Short of having real time outdoor testing data for all the materials and combinations of interest to the modern PV marketplace, there is a realistic testing approach that is able to reduce the risk of potential premature failure.
PV arrays are being installed around the globe, the impetus fueled by government subsidies and the current political climate. Many of these installations have been completed despite the fact that they are not financially competitive with other sources of electric energy. This situation, however, is beginning to change and the pressure to bring down the cost of these systems has become significant. In addition, such installations are coming to be viewed as part of a competitive environment, and they are expected to pay for themselves without subsidies and feed-in tariffs (FIT). To do this, they will have to last 20 to 25 or more years.
Achieving Competitive ROI
Many manufacturers currently warrant their systems to produce electricity at 90% of their rated capacity after 10 years in the field and/or 80% after 20-25 years. To realize a competitive ROI for these systems necessitates this level of durability, and the warrantees force companies to establish reserves or obtain insurance against claims in case the systems do not meet these promises.
How do we know PV systems will last for what amounts to an entire human generation? Unless, they are of the traditional crystalline silicon wafer and glass types, which have been made for many years, and for which there is 30+ years of field data, we don’t know. Even the current renditions of these c-Si systems differ from those produced 20+ years ago. To drive the cost of these systems down, there are efforts in many directions to utilize less expensive materials and construction than the traditional expensive c-Si modules. These include thin film Si approaches, CdTe, CIS types, dye-sensitized systems, OPVs, and GaAs. For all of these cell types our field data is nearly non-existent, at best amounting to five years or so. And, for new developments, we have to start all over again at time zero. Our objective is to find a way to demonstrate the long term durability of any new system without spending 20 years to do it. This is the challenge.
Current industry practice uses a certification test devised by the International Electrotechnical Commission (IEC), a standards-setting body, to identify infant mortality issues affecting PV systems. This is essentially the only common platform accepted by the major part of the PV industry, although it does not identify long term environmental durability and performance issues.
The IEC design type qualification test program measures a number of environmental stress factors, including high and low temperatures, high humidity and some cycling, exposure to UV light and some mechanical loads (Fig. 1). It uses eight separate representative modules for these tests, so that no single module goes through all the sequences. Following each exposure, the modules are visually examined for major visual defects such as cracks, bends, bubbles, and a variety of other descriptive items. At the end of each test sequence, the modules are expected to retain 90% of their rated output capability, exhibit no open circuit during or following the test, maintain low wet leakage current levels, and meet any other specific requirements anticipated for the modules. Failure of two or more modules during any part of the sequences means that the design is deemed to fail qualification.
|Figure 1. IEC PV module qualification test protocol.|
This scheme is based upon negotiation within the IEC committee of industry experts and was arrived at as the best set of options as a HALT (highly accelerated life test) to detect early lifetime failure issues. For this purpose, it does an adequate job, and the damp heat test (85ºC., 85% R.H.) is particularly challenging (although unrealistic) to PV modules, especially to workmanship issues. It follows the tenets, and has the limits of HALT testing. It does not attempt to simulate any field environment, only to find design and process flaws by any means possible . HALT testing tries to obtain fast failures by whatever means required and relates, to some extent, to infant mortality issues. HALT testing however, does not relate to accelerated environmental testing (AET), which identifies the chemical and physical changes occurring over a much more extended time frame.
For this purpose, the IEC qualification tests are performed and mistakenly accepted as indicators of future performance. What they do not tell us is what the effects of long term aging/weathering will be on these modules. To do this, the various stresses must be applied concomitantly, or the test represents the field environment unrealistically, and in ways unrelated to the real performance requirements. What is especially important is the fact that all of the IEC test sequences are performed without the application of full solar light, with the exception of one short outdoor test.
We can compare the concept of infant mortality with that of long term weatherability by reference to Fig. 2, commonly referred to as a bathtub reliability curve. The time frames of the service life exceed those to which the HALT applies by an order of magnitude. In a well annotated survey of the literature from 1975 onward, leading up to the creation of the IEC standards, Osterwald and McMahon of NREL (National Renewable Energy Laboratories) came to the conclusion that standard module qualification test results cannot be used to infer product lifetimes .
Noteworthy, however, is the work of Wohlgemuth  in which it was demonstrated that multiple cycles in an environmental cabinet (mimicking the IEC 200 thermal cycle test, but multiplying it several times) gave results that could be correlated with field data. The approach to be discussed later is supported by these data, which, unfortunately, are only available for silicon wafer-based systems.
The purpose of the method described here is to surmount this shortcoming of the HALT to identify durability issues anticipated to crop up much later in the service life of the modules under study.
A Realistic Approach to Accelerated Testing
Formal outdoor weathering testing has a hundred year history. Because of the lengthy test times required for meaningful outdoor exposure results, accelerated methods have been developed over the years by established industries in order to facilitate product development without extended waiting for results. Experience gained from many industrial systems has led to the conclusion that an accumulated damage model provides the best analog for real outdoor weathering. It is also known that most degradation of combinations of materials occurs during changes in the environmental conditions, not during steady state aging.
Figure 2. The ‘bath tub curve’ of product failure.
