Products & Equipment : Modules

Primary Category: 

As the solar industry has grown and become more sophisticated in recent years, stakeholders have been basing PV system and procurement decisions on total cost of ownership in addition to—or even instead of—installation cost. Ownership cost and asset valuation now often depend on the terms and conditions associated with the power purchase agreement, as well as how much electricity the PV system is expected to generate throughout its intended operating life, which is largely a function of component quality and system design and maintenance. As a result, PV systems are increasingly evaluated based on dollars per kilowatt- or megawatt-hour, rather than on dollars per watt.

In this article, we present a perspective on how PV module ratings have evolved to reflect the prevailing design and procurement criteria. Many in the industry agree that current nameplate ratings—specifically the power rating under standard test conditions (STC)—taken alone do not adequately represent PV module performance in today’s competitive marketplace. So what should buyers look for in PV module performance characteristics? Many buyers still rely on module ratings under PVUSA Test Conditions (PTC). Are PTC ratings adequate?

We review the motivations for and origin of the PTC module rating. We discuss how PV project valuation practices have evolved away from single-point capacity metrics in favor of more comprehensive measurements and simulation models for predicting PV module performance in the field. In the process, we highlight some promising new energy-based PV module rating methods that industry stakeholders would be wise to adopt going forward.

Evolution from STC to PTC Ratings

For many years, most PV system designers relied on the nameplate or STC ratings of PV modules as a proxy for expected project performance and the resulting return on investment. These ratings are based on PV module performance at standard testing conditions (STC), which are defined in IEC 61215 as 25°C cell temperature, 1,000 W/m2 global plane-of-array irradiance and a solar spectral irradiance at air mass (AM) 1.5. The problem with STC ratings is that the nameplate power rating of any PV module inherently reveals only one aspect of how that product will perform once installed. The reason for this is twofold: PV module—not to mention PV system—performance is largely a function of operating temperature and irradiance, and temperature and irradiance are highly variable and dependent on site location.

History of PTC ratings. Many PV system designers and installers use the PTC ratings system, which traces its origins to the PV for utility-scale applications (PVUSA) research project. The US Department of Energy, as well as local governments and utilities, sponsored this project. A primary goal was to construct, maintain and monitor grid-connected utility-scale projects for evaluation and testing purposes. In 1986, Pacific Gas and Electric Company (PG&E) commissioned the construction of an 86-acre solar farm just outside Davis, California. The large size of the installation (especially for its time) and the scope of its unique monitoring capabilities enabled PVUSA researchers to better understand what happens to PV module and system performance once a system is installed and connected to the grid.

The PVUSA research team developed the PTC rating to evaluate overall system performance over time. The team designed this rating for comparing plant performance as a power-output rating to contractual requirements for the PV system. The rating conditions—1,000 W/m2 irradiance, 20°C ambient (not cell) temperature and wind speed equivalent to 1 meter per second—represent the typical peak-production environment for an installed system in Northern California.

The research team determined the PTC rating by continuously monitoring the PG&E solar farm systems and performing regression analysis on the collected data, and then calculating the power output at rating conditions. This method eliminates the need to test under an exact condition or to scale a measurement taken under different conditions. Since the exact combination of temperature, wind speed and irradiance occurs only rarely, performing regression analysis on continuously monitored data enables more convenient and timely testing. This method also helps to avoid the larger uncertainties associated with scaling a single measured value under arbitrary conditions to a desired point.

System- versus module-level evaluation. The PVUSA research team makes clear its emphasis on system-level performance from its choice of rating condition values. For module-level ratings, direct measurement determines the module or cell temperature. For a system-level rating, measuring temperature is more complex because the operating temperature for cells and modules varies, sometimes dramatically, based on their location in the array and exposure to the elements. In addition, the operation of the inverter and other BOS components is temperature dependent.

Since the varying temperature of so many components influences the system’s behavior, directly measuring PV module temperature is not highly effective. For example, where do you place your thermocouple to measure the system temperature? Recognizing this complexity, the PVUSA research team cleverly circumvented the problem by defining the rating condition using basic environmental data: ambient temperature, wind speed and irradiance. These ambient conditions determine temperatures for all PV system components. The PVUSA team realized that it was not necessary to determine device temperatures to rate system capacity; indeed, it was easier to rate capacity based on ambient conditions.

Adoption of PTC Rating Methods

The research team developed the PTC rating as a tool for the evaluation and contractual acceptance of utility-scale PV power plants at a time when no standards for such testing existed. As the solar industry matured, standards gradually evolved. PV system performance evaluation standards used within the US have drawn heavily on the work of the PVUSA team. For example, both the ASTM International and the California Energy Commission (CEC) have effectively adopted the PTC rating method.

ASTM adoption. Two ASTM standards incorporate the PVUSA rating approach: ASTM E2527 (“Standard Test Method for Electrical Performance of Concentrator Terrestrial Photovoltaic Modules and Systems Under Natural Sunlight”) and ASTM E2848 (“Standard Test Method for Reporting Photovoltaic Non-Concentrator System Performance”). In both cases, the E44.09 subcommittee evaluated the PVUSA rating method against other methods for determining the power-output rating of an entire PV system over a relatively short period of time. The subcommittee chose the PVUSA rating approach for its use of ambient environmental conditions rather than device temperature, as well as its relative ease of implementation.

CEC adoption. In 1997, PG&E sold the Davis site to the CEC. Building upon the PVUSA team’s conclusions, the CEC subsequently decided to apply the PTC system-rating concept to PV modules to determine a financial incentive structure for PV projects. Since PV module performance drives system performance, assigning incentive dollars in proportion to PV module capacity (kilowatts) seemed both logical and relatively simple to administer. Additionally, the CEC wanted to put in place an incentive scheme that would reward expected performance based on realistic operating conditions.

In January 2007, the CEC launched the California Solar Initiative (CSI) program, which addressed industry concerns about the relevance of STC performance ratings by basing incentive payments on PTC ratings. While the industry welcomed the CEC’s decision to adopt PTC ratings, the practice created a new set of issues. Module manufacturers initially treated PTC reporting simply as a box-checking exercise that allowed them to become eligible for rebates in the CSI program. However, they soon realized that a subtle difference in PTC ratings could significantly impact the economics of PV projects in development.

To keep manufacturers from gaming the system, the CEC stopped accepting self-reported performance data in July 2009 and instead required data from third-party laboratories. To become eligible for rebates, PV module manufacturers needed to have their products independently tested at PVUSA test conditions. The premise behind the third-party test requirement was that independently verified PTC ratings would better represent real-world performance in Northern California. Thus, in principle, the PTC ratings reported to the CEC allowed PV system designers to make module-purchasing decisions with more confidence, at least for systems deployed in environments similar to Northern California.

Growing Concerns with PTC Ratings

For the past 7 years, the CEC has maintained a list of incentive-eligible PV modules on the Go Solar California website (see Resources). A large number of PV system designers and installers across North America have relied on the PTC ratings reported in this list to provide a more realistic indication of expected field performance than module nameplate ratings offer. Unfortunately, the future of this list is uncertain. While the CSI program was instrumental in driving PV market growth in California, particularly with respect to distributed generation, it is currently oversubscribed in most utility service territories and the incentive funds are largely exhausted. According to Terrie Prosper, the media contact for the California Public Utilities Commission, the Go Solar California site is funded through 2016.

In addition to uncertainty about the CEC list’s lifespan, the PTC data it contains may be suspect: Third-party testing laboratories report very different PTC ratings for modules with similar packaging (such as crystalline silicon PV cells sandwiched between a glass front and a plastic backsheet). PTC ratings can have a significant impact on PV project financing. Do variations in these ratings indicate that similar PV modules will perform differently based on differences in their composite materials, or do they indicate problems with the testing process itself?

Deriving PTC ratings. In his article “Changes to the PTC Module Ratings” (SolarPro magazine, October/November 2009), Blake Gleason points out that module PTC ratings are calculated rather than directly measured: “Certain module parameters are measured under specified conditions, and then those parameters are used to calculate the expected performance of the module under different conditions. The PTC rating is the result.”

Gleason continues: “Specifically, the nominal operating cell temperature (NOCT) is measured in the nominal terrestrial environment, which is described in ASTM E1036-96, Annex 1 technical standard, and is basically 800 W/m2, 20°C ambient and 1 m/s wind speed. The NOCT is used to calculate the expected module cell temperature under PTC, and then the temperature coefficient of power (measured separately) is used to determine the PTC power rating.”

Problems with NOCT procedures. To explain the potential for measurement error, we need to consider the standardized test methods used to determine PTC ratings. Test methods for crystalline silicon PV modules are based on the IEC 61215, Edition 2.0, test standard. Sections 10.4 and 10.5 cover the measurement of temperature coefficients and NOCT, respectively. However, testing labs interpret these sections differently, particularly 10.5, resulting in reproducibility error.

For instance, IEC 61215 calls for testers to attach cell temperature sensors by solder or thermally conductive adhesive to the backs of two solar cells near the middle of each tested PV module. Some labs interpret this direction by placing a thermocouple on the outside of the PV module backsheet, so as to not disturb the integrity of the laminate; other labs cut through the backsheet to place sensors as close as possible to the solar cell; and some labs do both, using multiple sensors. Both the test procedures and the type and quality of sensors vary, so the results may vary. A poster report by the CFV Solar Test Laboratory in Albuquerque, New Mexico (see Resources), indicates that even on clear days with low wind, module temperature measurements can vary by as much as 4°C based on thermocouple configuration.

To make matters worse, NOCT testing is conducted outdoors where ambient conditions can influence the results. For instance, light spectrum can vary significantly with latitude, altitude and humidity. Identical NOCT tests performed on the same PV module are likely to generate different test results at different locations or at different times of the year, such as in Germany in spring versus in the US Southwest in the middle of summer. This variability is problematic, as there are currently more than 35 CEC-approved laboratories located throughout North America, Europe and Asia.

NOCT variability. Industry stakeholders have scrutinized the NOCT variability for several years because it can influence PV project financing and viability. As early as 2010, the CEC reported that third-party NOCT values differed by as much as 10°C for similar rack-mounted crystalline silicon PV modules. This 10°C range represents a coefficient of variation greater than 20% and suggests that module selection alone could effect a 5% improvement in power. Given the similarity of the PV modules, the disparity in the third-party NOCT test results was surprising enough to warrant further investigation.

Researchers at the National Renewable Energy Laboratory (NREL) acquired three PV modules representing the range of NOCT values that approved third-party testing laboratories reported to the CEC. They tested these three modules side by side for 1 year in an outdoor environment in Golden, Colorado. The authors of a publication delivered at the 2012 IEEE Photovoltaic Specialist Conference, “Evaluating the IEC 61215 Ed. 3 NMOT Procedure Against the Existing NOCT Procedure with PV Modules in a Side-by-Side Configuration,” detail the test results (see Resources). Table 1 shows that in spite of the 10°C difference in NOCT that independent test laboratories reported, these three modules had very minor differences in NOCT values when tested at the same time. The repeatability and reproducibility errors evident in data reported by different testing labs cast doubt on the value of using the measured NOCT to derive the PTC rating.

While heat transfer theory, which suggests that similarly packaged PV modules will have similar NOCT values, can explain these side-by-side test results, some of NREL’s other findings were more surprising. For example, out of a 1-year test period in Golden, Colorado, only 25 days had the necessary environmental conditions to produce the number of data points required to determine NOCT as outlined in the product test standard. In addition, NREL researchers were able to model outdoor NOCT test results predicted under varying environmental conditions—including sky, ground and ambient temperature, and wind speed—and found that the possible range of results for the same PV module varied from 43.3°C to 51.9°C.

To put NREL’s findings into perspective, as of May 1, 2014, the CEC’s eligible equipment list included 332 multicrystalline PV modules with a 250 W nameplate rating. The average PTC rating for this population was 224.3 W, while the range of PTC values for this group was 16.5 W. Most of this variability may stem not from differences in the PV module design or quality of manufacturing, but instead from the testing laboratory or the local conditions at the time of the test. The CEC list of eligible equipment does not identify either the testing laboratories or the test dates.

To improve the accuracy of measuring module operating temperatures, NREL has investigated methods for improving the outdoor module testing. The decision to adopt this change lies with the committees developing the next editions of the International Electrotechnical Commission (IEC) standards, which proposes replacing NOCT with the metric nominal module operating temperature (NMOT). A key difference is that laboratories measure NMOT with the module operating at the maximum power bias condition instead of at open circuit. In the interim, we encourage PV system designers to consider similarities or differences in module construction when interpreting reported NOCT ratings.

Separating fact from fiction. While the PTC rating implies more-realistic device operating conditions, especially temperature, than STC does, it is clear that adapting a system-level metric such as PTC to PV modules can open the door for error and even for gaming the system. The variation in PTC ratings for PV modules made with similar cells and components is sometimes smaller than the measurement uncertainty. Furthermore, a company that measures the module operating temperature multiple times may be tempted to only record the most favorable value.

The input values and assumptions for modules have faced an intensifying level of scrutiny because they strongly influence PV project valuation. For example, a subtle improvement in the temperature coefficient for the PV modules used in a system can impact project valuation by thousands or even millions of dollars based on the modeled energy production. Though PV power ratings—such as a module’s or a system’s PTC rating—are valuable with respect to system design or acceptance, the real value of a PV project lies in its long-term energy-generating potential. For this reason, industry experts are currently focused on developing ratings and metrics designed to measure and predict energy generation at both the module and the system level.

Moving Beyond Single-Point Performance Metrics

While PV modules undoubtedly have a significant influence on overall PV system performance and value, they create value only as an integrated part of a larger system—and design, operations and maintenance factors, such as module tilt, shading or soiling, also influence system performance. To be most valuable, the metrics used to evaluate PV modules must reflect and support the metrics used to evaluate entire systems. The PTC module rating arose from a method for evaluating PV system capacity rating, so a logical place to look for appropriate alternatives to STC and PTC ratings is system evaluation practices, which heavily emphasize energy estimation.

The growth of the utility market sector, which is accustomed to more-sophisticated product performance curves, has made the limitations of PV module PTC ratings even more apparent. The PV industry urgently needs to create a practical and credible PV module energy rating system. To make such a system possible, industry stakeholders must continue to define more-comprehensive PV module performance metrics that capture performance over a meaningful range of environmental conditions. Further, the industry needs to adopt new product labeling and datasheet standards to ensure that manufacturers and third-party labs publish and report these PV performance metrics consistently.

More-comprehensive performance metrics. In 1982, a team from Arco Solar presented an “am/pm” approach—based on the concept of a standard solar day—for characterizing PV module behavior with respect to operating temperature, irradiance and air mass. In the 1990s, NREL, in conjunction with Endecon Engineering (which later became BEW Engineering, now a renewable energy division of DNV GL), expanded upon this work by developing a model that also considered load type, location and additional weather parameters. At the same time, Sandia National Laboratories developed an empirical PV array performance model based on outdoor PV module testing, known as the “King Method” (named after its creator, David King). In the US, PV designers and simulation software use the King Method to this day. (See “Production Modeling for Grid-Tied PV Systems,” SolarPro magazine, April/May 2010.)

IEC Technical Committee 82, Working Group 2 (TC82/WG2), the committee responsible for PV modules, also recognizes the need to standardize PV module performance metrics beyond STC. In January 2011, after more than 15 years of debate and development, the group published IEC 61853-1, “Part 1: Irradiance and Temperature Performance Measurements and Power Rating.” This first part of a proposed four-part IEC 61853 standard describes the requirements for evaluating PV module performance over a 23-element maximum power matrix at different temperature and irradiance conditions (see Table 2).

IEC 61853-1 serves as a guide for collecting measurements that labs can use to develop PV module parameters for performance simulation tools. For instance, an independent testing lab may refer to IEC 61853-1 for the sets of conditions under which to measure PV module performance. The lab then uses the resulting measurements to develop a model input file for the prediction of long-term energy production at an installation site. Since PV modules operate under a range of temperatures, irradiances and sunlight spectra, the 23-element performance characteristics matrix in IEC 61853-1 provides more-comprehensive rating information than a single-point value such as STC or PTC. Industry stakeholders will need to fully assess the uncertainty associated with the measurement of the characteristics matrix and module performance variability—within a production bin or under low-light conditions—to ensure that the matrix of measurements provides a more meaningful basis of comparison than the NOCT measurements described previously.

Meanwhile, the other three parts of the standard are at different stages of development. IEC 61853-2, approved in draft format, will undergo final committee review in the second half of 2014; it will further develop module parameterizations for PV performance simulation tools. IEC 61853-3 is in the early stages of development; it will present methods and models for predicting PV module energy production based on measurements obtained using Parts 1 and 2, as well as a method for stating a numerical energy rating. Work on IEC 61853-4 has yet to begin; it will define standard time periods and weather conditions for use in the energy rating calculations.

Many in the industry speculate that IEC 61853 Parts 3 and 4 will take years to complete, in part because some parties may lobby against them. PV module manufacturers may advocate for different standard weather condition sets to benefit their own technologies or products. Other committee members may advocate positions based on their testing capabilities or their in-house energy simulation models.

More-detailed labels and datasheets. Around the time of IEC 61853-1’s publication, DKE, the German nonprofit organization responsible for creating and maintaining standards covering the electronics industry in that country, issued a project progress report entitled “Energy Rating of PV Modules” (see Resources). Noting the slow progress on IEC 61853, the authors recommended integrating the module performance matrix from IEC 61853-1 into EU 50380, “Datasheet and Nameplate Information for Photovoltaic Modules.”

According to the DKE progress report, “The most obvious possibility to provide the data is the direct integration into the datasheet of the module in the form of the table.” Integrating the 23-element matrix from IEC 61835-1 into PV module datasheets is an intriguing proposition. Currently, PV module manufacturers typically provide performance data at STC and NOCT conditions, with a note regarding low-irradiance behavior. (The next edition of IEC 61215 is expected to require manufacturers to report module performance at 200 W/m2.) If manufacturers provide all 23 data points, then industry stakeholders will have access to more-transparent and more-reliable performance data. They can use these data to create meteorologically weighted matrices that represent module performance in different climates.

Alternatively, the DKE report notes, an energy label that describes performance based on parameters such as irradiance, temperature or spectral conditions could identify PV modules. Such a label would resemble the yellow EnergyGuide label found on appliances in the US, except that it would estimate energy production rather than consumption over the course of a year. The DKE energy label concept appears to have gained traction in some global markets, particularly in South America. While PV module performance at STC dictates the rating itself, labels also include estimated energy production based on typical annual weather conditions in the specific country.

Expanded requirements for PV module labeling and datasheet requirements may be coming to the US market as well. The Solar America Board for Codes and Standards (Solar ABCs) has published its power-rating recommendations in a report, “Module Power Rating Requirements,” as well as in “A Proposed Standard for Nameplate, Datasheet and Sampling Requirements of Photovoltaic Modules” (see Resources). Solar ABCs has submitted its proposal as an outline for UL 4703, which is awaiting technical review. If UL 4703 is approved, it would require that manufacturers report PV module performance at five rating conditions specified in IEC 61853-1 (as shown in Table 3), that they take these characteristics from a statistically representative sample population, as defined in ANSI/ASQ Z1.4, and that they measure after module stabilization to account for phenomena such as light-induced degradation.

Moving from Power to Energy

It is clear that the industry is slowly moving toward integrating more-detailed performance data into PV module datasheets and product labels. As stakeholders come to a consensus on standards for these parameters, PV system designers and installers need to consider how best to put these data to use. The solution lies in part with production modeling tools. The industry requires an accurate energy yield model to effectively rate and compare PV modules on the basis of energy instead of power. This needs to happen before the industry can transpose PV module power performance and standardized weather inputs into a predicted energy production statistic.

The rapid increase in utility-scale PV project development has prompted the proliferation and wide-scale adoption of PV system performance models—like those discussed in “Show Me the Model” on p. 66—for simulating energy yield estimates. These production models are based on project design parameters and assumptions, weather conditions and input values for system components, especially PV modules and inverters. Rather than relying on a single power rating to estimate energy generation, these sophisticated production models require a suite of parameter inputs and assumptions to describe PV module behavior. When an advanced performance model determines project value, module inputs and assumptions become the primary indicators of module performance, effectively replacing STC or PTC ratings.

Standards for PV system performance evaluation continue to evolve. For example, in 2013 ASTM published ASTM E2848-13 for determining the initial power capacity of a PV system. This standard is of great utility to industry stakeholders for system acceptance because commissioning agents can execute the test over a relatively short period of time (days or weeks). However, stakeholders also need to determine the energy-generating potential of a PV system in its first year(s) of service, as well as its stability over time after accounting for module degradation. Since initial power capacity may or may not correlate with energy-generating potential over time, longer-term PV performance evaluation methods are important, as Timothy Dierauf, et al., discuss in “PV System Energy Performance Evaluations” (p. 22). As the industry matures, we expect that both capacity and energy tests will play crucial roles in predicting and verifying PV system performance.


Brian Grenko / Yingli Green Energy Americas / San Francisco, CA /

Adrianne Kimber / Incident Power Consulting / Oakland, CA /

Sarah Kurtz / NREL / Golden, CO /


Deutsche Kommission Elektrotechnik (DKE) progress report, “Energy Rating of PV Modules,” August 2011,

Muller, Matthew, et al. “Evaluating the IEC 61215 Ed. 3 NMOT Procedure Against the Existing NOCT Procedure with PV Modules in a Side-by-Side Configuration,” 38th IEEE Photovoltaic Specialists Conference, June 2012,

Go Solar California, “Incentive Eligible Photovoltaic Modules in Compliance with SB1 Guidelines,”

Kurtz, Sarah, et al., “Analysis of Photovoltaic System Energy Performance Evaluation Method,” NREL Technical Report (NREL/TP-5200-60628), November 2013,

Sabuncuoglu, Fatih, et al., “Variability in NOCT Standard Test Results as Function of Day, Time of Day and TC Location,” CFV Solar Test Laboratory, February 2012,

TamizhMani, Govindasamy, et al., “Solar ABCs Policy Recommendation: PV Module Power Rating Requirements,” Solar America Board for Codes and Standards, March 2011,

TamizhMani, Govindasamy, et al., “A Proposed Standard for Nameplate, Datasheet and Sampling Requirements of Photovoltaic Modules,” Solar America Board for Codes and Standards, January 2012,

Primary Category: 

The following discussion is from a recent thread on SolarPro’s technical forum. Visit to post questions or join the conversation.

Original post from Terence: One of our installation crewmembers recently walked over a solar panel and the glass cracked. Is this covered by warranty? Per the manufacturer, Canadian Solar, the answer is no. The company sent me module-handling instructions that include “Do not stand, step, walk and/or jump on the module.”

I am just curious if this is true with all panel manufacturers. This is a first occurrence for me. Not that our crew walks over the panels all the time, but we haven’t had any breakage from other manufacturers.

Also, I have seen promotional videos, even at the SPI conference, that show installation crews walking all over the modules. Any insight is highly appreciated.

SolarPro (David Brearley): I cringe every time I see one of those photos. Installers should not stand on PV modules. Here’s the problem: Even if you don’t break the glass, the stress is likely to cause microcracks in the PV cells. Microcracks increase series resistance within the module, create hot spots and otherwise accelerate module degradation. Even something as innocuous as leaning on a PV module while torquing a mounting clamp can result in cell microcracks.

Marvin Hamon, PE: Module warranties typically cover manufacturer’s defects and loss of power only. In addition, there can be a long list of exclusions: for example, the warranty will be invalidated if you mount the modules at less than 10° tilt, if the PV array is less than 10 miles from the ocean or if you don’t follow the installation instructions.

So you are probably not covered by the warranty. However, a manufacturer will sometimes replace a module out of warranty just because it’s good business, good PR or you are a high-volume customer.

Terence: Gentlemen, thank you for the input. @Marvin Hamon: Some manufacturers have told me about the minimum tilt angle. I’ve always wondered, is this because of the snow load, or is there more to it?

@SolarPro: Very good point about microcracks. Assuming the performance dropped below the guaranteed production values, and it was established that it was due to microcracks, is there a way to differentiate between microcracks caused by localized pressure and those that come from the factory’s poor manufacturing processes or quality issues?

Marvin Hamon: Most framed modules have a lip around them that holds water at low angles. While I have not heard a specific reason for the low-angle restriction, I would think it has to do with not allowing water to pool on the module surface and possibly find a way inside. Water infiltration is a significant driver for several types of module failure.

If you send a module back for a power reduction replacement, and it is in good physical condition on the outside, you should get a replacement without too much trouble. If you send a module back with a forklift tine through it and try to claim a power reduction warranty replacement, I don’t think you will get too far.

SolarPro: Power warranty claims are notoriously difficult to pursue. You need high-quality testing equipment to identify the problem. The accuracy of the test results needs to be beyond reproach. You might need to hire a third-party O&M service to meet this criterion. Once you have proof that there’s a problem, you still need to negotiate the manufacturer’s claims process. It may require that you ship all of the questionable modules back at your own expense. Or the manufacturer might send you a quantity of new modules to make up for the lost power, regardless of whether you can integrate those modules into the existing electrical design.

Generally speaking, I think that loss of power due to microcracks would be difficult to trace back to a root cause. Is it a manufacturing defect? Or is it a handling or installation problem? If it were an isolated, low-value claim, a module manufacturer would likely provide a remedy according to its warranty terms.

However, if a system owner was trying to make a power warranty claim on a multimegawatt system and suspected microcracks as the failure mechanism, the monetary value of the claim might justify some very expensive forensics, such as aerial infrared imagery. These images can differentiate manufacturing defects from damage due to mounting. For example, Creotecc has some aerial infrared images that reveal a pattern of temperature variation across ground-mounted PV arrays that appear to result from module mounting or installation methods. These images are likely the type of data a module manufacturer could use to deny a warranty claim based on power loss due to microcracks. In other words, a certain pattern of failure could point away from the product manufacturing process and toward the system design or installation.

Marvin Hamon: It’s always good to keep in mind what a warranty really is: an insurance policy. We pay a premium for module insurance in the price of the module, and we hope we don’t need to use it. A warranty claim is a pain for everyone.

If a manufacturer is too difficult to deal with and gets a reputation for being strict about warranty claims, it risks losing business to a competitive company that is easier to work with. If it is too lenient in warranty claims, it risks being played by people wanting warranty replacements for modules they damaged.

To make it even more interesting, module warranties are too long to be useful. Even if you have a warranty claim 5 years from system install, the chances of getting warranty replacement modules that work in the old system are close to zero. Things are just changing too fast, and our industry lacks standards that would allow an easy interchange of modules. For a warranty claim before install or up to maybe 2 years after, you might be able to put together a workable fix. After that, it’s not so easy. For instance, try to find a drop-in replacement for an AstroPower 4 module panel. It’s been about 10 years since that company went bankrupt. That’s why in larger installations, owners put modules in storage for future replacements.

Primary Category: 

As the solar industry evolves, rigorous analysis and cost-benefit scrutiny are replacing rules of thumb. In this article, we investigate the relationship between module binning, or grouping modules based on specified power or current tolerances, and its impact on energy yields.

PV Module Binning

Output current and voltages vary slightly from one module to the next. Even best-in-class manufacturing techniques result in differences in module output values. A North American EPC recently flash-tested 90,000 300 W modules from a Tier-1 manufacturer. The results showed a variation in output voltages of 34.5 V–38.0 V and in currents of 7.89 A–8.73 A.

To account for output variations, PV modules have a power tolerance specification stating the potential deviation in actual power at STC from the module’s rated maximum power (Pmax). In addition, manufacturers group modules based on the power or current from each module’s flash-test results. This process of module binning groups modules with similar output characteristics based on a specified maximum percent variation. It enables manufacturers—and installers—to further differentiate modules with a smaller percent of variation in output tolerances, which reduces the losses associated with module mismatch.

Current versus power binning. Manufacturers generally group modules based on their power rating, which is also known as  power binning, although some sort modules based on their maximum power current rating (Imp), which is known as current binning. Power binning and current binning are not equal. Figure 1 shows that module output can vary in both current and voltage. Modules connected in series should operate at near-equal current output levels to reduce losses due to mismatch. Within a PV source circuit, variation in current—not voltage—results in mismatch losses.

Within a PV source circuit, different module voltages do not negatively impact one another because the sum of the voltages from each module connected in series equals the source circuit’s voltage. When the individual module voltages are normally distributed (as they are in Figure 1), the higher-voltage modules offset the lower-voltage modules in each string. This netting effect results in essentially no mismatch losses due to the voltage differences between modules.

However, binning modules based on their rated power can result in wide differences in module output current. Binning modules based solely on power results in a mix of relatively high-current with lower-voltage modules, and relatively low-voltage with high-current modules. So even the tightest power band of modules can have significant current mismatch. In Figure 1, for example, the entire sample has a power range of 12% and a current range of 6.5%. But as we tighten the power ranges to 1%, that reduces the current range to 3.5%.

Depending on the manufacturer, the binning tolerance (the range used to bin modules) typically varies from 2% to 10%. Most manufacturers use a 5% tolerance, meaning they group together modules that are within 5% of each other’s output values.

