Variation

PR TMY3+SAM

PR Climate+Meter

PR Ratio


27/66


8

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 hourlydata 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)


28/66


9

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 tonormal seasonal variation although cell temperature has been compensated directly in the SAM

PV model.


29/66


10

Figure 2.4: Plot of EPI, Actual kWh AC divided by Expected kWh AC

2.1.3 Performance Metric Calculation Methods

The method used is dependent upon the purpose of the assessment and must consider the costand benefit of the assessment. Generally, the level of accuracy and depth correlates to the level

of effort and cost.

0.50

0.55

0.60

0.65

0.70

0.75

0.80

0.85

0.90

0.95

1.00

0 1 2 3 4 5 6 7 8 9 10 11 12

PI

Month

Energy Performance Index

Meter PI


30/66


11

Table 2.3- Performance Assessment Methods Summary

Level of Effort Method Frequency

Minimal Utility bill compared to previous months billing, on-line

inverter monitoring if available.

Monthly

Minimal Instantaneous Power Performance Index (PPI) of actual AC

power output divided by expected AC power output includingderate factors. PExpected = PDC STC KIrrad KTemp KDerateThis method is termed the PKs method in this report. Plane of

array irradiance and cell temperature are needed. PActual isobtained by direct measurement or inverter display.

Periodic

Moderate Initial PPI and long-term Performance Ratio (PR) of specific

system trend of Actual Yield to Reference Yield.PR = (kWhAC/DCRated)/(kWhSun/1kW)

Periodic

Moderate 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).

Periodic,

depending on

level of

monitored dataavailable

Extensive 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.

Periodic,

depending on

level ofmonitored data

available

Extensive Proprietary algorithms Continuously

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)


31/66


12

One area that can be improved is the manner in which data is collected or monitored. Currentlythe 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 irradianceconditions. 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 Numerical Models of Expected Performance

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 systemdesign 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 METHOD REFERENCE

5-Parameter Determine equivalent circuit valuesadjusted for actual irradiance and

temperature. Used to calculate IV curveand MPP.

CEC, DeSoto paper

Sandia Empirical Use Sandia module tested data coefficients

and effective irradiance

Sandia, D. King 2004 paper

Efficiency Sunhours area efficiency SMA

Characteristic

Parameters

Determine coefficients from standard

module data, adjust for actual irradiance

and temperature

Rausehenbach eqns.

Fitzpatrick paper, North

Carolina Solar Center

Correction

factors, denotedPKs method

Use irradiance K factor on current and

temperature K factor on voltage, andtypical derate factors from PVWatts

SolarPro, Sun Light & Power

article

System Advisor

Model

NREL developed model that generates an

expected energy output based on TMY3

SAM developed by NREL

Direct Insolation

Solar Code

(DISC) model

Estimates direct normal irradiance (DNI)

based upon latitude, longitude, and global

horizontal irradiance (GHI)

Dr. E. Maxwell of the

National Renewable Energy

Laboratory

PVUSA Rating

Method

Uses a polynomial regression model in

order to predict power. Takes into account

irradiance, ambient temperature, and wind

IEA-PVPS T2-07:2008


32/66

speed.

A practical and readily availaSolar Advisor Model (SAM)

summarized below. More co

Sandia/King Model

The Sandia PV model is of th

obtained from specific outdoo

Rauschenbach Model

The model [Ref. 4] develops

and Imp, referred to as the ch

The characteristic parcalculate the value of

Voc and Vmp are indep

irradiance.

The values of Isc and I

by a factor equal to th

The model is accurate

curves.

The model has been te

on modules, and array

5-parameter model

A solar cell (or module orcircuit shown below. The

dependent current source

Rsh, and a series resistanc

Equivalent circu

The I-V relationship for t

The 5-parameters viz. IL,

SJSU

13

le computer program, which incorporates thvailable at NREL.gov website. Background

plete information is available in the literatu

PV module and based on empirical curve fi

r tests.

-V curve equations for a module, or array, b

racteristic parameters.

meters vary with temperature. Temperatureharacteristic parameters at operating temper

ndent of insolation. Isc and Imp are directly p

mp at a given irradiance are obtained by scali

ratio of operating irradiance to the standard

with less than 5 % error, when compared wit

sted on crystalline silicon, multi-crystalline a

s.

array of modules) can be modeled [Ref. 2] uequivalent circuit consists of a diode in para

L. The losses in the cell are represented by

e Rs.

it for the 5-parameter model

e above equivalent circuit is given by the eq

= 1 Io, Rs, Rsh and a, a modified ideality factor

SOLAR PROJECT - 007January 2012; Rev. C

first three models, ison the models is

e and references.

t equations with data

sed on Voc, Isc, Vmp

oefficients are used toture.

oportional to

g the reference value

irradiance.

h experimental

nd amorphous silicon

sing the equivalentlel with a light-

parallel resistance

ation below:

etermine the shape of


33/66


14

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 themodules 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 Measurement Methods of Actual Performance

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 SystemMeasurements

The actual AC and DC performance is measured using conventional voltage, current, and powerdevices. 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 toestimate 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 valueassuming 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 todetermine 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 live-


34/66


15

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 andirradiance data for use in calculating long-term Performance Ratio.

