Around a fifth of solar panels examined in a new study fail much faster than expected and some may last for only half their anticipated lifetime.
New research has uncovered a critical challenge in solar energy with the discovery that a considerable amount of solar panels degrade much more rapidly than expected.
Experts from UNSW have identified the reason behind the so-called 'long tail' in the probability distribution of the performance data after analysing information obtained from nearly 11,000 different photovoltaic samples globally.
The long tail appears on graphs showing the degradation rate per year of the panels, indicating that up to 20% of all samples perform 1.5 times worse than the average.
In other words, a significant number of panels do not degrade at a constant rate over a long period of time as might be anticipated, but instead lose energy or fail unexpectedly much sooner.
The discovery is important in terms of raising the standard of solar panels and making solar farms more cost-effective and reliable.
"Most solar systems are designed to last around 25 years, based on their warranty period," said Yang Tang, one of the authors of a paper on the subject published in IEEE .
"For the entire dataset, we observed that system performance typically declines by around 0.9% per year. However, our findings show extreme degradation rates in some of the systems.
"At least one in five systems degrade at least 1.5 times faster than this typical rate, and roughly one in 12 degrade twice as fast.
"This means that for some systems, their useful life could be closer to just 11 years. Or, in other words, they could lose about 45% of their output by the 25-year mark."
Why some panels fail faster
The UNSW team, including Dr Fiacre Rougieux , Dr Shukla Poddar , Associate Professor Merlinde Kay and PhD student Yang Tang from the School of Photovoltaic and Renewable Energy Engineering , analysed information collated previously by Dr Dirk Jordan from the US Department of Energy's National Renewable Energy Laboratory.
This is a collection of the annual production data from tens of thousands of photovoltaic systems produced globally and includes statistics on performance and maintenance.
The UNSW team found that although many panels do degrade smoothly and predictably, when the results are plotted on a graph there is a 'long tail' of samples that fail a lot more rapidly than should be expected.
This long tail is more than a statistical oddity. It especially poses a large financial risk for solar farms, where hundreds of thousands of panels are installed, since the data indicates there is a hidden cost associated with samples that do not perform as well or for as long as they should.
Importantly, it has also been shown that the extreme degradation observed in these panels is not related to the climatic conditions they are exposed to - ruling out the possibility that the data was being skewed by samples placed in extreme environmental locations such as very hot deserts.
Instead, the study found three major reasons for panels to be grouped in the long tail.
The first is interconnected failures. This highlights a scenario where different types of problems can interact with each other on an individual panel.
For example, if the backsheet (a protective layer on the rear of a module) is damaged, moisture can get in possibly causing a failure of the electrical junction box and other problems such as cell cracks or corrosion.
This domino effect, where the issues don't just add up but instead multiply, can be seen to make panels degrade much faster than predicted.
The second reason is rapid failure when modules are relatively new, dubbed infant mortality. Modules tend to have a slight recovery and slower degradation rate after the initial period of installation.
These panels likely have critical manufacturing defects or material flaws that are not discovered in quality control or testing and therefore fail rapidly - sometimes within just a few years of installation.
Finally, there are minor flaws that may not cause a problem initially, but then result in a sudden severe performance loss at a random point.
This could be a tiny hairline crack in a cell, or slightly imperfect soldering that goes unnoticed until complete failure.
Climate variation ruled out
Importantly, these three factors occur regardless of the location where the panels are installed.
This highlights the fact that climate is not a factor in the long tail of degradation rates, which was one hypothesis being considered.
Dr Poddar, a co-author of the paper, said: "A subset of the data shows information specifically related to solar modules in very hot climates which we know causes higher degradation.
"However, in other climates, when those hot regions are being excluded from the analysis, we see similar long-tail pattern in the probability distribution of performance degradation rate. This suggests that the issue is consistent regardless of where the panels are operating.
"Current testing standards focus primarily on three parameters: the modules' response to mechanical stress, extreme temperatures, and exposure to ultraviolet radiation - as well as often testing for humidity and response to a standardised amount of sunlight (AM1.5 spectrum).
"But when they are actually operating in real-world conditions there are so many different factors coming into play, and those cascading failures can be very significant.
"So I think we need to start thinking about different testing standards which would help to ensure we have more resilient types of modules."
The research is important for solar panel manufacturers and those spending many millions of dollars to build large solar arrays to provide clean energy.
The long tail phenomenon challenges the financial models that underpin the industry's growth, creating uncertainty in long-term energy yield and operational budgets.
If the real-world failure rate is much higher and panels degrade much sooner than expected, as shown in the UNSW study, there will be unforeseen expenditure required by operators for repairs or module replacements.
"With this research, we are hoping to make real impact in three ways," said Dr Poddar.
"We would like to get even more data from large-scale solar farms to analyse real-world failure rates in even more detail, so we can then make recommendations to the manufacturers of these modules.
"Secondly, we aim to understand different factors contributing to module failures in different climate types to develop early detection system and recommend manufacturers to improve design robustness.
"Thirdly, testing authorities should be informed of real-world degradation patterns across diverse climates and consider combining stress tests to better replicate outdoor operating conditions."