In addition, some manufacturers charge a higher price for modules grouped with narrower binning tolerances. For example, Trina Solar ( offers current-binned modules that it has binned to a range of 2% for a premium of $0.01 to $0.02 per watt. Some developers and EPCs rebin their modules on the project site to improve array performance. Most installers expect that rebinning modules—for example, from a 5% to a 1% range—will improve system energy yield 1%–2%. However, the magnitude of these benefits should be confirmed prior to paying a premium for smaller binning tolerances or rebinning modules on-site.

Calculating the Benefit of Smaller Binning Tolerances

As manufacturers group modules more closely based on their output characteristics, the losses due to mismatch of these characteristics decrease—but how much? To quantify the energy gains, we used HelioScope, a PV design and performance modeling software tool from Folsom Labs ( HelioScope simulates the behavior of each component in the system, including every module, conductor and inverter. The software fully models the system and is capable of predicting the losses associated with module mismatch. Understanding the magnitude of such losses allows project developers to assess whether tight binning or rebinning results in increased energy yields.

Impacts of tighter module binning tolerances. To model the impact of tighter module binning, we use a typical small commercial system consisting of 72-cell polycrystalline modules configured in groups of 12 modules per PV source circuit, with weather conditions defined by the Sacramento, California, TMY3 file. The baseline scenario assumes modules with a current-binning tolerance of 5%, compared to 1% for the enhanced system. The results show that the tighter tolerance has a small impact on overall energy production: about a 0.15% increase in system energy yield. In fact, as shown in Figure 2, even if the modules were current-binned at a 10% tolerance, this would result in mismatch losses just under 0.6%, compared to 0.16% for a 5% tolerance and 0.01% when binned at 1%.

Low effect. When rebinning polycrystalline modules from 5% to 1% current tolerance, the resulting 0.15% improvement in energy yield is much less than the 1%–2% that most installers and engineers expect. This illustrates a common misperception about module mismatch. While modules connected in series within a PV source circuit should operate at near-equal current levels, the differences in current do not translate to a 1-to-1 power loss. Figure 3 shows the nonlinear relationship between current and power at the peak power point.

When we plot a module’s current versus power, the region around the module’s MPP is relatively flat. As a result, within a PV source circuit, as the underproducing modules reduce a high-output module’s current, that module’s voltage increases. Voltages increase in modules with high-operating currents as modules with lower-operating currents effectively decrease their current output. Due to this voltage compensation, small disruptions in a module’s operating current have an even smaller effect on its power output. Farther away from their operating point, modules begin to have greater mismatch loss, as seen in Figure 3 (note that the slope increases with larger ranges).

Sensitivity analysis. Figure 4 shows the energy gains due to rebinning from a 5% to a 1% tolerance for various system designs and components. In each scenario, the impact from module binning on energy production is small. There are two scenarios worth noting. A PV source circuit with 18 polycrystalline modules connected in series—such as in a 1,000 Vdc array configuration—incurs a 0.19% increase in energy yield from rebinning. Alternatively, we see that First Solar’s CdTe modules benefit the least from rebinning from a 5% to 1% tolerance—only about 0.09%.

The more modules per PV source circuit, the more likely the circuit is to experience losses due to module mismatch: The modules in series must have identical current, and more modules means more data points. The lower fill factor of CdTe modules makes them more tolerant to mismatch, since the module’s power does not change as much based on changes in current near its maximum current point. For both CdTe and crystalline-based modules, the current-power curve is flat near a module’s MPP, and it is even flatter for a CdTe module.

Financial Implications of Binning

The bottom line is that tighter module binning can lead to a slight improvement in a system’s energy production, but not as much as expected. Although module type and the number of modules per PV source circuit have noticeable effects on the amount of energy gained by rebinning from 5% to 1%, our analysis indicates an increase in energy production of only 0.09%–0.19%. Understanding the energy impacts from binning enables developers to assess how much they should be willing to pay for a tighter module binning tolerance.

A common industry guideline is that a 1% improvement in system energy yield is equivalent to an up-front net present value of $0.03–0.04/Wp. If we apply this to the modeled 0.15% energy increase from rebinning, the up-front net present value would be $0.006/Wp. While this is slightly lower than the $0.01–$0.02/Wp that Trina Solar charges for current-binned modules, it is close. For a 1 MW array with 4,000 modules, this 0.15% of energy yield would be worth $4,500–$6,000 per year. At $40 per labor hour, this total provides a maximum budget for 100 to 150 hours of labor. It seems unlikely that an on-site installation team could test and sort 4,000 modules in that time.

Ultimately, the benefit of rebinning modules depends on kilowatt-hour value for PV generation and the labor cost to test and sort modules. As the prices for PV systems decrease, the benefits of tighter tolerances become harder to justify economically.

Informing Module Choices

Understanding the minimal impact that module mismatch has on energy losses is also valuable when selecting a module. For example, is a module with a 0% to +1% power tolerance preferable to one with 0% to +5%?

When module power tolerances are positive, the average module power is higher than the datasheet rating. While both examples above have the same lower end of the range, the +5% modules have a midpoint of +2.5%, compared to +0.5% for the alternative module—resulting in a 2% net difference. Given that systems with modules with a current-binning tolerance of 5% produce 0.15% less energy due to module mismatch than a system with 1% current-binned modules, we expect the modules with the 0% to +5% power tolerance to outproduce those with a 0% to +1% power tolerance by 1.85% (2.0% − 0.15% = 1.85%).


As the industry matures, PV developers are looking for ways to improve their ROI by reducing costs and increasing production. However, in the case of rebinning modules, it is very unlikely that the improvement in energy production justifies the labor cost, particularly if the testing and sorting is done in the field. Binning practices are just the first of many PV design strategies that must be put to a more comprehensive cost-benefit scrutiny.

Paul Grana / Folsom Labs / San Francisco, CA /

Paul Gibbs / Folsom Labs / San Francisco, CA /

Primary Category: 

Discerning buyers understand that seemingly identical PV modules may perform very differently over 20 or 30 years. But how can they take that knowledge to the bank?

On the opening day of Intersolar North America 2013, NPD Solarbuzz announced that installed PV capacity in the US had recently surpassed 10 GW. While the compound annual PV capacity growth rate has exceeded 50% since 2007, roughly 8.3 GW of capacity was installed between Q1 2011 and Q2 2013. This impressive build-out is largely due to a rapid expansion of the utility sector and collapsing module prices, both of which have contributed to dramatic PV system price declines.

According to “US Solar Market Insight Report: 2012 Year in Review,” part of a series published quarterly by GTM Research and the Solar Energy Industries Association (SEIA), more than 3.3 GW of PV capacity was installed in the US in 2012. With the utility sector accounting for 53% of this capacity, the weighted average PV system price came in at $3.01 per watt, which is lower than the average selling price (ASP) for a PV module circa 2007 or 2008. By comparison, the ASP for a PV module fell from $1.15 per watt in Q4 2011 to $0.68 per watt in Q4 2012, a precipitous 41% decline, before leveling off somewhat at $0.64 per watt in Q1 2013.

Arguably, there has never been a better time to purchase PV modules. Prices are hovering at or near record lows. But at what cost to the industry? While falling module prices are generally seen as a sign of progress, the rate at which module prices have fallen over the last 18–24 months is potentially problematic. In “Solar PV Profit’s Last Stand” (, March 2013), Paula Mints notes that a “42 percent decrease in price in one year is not progress by any logical assessment,” because it is not “representative of true learning.” True progress, Mints suggests, is sustainable, as when price declines generally track a learning curve or a technology road map. She likens the current market conditions, in which manufacturers are pricing at a loss to garner or preserve market share, to a nightmare.

If this is a nightmare, it is one that some companies will not wake up from. Since 2008, Greentech Media has followed the fortunes of more than 200 VC-funded solar startups, publishing regular updates on the growing list of shuttered or restructured solar firms (see “Rest in Peace: The List of Deceased Solar Companies” for the most recent article). The list includes some well-known crystalline silicon (c-Si) PV module manufacturers, including Bosch, BP, Evergreen, Q Cells, SCHOTT Solar, Solon, Siliken and Suntech Power. According to a recent GTM Research report “Global PV Module Manufacturers 2013: Competitive Positioning, Consolidation and the China Factor,” more companies are expected to join this list. Having analyzed the facilities, financial health and business models for more than 300 module manufacturers, GTM Research concludes that 180 existing manufacturers will either go out of business, exit the market or undergo an acquisition between 2013 and 2015.

While module manufacturing is undoubtedly ripe for consolidation, our “2013 c-Si PV Module Specifications” table (see sidebar) suggests that the real shakeout is yet to come. This table includes abbreviated specifications for 849 module models from 62 manufacturers. Surprisingly, these totals have remained substantially unchanged since SolarPro magazine published module data for 2012. All of these PV modules are certified to UL 1703 and eligible for rebates under California Solar Initiative programs; all of the manufacturers have established North American distribution channels.

Given the challenging competitive environment facing module manufacturers, purchasers are wise to keep this adage in mind: caveat emptor, “let the buyer beware.” In last year’s c-Si PV module article (SolarPro magazine, October/November 2012), Publisher Joe Schwartz focused on manufacturer profiles, addressing one major concern for purchasing agents—namely, “From whom am I buying?” This article takes a closer look at module reliability, quality assurance and performance. With the help of a cross-section of industry experts and stakeholders—representatives from module manufacturers, testing labs, material suppliers, project developers and plant constructors—I address an increasingly important follow-up question: “How do I know that the c-Si PV modules I am buying today will last for 20 or 30 years and provide the performance that I am banking on?”

Beyond Certification and Qualification Testing

An important first step for anyone purchasing PV modules is to determine who performed what product testing. For example, Nationally Recognized Testing Laboratories (NRTLs), which are private-sector organizations recognized by OSHA, currently offer two types of testing programs for c-Si PV modules: safety certification and design qualification.

Since certification and qualification testing programs are based on published product standards and performed by testing entities that are subject to external audits and inspections, it is tempting to view all listed c-Si PV modules as inherently comparable commodities. While this may be true from an AHJ’s point of view, common sense—not to mention a growing body of technical literature (see Recommended Reading sidebar)—suggests that this is a naïve assumption for project developers, contractors and owners. For reasons that I discuss later, neither safety certification nor design qualification testing is intended to predict long-term module reliability.

However, as the PV industry has matured, customers have become more sophisticated. Buyers now demand testing programs and reporting protocols that better differentiate products according to long-term reliability and performance in the field. Industry stakeholders and working groups have responded accordingly.

Many testing labs now offer comparative accelerated testing programs, and these parallel efforts are expected to eventually culminate in a standard comparative rating system. Meanwhile, a PV industry–specific version of ISO 9001, which addresses quality assurance in manufacturing, was proposed in June 2013. Last but not least, the Solar America Board for Codes and Standards (Solar ABCs) has published proposed nameplate, datasheet and sampling requirements for PV modules. If the solar industry can successfully implement accelerated testing, quality assurance and labeling standards, these will augment existing safety and performance testing programs in promising ways.

Safety certification. To gain access to the rapidly growing North American solar market, solar module manufacturers technically need only safety certification for their products. For flat-plate PV modules, the relevant safety standard is UL 1703. What buyers need to remember is that the intent of this safety standard is to prevent hazards to persons and property, such as electrical shock or fire. While a product that is certified to UL 1703 has successfully passed an essential series of safety tests, this standard does not require that an NRTL verify the module as actually functional at the end of the test sequence. Therefore, safety certification alone is not adequate as a means of qualifying product performance or reliability in the field.

Design qualification. Module buyers and purchasing agents can raise the bar by identifying products qualified to performance standards that the International Electrochemical Commission (IEC) developed. For c-Si PV modules, the relevant design qualification test is IEC 61215, which was also adopted as UL 61215. This performance standard not only includes tests and measurements that are not part of the product safety certification process, but also determines whether the test modules remain functional. For example, this standard ensures that test modules operate above a minimum output power threshold (90% of PMAX at STC) after the final light soaking.

Additional tests performed as part of the design qualification process include electrical safety tests (insulation resistance and wet leakage current), environmental tests (thermal cycling, humidity freeze and damp heat), performance tests (PMAX at STC and low irradiance, temperature coefficients and nominal operating cell temperature [NOCT]), and various stress tests (UV exposure, hot-spot endurance and mechanical load tests). While IEC 61215 also includes an outdoor exposure test component, the duration of outdoor exposure is limited to just 60 kilowatt-hours per square meter. For a product intended to withstand 20 or 30 years of exposure, this is an exceptionally short test period, equivalent to about 10 days outdoors in Phoenix, Arizona. While the outdoor exposure test may identify problems other laboratory tests do not reveal, its short duration prevents it from being a useful indicator of long-term reliability.

In a presentation delivered at the PV Module Reliability Workshop 2010, hosted by the US Department of Energy, Dr. Govindasamy TamizhMani of the TÜV Rheinland PV Testing Lab pointed out: “The primary goal of the qualification testing is to identify the initial short-term reliability issues in the field.” While long-term reliability is increasingly important for contractors and investors, IEC 61215 was never intended to address this issue. In “Basic Understanding of IEC Standard Testing for Photovoltaic Panels,” a white paper published by TÜV America, authors Regan Arndt and Robert Pluto clarify: “Reliability is neither defined nor covered by the existing IEC standards. The lack of reliability standards is partially due to the fact that, to date, there are not enough statistical data collected from the PV fields.”

Comparative accelerated testing. In 2011, NREL and other industry stakeholders founded the International PV Module Quality Assurance Task Force. One of the task force’s primary goals is to develop a comparative rating system for PV module durability. According to a framework outlined at a regional meeting of the task force in February 2013, comparative tests are defined as “accelerated tests that differentiate products according to their long-term durability in the field for a specific-use environment.”

The International PV Module Quality Assurance Task Force is in the process of defining a specific comparative testing protocol. Conceptually, individual tests will be similar and in some cases identical to those used for qualification testing, but they will be applied over a longer duration. Further, the tests will be applied in a specific sequence and module performance will be quantified after each test. Once a test protocol is established, it will be fine-tuned over time. The task force expects that the testing will evolve as the industry better understands the science behind the types of failure and wear-out mechanisms observed in the field.

One of the task force’s end goals for the comparative testing standard is a rating system for PV modules that is easy for customers to understand. The authors of “A Framework for a Comparative Accelerated Testing Standard for Photovoltaic Modules” propose a comparative rating system, as shown in Table 1, based on three IEC climate zones: moderate; warm damp, equable; and extremely warm dry. For example, the relative durability of a module might be indicated using an A, B or C scale (best, better or acceptable) depending upon the general installation climate (temperate, tropical or desert). This proposed module rating system is analogous to the Uniform Tire Quality Grading System, which allows consumers to quickly compare tires based on tread wear, traction and temperature ratings. The authors of the comparative testing framework for PV modules would also like the rating system to account for additional environmental stresses that vary by location, such as snow, wind, salt and farm chemicals.

Collectively, this information will prove valuable for large fleet owners and operators. It will effectively provide customers with a means of improving the accuracy of their quantitative predictions of product lifetime in specific applications. Unfortunately, the International PV Module Quality Assurance Task Force is unlikely to publish a single international standard for comparative accelerated testing in the near term. The process will move at the methodical pace typical of volunteer efforts designed to build international consensus. (A website for the task force [] provides information about how to volunteer for one of the ten task groups.)

Meanwhile, private-sector organizations are to some extent filling the void. The Fraunhofer Institute, for example, has started testing PV modules to its proprietary PV Durability Initiative; it completed the first results in June 2013. The Renewable Energy Test Center offers the Thresher Test, a model reliability test protocol that it helped to develop. Intertek offers a module bankability service to its customers. Besides the fact that these test protocols are not identical, making a side-by-side product comparison impossible, each manufacturer releases the results of these voluntary tests at its own discretion. This lack of transparency has created a business opportunity for companies to provide product testing and manufacturer qualification services direct to module buyers. PV Evolution Labs ( and SolarBuyer ( are two of the companies active in this market niche.

Quality assurance. Task Group 1 of the International PV Module Quality Assurance Task Force is focused on generating guidelines for manufacturing consistency. This task group determined that the best way to address quality assurance within the manufacturing environment was to develop a PV-specific version of ISO 9001, a generic standard for quality management systems in manufacturing. As with product certification, an independent third party audits and verifies ISO 9001 compliance.

As described in an NREL Technical Report, the goal of the supplemental requirements to ISO 9001 “is to provide a guideline for manufacturers of modules to produce modules that, once the design has been proven to meet the quality and reliability requirements, replicate such a design on an industrial scale without compromising its consistency with the requirements.” While it will be some time before the IEC and ISO standards organizations adopt these supplemental requirements, the proposal is available in the meantime as a resource to PV module manufacturers wishing to develop or improve quality management systems. The task group welcomes industry feedback—positive and negative—regarding the process of implementing the draft standard.

Labeling and sampling. In March 2011, Solar ABCs published a recommended policy report regarding PV module rating requirements (see sidebar). This report not only educates consumers about the potential for discrepancy between nameplate ratings and the output characteristics of delivered PV modules, but also proposes a method for tightening the tolerance between these values, which may reduce financial risk and increase investment in PV systems. Subsequently, in January 2012, Solar ABCs released a proposed standard on “Nameplate, Datasheet, and Sampling Requirements of Photovoltaic Modules,” similar to EN 50380, a standard developed by the EU in 2003.

These proposed labeling and sampling requirements seek to improve the accuracy and consistency of production and financial models, which is of great interest to project owners, developers, constructors and investors. At present, for example, many module manufacturers publish PV module ratings at STC only. However, IEC 61853-1 requires manufacturers to quantify module performance under four additional test conditions: high temperature, NOCT, low temperature and low irradiance conditions.

The proposed Solar ABCs standard would require manufacturers to provide these additional data on the product datasheet. Incorporating these data into PAN files will improve the accuracy of production and financial models for plant performance. The proposed standard also includes statistically valid random sampling requirements intended to reduce uncertainty in the reported data. Collectively, these labeling and sampling requirements minimize risk, which will make it easier and less expensive to finance solar investments.

Can You Bank on It?

While a lot of progress is clearly being made in the areas of comparative testing, quality assurance and labeling standards, the development process is slow and steady. Market conditions, meanwhile, are fast-paced and dynamic. This means it is more important than ever for module buyers to do their due diligence.

Module quality and reliability is such a hot topic right now that it was even the subject of a provocatively titled article in the New York Times. Published on May 29, 2013, “Solar Power’s Dark Side” by Todd Woody (the online version is titled “Solar Industry Anxious Over Defective Panels”) quickly became a hit in the anti-solar blogosphere. After all, data quoted in the article suggest that defect rates for PV modules are very high—between 5.5% and 34.5%, depending on the source of the data—and that 80% of PV systems in Germany are underperforming.

While numbers like these make for exciting copy, they do not tell the whole story. In the comments section accompanying the online version of Woody’s article, commenters identifying themselves as veterans of the solar industry maintain that they have never experienced a single module failure. For example, Canute from Central Vermont writes: “I have been a PV installer since 1998, and I have never had a module failure. My colleagues report a few, but limited to particular batches of modules from a small number of manufacturers.” This is echoed in a comment by Jim Jenal from Pasadena, California: “We have been installing solar since 2007, and we have never had a solar module failure. Not one.”

Depending on whom you talk to, PV modules either never fail or they routinely demonstrate unacceptably high defect rates and pose a high risk of failure. While these options may at first appear to be mutually exclusive, my interviews with industry experts suggest that this is not necessarily the case. Given the downward price pressure in the market and the fact that international standards are not yet in place for reliability, quality and performance, either outcome is possible. Your results will likely vary based on how robust your process is for qualifying PV module manufacturers and their products.

You should ask your supplier if it has an extended reliability-testing program in place. If so, who oversees the program? Does it have quality management systems in place? If so, are these audited? Does it have independently verified PAN files for use in PVsyst or other modeling programs?

If your supplier cannot answer these questions or if its answers are unacceptable, look elsewhere. While PV modules may not be commodities, you can choose from many manufacturers. Be aware that the module tiering system is primarily a measure of bankability, of a manufacturer’s access to nonrecourse debt financing. While companies often use tiering categories for marketing purposes, they are not intended to substitute for technical due diligence.

Industry Perspectives

In the immediate wake of the Woody article, veteran tech editor and blogger Tom Cheyney posted a meditation on the SolarCurator blog entitled “Pachyderm in the PV Parlor.” In it, he likens concerns about PV module quality and reliability to the proverbial elephant in the room. However, this is less a case of an obvious problem that no one wants to discuss so much as a complex topic that is the subject of intense scrutiny—one that involves material science and statistical probabilities that only lifecycle testing can verify. Since we cannot adequately describe this elephant from a single perspective, I reached out to industry experts throughout the value chain to get their perspectives on module quality and reliability.

Sarah Kurtz, PhD

Principal scientist, NREL

SP: In addition to being a principal scientist at NREL, you manage its PV Reliability Group. Do you think the PV industry has a reliability problem? If so, how big a problem is it?

SK: There is ample evidence that PV products can have exceptionally high reliability. But that does not make them immune to the sorts of quality issues that have popped up at one time or another in almost every product on the market. If a customer asks for the lowest price and never asks about quality, that customer is taking a risk—it does not matter whether the product is a PV module or a car or baby formula.

The reliability numbers in Todd Woody’s New York Times article could be a reality for customers shopping on price only. But if a customer is shopping for a high-quality product and asking for evidence, then statistics show that PV modules can deliver very high quality. For example, a presentation at the 39th IEEE PV Specialists Conference on June 17, 2013, described a return rate of 0.44% percent for a fleet of 3.4 million front-contact silicon modules installed for an average of 4.9 years. A companion fleet of more than 8 million back-contact silicon modules was described as having a return rate of 0.005%. Although this fleet has been in the field for only 2 years, the low return rate contrasts with the much higher defect rates quoted in the New York Times article.

These data represent a very large sample set, more than 1 GW. However, they are also reported by SunPower, which has gone to great lengths to emphasize quality. While the PV industry has largely focused on cutting costs in recent years, many PV companies are retaining an emphasis on reliability. Products made by companies with a strong focus on reliability are likely to cost a little more up front. But they are also likely to deliver solar electricity at a lower cost in the long run, as the products retain high efficiency and low failure rates throughout the warranty period.

Of course, I have not answered your real question, which is whether the industry is headed for a major setback because of massive failures. I can document that many companies have excellent statistics; I can also document isolated instances of massive failures. While I do not have data representing the experience across the board, I think this is of little consequence to the PV customer. If only 0.1% of the PV modules installed in 2013 fail, we are still talking about tens of megawatts of PV capacity globally. No one wants to be the customer who bought those megawatts.

SP: How will the work the International PV Module Quality Assurance Task Force is performing improve the situation?

SK: The task force is working toward a universal set of standards that will make it much easier for PV customers to assess how well a given product will meet their needs. This is not as simple as labeling this module “good” or that module “bad.” The challenge is to be able to identify which module will last the longest in a given application. A module that survives the heat of the desert may be different from one that survives snow and ice. The module that survives a prolonged steam bath might not survive extended UV exposure.

If we can improve our ability to test products for each climate, it also provides manufacturers with an opportunity to reduce costs for a specific application. For example, modules that do not need to withstand snow or hurricane-force winds may be made with thinner frames and glass, reducing materials, shipping and installation costs.

As the community works to squeeze the cost to the lowest possible level, having a tool to identify when the squeeze is helpful and when the squeeze went too far will enable design optimization. Eventually, I expect that PV prices and insurance rates will routinely reflect a product’s demonstrated durability. For example, when I buy shingles for my roof in Colorado, I get lower insurance rates if I buy the shingle with the best hail rating. As the solar industry matures, customers, financers and insurance providers have similar needs. Being able to differentiate the expected outcome for the intended application will enable us to make choices that minimize cost in the long run.

While a comparative testing standard will obviously make it easier to identify modules that will work well in a given climate and mounting configuration, quality assurance during manufacturing is even more important right now. It does not matter how good your design is if the manufacturing process does not duplicate it. Deviation from the intended design increases risk.

SP: Do you have any recommended best practices for PV module buyers for whom quality and reliability are a primary concern?

SK: Just as you typically check references before engaging in any significant purchase, you should evaluate manufacturers according to their commitment to quality and reliability before making a sizeable PV purchase. Ask manufacturers to show you their warranty return rates and data documenting the field performance and degradation rates that they have demonstrated. Ask other customers about their experience with the company. Ask third-party organizations to do independent checks. The level of scrutiny should be appropriate for the size of the purchase.

In other words, customers should treat an investment in PV prudently, like they would any other investment. There is nothing magical about PV that makes it impossible for modules to fail. In conversations with suppliers, customers should communicate their desire to have a product that will last. Further, customers need to demonstrate their willingness to pay a little more up front for a product that can be demonstrated to be superior, since this can reduce their cost per kilowatt-hour in the long run.

Sunny Rai

Regional vice president, Intertek

SP: In your role as a regional vice president for renewable energy at Intertek—which is one of the NRTLs that tests c-Si PV modules to UL 1703 and IEC 61215—do you see any evidence that the PV industry has a reliability problem? If so, how big a problem is it?

SR: While I think reliability is an issue, it is not at a level that is very alarming. There was a time when every other small manufacturer across the world was getting into this business and no quality-control watchdog system was in place. Then we had multiple new manufacturers coming to Intertek every month with new products to test. Now we see fewer new products and more modifications to existing products. With the slowdown that we have had over the last year or so, there is a lot of consolidation in the industry, which has actually helped us in terms of potential reliability problems. Smaller manufacturers that did not have module quality programs in place are now exiting the market.

But that does not mean we do not have a problem. Reliability is still an issue, even with some well-known manufacturers. The problem is not intentional. Rather data that was not available 4, 5 or 6 years ago with regard to field performance is now available. I have seen pictures of modules out in the sun for 2 or 3 years that show more significant product fatigue than was expected. It is not a complete failure, but it is a concern. This has made buyers and investors more wary. They want additional testing performed. They want additional surveillance.

That is where quality-control programs offered by Intertek come in. Instead of just doing certification testing, we can do extended reliability testing or perform quarterly inspections or preshipment inspections. These quality controls are available for modules, inverters, racking systems and other products.

SP: Is there a need for internationally recognized standards for module reliability and quality assurance in manufacturing? If so, is Intertek involved in efforts to develop these standards?

SR: Yes, the industry needs comparative testing and quality management standards. Otherwise you have 20 different flavors of testing out there. Certain labs have proprietary programs; certain buyers and EPCs have their own qualification requirements. Then there are large manufacturers that have designed their own test programs. All of this makes an apples-to-apples comparison impossible for customers.

Staff from Intertek sit on several working committees to write more robust quality standards. We participate in the International PV Module Quality Assurance Task Force working under NREL and a similar effort under IEC. Since we are on those committees, we know the direction that things are heading. So in the absence of those new standards being published, we have developed our own reliability and testing program as a way to help the industry in the short term.

Our program involves extending existing tests defined by IEC standards—increasing duration or temperature or number of cycles—as well as performing additional tests based on our understanding about how products are performing in the field. For example, certification does not require tests for potential induced degradation or light-induced degradation, but we see that they are an indicator of product quality and performance.

The other thing we can do is test module performance under additional test conditions. Besides STC and NOCT, we can test to see how modules perform under high temperature conditions, low temperature conditions and low irradiance conditions, and we can verify temperature coefficients. We can then use these results to write more detailed PAN files for production modeling programs like PVsyst.

SP: Do you see any evidence that buyers are sophisticated enough to differentiate products according to quality and reliability?

SR: There are clearly buyers saying, “If you want me to consider buying your module, then go through this program and give me these results.” In the competitive environment that we are in right now, manufacturers are out of the market if they do not do the testing.

SP: If buyers are concerned about module quality and reliability, what should they be looking for?

SR: Any manufacturer that has gone through some type of quality process will publicize that fact on its website or in its brochures. Assuming that quality or reliability programs are in place, look to see who is doing the testing and providing the data. At one end of the spectrum, you have NRTLs—like Intertek or TÜV or UL—that are accredited and audited to make sure equipment is calibrated and personnel are qualified; on the other extreme, you have manufacturers that are doing their own testing. Companies like PV Evolution Labs fall somewhere in the middle. If I were a buyer, I would first ask if quality and reliability data are available and then evaluate the source of the data.

Jenya Meydbray

CEO, PV Evolution Labs

SP: You were featured in Todd Woody’s New York Times article, which describes the solar industry as “facing a crisis of quality.” To what extent do you agree or disagree with this characterization?

JM: At PV Evolution Labs [PVEL], we see only a small portion of the modules designated for deployment in US projects. These projects are developed by sophisticated buyers and investors who require third-party technical due diligence focused on PV module quality. Consequently, I am hesitant to draw broad conclusions about industry trends regarding PV module quality.

Having said that, we regularly perform vendor qualification and statistical batch testing services and have seen roughly a 5%–10% failure rate. Failures are typically caused by poor quality control and occasionally by low-quality materials or incompatible material combinations. From our perspective, the problem of low-quality modules is not pervasive. However, we do occasionally encounter defects, and the buyers are very glad to have caught these defects before they impact their return on investment.

SP: Are buyers sophisticated enough to differentiate products according to quality and reliability?

