Weekend Read: Temper tantrum

From pv magazine 12/23-01/24

A worrying trend is emerging in large-scale PV as project owners report high levels of broken module glass. Few of them are willing to share details from specific projects at this stage but quality experts say reports have been received relating to hundreds of megawatts of installed PV generation capacity. And the problem could be much bigger.

Broken glass details vary but appear to share factors in common: All concern dual-glass-module projects featuring 2 mm glass on both sides of panels. None have been exposed to strong wind, hail, heavy snow loads, or other events that might explain the broken glass. Data thus far suggest the problem is not confined to any region or solar technology. “We have seen reports of unexplained broken glass coming from Brazil, Chile, Australia, the US, and elsewhere,” said Tristan Erion-Lorico, vice president (VP) of sales and marketing at solar equipment testing lab PVEL. “This isn’t region specific, isn’t system-type specific, and it isn’t manufacturer specific. That’s why this is so concerning.”

Manufacturers, quality assurance providers, certification bodies, and the PV research community have noted the issue. Theories exist about the root of the problem and best course of action. One module manufacturer described a project where owners of a PV plant approached it to report of widespread broken glass.

The manufacturer began an investigation at the project site and was able to prove the breakage was caused by lawn-mowing robots that flicked up rocks from the ground hard enough to damage module glass. In a recent pv magazine webinar, Thomas Weber, of quality assurance experts PI Berlin, shared details of multiple investigations where it was able to identify causes including improper installation and use of the wrong clamps.

In other cases, investigations have not been able to reach such a clear conclusion and there is likely a combination of causes out in the field. Most of the possibilities boil down to two things – weaknesses stemming from the use of larger modules with thinner glass, and a possible gap in testing and certification standards which modules that are highly susceptible to glass breakage have slipped through.

Module glass

With most modules now generating power from both sides, and a long history of reliability problems with polymer backsheets, many manufacturers have made the switch to using glass on both sides. This has plenty of advantages but also necessitates a switch to thinner glass in order to keep the weight down – glass-backsheet modules typically use 3.2 mm thick glass, while glass-glass products slim this down to 2 mm per sheet.

The thicker glass can be fully tempered, greatly increasing its strength but the tempering process relies on a temperature difference between the outer surface and inner portion of the glass. A thickness of just 2 mm does not leave enough of an inner portion between the two surfaces for the glass to be fully tempered (at least cost-effectively).

At the same time as glass has been getting thinner, frame designs have also been shrinking while the modules themselves get larger, leading to a significant change in module design that appears to have been widely overlooked: That glass is now playing a much more significant role as a load-bearing material in panels.

“If you’re using the glass as part of the support structure that’s actually bearing significant load, the failure mechanism you’re confronting is brittle fracture,” said Henry Hieslmair, principal solar engineer at DNV. “Preventing that depends on how well you control defects along the edges and surfaces, and that can have a huge variation. So you could easily pass a few modules in a load test but then have something like 1% fail in the field.”

Module manufacturers state that they carefully qualify glass suppliers and test products before using them in mass production. “We qualify glass from many aspects such as size, durability, reliability, bending strength, the drop-ball test, hail test, appearance, etc,” a representative of module manufacturer Trina Solar told pv magazine in a statement. “With each glass type from each supplier, we go through comprehensive reliability tests beyond [the] industry standard within Trina Solar’s own lab as well as third party labs for certification. We do regular quality checks for incoming glass deliveries from suppliers. We also do regular quality and reliability test checks for our modules in mass production. It is a fully controlled process.”

There are signs that the move to thinner glass and larger modules may mean current testing standards don’t take the role of glass into account well enough. “In my mind, the root cause is that, as an industry, we’ve been treating glass as a commodity,” said PVEL’s Erion-Lorico. He explained that International Electrotechnical Council (IEC) standards currently allow for module makers to switch glass manufacturers without a need to retest for certification as long as the thickness and high-level strengthening process haven’t changed. With thicker, fully tempered glass, this may have made sense as there was very little difference between manufacturers but with various types of “heat strengthening” on the market, that may no longer be the case. “Now we’re seeing that glass from different manufacturers could have significantly different strength and mechanical properties,” added Erion-Lorico.

