ERIC FITZ, Navigant Consulting, San Francisco, CA USA
The PV industry is increasingly focused on module reliability. However, misuse of the terms reliability and durability, immature long term reliability prediction tools, and general lack of empirical data make it difficult for the industry to characterize the risks. While PV module qualification tests have progressed tremendously, they can't by themselves fully predict lifetime reliability. Given these issues, investors, developers and manufacturers need to consider module reliability in their risk management portfolios.
Reliability vs. durability
Reliability is a measure of a system's or component's probability of failure over time where failure is defined as a discrete (and most often sudden) event that has immediate and typically catastrophic impact on performance. A highly reliable PV module has a low probability of such failure over its life.
Durability is a measure of how well a system or component resists performance degradation over time. Unlike reliability, durability is characterized by a continuous gradual loss of performance. In today's market, PV modules are considered to be durable if they maintain a typical warranty offering of at least 80% of its original performance after 25 years.
In summary, reliability (the focus of this article) is a sudden total loss of performance; durability is a gradual performance decline.
PV module reliability issues
At first glance, a PV module seems like a pretty simple product: a rectangular panel that produces electricity when exposed to sunlight. However, a module is comprised of dozens of materials and components, such as: back contact sheet, semi-conductor, transparent conductors, encapsulates/sealants, cell-interconnects/solder joints, glass, and bypass diodes.
|Figure 1. Failure rate comparison of thin-film modules for 1997-2005, 2005-2007, and 2007-2009 from IEC 61646 .|
Each module component, sub-component and/or underlying material has various and sometimes independent failure mechanisms that could lead to poor module reliability. Environmental stressors that can cause problems or exacerbate manufacturing defects include the following:
- Rapid thermal cycling due to sun and passing clouds;
- Extreme temperatures in winter and summer;
- Mechanical loading caused by wind and snow; and
- Moisture infiltration due to high humidity.
These conditions can result in thermal expansion cracks, material warping and discoloration, conductor corrosion and contamination. The National Renewable Energy Laboratory (NREL) has pointed out that reliability engineering is a deep and active field. Advanced Product Quality Planning, Design Failure Modes Effects and Analysis, Fault Tree Analysis, Design for Manufacturability, and Design Review Based on Failure Mode are only a few of the methods used by module manufactures to ensure reliable products . While these are useful tools for designing and manufacturing robust PV modules, they do not predict the long term module reliability performance.
Predicting module reliability
Lifetime reliability prediction leverages expertise from multiple fields and requires complex experimental setups. Accelerated life testing (ALT), PV module qualification testing, and historical field data all play an important role in developing accurate predictions .
ALT describes a broad group of qualitative and quantitative methods used to predict the probability of failure and/or failure mechanisms of a given product. Since it is too costly and impractical to collect field data for 25 years, accelerated test methods are often used to predict module performance. These tests rapidly cycle temperatures, humidity, mechanical loads, light intensity, etc., to simulate environmental stress experienced by a module over its lifetime.
PV module qualification tests such as International Electrochemical Commission's (IEC) 61215  for crystalline-silicon (c-Si) modules or IEC 61646  for thin-film modules are two standards that rely heavily on ALT. Figures 1 and 2 show failure rate data vs. IEC tests from several generations of thin-film and c-Si modules. While it is clear what conditions were present when a failure occurred, it is often difficult to discern the root-cause component, sub-component and/or material that led to the module failure. Many industry experts are also quick to point out that the qualification tests (and ALT in general) are not predictors of long-term reliability on their own. They are only useful for identifying defects that are likely to arise in the first few years of field operation .
|Figure 2. Failure rate comparison of crystalline silicon modules for 1997-2005, 2005-2007 and 2007-2009 from IEC 61215 .|
In addition to ALT and required qualification tests, the industry relies on historical field data as a basis for prediction. However, other than research completed at national labs, there is a limited body of publicly available data on module reliability. BP Solar is one of the few manufacturers who have willingly shared data. From 1994-2002, BP had ~2 million Solarex c-Si modules in the field. Over this period a 0.13% module failure rate was observed . More recently, using a combination of historical data from Tucson Electric Power's fleet of ~5MW of Schott Solar c-Si modules and computer modeling, Sandia National Labs has forecasted a module replacement rate of 0.05% per year .