JM: Buyers and investors across the industry are currently developing their technical-diligence programs. Historically, price and brand dominated buying decisions. Today, we do see more and more buyers introducing testing programs for qualifying new module types. At PVEL, we strongly believe that measuring critical PV module behaviors—rather than using industry averages and assumptions—results in higher certainty of meeting plant performance predictions. This has implications for cost of capital, refinancing and bringing new risk-averse sources of capital to this industry.

Regulatory standards like UL 1703 and IEC 61215 set a minimum bar for product safety and performance. They are designed to screen out gross design flaws. Every module in the market must pass these tests, so they cannot be too onerous. But different end markets have differing product reliability and performance needs. One current gap is that these tests are performed on prototype samples before the volume production lines ramp up, so they do not capture manufacturing quality-control issues. PVEL provides statistical batch testing services to ensure quality on a batch-by-batch basis.

The solar market is increasingly migrating to more risk-averse debt investors who are underwriting long-term cash flows that require more certainty. Long-term cash flow certainty requires that the PV system perform for 20-plus years. PVEL provides a higher bar of reliability, performance and quality assessment for module buyers and investors that have more strict long-term performance expectations.

SP: What do module buyers need to understand about the relationship between module quality and reliability?

JM: We typically refer to product quality as consistency across tens or hundreds or thousands of modules. For example, to build statistical certainty that specific module batches going into our clients’ projects are defect free, PVEL offers statistical batch testing. We think of product reliability as characterizing how the module ages under environmental stress, if it is built to spec. The bill of materials, product design and manufacturing recipes largely define this. We offer vendor qualification services to benchmark many module types against each other and to understand aging behavior.

SP: Do you have any recommended best practices for PV module buyers for whom quality and reliability are a primary concern?

JM: These buyers should work with PVEL and SolarBuyer. Together, we have a large number of PV module vendors that have gone through some or all of our approved vendor program, which includes factory audits, reliability testing and performance testing. Buyers can gain valuable insights into product performance and factory quality control by participating in this program, often at minimal cost. All individual projects greater than several MWs should also undergo statistical batch testing to ensure that the module supply is free from manufacturing defects. SolarBuyer offers several additional services to oversee quality issues on the factory floor. We can customize our engagement to the customer’s risk tolerance and available budget.

Brian Grenko

VP of operations, Yingli Green Energy

SP: Do you think that the solar industry is “facing a crisis of quality”?

BG: While the inevitable shakeout of companies throughout the solar value chain—from system developers to equipment and component suppliers—should ultimately advance our industry, it has resulted in a fair amount of collateral damage. Arguably, these consequences are typical of a technology adoption lifecycle. Industry oversupply and stabilizing silicon costs have forced PV module manufacturers—and subsequently their component suppliers—to find new ways to take cost out while the elasticity of market pricing is put to the test.

Based on a number of recently published articles, including the New York Times piece, it would seem apparent that not all PV modules are created equally. But that notion should come as no surprise to anyone in our industry. PV modules are not commodity products and never have been. At the same time, solar energy is becoming mainstream and is consequently commanding more attention.

The amount of installed solar in the US has increased by an order of magnitude within the last 5 years. As anyone who has ever owned a smartphone or laptop computer knows, quality problems are a byproduct of technology maturation. The ways in which companies respond to these issues and provide value are ultimately how customers measure them. The prompt and proper identification, management and communication of quality problems are critical to legitimizing PV technology as a sustainable, competitive form of power generation. Those who take shortcuts should be exposed.

With that said, it is evident that certain industry players quoted in these articles have opportunistically aligned themselves to proliferate a coordinated campaign of fearmongering while attempting to create market pull for their own products and services. Our industry is facing a crisis of integrity as much as, if not more than, a crisis of quality. I am confident that most people see right through this. While some may gravitate towards negative headlines like “The Dark Side of Solar,” the reality, as many of us know, is that our industry on the whole is actually flourishing and creating downstream jobs, despite consolidation.

SP: Is there a need for internationally recognized standards for module reliability and quality assurance in manufacturing? If so, is Yingli engaged in these efforts?

BG: A large number of industry stakeholders have been very seriously looking into these questions, and Yingli is absolutely engaged in these efforts. The fundamental challenge is whether we are able to generate expected field failures based upon the environmental conditions generated with the testing equipment available to us. For instance, it is financially impractical to accelerate conditions like UV exposure with temperature cycling at the same time. So, given the consensus that the IEC 61215 protocol is not sufficient to adequately qualify PV modules, various groups have focused on either extending current IEC 61215 tests outward by multiplication factors or combining different test cycles in sequence or both.

In 2011, Yingli participated in a large industry working group engaged on this topic, the result of which was a test protocol commonly referred to as the “Thresher Test.” We are currently utilizing this procedure in the evaluation and qualification of new components and PV module materials at our PV testing laboratory in the San Francisco Bay Area. We have also participated in the International PV Module Quality Assurance Task Force, sponsored in part by NREL, which is developing different tests specifically focused on different known failure modes. We fully support these initiatives and trust that new standards will ultimately become adopted and consequently improve the industry.

SP: Is the market sending a mixed message to PV module manufacturers? Are customers willing to pay more for a higher-quality, more reliable product?

BG: I believe customer expectations are pretty clear in this regard. With no moving parts, PV modules are expected to perform as predicted and operate for 25 years or more with little to no maintenance. Companies that build and operate PV systems bank on this when they wrap system performance or look to sell completed projects. The last 5 years in particular have made these companies—as well as project investors and end users—wonder if any solar companies will last as long as their product warranties. Sophisticated buyers conduct a thorough due-diligence process before selecting equipment. Further, manufacturers with a proven track record for delivering high-quality products on time can definitely command a premium. PV module manufacturers that are able to offer value-added products, such as PV module fleet performance guarantees and third-party product insurance, can also realize more business opportunities.

Raju Yenamandra

VP of business development, SolarWorld

SP: You have been working in the US PV industry for more than 33 years. What is your perspective on module reliability? Does the industry have a problem? If so, how big a problem is it?

RY: Historically, solar modules have been extremely reliable. The origin of that reliability is that companies like ours teamed up with the Jet Propulsion Lab [JPL] in the late 1970s and developed a protocol for reliability testing, which was incorporated into product manufacturing. This is what created the standards that we have today. The explosive growth in the industry in recent years is in part due to the available supply of manufacturing equipment, and new market entrants can copy existing processes.

The problem arises with falling prices. PV modules are the generator at the heart of every PV system. They are manufactured by a large number of companies, every one of which lost money in 2012. So there is obviously an imbalance in the supply chain. Different companies have responded to this imbalance in different ways.

At SolarWorld, our strategy is built around our 37-year reputation for unparalleled reliability, and we are not going to sacrifice that. For example, we have had 40 or 50 EVA [ethylene vinyl acetate] suppliers come to us saying that they could reduce our costs. We narrowed those companies down to eight, and ultimately to two. Why? Because we performed tests, tests and more tests. New materials have to be equal to or better than existing materials for us to replace any component.

Companies that have never been through a downward business cycle potentially see things differently. Costs are high and prices are low. They may have relatively stagnant technology, meaning they cannot increase module or production efficiency to keep pace with falling prices. So how can they decrease their losses? They can increase throughput by lowering the acceptable quality level for their factory. They can decrease costs by substituting cheaper, unproven materials. Or they can cut short processes—like EVA curing—to increase throughput beyond what is realistically possible. They may do a paper analysis, realize they can save some money and press “Go” without doing any testing.

What scares me is that the sweetness of the low price will one day be completely overshadowed by the bitterness of failures. What if a company supplying a large quantity of modules has wide-scale failures related to the use of unproven materials? If products fail, the media—especially people who are not within the industry—will see that particular solar projects are failing and conclude that solar has failed. The media will paint the industry with a big broad brush. If I am a banker reading those reports, I do not know all of the details. But do I want to fund solar projects? Now I am not so sure. I need more data, more proof.

In the short term, I do not see module failures to be a huge problem. They are more of a nuisance, a hassle that will leave a bad taste in the mouth of customers. But I hope and pray that we do not have mass failures in the long term. If we do, that will be a huge disservice to the industry, which took a long time to grow to where it is today.

SP: Is there a need for internationally recognized standards for module reliability and quality assurance in manufacturing?

RY: We do not need more regulations or more testing so much as we need a self-regulating approach. A manufacturer can circumvent any new standard that you create. It can make a special module for the testing and get the product approved, but go right on doing a shoddy job with the manufacturing process itself.

Our call to the industry is for every company to make sure that the quality of the product it puts out is very good—and please do not send cherry-picked modules for testing. Pick modules from your production lines. The intent of testing is to maintain product quality. That intent should not be lost for short-term gains.

SP: Is the market sending a mixed message to PV module manufacturers? Are customers willing to pay more for a higher-quality, more reliable product?

RY: I think they will. Product prices are pretty tightly banded together. When viewed as a percentage, we are talking about a single-digit difference in module prices. This difference is even smaller when viewed at the system level. As module prices fall, the PV generator makes up a smaller percentage of total system cost.

However, in terms of system performance—as measured by the levelized cost of energy, for example—modules are the most critical component. Increasing the probability of failure in a critical component, one that makes up a shrinking percentage of total system cost, is a risk that is simply not worth taking. What if the project stops working after 10 years? Now all of a sudden the project economics are upside down.

There is no reason to sacrifice quality for price. Investors should first look for a proven, qualified product, which requires documented test results, and only then look at the price point.

SP: Do you have any recommended best practices for buyers for whom quality and reliability are a primary concern, especially low-volume purchasers who are unable to afford third-party due-diligence services?

RY: Consumers should ask basic questions: How long have you been in the industry? How long have you been manufacturing modules? How many recalls have you had? What is your warranty return rate? Most consumers do not need to have a thorough understanding of testing protocols. Instead, they should gather commonsense information. This is a very simple way of minimizing the risk of making a product selection mistake.

Conrad Burke

Global marketing director, DuPont Photovoltaic Solutions

SP: Do you think that the solar industry is “facing a crisis of quality”?

CB: I thought that Todd Woody wrote a pretty balanced piece. It was very thorough. He interviewed a good subset of constituents: developers, manufacturers, material suppliers and testing entities. What may not come across in the article is that module quality problems are a recent phenomenon. PV modules have performed extremely well for decades. It is really only in the last 2 years that we have started to see deviations that are entirely driven by market forces. It appears that some financially distressed companies are cutting corners. These are companies fighting for survival and engaging in a race to the bottom. Since this problem does not go back many years, I am confident that it will be corrected.

SP: Is there a need for internationally recognized standards for module reliability and quality assurance in manufacturing? If so, is DuPont engaged in these efforts?

CB: While everyone wants results now, I am not convinced that extended reliability tests are the best way to go about getting those results. These tests include things like thermal cycling, humidity freeze cycling, damp heat exposure, mechanical stress tests and so forth. However, failure modes are quite difficult for a scientist to understand and emulate. It is even more difficult to correlate the results of extended stress tests with module performance in the field in a way that is statistically valid.

The truth is that you cannot point to any series of tests that can predict the long-term reliability of a solar module, except maybe one. The testing that was done in the late 1970s by the JPL, which was contracted by the DOE [Department of Energy], is the most comprehensive work that has ever been done to qualify PV module materials and product reliability. While the IEC and UL have done some great work in the meanwhile, unfortunately the best predictor of long-term reliability is time.

There is really no statistically valid alternative to performing a lot of tests over a long period of time. That is why DuPont has its Fielded Module program. We are actually going out and buying modules that have been in the field for up to 30 years—often manufactured by companies that do not exist anymore—and testing them to characterize things like power degradation and material integrity. When you look at those data, modules that conform to the original JPL specifications perform very well, almost exclusively.

DuPont is also involved in setting global product standards and in ongoing efforts with NREL, DOE, SEIA and other trade organizations. We are also involved in new initiatives like the truSolar Working Group, which was established in January 2013 with the goal of developing solar project standards for quality and durability. The intent is to reduce risk and make it easier and less expensive to finance projects. DuPont is a founding member of truSolar, along with other industry leaders like SMA America, ABB and PanelClaw, as well as finance, insurance and ratings entities like Booz Allen Hamilton, Assurant and Standard & Poor’s; the DOE has even joined the consortium.

SP: Have you seen analogies to the recent panel quality problems in other industries? Is this a typical market response to falling prices?

CB: A bit of shortsightedness is out there that I, quite frankly, have not seen in other industries. I am speaking as someone who lives in Silicon Valley, who has seen boom-bust cycles before. Through the turmoil of the semiconductor, the computer or the telecommunications industries, I never saw successful companies cutting corners. I also never saw low-quality suppliers surviving and doing well.

There is a myth in the solar industry that warranties and insurance will cover all risks. That is not necessarily the case. DuPont has acquired some operating solar plants around the world that help run some of our own operations. We actually have a couple instances where modules have started to deteriorate prematurely. In one case, the solar module  manufacturer is already out of business—so end of discussion there. In the other case, the finger-pointing between the installer and the module manufacturer is never-ending—so we gave up trying to resolve the issue.

What DuPont is dealing with on a small scale is indicative of some of the industry problems. When projects are flipped, a bit of passing the buck or kicking the can down the road is going on. I do not want to overdramatize the problem, but as solar becomes an increasingly larger percentage of our national energy portfolio—solar power generation in the US is up to 10 GW right now—we can ill afford a black eye.

SP: Are buyers sophisticated enough to differentiate products according to quality and reliability? If so, are those buyers willing to pay more for higher-quality, more reliable products?

CB: DuPont is increasingly active in the downstream markets. We are proactively educating customers that every module is not the same, and you need to know what you are buying. Those efforts seem to be working. We are starting to see project developers, bankers and so forth ask more hard questions like: What materials are used in these modules? What evidence is there that those materials will last as long as you say they will?

Sophisticated customers understand that they are not buying kilowatts, but rather kilowatt-hours. To get 2 decades’ worth of kilowatt-hours, as assumed by the finance model, the modules have to last. We are slowly getting to a place where customers understand just how important this is. Modules make up an increasingly smaller percentage of overall system costs. A laser focus on up-front module costs simply does not make sense in today’s market. Return on investment will have more to do with how the system is performing after 10, 15, 20 or 30 years.

Jeff Wolfe

Principal, Jeff Wolfe Consulting

SP: You have been working in the PV industry since 1998, and over that 15-year period you have bought and sold a lot of PV modules. Does your experience suggest that the industry is “facing a quality crisis"?

JW: The PV industry is not facing a broad quality crisis. As the founder of groSolar, I purchased 100 MW of products from more than a dozen companies, from China, Europe, Korea, Taiwan and the US. I have visited a couple dozen PV manufacturing plants on three continents. I have seen very high-quality manufacturing and quality-control processes on every continent.

Over a 15-year period, we experienced a very small percentage of module failures, either at installation or later. The percentage of failures is well below 1%, perhaps as low as 0.1%, which is impressive considering that we handled close to half a million modules. Failed diodes account for the highest occurrence of module failures. Frame failures are probably the second most common failure mode in our experience, but the frames may have been loaded beyond their rating. In 15 years, only a few dozen modules have failed to output rated power, absent a diode failure.

When I refer to a module failure, I am referring to a module that is no longer producing power within the manufacturer’s specified power tolerance. This does not refer to modules that have developed an aesthetic issue, which is also rare, or to modules that perform below 100% of nameplate but are still above the warranty rating. Our primary concern is the electrical performance of the module, and this has not been a significant issue.

Historically, I have purchased modules from reputable manufacturers whose factories I have visited. It is possible that there is a “quality issue” from lesser manufacturers. There may be companies out there that cut corners or manufacture in a poor environment or do not understand the manufacturing processes well enough to control quality. As with any other product, it is important for customers to buy products made by a reputable manufacturer.

SP: Is there a need for internationally recognized standards for module reliability and quality assurance in manufacturing? If so, is the scope and speed of the industry’s response to this need adequate?

JW: Creating and implementing a worthwhile international reliability test standard for PV modules is an unnecessary exercise in bureaucracy. PV modules are exposed to an extremely wide set of conditions, from desert heat to arctic chill. Creating a single test regimen that would sufficiently test and qualify a product under all conditions would be an extremely lengthy and ultimately unproductive exercise, potentially providing a false sense of comfort for customers. As with any major long-term purchase, customers should understand what qualities they want and need, and then work with a qualified manufacturer or distributor to provide a suitably qualified product. This is how it works in other construction industries. The solar PV industry does not gain anything by being different.

SP: Is the market sending a mixed message to PV module manufacturers? Are customers willing to pay more for a higher-quality, more reliable product?

JW: There is a sector of the market that is sending a very clear message to the manufacturers that nothing matters except price. These buyers are getting what they want. It is not necessarily an unsound practice, as the industry has not seen wide-scale failures.

Another sector of the market is demanding and receiving products constructed with the most time-tested materials and processes to absolutely minimize the already low risk of failure. Most people consider this to be a sound practice.

Different quality levels are to be expected in any industry. This is part of the maturation of the industry into differentiable products. Of course, the cost difference between premium materials and less-tested materials is in the order of 1 to 3 cents per watt. Even in today’s environment, this number is low enough that it is lost in the noise relative to other pricing issues.

SP: Do you think that buyers are sophisticated enough that they can differentiate products according to quality and reliability?

JW: Many buyers are not sufficiently sophisticated to discern potential quality differences. Many potential buyers should not be building projects, as they have insufficient background in general engineering and construction technologies to properly specify, select and apply the technologies to create a system with optimal performance. The root of the problem is that too many buyers make an insufficient investment in high-quality engineering services; buyers need help determining proper specification and construction practices.

I qualify modules through an iterative series of discussions with the manufacturer and then factory visits. I have visited almost every factory I have ever purchased products from. I have also visited factories where I decided not to purchase modules, even from multinational companies. Depending upon the project and the need, I may then specify the key module components and acceptable manufacturers for the project, and require certifications for the bill of material and fabrication. Lastly, on-site nondestructive testing is available to ensure that what was specified and paid for was in fact provided.

SP: Do you have any recommended best practices for PV module buyers for whom quality and reliability are a primary concern, especially low-volume purchasers unable to afford third-party due-diligence services?

JW: I recommend that buyers make an investment in product knowledge. This can be done in many ways. I invested time and money visiting factories and working with material providers. Those who cannot visit factories, or do not want to spend that time and energy, should purchase modules from a trusted supplier or distributor, one that has made an investment in qualifying products and manufacturers. Customers should also understand that the distributor is there to meet their stated needs. If customers are saying that price is all that matters, then that is what will be provided.

Buyers who shop every deal for the lowest price are saving very little after their time and added risks are considered. They are also not gaining the benefits of having a standing relationship with their supplier. Forming a strong relationship with a distributor or a few manufacturers pays big dividends if there ever is a problem. Material specifications can be changed, on purpose or not, with or without notice. Keeping a key contact in the C-suite can help buyers mitigate any situation that may occur.

Hugh Kuhn

Principal, Solar Advisory Services

SP: Do you think that the solar industry is “facing a crisis of quality”?

HK: This industry is not unlike any other growing industry with hundreds of suppliers and evolving quality standards. Consider the computer memory or disk storage industries of 25 years ago. It takes time to separate the wheat from the chaff. The module firms en masse have done the industry a disservice by not aggressively challenging the notion that solar panels are commodities. Given the number of variables involved in materials selection and manufacturing processes, nothing could be further from the truth.

SP: From your point of view as a solar technical due-diligence specialist, is there a difference between quality assurance and product reliability? To what extent does a typical bankability analysis today take quality assurance and product reliability into account?

HK: I have found a large blurry space in the industry related to quality assurance, product reliability and the extent to which bankability analyses do or do not address either issue.

Quality assurance means that the manufacturer is doing what they say they are going to do on the manufacturing plant floor—take materials A, B and C and put them together in a certain way every time. This is ISO 9001 stuff. It has nothing to do with the quality of A, B or C, nor does it have much to do with the reliability of the end product—at least not long-term reliability—but rather it has to do with the ability of the plant to be very consistent.

Product reliability has to do with the actual decision to combine A, B and C together, understanding whether A and B are actually compatible with each other over time. For example, a backsheet might work fine with some formulations of EVA encapsulates, but not work well with others. Different backsheets differ in terms of compatibility. There are formulations of EVA that can handle UV over time, and formulations that brown in the first year. Inside every solar panel, a myriad of different component variables must work together under harsh environmental conditions—UV light, high temperatures, humidity, wide daily temperature swings, salt and farm chemicals in the air—and over a long time.

While building a PV system is not rocket science, the materials science that goes into building a PV module is pretty close to it. Unfortunately, I have yet to see a bankability study that discusses the pros and cons of the backsheet laminate used, the formulation of EVA, the quality of the solder flux used or the testing done to ensure that the resulting end product is reliable and durable. A module firm can have awesome quality assurance and build a product that passes low-bar certification tests every time, but the product reliability and durability over time could be pathetic.

SP: Is there a need for internationally recognized standards for module reliability and quality assurance in manufacturing? If so, is the scope and speed of the industry’s response to this need adequate?

HK: Yes, there is a need for an international module reliability and durability test standard. This is not to say that there should be a higher bar of certification, but rather that there should be a set of tests that provide trusted additional information about how a module will perform over time in this or that environment so that buyers can make their own decision on which they want and what they are willing to pay.

Starting in early 2010, when I was in a position of trying to make informed buying decisions for my company, I was frustrated by the lack of long-term performance data available. I was not buying watts, but rather watt-hours over a 20-year period. Yet module firms were selling their product on a dollar per watt basis. A few firms wanted to charge more because “their product was of a higher quality” and would thus supposedly perform much better in the later years of ownership. I suggested that if they could quantify this claim, I would be happy to consider paying more. It is a pretty easy net present value calculation to run the additional up-front cost against the value of the “added” electricity I would be selling. Sadly, none of the manufacturers could quantify the performance difference, though several tried.

As a result I pulled together a group of module firms, testing labs, folks from NREL and elsewhere, and we collaborated informally to develop what we called the “Thresher Test,” which is a series of accelerated chamber tests based on IEC 61215 but expanded in ways that we all felt would provide better long-term performance data for buyers. It was not a pass-fail test, but rather a means to gather and provide comparative data to prospective buyers so that informed decisions could be made. The Thresher Test is run in several testing labs today—with various modifications such as added UV preconditioning or damp heat under bias—as required by buyers who are trying to make the same decisions I was several years ago.

Today, NREL is leading the charge to evolve the initial Thresher Test effort into a true industry standards effort via the work of the International PV Module Quality Assurance Task Force headed up by Sarah Kurtz. International standards bodies work slowly. I wish they could work faster on this particular issue, as many gigawatts of projects are being installed each year. We can only hope the module buyers for those projects did their due diligence.

SP: Is the market sending a mixed message to PV module manufacturers? Are customers willing to pay more for a higher-quality module?

HK: Absolutely. Due to the immaturity of the industry, project financing bodies are still relying on reports from independent engineers—like Black & Veatch, DNV KEMA, Enertis Solar, SAIC—and limited bankability studies to approve module suppliers. Instead of asking “Is this product any good?” they are asking for a “bankable” warranty at the lowest cost. The problem is that you get what you pay for, and it is absurd to expect a 20-year warranty on the lowest-cost product. What the market should be saying is: “Prove to me that the product is well made—using quality materials, tested properly for long-term durability given the expected stresses of my project environment—and I will pay you properly for providing me with that confidence.”

SP: Do you have any recommended best practices for PV module buyers for whom quality and reliability are a primary concern, especially low-volume purchasers?

HK: Do not buy the lowest-cost module just because it has a warranty. Buy modules from firms that have been in business for many years, with an installation track record and a track record of customer support that includes standing behind their warranty. Consider using a microinverter or smart module architecture. That way one failure will not mess up your entire array; you can mix and match modules in the future, if need be. Buy a few spares if you are able to.


David Brearley / SolarPro magazine / Ashland, OR /

Specification Aggregation

Doug Puffer / SolarPro magazine / Ashland, OR /

Joe Schwartz / SolarPro magazine / Ashland, OR /

PV Manufacturer Contact

1SolTech / 888.598.0295 /

aleo solar / 303.325.3917 /

Antaris Solar / 855.484.3041 /

Astronergy / 650.392.2777 /

Auxin Solar / 408.868.4380 /

AXITEC / 856.701.2187 /

BenQ Solar /

Canadian Solar / 888.998.7739 /

Centrosolar America / 877.348.2555 /

CNPV Solar Power / 212.359.0205 /

Colored Solar / 877.526.2462 /

Conergy / 888.396.6611 /

Eclipsall Energy / 416.716.3390 /

ecoSolargy / 877.808.4213 /

Eoplly New Energy / 650.225.9400 /

ET Solar / 925.460.9898 /

Grape Solar / 541.349.9000 /

GreenBrilliance / 707.657.0090 /

Hanwha SolarOne / 714.689.6868 /

Helios Solar Works / 877.443.5467 /

Hyundai Solar /

Isofoton / 419.591.4360 /

Itek Energy / 360.647.9531 /

JA Solar / 408.586.0000 /

Jinko Solar / 415.402.0502 /

Kyocera Solar / 800.223.9580 /

LDK Solar / 408.245.0858 /

LG Electronics / 855.854.7652 /

Lightway Solar / 732.602.1930 /

Lumos / 877.301.3582 /

MAGE Solar / 877.311.6243 /

Mitsubishi Electric / 714.220.2500 /

Motech Americas / 302.451.7500 /

NESL USA / 800.242.7114 /

Panasonic / 408.861.8424 /

Perlight Solar / 424.242.8068 /

Phono Solar / 855.408.9528 /

REC Group / 877.332.4087 /

ReneSola / 415.852.7418 /

Renogy / 909.517.3598 /

RITEK Solar / 800.823.2505 /

Sharp / 800.765.2706 /

Silevo / 510.771.1360 /

Silfab / 905.255.2501 /

Silicon Energy / 360.618.6500 /

Solaria / 510.270.2500 /

SolarWorld / 805.388.6590 /

SOLON / 520.807.1300 /

Sonali Solar / 201.297.1177 /

SunEdison / 888.786.3347 /

Suniva / 404.477.2700 /

Sunowe / 408.785.0836 /

SunPerfect Solar / 408.273.4534 /

SunPower / 800.786.7693 /

Sunpreme / 408.245.1112 /

Suntech / 866.966.6555 /

Talesun Solar / 800.276.0823 /

Trina Solar / 800.696.7114 /

Upsolar / 415.263.9920 /

WINAICO / 646.520.7673 /

Yingli Green Energy / 888.686.8820 /

ZNSHINE Solar / 606.364.8052 /

Primary Category: 

The NEC recognizes a variety of conduits and tubing that hold wires and cables. How do you select the right raceway type for a particular application, and how do you design and install it efficiently, in a Code-compliant and long-lasting manner?

PV systems, like other electrical power systems, use electrical conductors to route power from sources to loads. It is often necessary to enclose these conductors and protect them along their path. The 2011 NEC lists several methods of getting conductors from here to there, including about 20 different types of raceways and almost as many types of cable, a multiconductor assembly with a covering or jacket.

In this article I focus on raceways commonly used for PV systems. In Part One, to aid in specification and selection, I describe various types of raceways and discuss the differences among them. In the next issue, Part Two will present practical design considerations regarding specific applications of given raceway types and cover installation techniques.

Definitions and Jargon

Unfortunately, the topic of raceways is an area full of trade shorthand and jargon. I can clear up some of this terminology.

Solid or stranded electrical wires are generally known as conductors. Individual conductors can appear by themselves, as in USE-2 underneath a PV array, or several can be assembled in the factory into a cable or installed together in the field into a raceway.

Conduits, tubing and even square wireways are all types of raceways, according to the NEC. However, the NEC does not consider auxiliary gutters raceways, although they are similar to square wireways. While Code does not define the terms conduit and tubing, it does define the term raceway and includes the types of conduit and tubing discussed in this article. For clarity, I use the term raceway as defined in Article 100 to generally indicate “an enclosed channel of metal or nonmetallic materials designed expressly for holding wires, cables, or busbars.”

The trade often refers to particular raceway types with shorthand. For example, among the common flexible raceways, flex usually refers to type FMC (flexible metal conduit; see NEC Article 348), liquidtight usually refers to type LFMC (liquidtight flexible metal conduit; see NEC Article 350) and Sealtite usually refers to type LFNC (liquidtight flexible nonmetallic conduit; see NEC Article 356).

The real confusion happens with the nonflexible circular raceways: These common raceways are rigid (not flexible), but the term rigid often refers to type RMC (rigid metal conduit; see NEC Article 344). Thinwall refers to type EMT (electrical metallic tubing: see NEC Article 358), which in turn is slightly confusing because it is called tubing instead of conduit. Fortunately, both tubing and conduit are types of raceway. Schedule 40 and Schedule 80 usually refer to PVC (polyvinyl chloride) conduits, as per NEC Article 352. Steel or galvanized are ambiguous terms because either could refer to RMC, IMC (intermediate metal conduit; see NEC Article 342) or EMT. The term rigid is even more ambiguous because it could include both metallic and nonmetallic raceway materials, such as PVC. Technically, RMC can also be made from stainless steel, red brass or aluminum, but the vast majority of RMC is galvanized steel.

