J.P.M. Thank you!
Mounting went OK. I dropped a 1-5/8" galvanized fence post through the roof, plumbed it and secured it to a truss... it isn't going anywhere. I tried using a compass app to make sure it was facing true south, although I had a hard time finding a reference surface on the station that I trusted... it certainly isn't exact. I ended up just visually matching the azimuth of the array, which should be within a degree or two of south and good enough to get me going. Doubled up the flashing around the penetration with conventional vent pipe seals, an EPDM layer on top to protect the PVC layer underneath from UV.
The temperature sensor radiation shield (fan aspirated) is sitting a couple inches off the roof, about 8 inches north / upslope of the line of panels, somewhat west of the center of the array. The shingles are energy star, so the roof doesn't get quite as hot as it did with conventional shingles, although until I add a few thermistors (leaning towards Tartssensors at this point) I can't comment on the difference in temp under the array vs where the station is measuring it.
Data are stored in 1 min increments, but PVOutput's minimum interval is 5 so I'm manually averaging points before upload. I took some care to keep the average synchronized with the timestamps of my generation meter, and so far, it looks good. As I get deeper into it, I can build in some more robust synchronization checks. At some point, I'll probably build a database or find a hosting site more weather focused, but getting the irradiance data overlaid with the array power on PVoutput is a nice visual. Running the data through a diffuse fraction model and splitting it up into DNI/DHI is probably the first processing step I want to figure out. Once I'm confident in that, working through the HDKR with environmental contributions will be interesting.
The station didn't end up perfectly level, and I can see some some preference in the rain gauge tipping bucket, it takes about 5 ml to tip one way and about 6 ml to tip the other. I ran out of time before I could check those numbers against what the calibration should be. It is nice that they designed the radiation sensor mount to have its own leveling adjustment.
The wind cups are below the peak of the roof, and should do a decent job of representing the wind felt by the array. Although the cups are above the rain gauge, it isn't by much, and I wonder if measurement of the wind out of the south will be affected. I put the bird spikes in, but they are really close to the radiation sensor... probably doesn't make a difference, but having a surface that can reflect so close caught my attention (even if it is very small diameter and should produce very little specular reflection). The birds should prefer the power lines near my house, so I might just pull the spikes back out until it is clear they are needed.
You can see how sharply the ridge to the west of the house (and the trees on it) knock out the power in the late afternoon / evening. That will get worse as we get into spring, before cleaning back up again in summer. The shade gradually crawls across the entire array, so the generating power is incrementally dropping out one panel at a time, leading to the softer decline in overall array power compared to the sharp drop of the sensor.
I probably won't be able to get a station picture for another couple weeks, time at home during daylight hours is limited (tangentially, why TOU works so well for me!). I'm sure I made some non-optimal installation decisions, but tried to balance the serviceability and aesthetics, and still get an accurate representation of the conditions at the array.
Mounting went OK. I dropped a 1-5/8" galvanized fence post through the roof, plumbed it and secured it to a truss... it isn't going anywhere. I tried using a compass app to make sure it was facing true south, although I had a hard time finding a reference surface on the station that I trusted... it certainly isn't exact. I ended up just visually matching the azimuth of the array, which should be within a degree or two of south and good enough to get me going. Doubled up the flashing around the penetration with conventional vent pipe seals, an EPDM layer on top to protect the PVC layer underneath from UV.
The temperature sensor radiation shield (fan aspirated) is sitting a couple inches off the roof, about 8 inches north / upslope of the line of panels, somewhat west of the center of the array. The shingles are energy star, so the roof doesn't get quite as hot as it did with conventional shingles, although until I add a few thermistors (leaning towards Tartssensors at this point) I can't comment on the difference in temp under the array vs where the station is measuring it.
Data are stored in 1 min increments, but PVOutput's minimum interval is 5 so I'm manually averaging points before upload. I took some care to keep the average synchronized with the timestamps of my generation meter, and so far, it looks good. As I get deeper into it, I can build in some more robust synchronization checks. At some point, I'll probably build a database or find a hosting site more weather focused, but getting the irradiance data overlaid with the array power on PVoutput is a nice visual. Running the data through a diffuse fraction model and splitting it up into DNI/DHI is probably the first processing step I want to figure out. Once I'm confident in that, working through the HDKR with environmental contributions will be interesting.
The station didn't end up perfectly level, and I can see some some preference in the rain gauge tipping bucket, it takes about 5 ml to tip one way and about 6 ml to tip the other. I ran out of time before I could check those numbers against what the calibration should be. It is nice that they designed the radiation sensor mount to have its own leveling adjustment.
The wind cups are below the peak of the roof, and should do a decent job of representing the wind felt by the array. Although the cups are above the rain gauge, it isn't by much, and I wonder if measurement of the wind out of the south will be affected. I put the bird spikes in, but they are really close to the radiation sensor... probably doesn't make a difference, but having a surface that can reflect so close caught my attention (even if it is very small diameter and should produce very little specular reflection). The birds should prefer the power lines near my house, so I might just pull the spikes back out until it is clear they are needed.
You can see how sharply the ridge to the west of the house (and the trees on it) knock out the power in the late afternoon / evening. That will get worse as we get into spring, before cleaning back up again in summer. The shade gradually crawls across the entire array, so the generating power is incrementally dropping out one panel at a time, leading to the softer decline in overall array power compared to the sharp drop of the sensor.
I probably won't be able to get a station picture for another couple weeks, time at home during daylight hours is limited (tangentially, why TOU works so well for me!). I'm sure I made some non-optimal installation decisions, but tried to balance the serviceability and aesthetics, and still get an accurate representation of the conditions at the array.
Comment