The system devised to achieve this realistic approach to accelerated outdoor weathering is the Atlas 25PLUS Program. It incorporates all of the critical outdoor stresses, light, heat and moisture, into a comprehensive collection of exposures of a test module. It also includes the sequencing of changes to these variables as analogs for the diurnal and annual seasonal cycles. Conditions representing a global environmental composite are provided in a “black box” approach, although special adaptations can be made for such specialized environments as marine, agricultural (dust, dirt, chemicals), dry desert, humid tropical, or alpine. The term black box is used to indicate that the same conditions pertain within the box, regardless of what material is introduced into its environment. The concept is that Mother Nature does not care what the subject materials consist of, she provides the same stresses regardless. This program delivers the key weather and climate stresses at the levels found in nature, in ways that simulate both short term daily and longer term seasonal cycles.
The program was founded upon several key considerations: 1) Utilize existing published and internal findings regarding PV degradation and failure; 2) Utilize existing data on accelerated PV module testing; 3) Incorporate fundamental outdoor and laboratory weathering principles learned from other industries; 4) Deliver environmental stresses within realistic boundary ranges; 5) Deliver stresses simultaneously in a fashion as close to that imposed by nature as possible while employing known acceleration methodologies; and 6) Leverage existing test methods for economy wherever possible.
The program is applied to one primary and two secondary modules, the latter two mounted on outdoor solar tracking racks in Florida and Arizona. The primary module is exposed as follows:
UVA/UVB light exposure is used to provide activation of some semiconductor formats that require pre-activation as well as to provide radiation for such phenomena as Staebler-Wronski degradation of amorphous thin film silicon systems, and some preliminary photo-degradation of polymer components.
Condensing salt fog exposure followed by condensing humidity provides accelerated marine and alkaline desert corrosion of connectors, micro-inverters, etc.
The module then enters the core element of the test in which the following tests occur inside the chamber : a.) accelerated exposure to full spectrum solar radiation, temperature and relative humidity cycling (that is nature derived) for seven days of each ten day sequence; b) accelerated cycles of solar temperature and humidity under freeze/thaw conditions typical of temperate climates for the other three days. Additionally, the full spectrum solar load applied during all of these cycles provides much more realistic stresses than does basic environmental chamber cycling, as it adds the elements of photo-degradation and forward bias (modules are tested under resistive load). Heating by the unidirectional “solar” beam also causes non-uniform IR stresses that cannot be duplicated in heated cabinets. This so called “laboratory core testing” imposes thermal, solar, humidity cycles, each of which has the equivalent effect of two to three days of real world stresses. The total number of full environmental cycles is approximately 1300.
Mounting on an outdoor Arizona track rack for the peak solar months of summer.
Intervals during balance of year consumed in track rack exposure to Arizona sunlight.
The two outdoor modules, which are on track racks in South Florida and Arizona, are used to give early indications of true outdoor response of the subject modules, and are evaluated by the same criteria as the primary laboratory module.
At several intervals throughout the testing, evaluation of the module is done in terms of the same visual physical manifestations as in the IEC certification, with the additional evaluation of IV curves and PMAX, and infra-red evaluation of hotspots while the module is forward biased. Development of any of the normal symptoms of pre-failure requires assignment of the module to the category of “disqualified,” based upon the criteria established by the end user.
In its entirety, the program requires approximately one year to complete, providing information specifically designed to predict performance in the long term aging, accumulated damage region of its service life.
What Do We Get from This Program?
The approach described above was specifically designed to answer the question “Will my module last in the real outdoor environment?” As such, it has been formulated using the wisdom of industry experts from around the world, and the internal experience of weathering experts from a world leader in outdoor and laboratory testing of the environmental durability of materials.
Short of having real time outdoor testing data for all the materials and combinations of interest to the modern PV marketplace, this approach provides a method for reducing the risk of potential premature failure of these systems. It applies lessons learned from the evaluation of multitudes of other industrial materials over the course of almost 100 years of testing and correlation of real world with laboratory test environments.
Evaluation of PV modules using this system will provide a preview of the behavior of any PV module beyond the infant mortality period and into the long term aging time frame of accumulated damage and long term performance, offering assurance of at least ten years of useful service and giving indications of performance beyond that. Although it is designed to be conducted over a one year time frame, extending the test into a second year can give assurances of durability approaching 20 years.
Although the program described above is based upon what information and experience is currently available, and has been crafted to produce an advanced accelerated weather aging protocol for the evaluation of PV modules, it is still subject to the limitations of the real world.
1. Gregg K. Hobbs, “Accelerated Reliability Engineering: HALT & HASS,” John Wiley & Sons, April 10, 2000.
2. “History of Accelerated and Qualification Testing of Terrestrial Photovoltaic Modules: A Literature Review,” 2008, www.nrel.gov/pv/performance_reliability/pdfs/osterwald_pip_review_2008.pdf
3. J. Wohlgemuth, “Reliability of PV Systems, Reliability of Photovoltaic Cells, Modules, Components and Systems,” edited by Neelkanth G. Dhere, Proc. of SPIE, Vol. 7048, 704802-1, (2008).
David P. Dumbleton received his BA in Chemistry from Northwestern U., BSChE from the U. of Wisconsin, and his MS and PhD degrees from Georgia Tech and is a senior consultant at Atlas Materials Testing Technologies, Chicago, IL 60613 USA; ph.: 773-289-5871; firstname.lastname@example.org