This article—and hopefully PV project specifications and plan sets—refers to the various raceways by their NEC designators. If you are bidding on a project with drawings or specifications calling for an ambiguous raceway type such as “rigid,” make sure to get formal, written clarification through the Request for Information (RFI) process to determine the intent of the specification. If you assume one type of raceway and the intent was for the other, you may substantially over- or underbid.

Raceway Options

PV systems are generally similar to other electrical power systems, but they do have certain features that affect raceway selection and installation. For example, the bulk of a typical PV system is installed in an exterior location with high exposure to sunlight and other environmental phenomena.

Different raceway types have different attributes, and it is incumbent on the system designer to specify the product most appropriate for a given project and application. In general, the three most common nonflexible raceway types are EMT, RMC and PVC.

EMT. EMT is a thin-walled, rigid raceway. It is almost always made of steel. Corrosion resistance comes from electroplated zinc on the outside and an organic coating on the inside. The pipe is not threaded, so exterior fittings are compression type and interior fittings are setscrew type. Installers can readily bend EMT up to trade size 1¼ inch in the field with hand benders. EMT is lightweight and relatively inexpensive, and offers substantial protection from mechanical damage. Therefore it is a very common raceway for both PV systems and general electrical applications.

RMC. RMC is a thick-walled, rigid raceway almost always made of galvanized steel. The manufacturer threads the ends of each length, and typical assembly involves threaded couplings. RMC is heavier than EMT and more difficult to bend. When cut, RMC requires field thread cutting. While it provides a high level of protection against mechanical damage, it is significantly more time-consuming to install than EMT. It is often used in locations where mechanical protection is required, such as shallow trenches, aboveground areas in close proximity to vehicular traffic, utility service entrances and high-traffic roof areas.

RMC’s almost identical cousin, IMC, is interchangeable in application with RMC. It has a slightly thinner wall than RMC and is always made of galvanized steel. The steel used for IMC is stronger than that used for RMC, so despite the thinner wall, it provides equivalent overall protection. IMC is about two-thirds the weight of RMC and has a slightly larger cross-sectional area available for conductor fill. Many in the field report that IMC is more difficult to bend and to thread properly than RMC, as well as less readily available in most areas.

PVC. PVC is a polymeric (plastic) rigid raceway. It typically comes in two thicknesses: Schedule 40 (standard) and Schedule 80 (thicker wall). The material has a high inherent corrosion resistance. Lengths of PVC raceway are coupled using PVC cement. PVC can be field-bent by careful heating or assembled with prebent elbows. PVC raceway is low cost and lightweight, and can be installed quickly. It is most often used underground, where it is not subject to impact damage or sun exposure.

Impacts on Raceway Selection

Choosing the most appropriate raceway for a particular application depends on a variety of considerations, including cost, durability, mechanical characteristics, thermal expansion, bonding requirements, voltage losses and embodied energy.


The material cost per foot for RMC is about twice that for EMT. RMC has a wall thickness about 2.5 times that of EMT, so a 10-foot stick of 1-inch RMC weighs 16 pounds, versus a 10-foot stick of EMT at 6.5 pounds. RMC costs more to ship and handle. Further, RMC needs to be threaded on both ends—an additional cost. PVC material costs are similar to those of EMT.

The labor involved in installing EMT and PVC is similar, but using RMC is more time-consuming because bending is more difficult and cut ends require thread cutting in the field. Where installed aboveground, PVC requires two to three times as many supports as EMT or RMC, which adds cost for materials and labor to install the supports. For a rooftop installation, the pier supports can be quite expensive, so the increased quantity of supports required for PVC may add significant cost.

In the end, galvanized-steel RMC costs about twice as much, in terms of both labor and materials, as galvanized-steel EMT and PVC. For project-specific analysis, RSMeans publishes annually updated construction cost data based on national averages, which is useful for estimating material and associated installation costs. RSMeans’ Electrical Cost Data ebook includes a comprehensive list of electrical systems and equipment, as well as a brief section on PV applications (see Resources).


Several factors determine a raceway’s durability. For high-quality PV systems, the common functional requirement is that the raceway system protect conductors for the system’s lifetime. This requirement means that the raceway system itself should remain in good condition, without loss of key properties, for the average 25-year life of a PV system. Here are a few considerations to keep in mind for a long-lasting raceway.

UV resistance. UV exposure does not affect the material properties of metal raceways, but you need to evaluate all polymeric raceways—including PVC, LFMC and LFNC—for their resistance to UV degradation. The PVC used for electrical raceways includes additives intended to make the finished product relatively stable in sunlight. In practice, PVC raceway exposed to UV undergoes a reaction, often referred to as browning, in which the outer layer weakens and discolors. According to tests by PVC manufacturer organizations, this does not affect the tensile strength and modulus of elasticity, but it reduces the impact strength. One study by the Uni-Bell PVC Pipe Association, The Effects of Ultraviolet Radiation on PVC Pipe (see Resources), shows a continuous decline in impact strength over a 2-year test period. The study states that it is intended to examine the UV exposure of pipes stored temporarily (up to 2 years) in the sun but eventually installed underground. Clearly, this intended use is quite different from permanently installing PVC on a rooftop, where raceways may be exposed to UV radiation for 25 years or longer.

The relevant UL standard for rigid PVC electrical raceway, UL 651, and the UL standard for polymeric materials used in electrical equipment, UL 746C, describe Weather-Ometer testing that includes only 720 hours of accelerated UV testing. This is roughly equivalent to 2 years of full sunlight exposure, depending on location. There is some interest among experts concerned about long-term outdoor performance of nonmetallic raceways in extending this testing for longer periods, which would make it more pertinent to PV rooftop installations.

You can limit UV exposure by painting polymeric conduits a light color with water-based latex paint appropriate for exterior use. You must thoroughly clean the conduit before applying the paint. Painting is a good solution for short lengths of exposed PVC, LFMC or LFNC, such as when you use PVC underground with short stubs emerging into sunlight before terminating in the bottom of an enclosure, or when you use a short section of LFMC or LFNC as an expansion joint on a rooftop conduit run.

Chemical resistance. Whereas PVC suffers degradation when exposed to UV, it is quite resistant to corrosion from other environmental forces. Steel, however, is susceptible to oxidation in the presence of water, which it is commonly exposed to on rooftops, underground and in other exterior locations. Salt water, commonly found in the air near coasts, is even more corrosive.

Fortunately, steel raceways are commonly manufactured with protection against corrosion, most often in the form of galvanizing. Both EMT and RMC are made of galvanized steel, which provides enough corrosion protection for all but the harshest environments. In highly corrosive marine areas, however, electroplating or even hot-dip galvanizing may not be enough long-term protection for steel raceway. In those locations, you should specify RMC in stainless steel, which is very expensive, or more commonly RMC with a PVC coating. However, assuming UV exposure in addition to salt-water exposure, the PVC coating will degrade and eventually allow salt water to access the steel.

Another corrosion-resistant option is aluminum RMC. Aluminum oxidizes, but in contrast to steel oxidation (rust)—which tends to weaken the metal and flake off, exposing more steel to further oxidation—aluminum oxide forms a strong, protective layer for the underlying aluminum. In a corrosive environment, stainless steel or aluminum fittings should be used with any aluminum raceway.

Chemical resistance can also be a factor where raceways are installed in direct contact with earth or encased in concrete slabs or piers. In these environments, PVC provides the best corrosion resistance. The NEC permits you to install IMC and RMC in potentially corrosive environments where the raceway is “protected by corrosion protection and judged suitable for the condition”; see NEC Sections 342.10(B) and 344.10(B)(1). EMT is granted the same permissions, yet the corrosion protection must be “approved as suitable for the condition”; see NEC Section 358.10(B). Corrosion protection tape is often used on IMC, RMC and EMT in corrosive environments and is generally a good practice on sweeps where the raceways emerge aboveground. PVC-coated RMC is another alternative where additional corrosion protection is necessary.

Fire resistance. Steel survives high temperatures longer than PVC. I hope that your installations never experience an external fire or an internal arcing fault in the raceway—but if any of them do, you can breathe easier if you have specified steel raceway.

Mechanical Characteristics

In general, stronger raceway materials and thicker raceway walls provide better mechanical protection for conductors. Thus, all else being equal, steel is stronger than aluminum, and aluminum is stronger than PVC. Thick-walled steel RMC and IMC are stronger than thin-walled EMT. In underground installations where the raceway is subject to physical damage, NEC Section 300.5(D)(4) explicitly allows RMC, IMC and Schedule 80 PVC. It does not allow EMT and flexible conduits.

Interestingly, NEC Section 358.12(1) disallows EMT where “subject to severe physical damage,” (emphasis added) whereas Section 352.12(C) disallows PVC “where subject to physical damage unless identified for such use.” The UL White Book makes it clear that Schedule 80 PVC is identified for use where subject to physical damage, but Schedule 40 PVC is not. However, the Code leaves “subject to physical damage” and “severe physical damage” up to interpretation. Commonly, areas above ground are interpreted as subject to physical damage if there is any chance of vehicular presence, such as in driveways, parking lots, warehouses with forklifts and so on.

Many AHJs require at least Schedule 80 PVC where the raceway emerges from underground, whether or not there is  exposure to vehicles, and further require RMC when there are vehicles present. For PV systems, another common location potentially subject to physical damage is the rooftop. During installation and maintenance of rooftop equipment, including both the PV system itself and HVAC or window-washing equipment, there may be significant foot traffic on the roof. Raceways installed a few inches above the surface tend to get kicked and stepped on with some regularity, especially where the raceways cross walkways. These areas may require more robust raceways or physical protection such as ramps or steps.

In addition to losing impact strength with increased UV exposure, PVC becomes brittle in cold temperatures, making it subject to damage. At the other temperature extreme, both NEC Section 352.12(D) and UL 651 Section 1.2.3(a) limit the use of PVC to ambient temperatures up to 122°F (50°C) only. According to Thermal Expansion and Contraction in Plastics Piping Systems PPI TR-21/2001, published by the Plastics Pipe Institute (see Resources), even at 100°F (37.78°C), PVC loses 40% of its stiffness, or elastic modulus. Further, according to the same publication, PVC’s effective stiffness decreases significantly with increased duration of loading, so a PVC raceway tends to sag more over time. Metal raceways do not have any practical temperature limits, either hot or cold, and do not exhibit effective loss of stiffness under extended loading.

One common loading application for raceways is a horizontal run supported at certain intervals. Stiffer raceways show less deflection for a given span between supports, and stronger raceways allow longer spans without exceeding the yield stress or failing. The Code takes raceway stiffness into account when dictating minimum support spacing. PVC must be supported every 3 feet for sizes up to 1 inch, every 5 feet up to 2 inches, and a maximum of every 8 feet for 6 inches. In contrast, EMT of any size requires supports every 10 feet, and RMC can span 10–20 feet between supports, depending on its size.

Buried raceways are not susceptible to vehicle impact, and they are effectively supported along their length, but stronger raceways better survive localized compaction from heavy equipment, soil shifting and settling, and motion resulting from seismic activity.

Fittings and assembly quality. In addition to the material properties and geometry of the raceways themselves, the difference in fittings used to connect them creates significant variances in mechanical durability among assembled raceway systems. Typically, RMC is assembled with threaded fittings, EMT is assembled with either compression fittings or setscrew fittings, and PVC is solvent welded. All of these methods are long lasting and reliable, if you assemble them with excellent workmanship. However, some are more susceptible to field failures than others.

Solvent-welding PVC sections can be problematic in several ways. Bell-end raceway sections, intended to save labor and materials by eliminating a separate coupling piece, often do not have the tight tolerance required for a perfect joint. Even with good materials, the quality of a joint is highly dependent on the skill and attention of the installer. The mating surfaces must be deburred, cleaned and primed properly (if required), and the appropriate amount of the correct solvent cement must be applied evenly. Then the joint must be pressed together, rotated slightly and held for at least 30 seconds to prevent it from pushing itself apart as it cures. If the installer improperly executes—or ignores—any of these steps, that compromises the joint. Unfortunately, even close inspection does not reveal shortcomings until a few thermal cycles or a moderate amount of stress (including wire-pulling forces) cause the joint to fail, leaving the conductors susceptible to damage. Another minor downside to these welded joints is that they are not reversible. If a joint needs adjustment after welding, you need to cut out and replace the raceway section with the joint.

With exterior-rooftop EMT raceway used in PV systems, installers join sections of EMT using compression couplings. These fittings also rely on good workmanship and are difficult to inspect at a glance. To make a good joint, the tubing must be cut square, deburred and inserted cleanly through all of the rain-tight gaskets and compression rings. The fitting must be a compression fitting listed for use with EMT. Slightly larger compression fittings listed for RMC do not work. The installer must make the fitting wrench tight while ensuring that the tubing remains bottomed out in the fitting and, if the tubing has bends in it, that it does not rotate as the fitting nut is tightened. Again, if the installer forgets any of these steps or does them improperly, the joint may appear acceptable at a glance, but sharp exposed burrs or edges can damage conductor insulation during the wire pull or the fittings may pull apart when stressed. Setscrew fittings used for indoor applications of EMT have similar requirements and shortcomings.

The combination of the need for excellent workmanship with the difficulty of accomplishing a thorough inspection of an installed raceway system gives John Wiles, Code guru and senior research engineer at the Southwest Technology Development Institute at New Mexico State University, cause for concern: “The only time that we get reports of a conduit problem is when there has been a failure leading to a fire or other issue. As PV systems age, issues with improper installation of metallic conduits increase.”

The threaded couplings used to join sections of RMC and IMC are the strongest and most reliable of the various connection methods. Cut sections require more labor and tools to field-cut threads—again, after cutting the pipe square and deburring it. As long as the pipe ends have threads and you have selected the correct fittings, the mechanical integrity of the joint is all but assured once you have engaged enough threads. When you have made the joint wrench tight, only a thread or two is visible. These joints are easy to inspect visually, and they are tremendously strong.

Thermal Expansion

Most of us who have spent time on rooftops have often seen long runs of wavy raceway and broken sections. Wavy or buckled pipe does not usually result from the failure to set a straight string line when the pipe was installed, but rather is a clear manifestation of thermal expansion and contraction. All raceway materials are subject to thermal expansion, but some much more so than others.

Two results are common when high–thermal-expansion raceways are installed without properly designed and installed expansion fittings: First, when the raceway expands against fixed supports or terminations, it causes high stress on the raceway and fittings, which can cause buckling and cracking. Second, when the raceway cools and contracts, the resulting tensile stress can pull fittings apart. In either case, the raceway has failed and exposed the conductors it carries to damage.

The linear coefficient of thermal expansion is a material property. Basically, when the atoms in a material get warmer, they vibrate more and take up more space, thereby causing expansion of the bulk material. The coefficient for a given material is expressed in fractional change in length (strain) per degree of temperature change of the material. Common building materials expand on the order of 10 units of length per million units of length for every 1°F of temperature change, or 10 × 10-6/°F. A convenient way to express this change is in inches of expansion per 100 feet of length. For a 100°F temperature change—commonly seen on roofs—thermal expansion of common materials is as follows:

  • Steel 0.8 inches
  • Concrete 0.8 inches
  • Wood (parallel to grain) 0.3 inches
  • Aluminum 1.6 inches
  • PVC 4.1 inches

Calculating raceway thermal expansion. NEC Article 300, “Wiring Methods,” has a section that applies to all raceways called “Raceways Exposed to Different Temperatures.” Section 300.7(B) requires expansion fittings “where necessary to compensate for thermal expansion and contraction.” The raceway-specific articles come later in Chapter 3. Article 352 governs PVC raceways and has specific requirements for mitigating thermal expansion because PVC has such a high coefficient and causes a bigger problem if thermal expansion is ignored. Section 352.44 requires expansion fittings for PVC raceway when the expected length change due to thermal expansion is ¼ inch or greater. Table 352.44 lists the expansion coefficient for PVC and shows calculated length-change values for different assumed temperature changes.

For steel and aluminum raceways, it is up to the designer and the AHJ to interpret the definition of the “where necessary” language in NEC Section 300.7(B), although Code does provide an informational note after Section 300.7(B) that suggests using rounded multipliers for steel (0.2) and aluminum (0.4) based on the PVC chart in Section 352.44, because the coefficient of thermal expansion for PVC is approximately 5 times that of steel and 2.5 times that of aluminum. This is one of the primary reasons why PVC raceway is typically not specified for rooftop use in PV systems.

Among common raceway materials, steel has the smallest coefficient of thermal expansion, which makes it most similar to the steel, concrete or wood buildings it is fastened to.  Specifically, for rooftop raceway applications—which tend to be exposed to severe temperature changes—you can compare the raceway linear coefficient of thermal expansion to that of the roof framing. In most buildings, the roof framing, which is often insulated from the exterior or moderated by conditioned space below, does not experience as large a temperature swing as does a raceway installed a few inches above the roof surface. In addition, common roof framing materials have a relatively low linear coefficient of thermal expansion, meaning with less temperature change they tend to expand and contract less than rooftop raceways.

The equation used to calculate thermal expansion and contraction is:

delta L = alpha × delta T × L

where delta L is the length change of the raceway, alpha is the coefficient of thermal expansion of the material, delta T is the temperature change and L is the initial raceway length. This is always an approximation because alpha typically changes slightly with temperature. The NEC provides coefficients of thermal expansion for PVC, steel and aluminum raceways (see Table 1).

Consider a raceway running 200 feet along a parapet wall from a combiner box to a pull box (see Diagram 1). The location’s ambient temperature ranges from 10°F to 90°F. Assuming that the building has no expansion joints between the boxes, the building itself (and therefore the distance between the boxes) will expand by as much as 1 inch. Depending on the framing materials and the level of insulation and interior temperature conditioning, which reduces the temperature change to which the framing is exposed, the building may expand as little as ¼ inch. Since the raceway is a few inches above the roof surface and is exposed to direct sun, it can reach 120°F, for a total temperature change of 110°F (120°F − 10°F). For a steel raceway, the length change of the raceway would be 0.143 feet (0.650 × 10-5 × 110°F × 200'), or 1.72 inches, which is about ¾ inch to 1½ inch more expansion than the building itself will experience.

For an aluminum raceway, the length change would be 0.286 feet (1.3 × 10-5 × 110°F × 200'), or 3.43 inches. Finally, for a PVC raceway, the length change would be 0.744 feet (3.38 × 10-5 × 110°F × 200'), or 8.92 inches. All three raceway types would require at least one expansion fitting—and PVC would require at least two-—as well as associated sliding supports and guides to allow the expansion fittings to operate properly.

When burying PVC pipe, the temperature change underground is generally assumed to be relatively small. However, if you assemble the PVC raceway and solvent-weld it out of the trench, in the sun, and then drop it into the trench and immediately bury it, it may develop high tensile forces when it cools to below-grade temperature after you have fixed it in place. These tensile forces may cause the weakest joint to pull apart underground, resulting in a failed raceway system. You can easily avoid this problem by allowing the assembled raceway to reach the cooler underground temperature before backfilling and compacting.

Bonding Requirements

Another difference between raceway systems relates to grounding and bonding. Polymeric raceways are not conductive, so they cannot be used as the primary or supplemental equipment-grounding conductor. In addition, they are nonferrous, so they have no effect on the impedance of fault-clearing circuits or dissipation of high-current surges to the grounding-electrode system. In contrast, steel raceways are conductive, which has two implications. First, it means they need bonding to prevent unintentional energizing of the raceway and subsequent shock hazard. For most PV dc circuits, as well as 480 Vac and above, installers must pay special attention to bonding any metal raceway because NEC Section 250.97 essentially requires the use of grounding bushings (with some exceptions) for all circuits above 250 V to ground. Second, the fact that steel raceways are conductive also means that they can provide either the primary or supplemental ground-fault current path to help clear faults by opening overcurrent protection devices.

You typically cannot use flexible steel raceways, such as FMC and LFMC, as the primary equipment-grounding conductor (EGC), except for very short lengths, whereas rigid steel raceways are permitted as the sole equipment-grounding conductor; see NEC Sections 250.118(6) and (7). However, best practice is to include an insulated EGC along with the current-carrying conductors inside the raceway so that the fault-clearing function of the EGC is not dependent on every fitting in the raceway system. Especially on a long run across a roof, where the raceway may be subject to all manner of challenges to its integrity, it is critical to install the nonraceway EGC inside the raceway. The addition of this “above Code” conductor also reduces the impedance of a potential ground-fault circuit and gives PV ground-fault detection devices a better chance of doing their job and preventing fires.

By definition, steel raceways are ferrous, which means they have magnetic interaction with both intentional and unintentional current flowing in the conductors inside the raceways. Among other considerations, this means that if any grounding electrode conductors (GECs) used in a PV system, including dc GECs, are contained in steel raceways, then NEC Section 250.64(E) requires bonding of both ends of the steel raceway to the GEC to prevent the raceway from restricting the flow of current from a lightning strike—or any high-current surge—to the grounding electrode and into the earth. Aluminum raceways have the benefit of being conductive without the drawbacks of magnetic interaction with fault currents. Therefore aluminum or PVC raceways are recommended for protecting GECs. See NEC Article 250 and especially the Soares Book on Grounding and Bonding, by the International Association of Electrical Inspectors, for more details on this topic (see Resources).

Voltage Losses

Choice of raceway material can affect the amount of voltage drop on ac inverter output circuits, especially with larger circuits (300-kcmil and above) and in newer systems with grid-stabilizing inverter features such as power factor correction. See “Voltage Drop in PV Systems,” SolarPro magazine, February/March 2010, for more information on voltage drop in general, and for a comparison of the effect on voltage drop of PVC, aluminum and steel raceways.

Embodied Energy

For certain projects, factors such as embodied energy and recyclability are important design criteria. In those cases, lightweight EMT steel raceway may be a better choice than aluminum or PVC. Steel has about ¼ the embodied energy of PVC on a per-kg basis, and about 1/8 that of aluminum. Typically, for a given size of raceway, Schedule 40 PVC is about 1/2 the weight of EMT, or about 1/5 the weight of steel RMC. Steel and aluminum are readily recyclable, whereas only a tiny fraction of PVC gets recycled.


Once you have assessed the benefits and limitations of raceway options for PV installations, raceway selection can proceed. Making the selection is quite valuable for bidding and estimating projects, yet the raceway system is only as good as the installation. Poor installation techniques can result in raceway issues that may damage conductors during the wire pull or raceway failures. Proper installation of the raceway system and the conductors it holds requires design, planning, preparation and execution strategies that I will expand upon in Part Two of this article in the next issue.


Blake Gleason / Sun Light & Power / Berkeley, CA /


Electrical Cost Data, RSMeans, 2013,

The Effects of Ultraviolet Radiation on PVC Pipe, Uni-Bell PVC Pipe Association,

Thermal Expansion and Contraction in Plastics Piping Systems PPI TR-21/2001, Plastics Pipe Institute,

Soares Book on Grounding and Bonding,  International Association of Electrical Inspectors, 2011,

Primary Category: 

The past year has been a tumultuous one for PV manufacturers, marked by notable events such as the escalation of the trade case against China and the exit of several manufacturers from the market, including SCHOTT Solar. Market analysts are projecting that global module supply will exceed global demand by 100% in 2012, and all indications point toward a significant consolidation of the PV manufacturing industry over the next 2–3 years. While the PV manufacturer landscape will likely change, perhaps dramatically, in the next few years, integrators are currently faced with more module options than ever before.

As part of our research to determine which manufacturers and models met the criteria for inclusion in our 2012 c-Si module specifications dataset, we reached out to all manufacturers with products listed to the UL 1703 standard that were also eligible for the California Solar Initiative (CSI) program. We consider the latter requirement to be a good indication of a manufacturer’s commitment to maintaining or establishing a presence in the North American market. We used the information we collected, coupled with information published on the various manufacturers’ websites, to develop profiles for each manufacturer that responded to our queries. We include background details for the companies and their products, as well as recent technological or business developments. For integrators interested in sourcing domestically made products, we present ARRA and Buy American Act compliance details for manufacturers that are producing products in the US. Products that have not been tested to IEC standard 61215 are noted.

The manufacturer profiles presented in this article compliment our 2012 c-Si module specifications table, which contains more than 900 c-Si PV models currently available in the US. Products included in the dataset have rated outputs of 200 W or higher and were listed as eligible for the CSI program per SB1 Guidelines as of June 15, 2012. An additional requirement is an established physical presence in the US that includes sales offices and product warehousing. Due to the large number of manufacturers and module models that met these criteria, the version of the dataset published in this article is abbreviated. The full dataset is available in Microsoft Excel format by clicking HERE and includes additional specifications, such as temperature coefficients, connector type, series fuse rating, frame and backsheet details, and materials and performance warranty information. SolarPro permits and encourages the integration of the Excel-based data with databases your company has developed for internal product tracking and system design activities.


Advanced Solar Photonics. Advanced Solar Photonics (ASP) is a US-based manufacturer of poly and mono c-Si modules. The company is a wholly owned subsidiary of BlueChip Energy. Headquartered near Orlando, Florida, in Lake Mary, the privately held firm commenced manufacturing and product sales in the US market in 2009. ASP purchases cells and assembles modules ranging from 175 W to 410 W. According to the manufacturer, the company has a production capacity of 100 MW and is currently the second-largest producer of c-Si modules in the US. ASP products are Buy American Act compliant for government, military and municipal projects. ASP recently launched a new series of bifacial doubleglass modules at Intersolar North America. The new module line was not listed as eligible for the CSI program at the time of publication.

Astronergy. Astronergy is a subsidiary of the Chint Group, a Chinese low-voltage electronics supplier headquartered in Hangzhou, China. Astronergy began manufacturing modules in 2006 and introduced product to the US market in 2008. The company reports a global PV manufacturing capacity of 700 MW. Astronergy’s US sales office is located in South San Francisco. Warehousing locations include South San Francisco and Hayward, California, as well as Edison, New Jersey, and Wilmington, North Carolina. Astronergy offers poly and mono c-Si products with power outputs up to 310 W. A separate subsidiary of the Chint Group, Chint Power Systems, has been increasingly active in the North American string and central inverter market in 2012.

AU Optronics. Formed in 2001, AU Optronics (AUO) is currently one of the largest manufacturers of liquid-crystal displays worldwide. The publicly traded Taiwanese company extended its market into the solar industry in 2008, and is currently active in polysilicon, ingot, wafer, cell and module processing and manufacturing. AUO operates three PV manufacturing facilities located in Taiwan, China and the Czech Republic. In 2012 AUO established a new division and solar brand, BenQ Solar. The division is active in residential, commercial and utility-scale project development on an international basis. BenQ’s US headquarters and sales office is located in Milpitas, California. AUO and BenQ Solar recently partnered with microinverter manufacturer SolarBridge to launch a listed ac module, the AC Unison.

Canadian Solar. Founded in Ontario, Canada, in 2001, Canadian Solar is a vertically integrated provider of PV ingots, wafers, cells and modules. The publicly traded corporation was listed on the NASDAQ Exchange in 2006. The company’s US sales office is located in San Ramon, California, with warehousing locations in Pittsburg, California, and Cranford, New Jersey. The firm established PV manufacturing facilities in 2001 and introduced product to the US market in 2009. Canadian Solar modules carry a noncancelable product warranty backed by A.M. Best–rated insurance companies, and the company has taken steps to differentiate its products, including its NewEdge module line, which allows full integration with the railless mounting and auto-grounding solutions offered by Zep Solar.

Centrosolar America. Centrosolar America’s parent company, Centrosolar Group AG, is a publicly traded PV manufacturing firm headquartered in Munich, Germany, with a global PV manufacturing capacity of 350 MW. The firm was formed in 2005 and began offering products in the US market in 2008. Centrosolar’s B Series, C Series and E Series modules are engineered in Germany and manufactured in the US (through contract partners), Germany and Asia, respectively. Centrosolar America operates sales and warehousing facilities in Scottsdale, Arizona; Fremont, California; and Edison, New Jersey. CentroClub, Centrosolar America’s channel program for solar installers, offers sales and technical training, financing products that include a residential lease, and engineering services. The company also offers pre-engineered module and BOS equipment packages.

CNPV Solar Power. Headquartered in Strassen, Luxembourg, with vertically integrated ingot, wafer, cell and module manufacturing in Shandong Province, China, CNPV Solar Power was founded in 2006 as the CNPV Dongying Photovoltaic Power Company. CNPV Solar Power entered the North American market in 2010. The company reports a manufacturing capacity of 600 MW and operates a US sales office in Boulder, Colorado, with warehousing in Denver. To mitigate the financial impact of the current tariffs on Chinesemanufactured modules, CNPV recently announced that it plans to utilize third-country, non–tariff-exposed cells for modules destined for the US. CNPV Solar Power’s modules are backed by third-party insurance.

Colored Solar. Colored Solar is a private PV start-up based in Ventura, California. Colored Solar’s parent company, Gold Coast Solar, is hoping to capture a specific segment of the residential and small commercial market by offering colored modules with improved aesthetics. In 2012, the company launched a line of ETL-listed and CSI-eligible modules available in several colors ranging from forest green to tile red. Colored Solar reports module efficiencies ranging from 13% to 15%. The company conducts R&D and module assembly in the US, and its products are Buy American Act compliant.