2.4 Performance Shortfall Investigation:

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 irradianceand 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.


35/66


16

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

2.5 Financial Model to Evaluate Cost/Benefit of Maintenance:

An economic analysis is needed to determine the cost and benefit of possible maintenanceoptions.

a) Calculate Present Worth (PW) of cost of maintenance on an expected schedule for aminimum of a 1 year period or through the maintenance interval for major maintenance.

b) Calculate PW of savings expected from increased system output (kWh) for the 1 yearperiod.

c) If PW for maintenance is less than PW of savings, it is cost justified to performancecleaning 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.

String 1 String 2 String 3

Isc Ratio 1.002 0.996 0.996

Voc Ratio 1.018 1.013 1.013

Imp Ratio 1.093 1.075 1.071

Vmp Ratio 0.992 0.985 0.980

FF Ratio 1.063 1.049 1.040

Power Ratio 1.085 1.058 1.050

0.800

0.850

0.900

0.950

1.000

1.050

1.100

1.150

1.200

1.250

RatioofActual/Expected

String Parameter Comparison


36/66


17

3.0 Application of Above Methods to Existing Systems

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.


37/66

1

Power Performance Index Spreadsheet, page 1

PV SYSTEM FIELD PERFORMANCE ASSESSMENT:Customer/Location: Accurate Solar Assessment Date: 6/21/2011 SJSU Team

SYSTEM DESCRIPTION:

System Configuration (per inverter): Module nameplate data and rated parameters at STC (1000 W/m2

, 25C, AM1.5):

Number of modules/string = 11 Mfg = Schott Voc (volts) = 4301 VoltageTem

Number of strings = 3 Model = SAPC 165 Isc (amps) = 5.46 CurrentTem

Module tilt angle (degrees) = 60 Power = 165 Vmp (volts) = 34.6 PowerTemp

Module azimuth (degrees) = 176 Matl Type= mc-Si Imp (amps) = 4.77

Total Rated DC Power at STC (W) = 5445 [Product of module power x number modules x number strings]

FIELD INSPECTION ESTIMATED DERATE FACTORS:Estimated module related derate factors: Estimated string related derate factors: Estimated system related derate fact

Module mismatch = 0.97 DC Voltage drop = 0.99 Inverter efficiency = 0.96

Module Soiling = 0.93 MPPT factor = 0.98 AC voltage drop = 0.99

Mfg Tolerance = 0.92Shading = 1 Age(years)= 6 0.965

System Derate Factor = 0.738 [Product of above 8 factors, refer to PVWATTS for explanation.]

FIELD MEASUREMENTS FOR PKs METHOD: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 KDF5.Measure actual AC output power (kW)

6.Calculate ratio of Measured kW / Expected kW, define as Power Performance Index (PPI)


38/66

2

Power Performance Index Spreadsheet, page 2

PV SYSTEM FIELD PERFORMANCE ASSESSMENT:

Customer/Location: Accurate Solar Assessment Date: 6/21/2011

INVERTER DESCRIPTION:

Mfg Pwr Rating

Model Input Voltage Rating

INVERTER POWER MEASUREMENTS

System measurements to be taken at AC disconnect switch or at inverter output connection box.

Irrad-iance

(POA)

Module

Temp Avg

Measured

String DC

Voltage

Measured

String DC

Current

Inverter

Measured

DC Power

Calc. DC

Power

Discrep.

Measured

vs Calc.

Meas. AC

Voltage line-

line

Meas. AC

Voltage line-

neut

Meas. AC

Current line

Meas. AC

Current neut

Meas

Pow

(W/m2) (C) (volts) (amps) (W) (W) (%) (volts) (volts) (amps) (amps ) (W

12:44 807 64 331 5.700 NA 1886 214.8 17 17 2108.

1:02 794 63 329 5.400 NA 1776 214.8 17.1 17.1 2120.

1:15 803 64 330 5.600 NA 1848 214.8 17 17 2108.

