• Print

Project-level quality: Reliability demonstration testing helps root out PV module failure mechanisms


Tom Cheyney
Tom Cheyney
Tom Cheyney, former senior editor of PV-Tech and Photovoltaics International, is now chief curator of SolarCurator.com and director of Impress Labs’ solar practice.

As impressive as the exponential expansion of solar power has been over the past few years, it’s important to remember the last part of that preceding phrase: the industry, as a large-scale global concern, is still in its infancy. During his presentation on module reliability and performance characterization at the recent PV Power Plant USA conference in Phoenix, PV Evolution Labs cofounder/CEO Jenya Meydbray shared some telling factoids. Less than 5% of the total installed solar capacity has been deployed for more than 10 years old, and less than 18% has been generating power (or not) for more than five years. Given the sheer numbers of modules involved and wide range of quality assurance levels—and historical field data samples showing a third of 18-24-year old modules degrading at a guarantee-busting 1% or more per year—no wonder he expects to see a whole lot of warranty claims in the next five years.   

Meydbray’s discussion focused on project-level quality assurance and module reliability demonstration issues. He believes that deviations in the manufacturing processes are the leading cause of most quality-related problems, not the design of the modules themselves.

With an all-over-the-map level of consistency in approaches to statistical process control (as some manufacturing mavens have told me, the terms “process” and “control” are not necessarily wedded in some PV factories) and increasing cost pressure leading to some producers cutting corners in their QA/QC efforts, the role of “module intelligence” takes on critical importance. Researchers, project developers, and end-users want to get an accurate, bankable idea of just how well a system is going to perform over its 25-year lifetime.

The three pillars of an effective project-level module QA program, according to Meydbray, are lot acceptance testing (usually done during shipping or construction), latent defect screening (also done up front), and ongoing degradation testing. He finds defect screening the most interesting of the trio, since one can screen for different failure behavior “right out of the box.”

“There are methods of stimulating all of these degradation mechanisms in the laboratory,” he told me in an email after the conference. “This latent defect screening can and should be utilized for large-scale PV power plant development.

Admitting that it’s hard to try and simulate 25-year field performance, he explained during his talk that a reliability demonstration testing (RDT) plan can be instituted that provides a good idea of what the relevant failure mechanisms look like. Along with NREL and Sandia, Meydbray’s Berkeley, CA-based lab has come up with such a plan, which comprehensively assesses failure modes on modules over a five-month test period.

The plan factors in a compilation of failure modes seen in the field and the identification of stress tests that simulate those fails, and then lists the analytical techniques needed to characterize those fails.

It’s not that hard to break a module, and there are many ways to do it. He listed twenty-some common failure modes, a table of infamy including myriad junction-box bugaboos, cracked cells and fractured glass, light-induced cell degradation, various corrosion scenarios, and backsheet delamination. Beauty shots of failed modules and systems from the field (some of which accompany this blog) revealed nothing good, and a whole lot of JPGs of bad and ugly pieces of PV gear.

The RDT plan incorporates correlations between the failure/degradation mechanisms and specific stress tests and then determines the appropriate characterization and analysis testing methods to assess the problems. For example, solder joint degradation (in various stages of propagation)—which he cites as the most common failure seen in the field, especially on older modules—can be sniffed out by running 600 cycles of thermal cycling (significantly beyond the number required for plain-vanilla IEC certification) and a 1K load test, followed by a combo of light and dark IV exams as well as a regimen of electroluminescence (EL) imaging.  

Recent failure trends have included a rash of substandard diodes from one manufacturer (leading to j-box failures) as well as an increasing number of potential-induced degradation (PID) problems and a sodium migration-based corrosion issue that is not factored into current test protocols. PV Evolution’s own tests have shown a wide range of results on this salty development, with some modules coming through “rock-solid” and others experiencing nearly total failure modes, according to Meydbray.

When modules do pass muster after undergoing the battery of reliability demonstration tests, that doesn’t mean their deployment equates with profit margins. As session moderator Steve Smith of Solivida Energy quipped, “bankability doesn’t always translate into a project’s backers making a lot of money.” But deploying reliable, robust PV panels capable of meeting and exceeding their 25-year warranty specs doesn’t hurt those prospects either.


  • Photovoltaics International 29th Edition

    Forecasting the evolution of a young, dynamic industry is by definition an uncertain business, and solar is no exception. Rarely, if ever, do the numbers broadcast by any of the various bodies involved in the PV prediction game tally, and even historical deployment rates remain the subject of hot debate. The paradox is that getting forecasts broadly right is going to become increasingly important over the next few years, particularly for those involved in producing the equipment that will support whatever levels of demand come to pass.



Solar Media


We won't share your details - promise!