Conergy. Conergy AG, a German module manufacturer, equipment wholesaler and project developer, established Conergy North America in 2005. While the company owns and operates module manufacturing facilities in Germany, the PH and PM Series module lines offered by its North American subsidiary are built to spec by non–company-owned first-tier module manufacturers outside the US. Warehousing locations include Sacramento, California, and Carlstadt, New Jersey. The company’s US sales office is headquartered in Denver. In addition to its built-to-spec modules, Conergy North America offers product distribution and financial services, as well as racking solutions manufactured by Conergy in the US.

ecoSolargy. Headquartered in Irvine, California, ecoSolargy is a privately owned module manufacturer and integration firm. The company does not manufacture modules in the US but is American owned. ecoSolargy’s mono and poly c-Si modules were commercialized in 1997 and introduced to the US market in 2009. Product is warehoused in Irvine, California. ecoSolargy offers lead referral and system design services, as well as system BOS equipment through partnerships that include Advanced Energy, KACO new energy and DPW Solar.

ET Solar. ET Solar is the US subsidiary of the privately held ET Solar Group headquartered in Nanjing, China. The company was founded in 2005 and introduced product to the US market in 2007. ET Solar has R&D teams located in Germany and China. In addition to its poly c-Si module line, ET Solar offers EPC services and provides turnkey systems that include inverters and racking for residential through utility-scale applications. ET Solar operates a US sales office in Pleasanton, California. Warehousing locations include Fremont, California, and Branchburg, New Jersey. The company recently partnered with microinverter manufacturer SolarBridge to launch an ac module for the North American market that has been listed to UL 1741 and UL 1703 standards.

Helios Solar Works. Established in 2011, Helios Solar Works is a privately held mono c-Si module manufacturer based in Milwaukee, Wisconsin. The company utilizes purchased cells to assemble products that meet Buy American Act and ARRA content requirements. Helios commercialized its module line in 2011. Notably, the current product line includes large-format 96-cell modules with power outputs of 400–420 W. Helios Solar Works has 50 MW of production capacity and recently formed a partnership with microinverter manufacturer Exeltech to develop and list a line of fully integrated ac modules.

Hyundai Solar. Hyundai Heavy Industries was founded as a shipbuilding business in 1972. The privately held Korean firm established Hyundai Solar in 2005 and entered the US market in 2010. Current cell and module manufacturing capacity is reported to be 600 MW. Hyundai Solar operates US sales offices in Englewood Cliffs, New Jersey; Torrance, California; Houston; and Mansfield, Ohio. Product is warehoused in Gardena, California, and Mansfield, Ohio.

Itek Energy. Launched in 2011, privately held Itek Energy is based in Bellingham, Washington. The company’s product meets Washington state’s “Made in Washington” requirements, which allow increased financial incentive payments for solar projects installed in the state that utilize equipment made there. Itek reports a manufacturing capacity of 4.3 MW and performs module assembly in Bellingham with purchased cells. The company currently offers a 60-cell 240 W mono c-Si product that at the time of publication had not been tested to IEC standard 61215. Itek has certified the use of Burndy WEEB clips as a means of providing equipment grounding for its module.

Jinko Solar. Headquartered in Shanghai, China, Jinko Solar is a vertically integrated manufacturer of ingots, wafers, cells, and mono and poly c-Si modules. The company was founded in 2006 and reports 1.2 GW of production capacity. In 2010, Jinko Solar was listed on the New York Stock Exchange and established a US sales office in San Francisco. Manufacturing facilities are located in the Chinese provinces of Jiangxi and Zhejiang.

Kyocera Solar. Kyocera Solar is a subsidiary of the publicly traded Kyocera, a global electronics group headquartered in Kyoto, Japan. Kyocera Solar is vertically integrated and controls all aspects of its PV manufacturing process, from silicon production to module assembly. The firm is one of the longest continually operating module companies in the North American solar market. Kyocera Solar began manufacturing modules in 1980 and entered the US market in 1983. The company currently operates a manufacturing facility in San Diego that produces Buy American Act–compliant modules. The company’s US sales office is located in Scottsdale, Arizona. Kyocera recently established a new warehousing location in Cherry Hill, New Jersey, to compliment its existing Scottsdale, Arizona, warehouse location.

Lumos. Privately held Lumos offers frameless modules with black, white or clear backsheets that it developed to provide a more aesthetic option than framed modules. The LSX module line is well suited for awning and overhead solar glazing applications and features a bolt-through fastening approach that fully integrates with Lumos’ proprietary low-profile racking system. While the modules are built to spec by a contract manufacturer outside the US, the integrated racking system is manufactured by Lumos in the US. The modules are not currently tested to IEC standard 61215. Lumos launched the LSX product in 2011 and has warehousing in Los Angeles and Boulder, Colorado.

MAGE Solar. MAGE Solar specializes in turnkey PV systems that include modules, mounting systems and leadingbrand inverters for residential, commercial and utility-scale projects. The manufacturer’s MAGE POWERTEC PLUS modules are available in ac and dc versions in the US market and feature 12-year 90% power and 30-year 80% power performance guarantees. MAGE Solar USA is headquartered in Dublin, Georgia. Its publicly traded parent company, MAGE Solar, was founded in 2007 and is based in Ravensburg, Germany. US warehousing locations include Dublin, Georgia, and Phoenix. The company reports a global manufacturing capacity of 50 MW. MAGE Solar introduced its products to the US market in 2009 and conducts module assembly of its Buy American Act–compliant products in Georgia.

MEMC. MEMC’s Silvantis P290 Series module line is the first product manufactured by a US company to achieve a 1,000 Vdc rating to the UL 1703 standard by CSA. The 1,000 Vdc–rated polycrystalline modules allow for more modules per source circuit and correspondingly lower BOS and O&M costs in utilityscale and distributed-generation projects. MEMC also offers the 600 Vdc–rated Silvantis M250 Series monocrystalline product line. MEMC is a publicly traded company that was founded in St. Peters, Missouri, in 1959. The company launched its PV product line in 2011 and conducts silicon, ingot and cell processing and manufacturing, as well as module assembly, in the US. MEMC has a manufacturing capacity of 400 MW and develops projects through its subsidiary, SunEdison.

Mitsubishi Electric. Headquartered in Japan with US sales offices and warehousing in Cypress, California, Mitsubishi Electric began manufacturing modules on a commercial scale in 1999 and released products in the US market in 2005. The company’s modules are manufactured in Japan. Mitsubishi Electric is a privately held company that offers both mono- and polycrystalline products that feature 100% lead-free solder. Additionally, Mitsubishi offers the predesigned Diamond Mount system. Available in 100 kW, 250 kW and 500 kW blocks, the Diamond Mount can be deployed in roof- or ground-mounted projects. The system includes modules, a ballasted racking system designed by P2, and all BOS components.

Motech Americas. Motech Americas’ publicly traded parent company, Motech Industries, is headquartered in Taiwan and reports a cell manufacturing capacity of 1.7 GW. The company’s US subsidiary is based in Newark, Delaware, where it assembles modules for the US market. All module product design, engineering and certification for Motech’s global module offer is based at Motech Americas’ Delaware plant, which previously housed R&D and module manufacturing facilities for AstroPower and GE. Globally, Motech Industries has 150 MW of module assembly capacity. The Delaware facility began assembling modules for the US in 2010. Motech Industries also manufactures a line of string inverters for the North American market with rated power outputs of 2.9–7.5 kW.

NESL USA. Located in Tualatin, Oregon, NESL USA is the North American office of Changzhou NESL Solartech, a privately held Chinese module manufacturer with 400 MW of production capacity. The firm commercialized its module products in 2005 and entered the US market in 2010. Warehousing locations are in Tualatin, Oregon, and Elizabeth, New Jersey. The company partnered with microinverter manufacturer SolarBridge in 2012 to offer a listed ac module in the US.

Panasonic. In 2010 the publicly traded Japanese Panasonic Corporation and Sanyo Electric conducted a share exchange that made Sanyo a wholly owned subsidiary of Panasonic. While the Sanyo subsidiary continues to manufacture the a-Si/c-Si HIT module line, in December 2011 the products were rebranded as Panasonic. With module efficiencies of 19% for the 240 W module and 18.7% for the 235 W module, Panasonic’s HIT products offer one of the highest module conversion efficiencies in the industry, second only to those of some SunPower models. Panasonic’s US sales office is located in Cupertino, California, and the company operates an ingot and wafer processing plant in Salem, Oregon. Panasonic commercialized its module manufacturing in Japan in 1997 and entered the US market in 2002. The firm reports more than 600 MW of global manufacturing capacity.

Perlight Solar. Perlight Solar was established in Wenling, China, in 2006 and introduced its products to the US market in the same year. In April 2012, Perlight announced the addition of its American-made USA Series modules to the CSI list of eligible products. Currently, all of the company’s USA Series and International Series modules are listed to the UL 1703 standard and are CSI eligible. Perlight Solar modules are manufactured in Texas and Wenling, China, with a combined annual production capacity of 500 MW for solar modules and cells. Module assembly, including tabbing, stringing, layup, lamination, framing and boxing for the USA Series modules, is conducted in Texas. This module line is Buy American Act compliant and can include as much as 90% domestic content. Perlight’s US sales offices are located in Torrance, California, with warehousing in Carson, California. The firm is privately held.

REC Group. REC Group is active in silicon, wafer and cell production as well as module assembly. The company is headquartered in Norway and operates 800 MW of production capacity worldwide. REC Group commercialized module production in 2007 and entered the US market in 2008. The manufacturer operates silicon materials plants in Moses Lake, Washington, and Butte, Montana, where it produces 100% of the silicon it utilizes for module manufacturing. REC’s wafer, cell and module production facilities are located in Tuas, Singapore. US sales offices are located in San Luis Obispo, California, with warehousing in Sacramento, California. REC reports that its Peak Energy Series modules have the industry’s lowest carbon footprint due primarily to the efficiency of its fluidized bed reactor silicon production process. The company offers a certification program that includes training, as well as a 2-year extension to the company’s standard 10-year product workmanship warranty for certified installers.

ReneSola. Founded in 2005 in China as Zhejiang Yuhui Solar Energy Source, ReneSola is a publicly traded, vertically integrated polysilicon, ingot, wafer, cell and module manufacturer. The company began manufacturing module wafers for utilization by other firms in 2007 and was listed on the New York Stock Exchange in 2008. In 2009, ReneSola acquired Wuxi Jiacheng Solar Energy Technology and achieved full vertical integration. ReneSola entered the North American market in 2012 and currently offers a full line of 60- and 72-cell poly and mono c-Si modules ranging in capacity from 250 W to 305 W. US sales offices are located in San Francisco, with warehousing in Oakland, California, and Newark, New Jersey. ReneSola reports a production capacity of 1.2 GW.

Renogy. Renogy is a privately held corporation headquartered in Baton Rouge, Louisiana, comprised of three divisions including solar, LED and biomass technologies. The vertically integrated company manufactures products outside the US, where it operates ingot, wafer, cell and module assembly facilities with a reported 500 MW of capacity. Renogy commercialized its module products in 2005 and entered the US market in 2007. In 2011 Renogy partnered with the Chubb Group of Insurance Companies, a US-based property and casualty insurer. Renogy modules are covered by a double warranty that includes the company’s product warranty and Chubb’s insurance services. Warehousing locations include Baton Rouge, Louisiana; Long Beach, California; and New York.

RITEK Solar. Taiwan’s publicly held RITEK Corporation is one of the largest global manufacturers of optical media. Its independent US subsidiary, RITEK Solar, commenced module manufacturing in Taiwan in 2008 and began offering products in the US market in 2009. RITEK Solar’s US sales office is located in Diamond Bar, California. Products are warehoused in City of Industry, California. RITEK Solar reports that its modules are tariff-free due to its Taiwanese manufacturing location. RITEK Solar also provides design and construction services for residential, commercial and utility-scale projects.

Samsung. Samsung is a privately held global products and services provider that is active in the electronics, machinery, chemical and financial services industries. The South Korean company commercialized its module production efforts in 2010 and introduced PV products to the US in the same year. The company’s US sales office is located in Irvine, California. Considering the scale of Samsung’s global presence in the electronics space, its commitment to PV manufacturing appears to be relatively limited. The company currently offers six mono c-Si models with rated outputs of up to 255 W.

Schüco USA. Otto Fuchs is the German parent company of Schüco International and its US subsidiary, Schüco USA. Schüco is a global provider of energy-efficient building technologies including door, window and facade systems, as well as solar water heating and photovoltaic products. Schüco USA operates sales offices in Union City, California, and Newington, Connecticut. With a reported 1 GW of PV production capacity, Schüco launched its module products in 1999 and entered the US market in 2002. The company offers poly and mono c-Si modules as well as BOS packages for residential and commercial projects. In addition, Schüco offers its partners financing products that include residential leases and 12-month interest-free loans.

Sharp. Sharp Electronics commercialized its PV technology in 1963 and began exporting modules to the US market in 2001. Sharp Electronics is a subsidiary of the publicly traded Sharp Corporation, headquartered in Japan. The company reports 1 GW of combined c-Si and tandem junction thinfilm silicon module production capacity, and also manufactures racking systems designed specifically for Sharp modules. In the project development arena, Sharp collaborates with BOS, system integration and construction companies to provide turnkey projects at the MW plus scale. The company’s sales office is located in Camas, Washington, and it has warehouse facilities in Tennessee and California. Since 2003, Sharp has operated a 160 MW module assembly and racking manufacturing facility in Memphis, Tennessee. For residential and small commercial applications, Sharp offers its SunSnap system, which packages Sharp 60-cell modules compatible with Zep Solar’s mounting and auto-grounding system with Enphase M215 microinverters.

Silicon Energy. Headquartered in Marysville, Washington, Silicon Energy launched its integrated module and racking system in 2009. The privately owned company manufactures nine module models ranging from 160 W to 200 W that meet Washington state’s “Made in Washington” requirements, which allow increased financial incentive payments for solar projects that utilize equipment made in the state. Silicon Energy modules are well suited for awning and overhead solar glazing applications. Differentiating features include doubleglass construction, Class A fire rating, integrated wireways for conductor management, an open lower-edge design that facilitates cleaning and snow shedding, and a high design load rating of 125 psf. Silicon Energy manufactures modules with purchased cells and reports a 4 MW production capacity. Product testing to IEC standard 61215 is pending. Sales offices and warehousing are located in Marysville, Washington, and Mountain Iron, Minnesota.

SolarWorld. Headquartered in Bonn, Germany, SolarWorld established its US presence with the acquisition of Shell Solar in 2006. The history of SolarWorld’s US subsidiary dates back to 1977 when the Atlantic Richfield Company purchased Solar Technology International to form ARCO Solar in 1977. Siemens acquired the business in 1990 and operated it until 2002, when Royal Dutch Shell purchased it. Currently, SolarWorld USA operates a 500 MW ingot, wafer, cell and module manufacturing plant in Hillsboro, Oregon, that accounts for approximately half of SolarWorld’s 1 GW global production capacity. The US plant supplies modules for the North and Latin American markets. Modules manufactured in its US facility are ARRA and Buy American Act compliant. SolarWorld recently expanded its manufacturing scope to include fixed and tracked racking solutions. The company offers an authorized installer program, pre-engineered systems and financing products.

SOLON. German PV manufacturer and project developer SOLON began restructuring via insolvency in 2011. In March 2012, Microsol, a United Arab Emirates–based cell and module manufacturer, acquired components of SOLON’s insolvent companies and began operating the company as SOLON Energy GmbH, headquartered in Berlin, Germany. SOLON maintained its US sales and module assembly facility, established in Tucson, Arizona, in 2007. In addition to manufacturing poly and mono c-Si modules, SOLON is active in project development and in mounting-system product design and manufacturing for commercial rooftop, carport and utilityscale solar plants.

Suniva. Suniva is a privately held American manufacturer of c-Si cells and modules headquartered in Norcross, Georgia. The company launched its product in the US in 2008 and currently reports 400 MW of production capacity. Suniva’s Optimus Series module lines are Buy American Act compliant and contain more than 80% domestic content. The company operates US sales offices in Georgia, California, New Jersey and Washington, and has established product warehousing in Georgia, California and South Carolina. Suniva recently added AEE Solar and Soligent (formed by the merger of DC Power Systems and Solar Depot) as distribution partners.

SunPower. Headquartered in San Jose, California, SunPower commercialized its module technology in 2004 and currently manufactures the highest-efficiency modules available in the US market. The company went public in 2005 and is traded on the NASDAQ Exchange. In 2011, Total–a French oil and gas producer–purchased a majority stake in SunPower. Most of SunPower’s reported 1.2 GW manufacturing capacity is located in the Philippines, with additional module assembly conducted in Milpitas, California. SunPower was one of the first US-based module manufacturers to develop an extensive authorized dealer network and has set a high bar for the industry in terms of brand management and marketing efforts. SunPower is active in the residential lease arena, as well as in commercial and utility-scale project development. In 2011, SunPower partnered with microinverter manufacturer SolarBridge to launch one of the first listed ac modules.

Sunpreme. Sunpreme commercialized and launched its module product line in 2011 and currently has 30 MW of production capacity located in Jiaxing, China. The privately held company utilizes metallurgical-grade silicon as the basis of its patented SmartSilicon cell technology, which combines thinfilm structures with a crystalline silicon substrate. Sunpreme has two product lines, which include the G60 series frameless, double-glass modules and a more standard aluminum framed F60 series. The company’s global headquarters, R&D, sales offices and warehousing are located in Sunnyvale, California.

Suntech. Suntech was founded in Wuxi, China, in 2001 and achieved a global production capacity of 2.4 GW, the world’s largest, in 2012. The company entered the US market in 2004 and was listed on the New York Stock Exchange in 2005. Suntech’s US sales office is located in San Francisco, and it has warehousing operations in California, New Jersey and Arizona. Suntech opened a US module assembly facility in Goodyear, Arizona, in 2010 that utilizes cells manufactured by the firm in China. The company offers an authorized dealer program in North America that includes technical, sales and marketing support.

Trina Solar. Trina Solar was established in 1997 and listed on the New York Stock Exchange in 2006. The firm is a vertically integrated manufacturer of poly and mono c-Si ingots, wafers, cells and modules. The company conducts all of its research, development and manufacturing at its facilities in Changzhou, China. Trina Solar has multiple product differentiation efforts under way. The company’s Trinamount platform utilizes the railless mounting and auto-grounding system developed by Zep Solar for pitched and low-slope roof applications. In July 2012, Trina Solar announced a partnership with Tigo Energy and the launch of the Trinasmart smart module solution, which integrates a Module Maximizer manufactured by Tigo Energy. Trina Solar’s US sales office is located in San Jose, California.

Upsolar. Headquartered in Shanghai, China, privately held Upsolar commercialized its PV products in 2007. It established a US subsidiary, Upsolar America, in 2009, and opened US sales offices in San Francisco in the same year. It began shipping products into the US in 2010. The company’s products include 60- and 72-cell poly and mono c-Si modules with rated outputs of 240–300 W, which are available with standard or Zep Groove frames. As of May 2012, Upsolar has only utilized cells sourced from Taiwan for all modules shipped into the US market. Upsolar’s PowerCLIP warranty provides noncancelable coverage from PowerGuard Specialty Insurance Services. The company recently debuted its Smart Module, developed through a strategic partnership with Tigo Energy. Upsolar operates warehousing facilities in California, Arizona and New Jersey, and currently has 400 MW of production capacity.

Yingli Green Energy. Founded in 1998, Yingli Green Energy has been a vertically integrated business that covers the entire PV value chain, from poly c-Si production to module assembly, since 2004. The firm, based in Baoding, China, began manufacturing modules in 2006 and has gained a strong market share in the US since its product introduction in 2009. The company currently has 4 GW of product deployed worldwide and a manufacturing capacity of 1.7 GW. Yingli’s product offering includes poly and mono c-Si modules that range from 165 W to 300 W and 190 W to 270 W, respectively. Notably, Yingli has 20 MW of PV capacity installed at its manufacturing facilities in China. The firm operates US sales offices in New York and San Francisco. Warehousing facilities are located in Carson, California, and Edison, New Jersey



Joe Schwartz / SolarPro magazine / Ashland, OR /

Doug Puffer / SolarPro magazine / Ashland, OR /

An expanded version of the 2012 c-Si module dataset is available in Microsoft Excel format at click HERE.

PV Manufacturer Contact

1SolTech / 888.598.0295 /

Advanced Solar Photonics / 407.804.1000 /

Astronergy / 855.882.7876 /

AU Optronics (BenQ Solar) / 855.236.7478 /

Auxin Solar / 408.868.4380 /

Bosch Solar Energy / 650.356.3100 /

Canadian Solar / 925.866.2700 /

Centrosolar America / 877.348.2555 /

China Sunergy / 415.229.7994 /

CNPV Solar Power / 800.365.6214 /

Colored Solar / 877.526.2462 /

Conergy / 888.396.6611 /

CSG PVTech / 562.266.8480 /

DelSolar / 888.880.8868 /

ecoSolargy / 877.808.4213 /

ET Solar / 925.460.9898 /

Gloria Solar / 650.961.6100 /

Grape Solar / 877.264.1014 /

GreenBrilliance / 888.365.2754 /

Hanwha SolarOne / 714.689.6868 /

Helios Solar Works / 877.443.5467 /

Hyundai Solar /

Itek Energy / 360.647.9531 /

JA Solar / 408.586.0000 /

Jinko Solar / 415.402.0502 /

Kyocera Solar / 800.223.9580 /

LDK Solar / 408.245.0858 /

LG Electronics / 201.816.2000 /

Lightway Solar / 732.602.1930 /

Lumos / 877.301.3582 /

MAGE Solar / 877.311.6243 /

MEMC / 636.474.7660 /

Mitsubishi Electric / 714.220.2500 /

Motech Americas / 302.451.7500 /

MX Solar / 732.356.7300 /

NESL USA / 800.242.7114 /

Panasonic / 408.861.8424 /

Perlight Solar / 424.242.8068 /

Phono Solar / 281.909.0644 /

REC Group / 877.332.4087 /

ReneSola / 415.852.7418 /

Renogy / 225.578.5182 /

RITEK Solar / 800.823.2505 /

Samsung / 800.726.7864 /

Schüco USA / 510.477.0500 /

Sharp / 800.765.2706 /

Silicon Energy / 360.618.6500 /

Siliken / 760.448.2080 /

SolarWorld / 855.467.6527 /

SOLON / 520.807.1300 /

Sun Earth Solar Power (Nbsolar) /

Suniva / 404.477.2700 /

SunPerfect Solar / 408.232.8088 /

SunPower / 800.786.7693 /

Sunpreme / 408.245.1112 /

Suntech / 866.966.6555 /

Talesun Solar / 800.276.0823 /

Trina Solar / 800.696.7114 /

Upsolar / 415.263.9920 /

Westinghouse Solar / 888.395.2248 /

Yingli Green Energy / 888.686.8820 /

Primary Category: 

PV modules mounted to racks or trackers should be designed to withstand wind forces as prescribed in the building codes, like any other important structure. However, there is a significant challenge in applying existing codes to commercial and industrial roof-mounted PV arrays.

Everyone is familiar with the potentially destructive forces that wind can exert on structures, whether from video footage or firsthand experience with structural damage from strong winds. The wind load on a structure depends on many factors, including the wind speed, the wind characteristics (turbulence and velocity profile), the geometry of the structure, the effect of surrounding objects and the height above the ground, to name just a few. A structural engineer must calculate wind forces or loads to ensure that a structure can resist wind conditions at a particular location.

The engineer should always determine wind loads on PV systems, even if local building departments do not require such an analysis. Wind-related failures of both roof- and ground-mounted systems have occurred—and more can be expected because wind loads are poorly understood. The spread of solar energy will depend on favorable public opinion, and a poor safety record is counterproductive. The longterm success of the solar industry depends on the deployment of systems that are structurally engineered to meet building codes. Unfortunately, this is not as easy as it sounds.

In this article, we discuss wind loads on sloped PV modules installed on standard open racks on a flat or low-slope roof. We also provide high-level guidance for other types of systems. We present some fundamentals of wind loading on rooftop PV systems, as well as challenges associated with applying existing building codes to this type of system. While building codes will eventually include improved guidelines for determining wind loads on PV systems, the process of developing these guidelines is likely to take years. Until then, structural engineers need strategies to reduce the guesswork in estimating wind loads, particularly for sloped PV modules on flat roofs, because building codes are least applicable to this type of system.

With support from its parent company, Det Norske Veritas (DNV), and its wind engineering partner, CPP, BEW Engineering has developed an analytical approach for determining wind loads on sloped PV systems mounted on flat and low-slope roofs. BEW analyzed the results of thousands of wind tunnel tests in the process of developing this method. While this approach is more complex than typical ACSE 7 methods, the partnership developed a free online design tool for designers and engineers to simplify the process and expedite the analysis. Here we discuss how this tool works and how to apply the results.

The Building Code and PV

While all US building codes have sections on wind loading, it is widely accepted that the most comprehensive guide for estimating wind loads on structures is Standard No. 7 of the American Society of Civil Engineers (ASCE), Minimum Design Loads for Building and Other Structures (ASCE 7). All other building codes allow for the use of this ASCE standard. The latest edition of this standard is ASCE 7-10, the 2010 edition. However, ASCE 7-10 has not been widely adopted, and the 2005 edition is still mainly in use. Therefore, we reference ASCE 7-05 in this article. By the time ASCE 7-10 is widely adopted, it is likely that a broader range of publicly available wind tunnel data will be available that will improve upon the methods described in this article. The wind loading content found in ASCE 7 was primarily developed to calculate wind loads on buildings, though a small subset of other structures, such as billboards and chimneys, is included. As a result, building codes do not provide clear guidance on how to calculate wind loads on PV arrays, unless these are shaped like buildings—for instance, PV carports.

With proper guidance— see “Wind Load Analysis Recommendations by PV System Type” (sidebar, below)—designers and engineers applying ASCE 7-05 methods can do a reasonable job of estimating wind loads on some flat-plate PV systems. For example, ground-mounted PV systems are very much like small open buildings— meaning buildings with no walls, such as carports—closely spaced together. In addition, wind load patterns on modules mounted parallel to and close to the roof—as is common in residential applications—may be estimated by calculating the loads expected on the exterior of the building cladding. In many but not all of these cases, building codes overestimate wind loads on these types of PV systems, and designers and engineers can use these conservative results with confidence.

Applying building codes to sloped PV systems on flat roofs presents significant challenges. While some data are provided in the ASCE 7 standard related to “rooftop equipment,” these were developed for equipment with a prismatic shape, such as chimneys and HVAC units, with no gaps between the equipment and the roof. These data are not applicable to roof-mounted PV. Designers are left with nothing to do but guess which tables and figures—for example, which building shapes—in the building codes best apply to PV systems. Many of the choices designers must make depend upon the type of building classification.

Building classification. ASCE 7-05 has three different classifications of buildings depending on the porosity of the walls (Section 6.2): open buildings, enclosed buildings and partially enclosed buildings. Most buildings are considered either enclosed or partially enclosed buildings.

The walls of open buildings must be at least 80% open. Carports are an example. Partially enclosed buildings are those in which a wall has openings that are larger than the openings on other walls. In this situation, large positive pressures can develop inside the building. Enclosed buildings are buildings that are neither open nor partially enclosed.

PV system classification. Rooftop PV arrays are like very small open buildings on top of very large enclosed or partially enclosed buildings. This is unchartered territory for building codes. Should the engineer apply the loads for a flat roof to tilted panels on a flat roof ? If so, that ignores panel tilt. Should the engineer use tilted roof numbers? If so, which ones: sawtooth or monosloped? Engineers must make a judgment call on whether to use the PV tilt or roof tilt, or perhaps some combination of the two. Should they calculate net design wind pressures using the components and cladding (C&C) method or the main wind force resisting system (MWFRS) method? ASCE 7-05 provides no guidance on these issues.

Whatever method engineers eventually choose, the resulting wind loads will likely exceed the amount of ballast that most roofs can support. Further, ballasted systems are typically engineered with custom aerodynamic features that the generic application of building codes cannot address. Therefore, manufacturers of ballasted systems must often rely on wind tunnel testing to verify that their systems have adequate wind resistance, particularly when they have implemented unique design features to lower the wind loads.

Wind tunnel testing. The requirements for proper wind tunnel testing are spelled out explicitly in Method 3 of ASCE 7-05. Additional guidelines are found in the article by Gregory Kopp and others, “Rooftop Solar Arrays and Wind Loading: A Primer on Using Wind Tunnel Testing as a Basis for Code Compliant Design per ASCE 7.” (See Resources.) Unfortunately, it is all too common for wind tunnel testing to fail to comply with these guidelines.

For example, to perform a wind tunnel test correctly for a roof-mounted PV system, a model of the building must be included. The dimensions of the model building and the parapet height need to be to scale. The slope, row spacing and height above the roof of the model PV array also must be to scale. If a friction coefficient is modeled, it should match that expected on the actual roof.