Module-tracker compatibility

The way modules are attached to trackers and fixed racking is another topic that pops up regularly in discussions about glass breakage. The issue here seems similar to that related to glass thickness:

Module designs have changed without an accompanying update covering how they are attached to racking. “In the last 10 years, a lot of systems have been attached only with edge clamps,” said DNV’s Hieslmair. “The glass was thicker, the frames were thicker, and we didn’t see a lot of glass breakage. Now, with these really large modules, the ratio of load bearing glass to frame is not the same.”

Module manufacturers provide a letter of compatibility for specific racking systems, which should be the result of rigorous testing. The signs are that small sample sizes and limited parameters for these tests mean they haven’t always picked up on susceptibility to glass breakage. “If any evidence of the ‘compatibility’ is provided, the evidence is one successful result of a single static mechanical load test with the module on the tracker mount, followed by a visual inspection,” wrote DNV in a blog post on the issue. “The tested load is then de-rated 1.5 times to obtain the design load. No additional information is provided in the compatibility letter, or elsewhere, describing any potential additional testing or analyses that may have been performed by the manufacturer to certify compatibility.”

Testing standards

Whatever root causes are eventually found for glass breakage, the next step is to find tests that can reliably spot the problem before systems are installed. That is something the solar industry is only just waking up to.

“If a module-tracker combination passes the static load test according to IEC standards, we say our module is compatible,” said CJ Fu, director of product solutions at Longi Solar US. “But, based on my personal experience, this may not be adequate to reflect compatibility in the field. There you can have some surprises, multiple stresses happening at once. We do feel that there is room for improvement in developing a more thorough compatibility testing sequence.”

Tracker manufacturers also confirm that they see interest in more detailed compatibility testing, thought not necessarily from all players.

“Some have requested supplemental tests to the IEC and UL standards, but ultimately that is at the discretion of the module supplier,” said Greg Beardsworth, senior director of product marketing at Nextracker. “It’s up to them to define the maximum loads they are comfortable with while our job is to define the loading for each application as accurately as possible. The standard is based on the static mechanical load test and it varies by manufacturer whether there is anything supplemented on top of that.”

Members of the PV research and reliability community have also noted the increase in reports of broken glass at solar projects and are looking for the right set of tests to address the issue but that will take time. “These codes and standards committees can take a long time to evolve but I wouldn’t be surprised if we see additional testing requirements on top of what we have at present,” added Beardsworth.

In the United States, researchers at the National Renewable Energy Laboratory (NREL) have begun investigating new tests and additions to standards, both for separate testing of glass quality and for module-tracker compatibility.

“Products are changing faster and faster and it’s hard to get ahead of the game,” said Ingrid Repins, senior research fellow in the photovoltaics reliability group at NREL. “This glass breakage has caught us by surprise, though I think we knew to some extent there were weaknesses and holes in the testing. Now we’ll try to understand root cause and design tests so that this doesn’t happen again.”

The lab is currently designing tests to investigate various hypotheses concerning glass breakage. Firstly, by looking into glass manufacturing. NREL scientists have begun using optical techniques to measure the temper and heat strengthening of glass, aiming to spot differences in module glass that has cracked.

“There is a hypothesis that it’s really the heat strengthening that’s varying across these different modules and that’s what our scientist will measure when we look at these modules that have cracked,” said NREL’s Repins. “Are they less heat strengthened than they should be? And is that varying across the modules that break or don’t break?”

In another project, NREL scientists are studying the way that cracks form in the glass, aiming to build this into existing models of the whole module. That way, they can model different scenarios and guide the most effective testing strategies – also an important factor given the number of possible component and material combinations and the need for larger test samples. “At this point we have research and we have questions but we don’t yet have answers,” added Repins.