Taking all of these pieces (ALT, qualifications test results and historical data) and building an accurate prediction for new products is a difficult problem. Due to the complexity of lifetime prediction, DOE's Solar Energy Technologies Program has held industry workshops four out of the last five years to discuss this issue . Even with the recent effort and progress, there is still a lot of work to be done.
|Figure 3. Total lifecycle PV plant maintenance costs for Sunpower's managed fleet .|
The big picture
The Electric Power Research Institute (EPRI) claims that the lifecycle O&M cost due to poor inverter reliability dwarfs that of PV modules (50% vs. 5%, Fig. 3). That means only $2-3/kW-year is spent on module O&M. EPRI further notes that module failure is responsible for less than 1% of energy losses, as shown in Fig. 4. However, there are three items to note about this data: 1) it comprises almost all flat-plate, c-Si modules, 2) SunPower is an established company and an industry leader in quality, and 3) SunEdison is a well established developer in the PV industry. Unfortunately, the limited data for thin-films is not as rosy. A Conergy developer noted that, "First Solar accounts for 2% breakage, though [Conergy is] seeing far less than 1% on most of [their] projects over the first year ." If the best-in-class thin-film manufacturer sees failure rates around 1-2%, failure rates could exceed 2% for new market entrants. As the industry expands and new technologies are introduced, failure rates could climb even higher.
On a residential scale, module reliability is even more important. For a small array (~4kW), a single module can represent 5% of total system output. Because these small systems miss out on economies of scale, capital costs are high and have longer payback periods. Without the benefit of full-time O&M staff (unless the homeowner has a third party service agreement) an undetected module loss can have major impact on financial returns.
|Figure 4. Relative frequency of PV system component failure and outage impact, SunEdison, Jan. 2008- Sept. 2009 .|
In the early 90s, 10 year warranties were typical. Today, almost all manufacturers offer 20-25 year warranties. But a 25 year warranty doesn't mean your project is protected. Will your module supplier be around in 15 years when you have problems? Do they fund an escrow account to ensure that if they are gone, you still will be protected? Do they rely only on IEC qualification tests to make claims about long-term durability? If they have only been around for five years, how can they claim that their modules last for 25? The increase in length of warranties is promising, but an investor or developer must carefully review the company providing it.
Module reliability and durability are separate issues that a potential investor must consider independently. PV modules can fail in many ways, but the risk seems relatively low based upon module reliability studies. However, little public data is available on actual field results to verify the studies. A project with narrow margins could be impacted by even very small losses in generation or increased O&M. Even when backed by a warranty, failure rates drive lost revenue for investors, developers and plant operators. As such, a potential investor should carefully consider a module manufacturer’s track record, qualification tests, reliability forecasts and warranty coverage before investing in a project.
1. S. Kurtz, "Photovoltaic-Reliability R&D Toward a Solar Powered World," NREL, 2009.
2. P. Hacke, "Test-to-Failure for Long-Term Performance Assessment," NREL, PV Module Reliability Workshop, Feb 18-19, 2010.
3. IEC 61215: Crystalline silicon terrestrial PV modules - design qualification and type approval.
4. IEC 61646: Thin-film terrestrial PV modules - design qualification and type approval.
5. W. J. Gambogi, E. F. McCord, H. D. Rosenfeld, R. H. Senigo, S. Peacock, K. M. Stika, "Failure Analysis Methods Applied to PV Module Reliability," DuPont de Nemours & Co., Inc., 2005.
6. J. Wohlgemuth, "Long Term Photovoltaic Module Reliability," NCPV and Solar Program Review Meeting, 2003.
7. E. Collins, et al., "Reliability and Availability Analysis of a Fielded Photovoltaic System," Sandia National Laboratories, 2010.
8. DOE: Solar Energy Technologies Program: Past Meetings & Workshops. eere.energy.gov. [Online] [Cited: January 2, 2010.] http://www1.eere.energy.gov/solar/past_meetings_workshops.html.
9. D. Williams, "Large Scale Operations & Maintenance: Well Planned O&M Helps Mitigate 30 Year Risk," SolarPro, June/July 2010.
10. T. M. Govindasamy, "Testing the Reliability and Safety of Photovoltaic Modules: Failure Rates and Temperature Effects," TÜV Rheinland PTL & Arizona State University, 2010.
11. "Addressing Solar Photovoltaic Operations and Maintenance Challenges," Electic Power Research Institue (EPRI), July 2010. Product ID: 1021496.
Eric Fitz holds an MS in mechanical engineering from the Georgia Tech, a BE in engineering sciences from Dartmouth College, a BA in physics from Colby College and is a Senior Consultant in Navigant Consulting's Energy Practice, One Market Street, Spear Street Tower, Suite 1200, San Francisco, CA 94105 USA; ph.: 415-356-7157; [email protected].