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

PKs METHOD

String 1 Derate Factor (KS) 0.73847793 Irradiance Factor (KI) 0.807 Temp. Factor (KT) 0.805



Building 2Quantity of

modules

STC watts

per module

Total STC

watts

Derate

Factor (KS)

Irradiance

Factor (KI)

Temp.

Factor (KT)

Predicted

watts ac

Measured

watts ac

Power Perf

Index (PPI)

(watts) (watts) (watts) (watts)

String 1 33 165 5445 0.73847793 0.807 0.8050 2612.19 2108.25224 0.807

String 2 33 165 5445 0.73847793 0.794 0.8100 2586.07 2120.65373 0.820

String 3 33 165 5445 0.73847793 0.803 0.8050 2599.24 2108.25224 0.811

Time


39/66


1

3.2 191kW Live Site System Performance Ratio (PR)

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.

PV SystemTest-

Interval

Nameplate

Capacity

Wh

Standard

Test

Conditions

Irradiance G

W/m

2

Total Solar

Irradiance in

Plane-of-

Array H

Wh/m

2

Total Useful

Output Wh

Performance

Ratio

PR=Yr/Yf

Arizona

Game

&Fish

02/21/11-

03/24/11191100 1000 253118 29960000 0.6194

SJ Tech

Museum

03/31/11-

05/01/11185000 1000 150900 20407900 0.7310

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


40/66

1

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

Various PR temperature compensation techniques to reduce seasonal (or daily in this case) variation wer

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 measu

25C for STC temperature, and the temperature coefficient from the module datasheet. This method had

variation than other methods.Weighted Temperature Comp PR used power weighting which gives weight to hours having higher powe

Formula used was Average Daily Temp = Sum of (hourly temp hourly energy) / total daily energy. Th

Ref. 3.

The Sunpower method in Ref 4 uses an irradiance weighted TMY annual average cell temperature that is

than an annual period and resulted in PR values that matched the Basic PR formula, therefore further ana

0

10

20

30

40

50

60

0.50

0.55

0.60

0.65

0.70

0.75

35 45 55 65

Insolarion(kWh)andTemp(C)

PerformanceRatio(PR)

Assessment Period (Days)

Daily PR

Ba

Te

We

Su

Me

Tot


41/66

2

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

Hourly PR values filtered to eliminate hours before 7 AM and after 5 PM which produced division by ze

determined that filtering should be in terms of power output when kWh AC > 0 rather than time.

0

0.2

0.4

0.6

0.8

1

0 20 40 60 80 100 120 140 160 180

HourlyPR

Assessment Period, Hours

Hourly PR Trend

Assessment Period 2/13/2011 + 14 days


42/66

3

Figure 3.2.3 Daily Output Linearly Proportional to Input Irradiation

Figure 3.2.4 Daily PR Showing Degree of Influence Due to Irradiation

y = 116.58x + 17.766R = 0.9842

0

500

1000

1500

0 2 4 6 8 10 12

ACOutput,kWh

Insolation, kWh/M^2

Daily Output vs Input

Daily Totals

Linear (Daily Totals)

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0 2 4 6 8 10 12

BasicPRRatio

Daily Insolation, kWh/M^2

Daily PR vs Daily Insolation

Basic PR

Linear (Basic PR)


43/66

4

Figure 3.2.5 Daily PR Showing Degree of Influence Due to Temperature

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

Figure 3.2.6 Daily PR Showing Noon Dip

0

0.1

0.2

0.3

0.4

0.5

0.60.7

0.8

0 20 40 60

BasicPRRatio

Daily Temp, C

Daily PR vs Daily Temp

Basic PR

Linear (Basic PR)

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

0 24 48 72 96 120 144 168 192 216 240

Series1


44/66


1

3.3 600kW Local System Energy Performance Index (EPI)

System Description

The solar photovoltaic (PV) system that was used for this study is a 595 kW system installed as acovered 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 between12/1/2010 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/CInverter: SatCon PVS500 480V

Time Zone: -8Altitude: 16 m

Azimuth: 93 (S = 0)

Tilt: 4

Figure 2. 600kW PV System

Performance Assessment ReviewBased 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


45/66


2

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 ProcedureLong-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 ProcedureWhen 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.


46/66


3

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-(/))


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



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



a, b = Empirically determined coefficients establishing the rate at which module temperaturedrops 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)



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.


47/66


4

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 3Glass/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

ResultsAs 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 causedby 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 ourcalculated 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 automaticallycalculate 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 MeanStandard

Deviation

PV Perf Assessment Report Draft Rev C

Documents

Transcript of PV Perf Assessment Report Draft Rev C