We’ve had a fair bit of snow this winter, which has been impacting the production of electricity from our solar installation. This was entirely expected, and according to NAIT, we should only see about 5% drop in production on average over the course of the winter due to snow. In the winter, the sun is so low in the sky and for such short days that the snow cover has a pretty minimal effect overall. Once spring hits, the winter performance won’t really matter at all!
After a pretty dismal December through to most of February, this warm spell has finally melted the snow cover. This photo was taken on the 22nd, two days before it fully melted.
There are a few interesting observations here. We can easily see that even minor coverage of a module significantly impacts the production. This is most apparent in the top row on modules 1.1.7 and 1.1.8, which have snow covering just a small portion of the bottom, but the output has dropped by half or more.
We can see that the snow guards are working effectively. In an ideal installation, there would be no snow guards and the snow would easily slide off. Due to our proximity to the playground, a robust snow retention system was installed. While this helps keep our children safe, it causes some extra snow accumulation. A worthy compromise by any standard!
The pattern of how the melting occurred is also interesting. You can see here, and I’ve observed on other days, that far left and right melt first, while the center stays covered. There may be poorer or degraded roofing insulation around the vents and chimney, which leaks heat from the building, warming the modules from below and melting the snow. The central area is clear of vents and the building insulation is probably better, resulting in more snow accumulation.
Another phenomenon we can see is that some modules, even though they are clear of snow, have very little output. You can see this most clearly in the bottom left (2.2.1 and 2.2.2) but also in the middle row (2.1.2 and 2.1.11-13). To explain this, we need to explain how the system is laid out. We have two inverters, each with two strings, for a total of four strings. For our purposes, each of these 4 strings effectively operate independently. Our inverters expect an input voltage of about 400V, and the current is the same through all modules in a string. With optimizer outputting with the same current, they need to adjust their output voltage to optimize the voltage.
On the far right, string 1.2 has all twelve modules free of snow and shadow. Each is outputting roughly the same at 33 Vdc, and the entire string is at 392V. The top string (1.1) has some shading and snow cover, so the lesser producing modules operate at a lower voltage, while the higher producing modules output at a higher voltage.
On string 2.2 (the bottom row), there are three modules outputting 60V, the max for the optimizers. There is still enough voltage to hit the minimum for the inverter, the current is very low (I calculate 0.1 Amps). This means that the production of the three clear modules is capped at 60V * 0.1A = 6W, which is close to what we see in the first image showing ~8W. If microinverters were used here instead of SolarEdge string optimizers, we would be able to get the full 80W out of each of these three modules, a difference of 68W.
Likewise on string 2.1 (the middle row), there is a bit less snow cover and the current higher at about 0.8 Amps. For the four optimizers at 59V, this means they are capped out at 59V * 0.8A = 47W, roughly what we see for the four top producing modules. For each of these, we are losing out on a potential 30W per module.
Overall, this particular issue doesn’t have much of an impact. At this snapshot in time we are looking at a loss of about 200W for a few hours a day for a few weeks a year. A rough calculation puts this at around 16 kWh, or about $1.50 per year in electricity. Clearly not a major impact, but an interesting exercise nonetheless!