Great balls of hail … the invisible PV peril
After a summer of bushfires and smoke haze, around lunchtime on January 20 Canberra experienced a massive hailstorm with winds of up to 117km/h accompanied by hail as big as golf balls. Much damage was reported; the ACT emergency services received almost 2,000 calls for assistance (more than three times the annual average) and at the CSIRO 65 glasshouses were lost and crops of wheat, barley, legumes and cotton were devastated. Just six days later, almost 30,000 insurance claims had been lodged, mostly for cars. The National Insurance Council declared the event a catastrophe in order to fast-track claims. So, how did PV arrays fare?
Initial results were mixed. Some systems had obvious damage and smashed panels were clearly identified by all and sundry. On social media, reports came in of people observing the output of their system before and after the hailstorm and declaring it to be fine, sometimes even improved as months of smoke and dust were washed away. At PV Labs we heard reports of infra-red cameras being used to look for damaged panels. While an obvious place to start, as we’ll discuss later, neither of these test methods are particularly appropriate to look for damage caused by hail.
What’s in a standard
PV panels are built to withstand hail. The hail test is a specific test (MQT 17) in a specific standard (IEC 61215). The purpose of the test is clear: “to verify that the module is capable of withstanding the impact of hail”. All modules sold in Australia should have passed this test.
The test involves firing 11 perfectly spherical ice balls at fixed velocities at different and specific parts of the panel. The minimum requirements are spheres of 25mm in diameter traveling at 23.0m/s. The module passes the test if it has no major visual defects, power output is unaffected and there are no cracks in the backsheet that would allow water (or electricity) to ingress.
Anyone who has played golf and many others too would understand that golf balls, and hence the hailstones in Canberra this January, were larger than 25mm in diameter. More alarmingly the home of the big things, Queensland, has not failed to deliver and in April this year reported on 8-10cm diameter hailstorms; reportedly “tennis or baseball-sized”.
What’s behind the glass?
So, if a module appears to have survived a hailstorm and the output of the system appears to be unaffected, why should we doubt appearances? In a word: micro-cracks.
The workhorse of a PV module is (usually) a silicon solar cell. The silicon itself is quite brittle and very thin (less than 200 microns is typical). It is possible for the silicon to be cracked, and if it is, it is very unlikely the cracks will be visible. They are, after all, “micro” cracks! It is also possible for the micro-cracks to exist in the silicon but for the different pieces of silicon to remain electrically connected, so that panel output is not immediately reduced.
Over time, however, as the panel heats and cools with the day and with the seasons, cracks can widen and (while still remaining invisible to the naked eye) pieces of silicon can become sufficiently detached from one another to result in a drop in output power. It is not possible to predict when this will occur.
As days and seasons take time to pass, damage caused by a hail event may not result in an immediate drop in output power. It is therefore crucial that if a solar system is suspected of being damaged and could be replaced under insurance that a proper test is done to determine if micro-cracks exist in the silicon. One such proper test is an electroluminescence test.
Testing for micro-cracks
We have heard anecdotal reports of people using thermal imaging to look for micro-cracks. Thermal imaging sounds alluringly high-tech, because it can be performed from a drone, without the need to remove panels from an array. The downside of using thermal imaging to detect micro-cracks is that it detects hotspots. It does not detect micro-cracks. Electroluminescence or photoluminescence are the best methods to detect micro-cracks.
Electroluminescence is a diagnostic process that was introduced in 2005 by T. Fuyuki for solar cells. In a manner of speaking it reverses the photovoltaic effect. Current is applied to the PV module and the resulting radiance levels are recorded using a camera capable of detecting near-infrared light. The result is an image that resembles an X-ray.
Electroluminescence can be done in light or in the dark. Dark measurements (either in a special dark room or done on a moonless night with minimal light pollution) are less expensive to perform and typically show more details. Whether or not the measurements should be done in-situ or by sending panels away to a laboratory will depend on the size of the system. For smaller systems, it is usually less expensive to remove the panels and test in a laboratory.
PV Lab Australia has teamed up with Capital Solar Maintenance in Canberra to test panels with potential hail damage, where Capital Solar Maintenance provides the panel removal and replacement and PV Lab provides the electroluminescence testing. The images below show two panels that appear undamaged to the naked eye, but the electroluminescence imaging clearly shows that somewhere between a third and a half of all cells have been shattered.
Micro-cracks aren’t the only system damage to consider. We found at least one case where removing panels for electroluminescence testing revealed that even though the panels themselves were undamaged by hail an associated DC isolator and conduit were damaged and in need of replacement.
What to expect from measurements
Deciding how many panels to test is the first step. There is no need to test every panel in a system as there are well developed statistical sampling methods that can be applied. We normally recommend using ISO 2859. As an example, consider a system with 100 panels. In this case, good statistics can be achieved by testing a randomly selected eight panels.
Correct image analysis is the final step. The two smaller images show zoomed detail of electroluminescence images at a single cell level. There are typically 60 or 72 cells per panel. The image on the left is a multi-crystalline silicon wafer and the crystal boundaries are clearly visible. The cell also has a single micro-crack (running vertically the length of the cell, slightly to the left of the halfway mark). Such a micro-crack is usually acceptable. In a 60-cell (or 72-cell) module it is common to accept five (or six) such cracked cells. The image on the right is a mono-crystalline silicon wafer. This cell has been shattered. Such damage is normally unacceptable.