Further, to model the boundary layer, most tests must be done at a small scale (1:30 to 1:50). Unless testing was done at the new Insurance Institute for Business & Home Safety wind tunnel (see Resources), it is unlikely that tests done at larger scale or at full scale properly model wind flow. Designers may wish to hire an independent expert to review wind tunnel reports from manufacturers. The independent expert should comment on the compliance with ASCE 7-05 requirements and the similarity between the models and the fullscale systems.

When done correctly, however, using wind tunnel tests to determine wind forces removes the guesswork involved with applying building codes to PV systems. The wind tunnel data that we have analyzed suggest that some types of PV array structures are overdesigned, while others may be significantly underdesigned relative to expected wind loads—especially corners and edges of sloped PV arrays on flat roofs.

Wind loads on PV systems. Wind flowing over PV systems applies forces to PV modules, fasteners, the racking system and the roof, if an array is roof mounted. All wind forces on a roof- or ground-mounted array must ultimately be transmitted through the structure to the ground. All of the structural components in a PV system have limits on how much wind force they can withstand, so it is critical that wind loads on PV arrays be determined for every PV system; further, all structural components in the system should be verified to resist expected wind loads—not to mention the loads on the PV module itself or the loads that the entire system transmits to the roof or ground.

Note that “every PV system” means any size residential, commercial, or utility-scale system; ballasted or structurally attached; and rack mounted, flat roof mounted, sloped roof mounted or ground mounted. Similarly, “all structural components” includes all fasteners; welds; steel; aluminum, wood, concrete, plastic and other structural members; and motors, gears and drive mechanisms for tracking systems.

Wind Load Fundamentals

The force and pressure of the wind are proportional to the velocity of the wind squared. The basic equation that converts wind velocity into pressure is shown in Equation 1:

Wind pressure = ½ρ x V2 x C                   (1)

where ρ is the density of air, V is the wind velocity, and C is a dimensionless coefficient that is typically measured for a specific object.

The equations found in ASCE 7-05 are based on Equation 1 (a fundamental equation from fluid dynamics), where the coefficient C is referred to as the pressure coefficient. The most challenging part to estimating wind loads on any structure is finding out which of the many pressure coefficients in ASCE 7-05 should be used. The pressure coefficient depends on many factors, including the shape of the structure and the tributary area of the structural component being analyzed.

Your firsthand experience with the force of the wind provides some understanding of what makes an object more or less aerodynamic. The evolution of the geometrical shape of cars, the curve of airplane wings, and the tuck of bicyclists and skiers has largely focused on improving aerodynamics so that these objects can move through the air with less resistance. This improvement is measured as a reduction in the pressure coefficient.

Since force is equal to pressure times the area over which the pressure is applied, wind force is determined by multiplying the wind pressure by a representative area. This area is referred to by structural engineers as a tributary area and by ASCE as the effective wind area; see the definition in Section 6.2 of ASCE 7-05.

The tributary area or effective wind area is generally understood as the area that a component structurally supports. For example, if there are four fasteners securing a 4-foot-by-8-foot sheet of plywood to a roof deck, each fastener has a tributary area equal to the total area of the plywood divided by the number of fasteners, or 8 square feet. The effective wind area can also be thought of as the area over which loads are transmitted and effectively resisted by the structural system. For PV systems, one way to think of the effective wind area is this: Suppose the PV array was placed on a surface without physically restraining the system from uplift. If one were to lift one PV module in the array, how much of the system would lift along with it without permanently damaging any of the components?

More than one effective wind area applies to most PV systems, depending upon the component under analysis. When analyzing the racking structure, the effective wind area may be relatively large—perhaps the area of 5 to 30 modules if the rack is rigid enough to support the applied loads from this many modules. However, the fasteners that secure the module to the rack have a smaller effective wind area. If one PV module is secured with four fasteners, each fastener has an effective wind area of one-quarter of the module area. The PV module itself has an effective wind area equal to the module area.

Components with smaller effective wind areas have higher wind pressures. This is because wind pressure distributions on structures vary rapidly with time and location on the structure. One PV module may be subjected to a high wind pressure while a module 4 feet away may have a much lower wind pressure. A rigid rack supporting both modules may be able to spread the load across the structural components. However, load sharing across individual PV modules is limited, and the fasteners that secure modules to the structure cannot spread loads. Similarly, a PV mounting system that is insufficiently rigid will not spread loads over a large area. It is very important to analyze wind loads on individual PV modules using the effective wind area for one PV module to ensure that wind loads do not exceed the module’s rating. There are applications where wind loads in excess of module ratings are a very real possibility.

As shown in Equation 1, wind force and pressure on any object is proportional to the wind velocity squared. Therefore, a wind speed of 60 mph creates wind forces four times larger than a 30-mph wind. This is important to remember when considering claims that a product can withstand a 90- or 120-mph wind speed simply because it survived a 70- or 80-mph wind event.

In addition, it should be noted that the basic wind speed provided in the ASCE 7-05 standard for any location in the US represents the free-stream velocity, which corresponds with unobstructed flow over open Exposure C terrain (as defined in ASCE 7-05) at 10 meters above the ground with an averaging period of 3 seconds. Further, the basic wind speed values in ASCE 7-05 represent the expected wind speed with a 50-year recurrence interval.

When you compare ASCE 7-05 wind speeds to measurements made in the field, be sure that you are comparing apples to apples. Wind sensors placed within an array at module height do not provide free-stream velocity measurements. In fact, a wind speed measurement taken anywhere in the vicinity of the PV modules could be significantly higher or lower than the free-stream velocity. Wind speeds averaged over any period other than 3 seconds are not 3-second gust wind speeds, as defined in the standard. Wind speed measurements taken at different heights and over different averaging periods can be modified using equations in ASCE 7-05 to determine the equivalent 10-meter, 3-second wind speed. However, to do this modification, the surrounding objects must not affect the measurement, and unfortunately this is almost never the case with wind sensors in PV arrays.

Wind Loads on Roofs and Rooftop PV

When wind flows around an object, the wind flow becomes disturbed and its qualities are changed. When wind flows over a building, this change is dramatic, and the effects of this disturbed wind flow are very different than for wind flow across the ground.

Consider a large retail building with a flat roof, classified as a low-rise building. A low-rise building refers to a structure of modest height and substantial girth: consumer retail stores, wholesale clubs, warehouses and distribution centers are common examples. Large flat or nearly flat (lowslope) roofs are characteristic of low-rise buildings. Decades of wind tunnel testing and field measurements have shown that worst-case wind loads on the roof occur when wind hits the corner of these types of structures. These so-called cornering winds also result in worst-case wind loads on roofmounted PV modules.

Conical vortices. When wind hits a building, the flow separates at the roof edges. However, the two zones of flow separation interact as they both reattach, and the result is the generation of conical vortices above the roof. These vortices can be thought of as horizontal tornadoes that originate at the corner of the roof and radiate at angles of 10° to 20° along the leading edges of the building. The position and strength of these vortices are a function of the wind direction, and strong vortices are present for wind approach angles of 25°–65° (45° ±20°).

Figure 1 shows the vortices that form along the edges of a roof without PV modules when wind hits the corner of the building. In between the vortices near the windward corner, there is a region of accelerated flow where wind speeds along the roof surface are 20% higher than those approaching the building at roof height. The presence of accelerated airflow as a result of cornering winds is one reason why ASCE 7-05 and other building codes can underpredict wind loads on PV arrays.

Wind tunnel testing has shown that the vortices along the edges of the roof during cornering winds are responsible for the highest peak pressures on the building envelope, which occur directly beneath the vortices as shown in Figure 2. While the accelerated flow associated with conical vortices does not have a big impact on the loading of the roof itself— which is what ASCE 7-05 and other codes were designed to take into account—it does have an impact on objects that protrude above the roof.

Conical vortices are responsible for the greatest wind pressures that occur across roof-mounted PV modules. However, the manner in which vortices load rooftop PV modules fundamentally differs from how they load the roof cladding elements that codes are based on. This is illustrated in Figures 3a and 3b, which provide typical patterns measured in the wind tunnel of net pressure coefficients (GCp values) for an array of moderately tilted (10°–15°) modules on a roof that measures 6h × 6h, where h is the mean building height. (Note that in many figures in ASCE 7-05, the pressure coefficient is given the variable name GCp instead of C.)

It is clear from Figure 3a that on one hand the vortex that forms along the north edge of the building during northeast cornering winds is “quiet,” having little discernible effect on the modules. (Note that the term north and all other directions in this article are defined based solely on the array orientation, which is assumed to face to the south.) The east edge vortex at the northeast corner, on the other hand, produces significant wind loads that peak at an angle of nearly 30° from the roof edge. This indicates that it is the interaction between the swirling and reattaching wind flows and the accelerated between-vortex flows that creates peak pressures for the modules, rather than transfer of the high pressure at the center of the vortex to the surface, as is the case for the roof itself.

Conversely, Figure 3b shows that the east edge vortex of the southeast corner vortex pair is quiet, creating little lift on the modules, while the south edge vortex creates substantial lift. In this case, the location of the lift is closer to the edge of the roof and more nearly beneath the vortex core axis.

If the modules shown in Figure 3 were rotated to face to the west, then patterns in each corner would be reversed. Any noticeable rotation of the modules relative to the building edges increases the lift associated with the two quiet vortices, and may worsen overall loads as well. More data is needed to evaluate wind loads when arrays are not aligned along the same axes as the building edges.

Impact of array layout. It is important to note that the pressure coefficients in Figures 3a and 3b are for an array that completely covers the roof, with no gaps other than typical spacing in the north-south direction between rows of sloped PV modules. In most cases, there are gaps between mechanical sections of a PV array. PV modules located along the edges of mechanical subarrays can experience increased wind pressure. (Note that while the NEC defines the term subarray as “an electrical subset of a PV array,” in this article we apply the term to mechanical sections of an array that are separated from other array sections by distances larger than the standard row spacing in the north-south direction or more than a foot in the east-west direction.)

Wind loads on PV arrays are sensitive to the physical array layout. This is one of the most critical differences between results obtained by applying the ASCE 7-05 standard to sloped PV arrays on flat roofs and actual wind loads. Figure 4 demonstrates this point. The colored blocks in the figure represent six subarrays, shown in blue with some yellow and red regions, installed on a roof surface, shown in white. The color gradients found in the subarrays represent pressure coefficients measured in a wind tunnel, with blue representing relatively low-pressure coefficients and red higher-pressure coefficients. This figure shows how the edges of mechanical subarrays experience much higher wind loads than the interior sections.

The ASCE 7-05 standard, which only considers corner and edge zones for the roof and not for individual subarrays, would not address these higher wind loads. The red lines in Figure 4 show the corner and edge zones as defined by ASCE 7-05. It is clear from this figure that peak pressures on the subarrays can occur in what ASCE 7-05 considers to be interior roof zones.

Any building code requires us to design a structure to withstand wind loads from all directions. So even though Figure 4 shows that the north rows have the highest coefficients from a northeast cornering wind, other wind directions can cause relatively high-pressure coefficients in the south, east or west edges of mechanical subarrays.

Effects not captured by ASCE methods. Capturing the effects of subarray spacing, parapets and rooftop objects in an analytical process inherently forces the process to become more complicated than the typical application of ASCE 7-05. It is clear from available wind tunnel data that peak wind loads occur at locations that are more than one building height, h, from the roof edges. The edge and corner roof zones prescribed in ASCE 7-05 typically extend only 0.4h from the roof edges, while wind tunnel data suggests that the corner and roof zones should extend 2h from the roof edges.

The effects of parapet walls on PV arrays are not characterized in ASCE 7-05 wind loads. While parapets reduce the peak wind loads at extreme corners and edges of the roof, they increase wind loads in interior areas. Modules that are very close to the parapet wall do receive some shielding benefit, provided that the parapet is close to or taller than the maximum height of the PV array.

Objects on the roof that are taller than the PV array, such as penthouses and HVAC units, can also provide some shielding benefit; however, in certain wind directions, the objects can generate vortices that increase wind loads in the vicinity of the object. This possibility is not captured in the ASCE 7-05 standard.

The alternative to complexity is to simplify the process by making conservative assumptions. However, making overconservative assumptions tends to drive up structural costs—perhaps making a viable project uneconomical— and may rule out other viable sites due to structural loading limitations.

Recommended Practice for Estimating Wind Loads

Capturing the effects of subarray spacing, parapets and rooftop objects in a recommended analytical process inherently forces the process to become more complicated than the typical application of ASCE 7-05. The other alternative is to simplify the process by making conservative assumptions. To avoid being overconservative, the approach that we present takes all of the above impacts into account. The bad news is that the approach is more complicated than applying ASCE 7-05. However, there are two pieces of good news: One, the results will be more reflective of actual wind loads on the array; two, a free online design tool called the DNV Wind Load Calculator for Sloped PV Arrays on Flat Roofs ( segments/solar_energy/index.asp) will perform the complicated steps for you.

The DNV Wind Load Calculator uses an alternative method of calculating wind loads that was developed by determining which parts of ASCE 7-05 match available wind tunnel data. Thousands of wind tunnel tests have been conducted on roof-mounted PV systems. While much of the data is confidential, the results provide insights into how to best apply existing data in ASCE 7-05 to the problem of estimating wind loads on sloped PV modules on flat roofs. The DNV Wind Load Calculator’s method is completely ad hoc; there is little justification for it on the basis of physics or the intent of the code. It just fits the data.

How it works. Recall that Equation 1 defined wind pressure as ½? x V2 x C. This fundamental equation from fluid dynamics is the basis of the method for calculating wind pressures on roofs described in ASCE 7-05. DNV’s method relies on this equation as applied by ASCE 7-05 as well, but provides values for the pressure coefficient, C, that are not part of the ASCE 7-05 standard.

Sections 6.5.3 through 6.5.10 of ASCE 7-05 provide detailed guidelines on estimating what is essentially the first part of this equation (½? x V2), which ASCE calls the velocity pressure. As detailed in the standard, the ASCE 7-05 value for velocity pressure includes some additional terms beyond air density (?) and wind speed (V). ASCE 7-05 also provides values for pressure coefficients, C, which are a function of the building size and shape, as well as the effective wind area or tributary area of the structural component under investigation.

While the ASCE 7-05 pressure coefficients are commonly applied to sloped PV arrays on flat roofs, wind tunnel data has shown that these coefficients are not generally applicable to this configuration. The DNV Wind Load Calculator provides alternate values for the pressure coefficient on PV arrays that take into account the PV tilt, row spacing, height above the roof and sheltering from wind due to nearby objects such as nearby subarrays or parapet walls.

For example, a parapet wall shelters nearby modules, but can increase pressures in the middle of the roof. The calculator incorporates the potential increase in wind pressure in interior regions of the roof when a parapet wall is present, as well as the accelerated flow around rooftop objects such as HVAC units. The velocity pressure determined in the calculator is exactly as prescribed by ASCE 7-05, but the coefficients are based on actual wind tunnel data.


The DNV Wind Load Calculator is posted on DNV’s website (see Resources) along with a user manual. The first step is to enter a series of inputs that are defined in ASCE 7-05: mean roof height, basic wind speed, directionality factor, importance factor, topographic factor, velocity pressure exposure coefficient and effective wind area. These inputs, which structural engineers are used to working with, are used to determine the velocity pressure according to ASCE 7-05 .

A second set of inputs to the DNV Wind Load Calculator defines the PV array geometry. These parameters include effective wind area or tributary area, PV tilt, row spacing, height above the roof and the location of the array section being considered. Several factors are used to describe the array location, including the location within a subarray (north row, south row, eastern or western perimeter or interior), the roof location (corner, edge or middle) and the distances to nearby objects. The calculator also asks for the parapet height(s) on the building and the dimensions of rooftop objects that are taller than the array.

For each mechanical subarray section—for example, a north row in the corner of the roof—the calculator provides a wind pressure in pounds per square foot (psf). This result is the estimated pressure that should be applied analytically in the upward and downward direction perpendicular to the PV module. The calculator also provides recommended offset distances from rooftop objects, such as HVAC units, which can create regions of accelerated flow around them.

Comparison to results from ASCE 7-05. Comparing DNV Wind Load Calculator results to those determined using ASCE 7-05 is complicated because there is no one correct way to apply the standard to sloped PV modules on a flat roof. However, we have compared DNV Wind Load Calculator results to those achieved using several interpretations of the ASCE 7-05 standard commonly used by structural engineers. These comparisons indicate that the DNV Wind Load Calculator predicts wind loads higher than ASCE 7-05 methods in some cases and lower in others.

In general, interior PV array sections that are close to the roof, tightly spaced and that can spread loads to adjacent areas—sections with a relatively large effective wind area, in other words—will have lower wind loads than predicted by ASCE 7-05. Exposed array sections, particularly the north row in the corner and edge zones of the roof, may have higher wind loads.

If wind loads are found to be higher than expected or desired, the designer may iterate through many of the inputs to identify design characteristics that reduce wind loads. For example, placing the array closer to the roof, reducing row spacing or PV tilt, or increasing the effective wind area all reduce wind loads. If wind loads are still too high, a preengineered aerodynamic product that has undergone wind tunnel testing in accordance with ASCE 7-05, such as a ballasted system with wind deflecting shields, may be a viable option.

How are the results applied? The DNV Wind Load Calculator yields a wind pressure for any array location on the roof. As with ASCE 7-05, it is up to the designer and structural engineer to ensure that representative sections of the array are evaluated to determine worst-case conditions, considering both uplift and downward forces. Whereas ASCE 7-05 typically results in different values for uplift and downward pressure, the DNV Wind Load Calculator provides one result that should be applied in both directions in two separate analyses.

It is important that designers and engineers determine loads on modules, fasteners, all components within the racking system and the applied loads to the roof. Loads must ultimately be transferred from the modules to the fasteners and racking system, and ultimately through the roof deck and building structure to the ground. This is common knowledge for most structural engineers. Remember that this likely involves the use of different effective wind areas based on the load-sharing capability of the component under analysis. The wind load rating of the module should not be exceeded.

Once wind loads are determined, structural engineers must apply appropriate safety factors and combine loads as required in ASCE 7-05 Section 2. In addition to wind loads, other loads such as snow, seismic and gravity (dead load) must be taken into account. Structural engineers must consider each of these loads separately and in combination to identify the worst-case loading situation.

The use of these results assumes that the structure is sufficiently rigid so that it is not considered a flexible structure. It is recommended that PV mounting systems have a natural frequency above 4 Hz to be considered rigid enough to prevent resonance. This is higher than the 1 Hz limit prescribed in the ASCE 7-05 standard, because the 1 Hz limit is intended for large structures not likely to shed vortices that can create an excitation (resonance) in smaller structures. Wind tunnel testing has shown that PV systems shed vortices with frequencies in the 2–3 Hz range, so structures should be designed with natural frequencies in the 4–5 Hz range.

Wind Load Analysis Recommendations by PV System Type

For PV system types not covered in this article, we provide the following general wind loading analysis guidelines as a starting place for system designers and engineers.

Ground-mounted systems:

  • Following the definitions found in ASCE 7-05, treat the rows of PV modules as open buildings with monosloped roofs and use the slope of the PV modules as the roof slope to apply the standard.
  • Do not use ASCE 7-05 tables for billboards and signs.
  • For trackers, check all possible tilts and forces on the drive system (gears, struts, motors and so forth).
  • Results will be conservative for interior rows, but conducting wind tunnel tests in compliance with ASCE 7-05 guidelines can reduce this conservatism.

Sloped-roof systems:

  • When PV modules are incorporated directly into the roof surface, they can be analyzed using ASCE 7-05 in the same way that a roof would be analyzed.
  • If PV modules are parallel to the roof surface and offset no more than 6 inches, a method developed by the Solar A BCs and described in the report “Wind Load Calculations for PV Arrays” may be applied (see Resources below).
  • Any application of ASCE 7-05 will likely be conservative for arrays that have gaps on the order of 1 inch or more between modules.
  • For PV modules sloped relative to a sloped roof, there is no known publicly available data; wind tunnel testing in compliance with ASCE 7-05 guidelines is recommended, or consider placing the modules flush to the roof.

Ballasted roof-mounted systems:

  • ASCE 7-05 and the DNV Wind Load Calculator will likely result in wind loads that cannot feasibly be resisted by ballast alone since most roofs cannot support more than an additional 5–10 pounds per square foot, which means that ballasted systems must incorporate aerodynamic features such as wind-deflecting shields and must be tested in a wind tunnel, as no other method exists for estimating wind loads.
  • ASCE does not allow the use of computational fluid dynamics (CFD) in lieu of wind tunnel testing; any CFD results must be validated against wind tunnel measurements.

Designers specifying ballasted systems should ensure that:

  • Wind tunnel testing was done in accordance with ASCE 7-05 guidelines.
  • Worst-case conditions—such as wet and icy—are used to determine the friction coefficient for specific roofs, since the system relies on friction between the roof and the array.
  • The system will not cause a safety hazard as a result of an earthquake.
  • The roof membrane is not subject to fluttering that could damage the roof or ballasted system.
  • Ballasted systems installed on ballasted membrane roofs comply with established roofing industry guidelines, such as ANSI/SPRI RP-4, “Wind Design Standard for Ballasted Single-ply Roofing Systems.”

Limitations of use. The DNV Wind Load Calculator can be applied to your project only if it meets the following conditions:

  • Flat or low-slope roof (less than 7°)
  • Low-rise building (less than 60 feet high and wider and longer than it is tall)
  • Modules are tilted between 1° and 35° from roof surface
  • Modules are flat plate
  • Modules are solid (not porous)
  • Modules are mounted close to roof surface (gap under lowest part is less than 18 inches)
  • Top of module sits less than 4 feet above roof surface
  • Modules are placed more than 5 feet from roof edge with no parapet or with parapet shorter than top of modules
  • No wind deflector on perimeter of array
  • Modules are aligned with building edges
  • No significantly taller structures are located near roof in question
  • Structure is significantly rigid (natural frequency greater than approximately 4 Hz)

Apply the results of the DNV Wind Load Calculator with caution. Due to limitations in available data and the broad range of possible sloped PV geometries, the tool cannot possibly cover all types of sloped PV systems on flat and lowslope roofs. Rather, it covers standard rack-mounted PV systems that do not have aerodynamic enhancements such as wind-deflecting shields on the edges of modules.

Ballasted systems rely on a wide variety of aerodynamic features to withstand wind loads. These features are product specific and in many cases are protected by patents. The wind tunnel data that does exist for these products belong to individual manufacturers, so it is not possible to present generic results for ballasted systems. (Recommended guidelines for using ballasted systems are included in “Wind Load Analysis Recommendations by PV System Type,” sidebar above.)

Local building departments may not accept the results from the DNV Wind Load Calculator. For example, the local AHJ may require the use of the analytical methods described in ASCE 7-05 if these result in higher loads. If so, we recommend that designers and engineers use the more conservative results unless wind tunnel testing done in accordance with ASCE 7-05 demonstrates lower loads.

Winding Up

There is no ideally suited method for calculating wind loads on roof-mounted solar modules in ASCE 7-05. Many methods have been suggested by subject matter experts and analyzed by industry stakeholders. However, every one of these introduces significant shortcomings when comparing the results to the real pressure patterns that have been measured in the wind tunnel.

Several efforts are currently under way that may provide new insights into the problem of estimating wind loads on PV arrays. For example, the Structural Engineers Association of California is expected to publish relevant recommendations soon. The Solar America Board for Codes and Standards has published suggested practices for estimating wind loads on flush-mounted, sloped-roof applications, and has set priorities for follow-up studies and reports.

It is likely that in the next 3–8 years, new tables or figures will be introduced to ASCE 7 that directly address roofmounted solar modules tilted up off the roof. Until then, the DNV Wind Load Calculator allows for the estimation of wind uplift forces on commercial and industrial rooftop PV arrays using values currently in ASCE 7-05.


Colleen O’Brien / BEW Engineering (a subsidiary of Det Norske Veritas) / San Ramon, CA /

David Banks / CPP / Fort Collins, CO /


American Society of Civil Engineers /

DNV Wind Load Calculator / solar_energy/index.asp

Insurance Institute for Business & Home Safety /

Solar America Board for Codes and Standards /

Structural Engineers Association of California /


ASCE 7-05 Minimum Design Loads for Buildings and Other Structures, American Society of Civil Engineers, 2006

Kopp, Gregory, Maffei, Joe, and Tilley, Christopher, “Rooftop Solar Arrays and Wind Loading: A Primer on Using Wind Tunnel Testing as a Basis for Code Compliant Design per ASCE 7,” SunLink,, 2011

Barkaszi, Stephen, and O’Brien, Colleen, “Wind Load Calculations for PV Arrays,” Solar America Board for Codes and Standards,, 2010

Primary Category: 

While indispensible to PV cell and module manufacturing processes, I-V curve tracers have historically held a limited role in the field. Now that the technology is more widely accessible, the role of curve tracers is expanding beyond the laboratory.

Silicon PV modules are highly reliable, but performance problems do arise, and the industry needs fast and accurate ways to detect them. The stakeholders in newly built systems want to verify that all the PV modules are of a consistent quality, that they were not damaged during shipment or assembly, and that the array is producing at the contracted capacity. These stakeholders would also like a permanent record of the as-built system performance, a benchmark for comparison as arrays age and degrade—particularly in cases where warranty negotiations are required. Later in the system’s life cycle, operations and management (O&M) or asset management companies want to evaluate the health of older arrays and have the ability to efficiently locate an ailing module.

These are all potential applications for I-V curve tracers, which can provide both a qualitative visual representation and a quantitative measure of PV performance. Curve tracing equipment was developed for testing transistors and diodes in the semiconductor industry. Now it is a workhorse in PV R&D and manufacturing, for use with both individual cells and modules. It also has a long history of use in field testing of PV arrays, a use that is likely to increase in frequency as more affordable and user-friendly products become available.

In an effort to demystify I-V curve tracers, here I explain how these devices work and how they can be used to commission and troubleshoot PV arrays. The basic characteristics of a healthy I-V curve are described, as well as characteristics that indicate the most common classes of PV array performance impairments. I present rules of thumb for the successful use of I-V curve tracers in the field, which is inherently more challenging than taking measurements in controlled settings like a factory or laboratory. I also provide tips on how to avoid common measurement and data analysis mistakes. When properly attained and analyzed, I-V curve traces provide the most comprehensive measurement possible of PV module or array performance.

I-V Curve Measurements

I-V curves or traces are measured by sweeping the load on a PV source over a range of currents and voltages. Curve tracers accomplish this by loading a PV module or string at different points across its operating range between 0 V and Voc. At each point, the output current and voltage are measured simultaneously. The load presented by the curve tracer may be resistive, reactive (typically capacitive) or electronic. Field test gear uses resistive or capacitive loading, whereas reference I-V test systems at research facilities tend to use electronic loads. The I-V curve may be swept in either direction.

In field test equipment, the actual I-V measurement sweep typically requires less than a second. However, there is a sweep speed limit for certain cell types. High-efficiency cell technologies from Sanyo, SunPower and other manufacturers cannot be swept arbitrarily fast. Because these cells store considerably more charge, more time is required for the cells to reach steady-state operating conditions at each point in the curve. A rough guideline is that the sweep rate for high-efficiency cells should not exceed 10 V per second per cell.


I-V curves, which appear on every PV module datasheet, represent all of the combinations of current and voltage at which the module can be operated or loaded. Normally simple in shape, these curves actually provide the most complete measure of the health and capacity of a PV module or array, providing much more information than traditional electrical test methods.

A normal-shaped I-V curve is shown in Figure 1 (above). The maximum power point (Pmp) of the I-V curve—the product of the maximum power current (Imp) and the maximum power voltage (Vmp)—is located at the knee of the curve. At lower voltages, between the knee and the shortcircuit current (Isc), the current is less dependent on voltage. At higher voltages, between the knee and open-circuit voltage (Voc), the current drops steeply with increasing voltage. The output current of a typical crystalline silicon PV module drops 65% in the upper 10% of its output voltage range. It is not uncommon for an I-V curve to be displayed with its associated power-voltage (P-V) curve, which is also shown in Figure 1. The value of power at each voltage point is calculated using the corresponding current from the I-V curve. The peak of the P-V curve (Pmax), of course, occurs at Vmp.

Fill factor. For given values of Isc and Voc, the powergenerating capability of a PV module or array is related to the squareness of the I-V curve. The two rectangular areas in Figure 2 illustrate this relationship. The more square (or rectangular) the I-V curve, the closer Imp and Vmp approach Isc and Voc, and the higher the output power. This relationship is also described by a figure of merit called the fill factor, expressed mathematically in Equation 1:

FF = (Imp x Vmp) / (Isc x Voc) (1)

A fill factor of 1.0 represents a perfectly square I-V curve, a physical impossibility but a useful reference shape. The two areas in Figure 2, and the numerator and denominator of Equation 1, are all products of current and voltage, with units of electrical power. Any physical effect that reduces the fill factor also reduces the output power of the PV module or string.

Modules with a given PV module part number should have very similar fill factors under similar environmental conditions. Fill factor does vary across cell technologies, ranging from 0.75 to 0.85 in crystalline silicon cells and from 0.55 to 0.75 for most thin-film cells. Fill factor can be also be reduced by several classes of PV impairments, which are described later.

