Using thermal imaging to detect solar roof module issues?

[Powderkeg]

It is widely known that utilities are rapidly moving to thermal imaging to detect problems with commercial solar installations. This technology when mounted to a drone can thermal photograph thousands of commercial solar panels in a matter of a few hours and identify bad modules with various issues as well as non-producing strings.

The question is whether this technology could also be applicable for solar roof customers? My initial reaction would be why not, although there are several factors that may make this difficult or impossible.

1. The heat signature as an example on a string that is not working appears cooler than a producing string. This would be an open circuit and thermal imaging hence easily detects this as cooler than a normal panel. The base construction materials are different of a solar roof module which may limit the heat signature.
2. The same issue surrounds defective cells within a solar roof module. This typically would be a short which is hotter than normal and may or may not show up with thermal imaging due to the difference in construction material.
3. The sunlight must measure at least 600 W/m2 with a solar irradiance meter for valid data. This is easy and quick with a commercial solar farm since panels are typically facing in the same direction. This is not true for residential where there may be several (North, South, East, West) mounting planes. This would require measurements at different times of day to extrapolate valid data for all the solar roof modules.
4. Artificial Intelligence (AI) is used to automate the examination of a commercial solar farm. This requires the solar panels be a specific size. Obviously, this could be a potential issue since solar roof modules are radically different in size. They may have to be examined manually which is fine for most residential users.

The reason I bring this topic forward is to potentially provide the homeowner with an alternate method of identifying solar roof issues than having Tesla roll a truck and randomly tear the roof apart until they find the problem. I refer to this as the “needle in a haystack” method. String(s) that are not working can be easily identified. In addition, if there is a single cell that is out on a solar roof module it may continue to function albeit with less output. You may very well have individual solar cells that are not functioning right now and be unaware of it. Thermal imaging can detect this.

Personally, my roof is 21KW and I am almost certain it has issues but have thus far been unable to find anything out of whack.

And lastly, I am not advocating contracting with a commercial drone firm to scan your roof. For most of us it would be cost prohibitive.

Has anyone here ever gone down this path? What are your thoughts?

 

[zanary]

Picking up a Thermal DJI Drone for home use is pretty inexpensive vs. the cost of the solar roof. You should consider it and then let use now how it works out. It think you're on to something here with helping determining if a panel or string or solar roof tile or string is working or not.

https://www.bhphotovideo.com/c/product/1730098-REG/dji_cp_en_00000414_sb_mavic_3_thermal_with.html?gclid=Cj0KCQjw8qmhBhClARIsANAtboen7IF7Pn_PqTjf-WZnEba1i-bbKHz24QAMcqP1aF1_YONHAPqCKzEaAncsEALw_wcB

 

[Powderkeg]

I have done a significant amount of background work since I started this post and thought that perhaps an update would be appropriate.

After purchasing a DJI Mavic 3T drone and a lot of experimenting here is what I have discovered:

Referencing the attached photo, it is very easy to determine which Tesla Solar Roof modules (mods) are active versus the solar glass passive units and sheet metal. The active mods show as a darker almost purple color. Solar glass appears as orange, and sheet metal is yellow colored. The installation at my residence is a two story home with just about 21Kw using slightly more than 350 mods. BTW, the attached photo shows the west side of 57 mods.

Looking closely at the photo there is one defective module in the second row from the bottom third from the right side. The white mark on the top right of this module is a piece of tape that was left on at install time and has no bearing on the problem. As was expected a defective module is a lighter color. In this case it is probably open however I have not had an opportunity to remove and test it with a meter yet.

Also during testing I had the opportunity to power off one of the three inverters. I was expecting since this created no current flow it would show up as open and several mods be identified by a lighter color. This was not the case and there may be two possible reasons why:

1.) Powering off the inverter does not stop current flow through the mods and the wires need to be physically disconnected from the inverter in order to simulate an open string. If anyone knows the answer to this it would be much appreciated.

2.) Unlike a solar panel the mod has what appears to be a stainless steel backing that the solar cells are fastened to. The theory is that this may result in the mod having a specific purple color regardless of whether it is energized or not. This theory however contradicts the presence of the defective mod as outlined above.

And lastly, I have not been able to simulate a shorted mod so the behavior is still a mystery.

Stay tuned......

 

[Vines]

I have done a significant amount of background work since I started this post and thought that perhaps an update would be appropriate.

After purchasing a DJI Mavic 3T drone and a lot of experimenting here is what I have discovered:

Referencing the attached photo, it is very easy to determine which Tesla Solar Roof modules (mods) are active versus the solar glass passive units and sheet metal. The active mods show as a darker almost purple color. Solar glass appears as orange, and sheet metal is yellow colored. The installation at my residence is a two story home with just about 21Kw using slightly more than 350 mods. BTW, the attached photo shows the west side of 57 mods.

Looking closely at the photo there is one defective module in the second row from the bottom third from the right side. The white mark on the top right of this module is a piece of tape that was left on at install time and has no bearing on the problem. As was expected a defective module is a lighter color. In this case it is probably open however I have not had an opportunity to remove and test it with a meter yet.

Also during testing I had the opportunity to power off one of the three inverters. I was expecting since this created no current flow it would show up as open and several mods be identified by a lighter color. This was not the case and there may be two possible reasons why:

1.) Powering off the inverter does not stop current flow through the mods and the wires need to be physically disconnected from the inverter in order to simulate an open string. If anyone knows the answer to this it would be much appreciated.

2.) Unlike a solar panel the mod has what appears to be a stainless steel backing that the solar cells are fastened to. The theory is that this may result in the mod having a specific purple color regardless of whether it is energized or not. This theory however contradicts the presence of the defective mod as outlined above.

And lastly, I have not been able to simulate a shorted mod so the behavior is still a mystery.

Stay tuned......

Cool work you are doing, I like it. I know that part of what makes this work with commercial drones is the good angle of the picture. If the IR camera is very far off axis, then the reading is reflected or dampened and not as useful. A friend of mine actually built a plane around a high-speed IR camera to do exactly this kind of overhead QA evaluation of PV fields.

This is why it would be hard to do a bunch of residential roofs easily and cheaply. I'd love to see someone figure it out though. You really don't need an amazing resolution IR camera if looking at regular framed modules even, as they are quite large.

1. Your system is probably not equipped with rapid shutdown, so there are no MCI to fail either. There probably isn't another means of disconnecting a single string that will do what you want. You might be able to disconnect individual strings from a fused combiner box if any exist.

 

[Tdriver]

Actually a good thermal drone like the DJI 3T Thermal runs $5500.00 and that is just a start point. I have had one on order for six weeks and am told I will not see it till mid July. The Ukraine military purchased 300 recently and I think they bought up all the inventory....they get priority.
My plan is to offer inspections in my Sun City community and volunteer for local Search & Rescue, and Fire efforts here in the desert.

 

[wwhitney]

Also during testing I had the opportunity to power off one of the three inverters. I was expecting since this created no current flow it would show up as open and several mods be identified by a lighter color. This was not the case and there may be two possible reasons why:

How long did you wait? I would expect it you turned it off at dawn, the difference would be visible at noon for sure. If you just waited 15 minutes, thermal equilibrium may take longer to be re-established. Try at least an hour or two, and then dawn to noon if that doesn't work.

Cheers, Wayne

 

[CrazyRabbit]

When open circuit the cells heat up. When cells are short circuit they are same as open circuit. When solar is working properly sun energy goes into making electricity and thus they run cooler.