Friday, 15 February 2013

More revelations from a snowy roof

Another snowy night, and around 20 cm on the roof this morning. This made my trip to the balcony rather exciting, although I was reassured as there was also 20 cm of snow on the terrace below to break my fall.

The sun was trying to poke its head through the clouds soon after it climbed over the mountains, so I cleared the bottom row of panels on the roof before breakfast, then waited for something to happen. There was no generation for quite some time, even when the sky was the blue, and as the sun was out and climbing. By 9:30 large drifts of snow had started sliding off the roof and crashing into the garden. There was still no electrical action, so I went out to look at the roof and around a third of it was clear of snow. Then I went up to the loft to see what the two power conditioners were doing. They weren't doing a lot. I tried switching them off and on again. I think this may have done something but perhaps it was just a coincidence.

I went back up there a little later and the one on the left was generating 0.22 kW while the one on the right was generating over 2. I had assumed that these power conditioners each dealt with half the panels on the roof, the one of the left dealing with the panels on the left, and the one on the right the panels on the right. The roof has 48 panels, Eight high and six wide. The connectors are at the top and bottom, so it makes most sense to connect them in series vertically. I remember they were talking about connecting the panels up in sixes, although I can't find any written evidence of this now. From what I remember, the bottom six panels of each row were connected vertically, then the top two rows of panels were connected in two arrays three wide and two high. A total of eight sets of six panels.

My only explanation for the difference in the two power conditioners is that the top two panels were both connected to the left power conditioner, leading to a lower generation for the panels at the top that were still covered in snow since the snow was falling from the bottom of the roof. Once all the snow melted from the roof, and the generation was higher, the power conditioners were generating around the same amount.

I have noticed that the two power conditioners generated different amounts. This can be explained by three things: differences in the performance of each panel, different lengths of wire and different temperatures.

Each panel produces a slightly different amount of electricity. Although they are rated at 190 watts, they vary in power between 190 and 200. In theory, if all the low generators are in one circuit and the high generators are in the other, there could be 5 Watts difference between the power going into the left power conditioner and the right one, but it's much more likely that the difference will be small and no more than 1 or 2 watts. Even if it is 5 Watts difference, that's only 0.05% out of the 9.12 kW.

The wires to the bottom of the panel are longer. Also, and conversely, there will be more wire in a straight vertical array of six, than an array two high and three wide. More wire means more resistance. Power loss is the resistance times the square of the current. The maximum current of the panels is rated at 5.62 Amps. I'm not sure what gauge they used, but if they used the sums in this EcoWho solar wire sizing calculator , they will have come up with a AWG 12, 2 millimetres gauge, 3 square mm, which according to the Engineering toolbox has a resistance of about 5 milliohms per metre. This would lose 0.17 Watts per metre. Altogether, the panels lower on the roof may have 10 metres more cable, then that's a 1.7 watt difference. This is going to be around 0.05% too.

Since air is flowing under the panels, taking heat off each one, the temperature of the panels at the top of the roof is going to be higher than those at the bottom. The air temperature in the channel gets over 60 degrees in the summer, and there could be a 5 or even 10 degree difference in the panel temperature. Since electrical efficiency drops by around 1% every degree or two, this could drop the output by a 1 or 2% for the circuit with the top arrays compared to the bottom.

Now I don't have any live data for the different generation going through each of each power conditioner, but if you press the right button, rather than the on/off switch, they display the cumulated generation, which is 7,198 kWh for the one on the left, and 7,262 for the one on the right. This is a difference of around 0.8%, so it looks like they connected the two top arrays into the same power conditioner. I suppose the odds of this were even if they had stuck the wires in at random.

When we were talking about the connection of panels I had tried to encourage some kind of optimisation in shortening the wires as much as possible, but the eyes of the contractor started to glaze over and they brushed away my suggestions of optimisation over the next half century of generation in favour of being able to make things easy for the afternoon they were clambering around on my roof.

Knowing what I now know about the rating of the power conditioners being an absolute limit rather than a rough level, so we lose power when the panels get any where near their maximum output, I would have probably pushed more strongly for six vertical arrays of eight panels.

The Power Conditioners will take up to 370 Volts and 24.5 Amps, although it's rated at 250 V, at which it is most efficient. Its maximum power is 4kW at 30 degrees C, and 3.2 kW at 40 degrees C.

The panel maximum voltage is 36.6 Volts. So you could put 10 in series and the voltage would still be under 370 volts. Six panels is only going to be 220 V; less than the rating. Eight panels would be 290 V if they are all producing their maximum voltage. Of course, they're not always going to be producing their maximum voltage, so that's likely to come out around 250 Volts. So I'm not sure why they were putting them together in sixes.

Six circuits of eight rather than eight circuits of six would have reduced the amount of wire on the roof by about 20 metres, which would account for about 0.1% of the power. This doesn't sound a lot, but when you multiply by 50,000 yen, it's one lunch per month. It would also have made the fitting substantially easier, but perhaps they had a good reason for doing it in that way. I can't see any charge per metre of wire used on the invoice, so that's not it.

Maybe when the ten year contract with the electricity company runs out and we look at alternative heat generation, we can fix the wiring on the panels at the same time.