Scaling curves. The I-V curve of a given PV module can be scaled to represent a string or array by simply rescaling the voltage and current axes. A building block analogy, as shown in Figure 3, is useful in troubleshooting PV arrays. When modules are placed in series, the curves are stacked horizontally by adding voltages for each value of current. When modules are placed in parallel, their curves are stacked vertically by adding currents for each value of voltage. The resulting overall I-V curve and max power point are the horizontal and vertical sum of the individual building blocks.

The building blocks can also represent cell strings within PV modules. This view is helpful in troubleshooting situations because, although PV cells are the fundamental unit of production, string voltage tends to be lost in jumps that correspond to the loss of individual cell strings. A step down in an I-V curve may indicate the loss of a building block or at least a reduction in current of one of the building blocks. The width of the step is a clue to how many modules or cell strings are affected by shading, failed bypass diodes or other problems.

Benefits of Curve Tracing

The benefits of curve tracing are substantial. In addition to measuring Isc and Voc, curve tracing also captures all of the operating points in between these values, including the current and voltage of the MPP and thus the maximum power value itself. The overall shape of the I-V curve can be analyzed to give clues to performance issues in ways that traditional test methods cannot. Further, the maximum output-power rating for individual PV modules or strings can be obtained without an inverter or the attendant uncertainties of the individual inverter efficiency.

David King, PV consultant and founder of DK Solar Works, has extensive laboratory and field experience with I-V curve tracers. King worked for 31 years in the solar energy departments at Sandia National Laboratories, where he managed laboratories for testing PV cells, modules and high voltage arrays, as well as overseeing system performance characterization and modeling activities. Based on that experience, King concludes that I-V curve tracing is a fundamental, required measurement throughout the PV industry, both indoors in cell or module manufacturing environments, and outdoors for the testing of modules, module-strings and large arrays.

“I-V curve measurements provide direct performance characterization and verification, as well as a diagnostic tool for periodic PV system performance assessments,” says King. “I-V curve tracing is the most informative measurement that can be performed on a PV module or array. The visual shape of the curve provides immediate diagnostic insight for a PV specialist. When coupled with the associated solar irradiance and temperature data, it provides a quantified comparison to expected performance.”

Though no other diagnostic tool can provide as much relevant information about PV component or system health, today’s commissioning agents and O&M technicians do not often use I-V curve tracers for their periodic performance assessments. According to Andrew Rosenthal, director of the Southwest Technology Development Institute (SWTDI) at New Mexico State University, the high cost of these tools has limited their use in the field.

“Curve tracing has not been more widely used in the industry because of the prohibitively high cost of most curve tracers,” Rosenthal says. He explains that one of the common field applications for curve tracing in R&D is to accurately determine the dc power rating for a PV system. “I-V curve tracing is a valuable tool when an accurate system rating is required,” says Rosenthal. “It is also a valuable tool for system troubleshooting when string or array performance is less than expected.”

Specialists engaged in PV system troubleshooting activities have long required access to I-V curve tracers, regardless of the cost. For example, Bill Brooks, principal at Brooks Engineering, first used an I-V curve tracer in 1988 and purchased a tracer of his own in the early 1990s. Brooks believes that the educational benefits of working with a curve tracer are hard to overstate. “I had the great fortune of learning about PV through the eyes of an I-V curve tracer,” he says. “I consider that education a critical part of my success in understanding and troubleshooting how PV systems operate.”

This educational component is one of the reasons Brooks is excited about the increasing availability of affordable and portable products. Companies like Amprobe, Daystar, EKO, HT-Italia and Solmetric all have curve tracers available in North America that are specifically intended for field-testing applications like PV system commissioning and troubleshooting; even more products are available in Europe.

“Now that these devices are so much more affordable, there is no good reason not to get one,” says Brooks. “For example, the cost for the PVA-600 PV Analyzer from Solmetric is in a range that makes it attractive to any company with a dozen or more employees. Unlike a new truck that depreciates the second it is driven off the lot, a curve tracer is an investment that helps the employees of that company understand their trade better and eventually become experts in their field.”

Isaac Opalinsky, technical trainer at SunPower, has long used curve tracers in training programs as an educational tool to help students understand how real-world effects, such as temperature, irradiance, shading and mismatch, impact system performance.

“Recently, we have been able to measure some of the I-V curves that had previously only been modeled, including scenarios with multiple strings, shading and intentional mismatch, for instance multiple orientations,” says Opalinsky. “It is satisfying for students to see how the things we discuss in the classroom are validated and replicated with real-world measurements. Even if the students are not the technicians who are likely to be performing O&M tasks in the field, an I-V curve tracer that can be used directly by the students can help create the ‘ah-ha!’ moments where abstract concepts are synthesized.”

While it is possible for technicians in the field to get a basic snapshot of system performance and diagnose many field failures using affordable and widely available tools like digital multimeters and clamp meters, Opalinsky notes that I-V curve tracers have two unique advantages. First, I-V curve tracers make it easier and safer to take Isc measurements. Second, curve tracers can reveal what happens to an array under load.

“The four key measurements that can be performed with a digital multimeter (Voc, Vmp, Imp, Isc) are inadequate if we want to get a picture of how the PV system responds to a varying load,” states Opalinsky. “Without disassembling an array, it can be difficult to determine if a perceived problem is just a function of varying environmental conditions, a single bypass diode that has failed or high resistance in a corroded connector.”

PV Array Performance Impairments

In order to identify potential problems using an I-V curve tracer, technicians need to be trained to understand the different classes of performance impairments, as well as the associated curve signature for each. There are five basic classes of PV array performance impairments: series losses, shunt losses, mismatch losses, reduced current and reduced voltage.

Series losses. Losses due to excess series resistance show up in the I-V curve as a decreased slope, or inward tilt, of the curve near Voc. An example is shown in Figure 4 (below). Series resistance effects are equivalent to adding a single external resistor in series with the PV module. The voltage drop across this resistor increases linearly with output current, reducing the output voltage. Since the current change relative to voltage in the I-V curve is much more pronounced near Voc, the decreased slope due to increased series resistance is most apparent at these higher voltages, near Voc.

Series resistance losses can be located internally, inside a PV module, or externally, in the array wiring and switchgear. Dr. Sarah Kurtz, principal scientist and reliability group manager at the National Renewable Energy Laboratory (NREL), notes that the most common cause of PV performance problems is probably increased series resistance. “This is often caused by internal interconnections beginning to crack or break entirely,” she explains. “Most of today’s modules have redundant wiring so that a single break doesn’t stop the current flow, but it still increases the series resistance.” Corroded or poorly connected array wiring can be an external cause of increased series resistance.

A solar module is typically divided into cell strings, each of which is shunted by a bypass diode. If series resistance in a module is very large, sufficient voltage can be developed across the series resistance so that the bypass diode turns on. An example of this effect can be observed when part of a module is shaded. In this case, the shaded cells are no longer predominantly current generators, but instead act as a dissipative resistive element. The bypass diode then shunts the current around the resistive cells, at the price of reduced string voltage and power in the partially shaded string.

Both series and shunt (or parallel) resistances degrade system efficiency by dissipating power. Since power dissipation can occur in a localized region, “hot spots” can develop. This can lead to thermal runaway where the high temperatures lead to greater power dissipation and, in some cases, damage to a PV module. If functioning properly, the bypass diodes help mitigate series resistance effects.

Shunt losses. Losses due to shunt resistance show up in the I-V curve as an increased slope, or downward tilt, of the curve near Isc. This is a region where the I-V curve is ordinarily very flat, if no shunt resistance is present. Shunt resistance effects are equivalent to connecting resistors across PV cells. As the cell voltage increases, the current through this shunt resistor also increases, reducing the module’s output current and power correspondingly.

Shunt losses are located mostly within PV modules and are caused by resistive paths between the cell’s front and back faces. Imperfections in cell material and faulty edge isolation can cause shunt losses. Cells that are cracked or damaged—sometimes when the metallization is added— can also cause leakage.

Mismatch losses. Substantial mismatch effects show up as notches or kinks in the I-V curve, as shown in Figure 5. More moderate effects show up as slope changes in the Isc leg of the I-V curve. The many possible causes include shading, uneven soiling, cracked PV cells, shorted bypass diodes and mismatched modules. Module mismatch can be due to differential aging effects, manufacturing tolerances or the mixing of different modules in the same string. It can also arise from one or more cell strings cutting out due to shading, bypass diode failure or the triggering of bypass diodes by other module level issues.

Reduced current. Reduction in the height of the I-V curve can be caused by uniform soiling, edge soiling (common in low-tilt, portrait-mode arrays), PV module degradation or weather conditions that reduce the input irradiance. Soiling directly impacts the height of the curve because it reduces the incident irradiance.

Reduced voltage. The width of the I-V curve is affected by module temperature. Poor air circulation, for example, can raise the module temperature and substantially reduce Voc and Vmp. Module degradation, shorted bypass diodes and other system problems can also reduce Voc and Vmp. The width of the I-V curve is relatively insensitive to normal soiling.


Each of the impairment classes described has a characteristic I-V curve signature, as summarized in Figure 6. The reduced current and reduced voltage impairment classes affect the height and width of the I-V curve. The other three impairment classes affect the overall shape of the I-V curve. Excess series resistance, decreased shunt resistance, and mismatch cannot be detected by simple open-circuit voltage measurements or clamp meter current measurements of individual strings. I-V curve tracers, however, provide a window into these failure modes that allows the PV technician to verify performance quickly and spot problems early.

“Some failure or degradation mechanisms cause internal changes to cells that cannot be seen with the naked eye,” explains NREL’s Kurtz. However, it may be possible to see the effects of these changes in I-V curve traces. She continues, “These changes may increase the series resistance, decrease the voltage and/or current, or may cause some shunting that makes the flat part of the I-V curve slope somewhat, decreasing the height of the knee, and, therefore, the output power.”

PV Performance Verification Process

Because field testing PV arrays requires working on or around energized circuits, personal protective equipment is required. It is also necessary to observe proper safety procedures, such as lockout-tagout. Read and follow NFPA-70E, Electrical Safety in the Workplace, for more information (See Resources). The process of isolating circuits for measurement varies depending on the details and scale of the installation.

Residential systems. In residential systems, the PV output conductors may land on terminal blocks in an inverterintegrated disconnect switch or in the inverter itself. While curve tracers can be connected to live PV strings as long as the measurement process is disabled, lifting live PV circuit conductors can be dangerous. It is advisable that residential system designs always include either fuses or dc disconnects that enable the isolation of individual strings from the inverter and from each other. PV systems require checkups and servicing as they age, so it makes sense for system components to include the means for isolating and connecting to individual PV strings quickly and safely. Some inverters, particularly those with fused dc inputs, already provide this capability.

Commercial systems. In commercial PV systems, curve tracer measurements are performed at the combiner box, as shown in the photo to the right. The combiner is isolated from the rest of the array and the inverter by opening its dc disconnect switch. Then the combiner box is opened and all the touch-safe fuses are lifted. Once the busbars are de-energized, the curve tracer’s test leads are clipped onto the busbars. Fuses are inserted one at a time for measurement of individual strings. Once an I-V curve is captured, it can be saved electronically. Some I-V curve tracers, like the Solmetric PVA-600, also allow the measured I-V curve to be compared via integrated software to a model I-V curve. The entire process typically takes 10–15 seconds per string.


Verifying PV array capacity requires a standard of comparison, regardless of the testing equipment used. The standard of comparison may be a contracted power value or the prediction result from a PV array model. In the case of commercial-scale PV systems, performance verification test limits and even the test equipment itself are often specified in the performance guarantee contract. In troubleshooting situations, the standard for comparison is often a neighboring PV string.

The most common standard for performance verification measurements in the field is the nameplate specifications for the PV module. Since these apply at STC, measured I-V parameters must be translated to an irradiance of 1,000 W/m2 and a cell temperature of 25°C. Curve-tracing instruments, however, can use PV models to predict the expected I-V curve shape. This allows the user to instantly verify performance or diagnose problems by looking for deviations between the measured curve and the expected curve predicted by the PV model or models.

Several types of PV performance models are commonly used for estimating array capacity. These models describe the performance of PV modules, strings and arrays. The three performance models most often used in the PV industry are (in order of most to least detailed): the Sandia PV array performance model, the 5-parameter model and the singlepoint efficiency model. These models are provided in NREL’s Solar Advisor Model (SAM) simulation software. Assuming data for a PV module is available, the first two of these models can be used to generate a predicted I-V curve, given sufficient detail about system components, array orientation and environmental conditions. This makes them ideal candidates for predictive models built into curve tracing equipment. The third model predicts the maximum power value and is a good backup when a PV module is not represented in either the Sandia or 5-parameter model databases.

Sandia model. The Sandia PV array performance model was developed by David King and his co-workers at Sandia National Laboratories in Albuquerque, New Mexico, and features over 30 parameters representing irradiance and temperature dependence, spectral response, angle of incidence and other effects (see Resources). It is the most descriptive of the three PV performance models and has the greatest potential to benefit the PV industry. According to Richard Bozicevich, VP of business development for TÜV Rheinland PTL in Pheonix, Arizona: “Applications for the Sandia model include system design and sizing, translation of field performance measurements to standard reporting conditions, system performance optimization and real-time comparison of measured versus expected system performance.”

The Sandia model database now contains parameters for more than 500 PV module model numbers. Under contract with Sandia and the US Department of Energy, TÜV Rheinland PTL has developed the in-house capability for measuring Sandia model parameters. Regarding the status of the technology transfer, Bozicevich reports that model validation is completed, and TUV Rheinland PTL has full capabilities to execute testing to the model for client samples.

“Top-tier manufacturers are starting to request the testing and are using the data to analyze their module performance,” according to Bozicevich. “However, the likelihood that these manufacturers will release the parameters to the Sandia database remains an open question.”

5-parameter model. The 5-parameter PV performance model was developed at the University of Wisconsin Solar Energy Laboratory. It is used by the California Energy Commission (CEC) to simulate PV system performance for its New Solar Homes Partnership. Patrick Saxton, senior electrical engineer at the CEC, reports that as of May 12, 2011, the CEC 5-parameter model database contains parameters for more than 4,900 PV module model numbers.

“The database has been growing at the rate of 300-400 new model numbers a month for at least the last year,” Saxton says. “Some fraction of these may be repeat entries for private labels. Model parameters are generated at third-party testing facilities using a single sample module, often at the time of UL 1703 certification.”

Single-point model. The single-point efficiency model predicts the maximum power value based on parameters normally listed in the PV module datasheet. The calculations for this method are familiar to installers who have used datasheet parameters to translate the maximum power value of a PV system to standard test conditions, or vice versa. (For more details on I-V parameter translation, see “PV System Commissioning,” October/November, 2009, SolarPro magazine.)

Translation of measured I-V curve data to STC conditions always introduces error. The magnitude of the error increases with the difference in irradiance or temperature that is measured versus the STC conditions. A way around this limitation is to use the Sandia model to predict the shape of the I-V curve and the values of the key performance parameters, taking into account instantaneous irradiance and temperature. This approach allows for immediate, high-quality assessments of string performance in the field.


Analyzing PV array performance data always involves comparison of the results with a specification or model, and may involve detailed analysis of variations of I-V curves across the population of strings. The shape of a measured I-V curve gives important clues to the causes of performance problems. Combined with a predictive PV performance model, I-V curve traces provide the most complete picture of the electrical health of a PV module, string or array.

While commercial-scale PV arrays yield huge amounts of string-level performance data, automated measurement, data collection and analysis can be employed to increase throughput and reduce operator fatigue and data recording errors. Further, automated analysis tools quickly summarize results and make it easy to spot nonconforming strings.

Figures 7a, 7b and 7c show three automated analysis tools for a particular array. The table view, shown in Figure 7a, lists the key performance parameters extracted from the measured I-V traces. These include the familiar Isc, Imp, Vmp, Voc and Pmax values, and also fill factor and the current and voltage ratios that represent the slopes of the lower and upper voltage legs of the I-V curve. If a fillfactor value is out of line, the current and voltage ratios give hints as to whether series or shunt resistance effects may be involved. Statistics for each column are indicated, including the spread of the values. The user can define the acceptable range of values; out-of-range cells in the table are shaded yellow to identify the outlying string.

The I-V curve overlay graph, Figure 7b, gives a quick visual indication of the I-V curve consistency across strings. Figure 7c shows distribution plots or histograms that provide insight that simple statistical parameters such as max, min, mean and standard deviation do not. The shape of the distribution plot can indicate whether the spread or deviation measured is the result of random module performance or environmentsensing variations, a problem with the measurement setup, or even the outcome of more systematic effects. For example, the performance verification data for one commercial rooftop array showed an unusual distribution of string Voc data. Further analysis led to the discovery of a large temperature differential between strings at the edge of the array compared to strings located away from the edges where air circulation was limited. This is common in large rooftop arrays where modules are packed in a tight formation. Field Applications for I-V Curve Tracers The main field-testing applications for I-V curve tracers are system commissioning, routine operations and maintenance, and troubleshooting performance problems. Benchmarking system performance is an important aspect of system commissioning and acceptance, and it is valuable whenever PV performance guarantees are used. Once a performance benchmark is established, taking routine I-V curve traces can make preventative maintenance activities more meaningful for array operators. In the event that unscheduled maintenance is required, I-V curve collection and analysis can help to quickly pinpoint problems.

SunPower’s Opalinsky believes that companies engaged in these activities should consider taking curve traces. “Anybody involved in the commissioning of PV systems—either as system owner, integrator or third-party commissioning agent—should consider I-V curve traces as a method of benchmarking system performance at the time of startup and for verifying performance in the future,” Opalinsky says. “Companies involved in maintaining and operating PV systems should consider having at least one person on staff who is trained to use an I-V curve tracer and interpret the results.”


Developers, PPA financiers, engineering, procurement and construction (EPC) contractors, and providers of O&M services all have a strong interest in verifying and optimizing the performance of a solar asset. Each stakeholder stands to benefit from a test measurement method that provides deep insight into PV system operation and potential problems.

By employing proper performance measurements in solar PPA projects, financial risk can be reduced and ROI increased. When the developer and PPA financiers want to be sure that a system is fully functional and operating optimally, they can require a complete commissioning report that includes the measurement of I-V curves for every string. Any deviations of actual performance from expected performance beyond some agreed threshold are then corrected before funds are released to the EPC contractor.

For its part, the EPC contractor establishes a baseline of data that can be used in the future if performance questions or contract disputes arise. By curve tracing each string and demonstrating that the system is fully functional at the time of commission, the EPC contractor can prove that it has met the installation electrical performance verification portion of its contractual obligations. The archived data is then referenced if there are performance issues in the future.

Performance verification is typically required by contract as part of the commissioning of new commercial systems, and it is likely to become commonplace for systems of any size, including residential, in the future. Recommissioning is appropriate at other points in the PV system’s life cycle, including a change of ownership, a trauma to the PV system (lightning strike, extreme wind, theft and so on) and array removal and replacement for reroofing.

Traditionally, performance verification of commercialscale arrays involves measuring and recording the string open-circuit voltages, as well as the string operating currents at the overall system MPP as determined by the inverter. The short-circuit current may also be measured. The drawbacks to this traditional approach are that the individual measurements and data recording take considerable time; they are limited in the performance issues they can identify; and they do not make independent maximum power measurements of each string.

I-V curve tracers overcome these limitations by integrating and automating the measurements and data recording, and by revealing all the performance issues. For example, open-circuit voltage measurements and clamp meter current measurements cannot detect excess series or shunt resistance, or module mismatch in a string, whereas curve traces can.


Troubleshooting may be triggered by a system owner’s complaint of poor production, an alarm thrown by a monitoring system or by observations made during a routine checkup. The technician may turn to an I-V curve tracer after reading the inverter display and checking dc voltages and currents with a digital multimeter (DMM) and clamp meter. In troubleshooting situations, compared to using a DMM or clamp meter, an I-V curve tracer can provide far greater detail in the data that it reveals and the records it keeps of performance before and after the repair. If a module warranty return is in order, curve tracing provides the most complete documentation.

The first step in troubleshooting with a curve tracer requires no PV model or reference standard, but only the measurement of a string’s I-V curve. Once the trace is complete, consider whether the curve has a normal shape. If it does not—if there are steps or notches in the curve—consider the following:

  • Is any of the string shaded, even a fraction of a cell?
  • Is substantial, uneven soiling from birds, dirt dams, lichens or tree litter present?
  • Are there burn marks on the front or back faces of the modules?

Comparing the curves of two or more strings is a good way to spot more subtle effects, like a softer knee in an I-V curve or reduced short-circuit current or open-circuit voltage. For a more objective test, translate the I-V curve to STC and compare key parameter values to the specifications on the modules’ product data sheet. For the most reliable and accurate test, compare the shape of the trace with the predictions of an onboard PV performance model such as the Sandia model or 5-parameter model.

Fill factor is an effective initial screen for any performance problems that show up as subtle changes in the slopes or softness of the knee in an otherwise normal looking I-V curve. If the fill factor is low, check the Imp/Isc and Vmp/Voc ratios relative to neighboring strings or the predictions of the PV model.

Series resistance signature. Reduced Vmp/Voc may indicate increased series resistance. The I-V curve signature for series resistance is a reduced slope or inward tilt in the leg of the curve between Vmp and Voc. Figure 8 shows an example of this condition. I-V curves taken for two adjacent strings showed a noticeable difference in series resistance. Further investigation showed that the source of the extra resistance was a single module that had several fingerprintsized burn marks scattered along several cells.

Increased series resistance does not always show physical signs on the module face or backsheet. Therefore, it may be necessary to successively break the string in two, using the half-splitting method to zero in on the damaged or degraded module. In this technique, a poor performing string is split into two substrings, and each substring is measured. Then the poor performing substring is again split and measured until the problem becomes obvious or the substring is reduced to an individual module.

Shunt resistance signature. Shunt resistance effects show up as increases in the slope of the leg of the I-V curve near Isc, but shunt resistance is not the only possible cause of this increased tilt. Tapered edge soiling (dirt dams) or slight shading that tapers gradually across a row of modules can produce a similar change in slope with no apparent bypass diode action.

Reduced current signature. If the I-V curve has a normal shape and width, but the Isc is lower than predicted by the PV model, check first for uniform soiling. Depending on the purpose of the testing, you may need to clean the array.

An accurate way to demonstrate the impact of uniform soiling is to measure the I-V curve before and after cleaning and compare the maximum power values. Do the test under clear sky conditions close to solar noon, so that the irradiance is constant. Measure I-V curves for two neighboring strings. One of these strings will be cleaned as part of the test; the other serves as a control to remove the effect of any irradiance changes. After cleaning one string, measure I-V curves for both strings again and observe how much the cleaning affected Isc and Pmax in the test string. If the control string showed changes as well, use these changes to correct the before and after results in the test string for a more accurate comparison.

Certain module failure modes may also reduce module current. NREL’s Kurtz describes two examples: “Especially for older modules deployed in hot, humid locations, some browning of the encapsulant may be visible, causing a somewhat decreased current but probably an undetectable change to the voltage. In addition, delamination can slightly decrease the current because of the reduced coupling of the light into the cell; in the long term, however, delamination can lead to corrosion and, eventually, catastrophic failure.”

Reduced voltage signature. If the I-V curve of a single string has a normal shape and height, but the Voc value appears to be low, calculate the difference between the measured and expected Voc or translate to STC. Note that this comparison is done automatically when a curve tracer with an integrated PV model is used. If the difference happens to be close to the Voc of a single module, then a module may be missing from the source circuit—perhaps bypassed or not wired in. If the difference is smaller than the Voc for a single module, one or more cell strings within the modules may be bypassed or not functioning properly.

Bypass diodes sometimes play a role in PV module failures, particularly when modules are designed using inexpensive or undersized bypass diodes. “Frequent partial shading can cause a bypass diode to operate constantly, shortening its life,” Kurtz observes. She notes that shade prone rooftop applications are particularly troubling in this regard. “The bypass diodes are stressed most when the module is partially shaded,” she adds. “If they overheat, they may burn out.” Array installation methods, such as direct installation on a roof, can also cause overheating.

Mismatch signature. Shading, although not a problem caused by the array hardware, provides a good example of mismatch behavior. The shaded cell produces less current. If the shading is severe enough, the bypass diode spanning that cell string turns on and shunts current around it. The I-V curve shows a step on its falling slope, the width of which corresponds to that cell string’s open-circuit voltage. The reduction of current at the step is proportional to the cell area that is being bypassed. Figure 9 illustrates the shading of different numbers of modules in two parallel strings. The normal-shaped curves correspond to no shading or to equal numbers of modules shaded in each string.

Unlike solar thermal collectors, the dependence of PV production on the actual pattern of shade is very nonlinear. A simple outdoor lab setup demonstrates this. A source circuit consisting of two PV modules is shaded with a rectangular piece of cardboard large enough to cover two adjacent cells, as shown in the photos above. The modules have 72 cells each, split across three bypass diodes. The setup and results are shown in the photos and associated I-V curves. In the first example, the cardboard covers two cells in the same cell string, causing its bypass diode to conduct and dropping the string’s voltage and output power by roughly one-sixth. In the second example, the cardboard is rotated to cover one cell in each of two adjacent cell strings, dropping the voltage and output power by twice that amount.

Taking Environmental Measurements

Accurate array performance verification requires careful selection and measurement of environmental conditions. The shape of an I-V curve taken in the field is determined in part by the irradiance in the plane of the array (POA) and the cell temperature at the time of measurement. Therefore, POA irradiance and cell temperature are often collected simultaneously with I-V curve measurements in the field. No measurement is exact. Random variations and systematic bias combine to create some level of uncertainty. This uncertainty is a function of the test equipment, the environmental conditions and the user’s measurement technique. However, through the proper use of appropriate irradiance and temperature sensors and careful screening of sky conditions, both random and systematic errors can be reduced.

Irradiance measurement. This measurement is used to determine the irradiance in the plane of the array at the time of the trace. Good irradiance measurements can be obtained by selecting a high-quality sensor that uses a technology similar to that of the array being tested and is designed for backside mounting. To ensure that the sensor is mounted in the plane of the array, attach the sensor to a bar that is in turn clamped to the frame of a PV module.

If the irradiance sensor is not mounted in the plane of the array, it presents a different area to the sun; this is a key source of irradiance measurement error. Reflected light is another potential source of measurement error. Be aware of possible sources of reflected light and try to locate the irradiance sensor at a location that is representative of normal operating conditions for the array.

While handheld irradiance sensors can be used in I-V curve testing, they are usually difficult to accurately position in the plane of the array. In addition, the sensor technology and packaging are often quite different from the PV modules themselves. This can introduce spectral and angle of incidence errors. The angle of incidence is the angle between an incident light ray and a line that is normal (perpendicular) to the PV module. The angle of incidence response of the irradiance sensor and the PV modules under test should be reasonably well matched.

Changing irradiance conditions is another potential source of error. Using an I-V curve tracer capable of making rapid curve sweeps can minimize these errors. Many curve tracers also offer optional sensor kits. When these kits are used, the curve tracer logs the irradiance measurement simultaneously with the I-V curve. Manual measurement and recording of irradiance and temperature can more than double I-V curve measurement time, and the delay between sensor readings and the I-V sweep can introduce significant random error. The preferred technique is to use a curve tracer with sensors that are triggered and recorded at the same time as the I-V curve sweep.

Temperature measurements. Temperature measurement for performance verification usually involves a thermocouple or resistive temperature device taped to the backside of the PV module. The sensor should be placed toward the center of the module, as the edges tend to run cooler than the rest of the array. High-temperature tape, applied with firm pressure, assures good thermal contact between the sensor element and the module backsheet.

Digital infrared (IR) thermometers are sometimes used for this purpose, but their accuracy is very dependent on the emissivity of the surface. Calibration of the IR thermometer can be accomplished using a side-by-side measurement of the same PV cell using the IR and thermocouple methods. The emissivity control on the IR device can then be adjusted to make the two temperature readings match. IR temperature measurements are typically less accurate when taken through the face of the module than when taken off the backsheet.

Using the Array as a Sensor

Array performance verification measurements, such as those taken during system commissioning, normally require the use of irradiance and temperature sensors so that these parameters can be recorded along with the I-V curves. However, much of the diagnostic testing can be done without use of external sensors. Another option is to calculate the temperature and irradiance from the measured I-V curve itself, using the array as a sensor. The mathematical basis for this feature combines formulas from the Sandia PV Array Model and the IEC 60904-5 standard on determining equivalent cell temperature (see Resources). The translation method relies on knowing the dependence of Isc and Voc on irradiance and temperature. Irradiance is calculated from Isc with a slight correction from Voc. Temperature is calculated from Voc with a slight correction from Isc.

“Array as sensor” is an optional operating mode in the Solmetric PVA-600 that can simplify the process and save time when readings from external sensors are not required. When the array-as-sensor method is used to provide irradiance and temperature values to a PV model, the predicted I-V curve is forced to align with the measured curve at Isc and Voc. Thus, the array-as-sensor approach is blind to the effects of uniform soiling and to degradation in Isc or Voc. However, the method is very helpful when examining the shape of the I-V curves. Since the endpoints of the predicted and measured I-V curves coincide, any deviation in the shapes of the curves is very easy to spot. The sensor values are effectively measured at the same time as the I-V curve, so comparing measured and predicted curve shapes is much less affected by wind and by rapid changes in irradiance. The determined temperature also represents the string as a whole, not just the temperature at the edge of the array where an actual sensor would be attached.

The temperature across a string can vary ±15°C because of different exposure to wind, reflections and racking. In many cases, therefore, it is actually more accurate to use the array-as-sensor mode to determine the temperature because it measures the average cell junction temperature across the entire string rather than one specific spot measurement on the back of a single module.

The array-as-sensor mode is also useful for checking the I-V curve shape of basic functional modules, particularly when deploying a full sensor kit is not practical. The user can also mix sensor modes, using the array-as-sensor mode to determine temperature, while capturing irradiance with an external sensor mounted in any open area at the same orientation as the array. The temperature will be reasonably accurate as long as all of the modules and cell strings are operating. This can be assured by checking that Voc is roughly consistent across strings.


Because a PV source responds to changing environmental conditions, verifying PV performance in the field is potentially challenging. DK Solar Works’ King explains, “PV module and array performance in outdoor conditions is continuously changing due to a large number of factors, including variations in solar irradiance level and spectral content, ambient temperature, wind speed, thermal heat capacitance of the modules themselves, module shading, soiling and so on.” Taking performance verification measurements under the recommended environmental conditions helps give consistent results when remeasuring the same site at a later date.

The rule of thumb espoused by Dr. Jennifer Granata, technical lead of the PV Test, Evaluation and Characterization group in the Photovoltaic and Grid Integration Department at Sandia National Laboratories, is to gather performance test data in a stable environment. “The ideal is to test during clear sky conditions with a stable irradiance level, stable spectrum and stable temperature, including wind effects,” she states. “This usually occurs in the 4-hour window centered at solar noon.”

Determination of irradiance typically has the most significant impact on the accuracy of PV performance measurements. The direct radiation component of sunlight is larger during the 4-hour window around solar noon. Since direct radiation measurements tend to be more repeatable than measurements of the diffuse radiation component, test results during this window tend to be more accurate and repeatable. The proportion of diffuse radiation is lower around solar noon than it is at other times of the day.

Temperature measurement errors also affect the results. Windy conditions cause rapid variation of array temperature. More importantly, wind—even a steady wind—can change the pattern of temperature across the array, making measured string performance look less consistent. Testing in a wind speed of 2 mph or less is a good guideline for PV performance measurements.

Brooks agrees with the necessity for stable conditions. He states: “Ideally we always want a cloudless sky with no variations or jet contrails. This is rare for most of the US, so we normally have to compromise. It seems like clouds follow curve tracers, but the key is to take I-V curves when both the irradiance and module temperature are stable. If either one is moving at the time of the I-V curve, the data is going to be dubious. If we are just trying to get a ballpark shot for simple commissioning purposes, slight changes are okay. Temperature changes, since they tend to be much slower than irradiance changes, are more tolerable. Also, temperature changes have a much smaller impact on the curve, so the data error may be small. Irradiance changes of more than 1% or 2% while the curve tracer is measuring the data results in bad data. A 10% change results in a curve that looks like a major malfunction is present on a perfectly operating array.”

While it is not always possible to avoid clouds, some clouds are worse than others. Large, slow-moving clouds located a significant angle from the sun in an otherwise clear sky contribute some additional irradiance from cloud effect, but this variation may be slow enough to be corrected for by the sensors. According to Bill Sekulic, master research technician at NREL’s PV Performance and Reliability R&D group, “Large or spotty cumulus clouds located at fairly large distances from the sun are generally not an issue while taking curves.” However, if there are fast-moving clouds near the sun, performance measurements should be postponed. Cirrus cloud cover is another showstopper, reports Sekulic. “Cirrus clouds cause irregular variations in irradiance, as well as a magnification of irradiance called cloud effect,” he says. “Because cirrus clouds usually occur at high altitude, they can give an appearance of clear sky conditions that masks irradiance irregularities and cloud effect magnification.”

Air mass (AM) 1.5 is one of the standard test conditions under which PV modules are specified. The earth’s atmosphere affects the power spectrum of sunlight, and at AM 1.5 the atmospheric path length is 1.5 times more than it would be at sea level with the sun directly overhead. Sandia National Laboratories’ Granata warns: “Although modules and arrays are rated under the AM 1.5 spectrum, the spectrum and irradiance can change rapidly as the sun moves through the AM 1.5 position, depending on location and time of year.”

It is also important to recognize that PV module I-V curves change shape as light levels change. It is difficult to accurately extrapolate an I-V curve at STC from a trace taken at low irradiance levels. Granata recommends taking field measurements under conditions that are close to the reference condition: “Another aspect of choosing the conditions for testing an array is how one intends to normalize the data. If normalizing to standard reporting conditions or PTC, being as close to those conditions as possible is recommended to minimize uncertainties when translating the data.”


Paul Hernday / Solmetric / Sebastopol, CA /


California Energy Commission, New Solar Homes Partnership /

National Renewable Energy Laborary Solar Advisor Model /

NFPA-70E, Electrical Safety in the Workplace, National Fire Protection Association /


“Photovoltaic Array Performance Model,” D.L. King, W.E. Boyson, J.A. Kratochvil, Sandia Report: SANDIA2004—3535, 2004

“Photovoltaic Devices—Part 5: Determination of the equivalent cell temperature (ECT) of photovoltaic (PV) devices by the open-circuit voltage method,” International Electrochemical Commission, International Standard: IEC 60904-5, 1993


Amprobe / 877.267.7623 /

Daystar / 575.522.4943 /

EKO / 408.977.7751 /

HT-Italia (US distribution by Hukseflux) / 631.251.6963 /

Solmetric / 877.263.5026 /

Primary Category: 

With too few modules in series an inverter cannot maintain an array’s MPP under high temperature conditions for the site, sacrificing energy harvest. Too many modules in series results in voltages above 600 Vdc, which can damage equipment, violate the NEC and void the manufacturer’s warranty.

Matching array output to inverter input is a critical step in PV system design. The primary goal of matching an array to an inverter is to ensure that the inverter can capture a high percentage of the available energy that the array produces during all of the environmental conditions anticipated at the site. Often a secondary goal is to maximize the inverter capacity so that the inverter will operate at or near full power during high irradiance periods without power limiting. It is important that power limiting occur only during exceptional or transitory conditions, not under normal operating conditions.

The array in a typical grid-direct PV system consists of one or more strings of five to 20 modules wired in series. The exact number of modules - the number in series and the number of strings - depends on the electrical characteristics of the module, the input voltage and current range of the inverter, and the expected high and low ambient temperatures of the site. In a well designed system, the array’s operating voltage, current and power output will be within the inverter’s operating range at all times.

Inverter manufacturers typically provide string sizing guidelines or online programs to assist in matching a particular array configuration to a specific inverter. The system designer needs to provide the record low and average high temperatures for the site. Some online string sizing calculators generate detailed results that integrators can use to optimize their designs. However, the main function of these programs is to calculate the maximum and minimum number of modules in series, providing designers with a range of acceptable array configurations.


The necessary calculations can also be done manually. This is an important skill to learn, especially for designers. Building integrated products are seldom included in string sizing calculators, and new products on the market may not immediately be added to the inverter manufacturer’s online calculators. On the roof, the ability to manually verify array configurations can avoid costly mistakes. Without wireless laptop connections, crews in the field cannot rely upon online string sizing tools. But solar professionals can, and should, master the steps detailed in this article.

To illustrate how to calculate these configurations manually, the following example assumes:

  1. a rooftop PV array mounted at the plane of the roof and elevated by 3–4 inches;
  2. an environment with an ambient temperature range of 0°–45°C;
  3. a 7 kWac inverter with an input voltage range of 250–600 Vdc, a MPPT range of 250–480 Vdc, a maximum dc input current of 30 amps and a California Energy Commission (CEC) weighted efficiency of 96%; and
  4. a crystalline PV module with the specifications listed in Table 1.


Maximum input voltage for an inverter is a hard stop design limit. Exceeding the maximum inverter operating voltage can result in catastrophic failure of the inverter and could, in some cases, result in NEC violations. Therefore, the maximum open circuit voltage is the most critical value to consider when designing a PV array.

The number of modules in series determines the array open-circuit voltage (Voc). Because voltage is temperature dependent, Voc must be temperature corrected in order to calculate the maximum PV system voltage (see 2008 NEC, Table 690.7 to the right). This temperature corrected Voc will determine the maximum number of modules allowable per series string. Conservative designers will use the record low ambient temperature at their site as the cell temperature for these calculations.

A good first step for designers is to determine the temperature differential for their site, in this case the low temperature differential. In our example the record low temperature for the site is 0°C, which is a differential of -25°C from Standard Test Conditions (STC). As cell temperature drops, output voltage increases. This effect is described by the appropriate temperature coefficient, which in this case is given as a percentage of Voc.

In our example, we have a low temperature differential of -25°C and a -0.34 %/°C temperature coefficient of Voc. The maximum module voltage (Voc_max) is then calculated as:

Voc_max = Voc + (temp differential x temp coefficient of Voc)
                = 32.9 V + ((T_min – T_STC) x (-0.34%/°C x Voc))
                = 32.9 V + ((0°C – 25°C) x (-0.34%/°C x 32.9 V))
                = 32.9 V + (-25°C x -0.1119 V/degree C)
                = 32.9 V + 2.80 V
                = 35.7 Vdc per module

The maximum dc input voltage for the inverter we have selected is 600 Vdc. The maximum number of series wired modules (Nmax) must be less than or equal to 600 Vdc divided by the maximum module voltage. This is calculated as:

Nmax ≤ 600 V ÷ 35.7 V
          ≤ 16.81
          = 16

A quick cross check of Table 690.7 for minimum temperatures of 4°–0°C yields a voltage correction factor of 1.10. The maximum allowed STC open-circuit voltage allowed by the NEC is therefore 600 Vdc ÷ 1.1 = 545 Vdc. With a module Voc_STC of 32.9 Vdc, the maximum number of modules in series allowed by Table 690.7 is 545 Vdc ÷ 32.9 Vdc, or 16.57. The example 16-module series string configuration meets NEC Table 690.7 requirements using both the manufacturer supplied temperature coefficient of Voc and the table correction factor.


Excessively low array voltage can have a dramatic negative impact on PV system energy production. If the array voltage falls below the minimum operating voltage, the inverter may not be able to track the array maximum power point. If this low voltage condition occurs rarely, such as on extremely hot afternoons only, the overall impact on energy production will be negligible. With extremely low array voltage, however, the inverter may shut down completely on hot days, which would result in substantial energy loss and very poor system performance.

The minimum operating inverter voltage is calculated using the module’s temperature corrected maximum voltage (Vmp_min) at the average high ambient temperature. In order to estimate the actual cell temperature, it is necessary to factor in the temperature rise (T_rise) resulting from the actual mounting conditions. In this case, T_rise is assumed to be 30°C based on empirical data for PV arrays mounted close to the roof surface. Other mounting methods that provide for better airflow around the array may result in a T_rise value of between 20°C and 25°C.

The temperature corrected, minimum expected module maximum power voltage is calculated as:

Vmp_min = Vmp + (temp differential x temp coefficient of Vmp)
                = 27.1 V + ((T_rise + T_max – T_stc) x (-0.47%/°C x Vmp))
                = 27.1 V +((30°C + 45°C - 25°C) x (-0.47%/°C x 27.1 V))
                = 27.1 V + (50°C x -0.1274 V/degree C)
                = 27.1 V – 6.37 V
                = 20.73 Vdc per module

The inverter’s minimum input voltage range in this example is 250 Vdc. The minimum series string must be greater than or equal to 250 Vdc divided by the Vmp_min. This is calculated as:

Nmin ≥ 250 Vdc ÷ 20.73 Vdc
         = 13

A string with 12 modules in series will have a minimum expected module maximum power voltage of 12 × 20.73 Vdc, or 249 Vdc. While close to the inverter’s minimum MPPT voltage of 250 Vdc, this expected value falls just outside the tracking window for the device. This suggests that 12 modules in series is a marginal design for this location. After the deleterious effects of other array derate factors are considered, this configuration would likely be determined unacceptable.


The maximum number of parallel strings that can be connected to the inverter without causing current limiting can usually be determined by a simple calculation without the need for temperature correction. To calculate the maximum number of parallel strings, divide the maximum inverter input current by either the module Imp or Isc, depending upon what the inverter O&M manual specifies. A maximum short-circuit input rating is not specified for our example, so we will use the maximum power current for our calculations.

The maximum number of paralleled series strings should be less than or equal to the maximum inverter input current divided by maximum power current at STC. This is calculated using the equation:

N ≤ 30 A ÷ 7.2 A
N ≤ 4.2
N ≤ 4

Using the voltage and current constraints calculated here, the designer can determine that the possible array configurations for this inverter consist of one to four strings of modules in parallel with each string made up of 12 to 16 series connected modules.


The last step in determining the array to inverter match is to examine the array maximum power. Some array configurations that do not exceed the voltage or current constraints could still result in power levels that exceed the inverter’s output power capabilities. If this happens, valuable solar- generated energy will be dissipated into the environment as waste heat or otherwise lost when the inverter is forced into power limiting.

The analysis and optimization of the array to inverter power match can be quite complex. Some of the variables involved include module mounting method, array performance in real world conditions, inverter efficiency versus input voltage, inverter efficiency versus output power and the statistical distribution of irradiance versus ambient temperature at the site. A full optimization is beyond the scope of this article, but array to inverter power match can be done reasonably well with some basic assumptions.

One way to match array input to inverter output power is to consider the array’s PVUSA Test Condition (PTC) rating. Manufacturers specify module power as STC ratings, but real world module performance is typically lower than would be projected. PTC ratings better reflect real world performance. In most cases, matching the PTC rating of an array to the inverter’s continuous output rating at 40°C is a perfectly reasonable design decision for maximum array generating capacity. The CEC maintains an online database of eligible equipment. The database, which includes PTC ratings of PV modules and weighted average inverter efficiency, can be found on the Go Solar California Web site (see Resources). The PTC rating for the module we are considering is 173.3 W. If this information were unavailable, a fixed module power derating factor of 0.90 × Pmp_STC could also be used. This fixed factor will tend to overestimate module output and therefore result in a conservative system design. Using the fixed factor of 0.9 in our example module would yield a power rating of 175.5 W. While this fixed factor guesstimate can be used for design purposes, it should not be used to estimate actual system performance.

After matching an array’s PTC rating to the inverter output power, consider factoring in the inverter’s CEC weighted efficiency. Because some of the array output power is lost in the inversion process, it is often acceptable to increase the array generating capacity to take this into account. Inverter output power in this case should be less than or equal to the array PTC rating divided by the inverter’s CEC weighted efficiency. In most climates, an array sized this way will leave very little inverter capacity on the table. Furthermore, the high irradiance conditions that could lead to inverter clipping are transitory, brief in duration and generally seasonal. The equation for calculating the maximum number of modules in this manner is:

Inverter Power ≤ N x PTC x CEC weighted efficiency
          7,000 W ≤ N x 173.3 W x 0.96
                     N ≤ 7,000 W ÷ 173.3 W ÷ 0.96
                     N ≤ 42 modules

Actual full power inverter efficiency is typically slightly lower than the weighted CEC efficiency. In addition, voltage drop, module mismatch and array soiling could result in other system losses. Array installation details are also relevant. A flat roof installation, for example, is ideally oriented to the sun for summer energy harvest, but high cell temperatures significantly derate peak power output. Cold and clear conditions at other times of the year are unlikely to produce theoretical high peak power outputs, because the sun’s incidence angle would not be ideal. This is a case where oversizing the array in the example may be acceptable. A further derate factor of 0.95 can be used to estimate the combined effect of these additional factors. This is meant to give a better real world array to inverter match and does not account for any limitations imposed by local rebate programs. This estimate is calculated as:

Inverter Power ≤ N x PTC x CEC weighted efficiency x 0.95
          7,000 W ≤ N x 173.3 W x 0.96 x 0.95
                     N ≤ 7,000 W ÷ 173.3 W ÷ 0.96 ÷ 0.95
                     N ≤ 44 modules

Using all of this information, the valid array configurations showing the total number of modules, the maximum estimated output power (Pac_out) and inverter utilization values are summarized in Table 2. Inverter utilization factors below 80% generally indicate that a different inverter might be more appropriate. In some cases, however, the customer may be planning a future expansion of the array, so a 1- or 2-string configuration would be appropriate in the interim. The performance penalty for a small array on a larger inverter during the interim period is typically very minor.


While we have just shown how to maximize an inverter’s PV array generating capacity, this does not mean that the goal of every design is to maximize PV input to the inverter. You may wish to consider limiting the array size, especially where the inverter is exposed to high ambient temperatures or the array to high irradiance. This will extend the life of the power electronic components in the inverter. Also, the peak efficiency for most inverters normally occurs at 50%–80% of rated power output, meaning that system energy harvest improves slightly when the inverter operates at lower power. Finally, inverter cost is a relatively small part of the total installed system cost. From an economic perspective, therefore, a slightly undersized array is often preferred to a slightly oversized array. Over the life of the system, the value of energy lost due to inverter power limiting or increased system inefficiency quickly outweighs the value lost by operating the inverter at slightly less than its rated output.


John Berdner / groSolar / White River Junction, VT /


Go Solar California / (Module PTC ratings are published here.)

PLEASE NOTE: As of August 23, 2010, this article has been updated for accuracy. Some comments, below, refer to an earlier version of the article.

Primary Category: 

The value of energy produced from utility-scale PV systems deployed throughout the Desert Southwest depends on the systems’ ability to match the seasonal and time-of-day utility loads. During times of increased demand, some utilities charge more per kilowatt-hour—effectively increasing the value of each kilowatt-hour produced by a PV facility—whether it is realized as an avoided cost or sold. As a result, project developers, engineers, owners and utility operators focus on designing and maintaining systems that maximize production during these high-value periods.

Coincidently, the Desert Southwest experiences a long dry season that corresponds with these periods of high demand and increased energy values. In the absence of significant rain events or regular cleaning, production losses due to soiling, also known as soiling losses, increase. In 2006 at the IEEE 4th World Conference on Photovoltaic Energy Conversion, PowerLight released an article titled “The Effect of Soiling on Large Grid-Connected Photovoltaic Systems in California and the Southwest Region of the United States.” The authors confirmed “a gradual but marked decrease in system performance through the dry season for systems in arid climates” and concluded that “performance losses due to system soiling are not constant through time, rather they depend on the amount and frequency of rain that falls on the array.”

Despite PowerLight’s conclusions, several of the commonly used PV production modeling tools, and thus system owners, assess the impacts of soiling on an annual basis. Consequently, system owners have typically addressed array soiling in one of two ways—either wash the array on a regular basis to limit the losses due to soiling, or forgo array cleaning and rely on rain events to keep soiling losses to a minimum. When based on annual soiling losses—as opposed to seasonal or monthly—these decisions fail to address short-term soiling impacts that may justify the cost of cleaning an array to maximize production during the high-value kilowatt-hour periods.

In this article, I review key soiling characteristics with historical weather data to simulate site-specific soiling losses on a monthly and annual basis for two sites in the Desert Southwest. The results confirm variation in soiling losses throughout the year for both locations, with average monthly losses in June and July—the high-value period—well above the annual average. In addition, prescribed cleanings are introduced to the model to understand their impact on annual soiling losses.

Measuring Soiling Losses

Soiling can be measured as either the rate at which contaminants accumulate on the module surface or the resulting decrease in production. Ultimately, we need to determine the decrease in system performance due to soiling loss. Assuming all other factors remain constant, comparing actual production values between a control subject and a soiled array is one way to determine soiling losses for a given site.

To simulate soiling losses over time, we must determine the rate at which soiling accumulates. Although soiling rates can be calculated in a variety of ways, a soiling rate that represents the daily percent decrease in production is most valuable for the purposes of PV production modeling. Once a soiling rate for a site has been established, it can be used with rainfall data to estimate past, present and future soiling losses.

As demonstrated in the PowerLight study, the “measured soiling rate” represents the slope of a linear fit curve applied to performance data between rain events. In other words, PowerLight’s study assumes that the percent change in performance over time—in the absence of rain or cleaning—equals the percent change in soiling losses over time.

In January 2013, First Solar published a paper in the IEEE Journal of Photovoltaics titled “Direct Monitoring of Energy Lost Due to Soiling on First Solar Modules in California” that details an alternative soiling measurement technique to determine site-specific soiling rates. The technique is based on a methodology proposed in “Solar Cell Arrays: Degradation Due to Dirt,” which was published in the Proceedings of the American Section of the International Solar Energy Society in 1989, and is intended to be “practical and automated … foregoing complex equipment such as IV curve tracers.” Rather than equating soiling rates to the increase in production losses per unit of time, the method First Solar uses compares production levels among a control module, a module that is not cleaned on a regular basis and the expected performance based on typical irradiance and temperature readings.

PowerLight’s and First Solar’s techniques can be used to establish site-specific soiling rates. This is quite valuable since soiling rates can vary within a region and may depend on the incidence of nearby human activities, such as traffic, construction, or airports. Since soiling rates do not take module cleaning into account, they must be modeled with local rainfall data to determine the actual soiling losses over a defined time period. To effectively predict soiling losses based on local weather and soiling rates, we must assume how much rain is required to completely—or at least significantly—clean an array.

Characteristics of Soiling

The PowerLight and First Solar studies, along with a study by Arizona State University (ASU), identify some key characteristics of soiling on PV arrays. In addition, McCarthy Building Companies has observed these trends and characteristics for several of the sites it monitors in the Desert Southwest. These trends are useful for estimating how soiling impacts PV production. Several of these characteristics serve as key assumptions in models used to predict soiling loss, also known as soiling estimating models.

Rain events. A paper based on a study conducted by ASU titled “The Effects of Soiling on PV Module and Radiometer Performance” concluded that 0.2 inch of rain is nearly equivalent to physically cleaning the modules and typically restores production levels to 99.5% of a cleaned module. In other words, after 0.2 inch of rain, soiling losses are reduced to 0.5%. First Solar was not able to validate this assumption in its recent study, and PowerLight found that “significantly more rainfall” is required on some systems “to completely clean the modules.” Factors like humidity and small, dusty rain events tend to negatively impact this metric— meaning more rain may be needed where such conditions exist.

Soiling rates in the Desert Southwest. The PowerLight study found that systems in California and the Desert Southwest experienced an average of 0.1% to 0.3% decrease in efficiency per day due to soiling. McCarthy’s analysis of several larger systems in the Phoenix metropolitan area revealed soiling rates between 0.04% and 0.07% per day. Lastly, First Solar observed “fairly constant” soiling rates in low desert regions of southeastern California without agricultural activities, averaging 1% per month or 0.03% per day. The study also shows that sites with more agricultural activity have significantly higher soiling rates.

Soiling rates increase with human activity. Although soiling rates remain somewhat constant per site in the Desert Southwest, rates vary from one site to the next. In the Central Valley, PowerLight observed daily soiling rates between 0.1% and 0.2%, with the higher rate occurring in areas with more human activity, such as urban environments, highways and airports.

Angle of incidence. Soiling losses increase as the angle of incidence between the sun’s rays and the module surface increases. For example, a 25° incidence angle has twice the losses of a normal angle, and a 60° angle has four times the losses. Soiling losses tend to be highest in the morning and evening, when the angle of incidence is greatest.

Tilt angle. The ASU study shows that soiling rates are not significantly impacted by the module tilt angle nor by modules mounted on tracking arrays. That said, the tilt angle does impact the ability of rainfall to clean the module surface. To maximize the cleaning effect when it rains, tracking systems should be stowed at a minimum of 5°, and preferably more, to allow the water to sheet off the surface of the modules. Fixed systems with a tilt angle of less than 5% experience relatively higher soiling rates than those with a greater tilt angle.

Estimating Soiling Losses

Combining what we know of the characteristics of soiling on PV arrays with local weather data enables us to estimate soiling losses on a monthly or even daily basis. System operators can use soiling estimating models to better understand the impacts on production over a specific time period, calculate the predicted loss in revenues due to underperformance and decide when the cost of cleaning the array is worthwhile. McCarthy utilizes a Microsoft Excel–based estimating model that assumes an incremental increase in soiling losses for each day between rain events. The soiling estimator relies on user-assigned daily soiling rates, typically set between 0.05% and 0.1%, and resets soiling losses to 0.0% after a quarter-inch of rainfall.

Site-specific analysis. A high concentration of utility-scale PV systems are installed near Gila Bend, Arizona, including First Solar’s 290 MW Aqua Caliente Solar Project, which is expected to come on line in 2014. A review of 20 years of weather data can be modeled with an assumed soiling rate of 0.1% per day to plot soiling losses on a monthly basis and show the variation in soiling losses per season. Understanding this variation provides system owners and operators the ability to identify short-term losses due to soiling. These losses can then be assessed against the corresponding electric rates to determine the impacts on ROI and whether cleaning the array is justified.

Graph 1 (above) plots monthly soiling losses over a 20-year period. The highest soiling losses occur in the summer months, peaking in June and reaching the annual low in September—presumably once the dry season ends. The average annual soiling loss for the time frame evaluated is 5.2%, but losses during the summer months are much higher. In this example, assuming a 5.2% decrease in production due to soiling during June and July could result in underproduction relative to monthly goals. Given that most utilities pay more for kilowatt-hours produced during periods of high demand, the financial losses to soiling during these months would be greater.

The large data set for this area also allows us to compare annual soiling loss values over the same 20-year period. Using the same assumptions as the previous model, the estimated annual soiling losses range from 2.5% to nearly 15%.

As illustrated in Graph 2, the annual soiling losses vary quite a bit from one year to the next. The majority of the years have soiling losses between 3% and 6%; yet in 2002, when it rained only twice within a 9-month dry stretch, estimated soiling losses were greater than 14%.

Variation by location. Designers frequently assume that soiling characteristics and the associated impacts on PV systems are consistent throughout the Desert Southwest. We can effectively compare two distinct areas using McCarthy’s soiling estimator and identify local or site-specific trends. In Graph 3, McCarthy compares the average monthly soiling losses between Gila Bend, Arizona, and Bullhead City, Arizona. Bullhead City is located along the border between Arizona and Nevada, just south of Boulder City, Nevada, and, similar to Gila Bend, is experiencing a significant amount of PV development.

Compared to Bullhead City, Gila Bend averages 1.5 additional rain events per year. The rain events are also more consistent and evenly distributed throughout the year. As a result, Bullhead City has an average annual soiling loss of 9.4%—4% greater than Gila Bend—with losses exceeding 10% in August, when kilowatt-hour values are at a premium.

The Value of a Good Wash

Soiling losses not only vary from one location to the next within the Desert Southwest, but also vary from one month to the next. Cleaning utility-scale arrays can be costly and may not be worth it. Soiling estimation models are an effective way to simulate a prescribed cleaning and determine if the reduction in soiling losses justifies the expense.

For example, for the Gila Bend area, we can simulate prescribing a single wash in early July so that the soiling loss returns to 0.0%. As we can see from Graph 4, with a single wash per year, the estimated annual soiling losses decrease. There is a significant reduction in loss in 2002, from nearly 15% to approximately 7.5%. Although a single wash reduces soiling losses in 2002 and reduces the average annual losses due to soiling by 1% over the 20-year period, certain years experience minimal, if any, improvements. Depending on the cost of cleaning relative to the increase in revenue, the model suggests that it might have been worth cleaning the array only in 2002.

If it rains immediately before or after a cleaning, that can nullify or minimize the benefits of a cleaning. When this occurs, as it did in 2000, cleaning the array does not improve soiling losses for the year. Although system operators could opt to forego cleaning an array after a rain, unfortunately, until we have foolproof weather forecasting, it is impossible to guarantee it will not rain shortly after cleaning an array—which makes choosing if and when to wash even more challenging. That said, Graph 4 shows that a single wash in July 2002 would have had a significant impact on production and therefore warrants a cost-benefit analysis.

When applying the same simulation to the Bullhead City region, the annual soiling losses are decreased by 3% and an additional 1.3% if a second wash is prescribed. Thus, cleaning an array in the Bullhead City region twice per year is estimated to reduce the annual losses due to soiling by almost 5% and may be worth the investment. Understanding the increase in production due to cleaning allows system owners to calculate the associated increase in production revenue, and compare that with the cost of cleaning the array.

Using the Soiling Estimation Model for Operations

Soiling estimation models are also valuable tools during the design and operational phases of a project. Modeling soiling losses based on local weather data and assumed soiling rates provides valuable information for production modeling, especially when systems are intended to meet seasonal or monthly energy demands. Likewise, the model is quite valuable once the system is in operation. You can simulate a prescribed wash to estimate how much it will increase production. With this information, you can conduct a cost-benefit analysis to determine if washing the array is worthwhile. Finally, you can design a soiling estimating model to notify system operators during periods of abnormally high soiling loss due to extended dry periods. Such notifications could prove valuable, such as at Gila Bend in 2002 when soiling losses neared 15%.

—Scott Canada / McCarthy Building Companies / Tempe, AZ /


Subscribe to Modules