Tweet
Thank you. The picture helps.
Also, I now have your tabular data that I didn't have before that helps me to better understand what might be happening. My guess looking at the tabular numbers, that while a NW orientation is unfavorable for PV to the tune of maybe 20% less annual output than a more southerly orientation, the 13 panel array doesn't look differentially shaded as much as the 10 panel array seems to be, with the 10 panel array having more differential shading among those 10 panels. My guess is that the 13 panel array does not have much differential shading among those 13 panels, one to the next, but the 10 panel array does.
I suppose high electricity rates in HI as well as the lower latitude can reduce the penalty in cost effectiveness brought on by a lousy orientation, but I still wonder if I'd do a NW orientation if I was in HI. But NOMB.
As for PVWatts and output: Do not be under the impression that PVWatts can predict output over short periods. It's a model for residential PV design based on long term "representative) weather. It is not a predictor of performance, particularly short term performance. It's a model that uses data that is (anyway) mostly synthetically generated (not in most any way actual data for over 90% of the data used) to be used to provide long term performance estimates over many years. As the PVWatts help screens will state, any single month's output from the PVWatts model can vary from an array's actual production for the same calendar period by as much as +/- 30% or so. That you got close is not unusual, and the volcano eruption may well have contributed to the output, but to think PVWatts can, in effect, predict long term weather is no more accurate than thinking it's possible to predict next month's weather by any method.
It's a lot easier for me to visualize how you would have a shading problem once I got a look at the tabular data, and looking at that data now, I tend to agree with Butch on the shading, at least for the 10 panel array. The 13 panel array may and probably is partially shaded, but to me that shading looks more orientation driven except for some possible shading from the trees on the east side.
While not trying to be the statistics police, your sample size is too small to be using standard deviation and assume it's a valid parameter. Something called a "T" distribution might have some validity if you're interested, Another way to view variation in a qualitative sense might be to look at each array's average panel output divided by each corresponding array's range of panel outputs (highest output - lowest output) as a measure of panel/panel variation over an array. For the 13 panel array that # is ~ 0.10. For the 10 panel array, that number is ~ 0.20.
I'd go out and look at the arrays during various times of the day and keep track of any shading. And, FWIW, keep track of hourly total output and compare it to the PVWatts hourly output on clear days and look for large differences in clear day hourly output which might be an indication of shading.
Apologies for the late response. Something unexpected potted up.
Good luck.
Also, I now have your tabular data that I didn't have before that helps me to better understand what might be happening. My guess looking at the tabular numbers, that while a NW orientation is unfavorable for PV to the tune of maybe 20% less annual output than a more southerly orientation, the 13 panel array doesn't look differentially shaded as much as the 10 panel array seems to be, with the 10 panel array having more differential shading among those 10 panels. My guess is that the 13 panel array does not have much differential shading among those 13 panels, one to the next, but the 10 panel array does.
I suppose high electricity rates in HI as well as the lower latitude can reduce the penalty in cost effectiveness brought on by a lousy orientation, but I still wonder if I'd do a NW orientation if I was in HI. But NOMB.
As for PVWatts and output: Do not be under the impression that PVWatts can predict output over short periods. It's a model for residential PV design based on long term "representative) weather. It is not a predictor of performance, particularly short term performance. It's a model that uses data that is (anyway) mostly synthetically generated (not in most any way actual data for over 90% of the data used) to be used to provide long term performance estimates over many years. As the PVWatts help screens will state, any single month's output from the PVWatts model can vary from an array's actual production for the same calendar period by as much as +/- 30% or so. That you got close is not unusual, and the volcano eruption may well have contributed to the output, but to think PVWatts can, in effect, predict long term weather is no more accurate than thinking it's possible to predict next month's weather by any method.
It's a lot easier for me to visualize how you would have a shading problem once I got a look at the tabular data, and looking at that data now, I tend to agree with Butch on the shading, at least for the 10 panel array. The 13 panel array may and probably is partially shaded, but to me that shading looks more orientation driven except for some possible shading from the trees on the east side.
While not trying to be the statistics police, your sample size is too small to be using standard deviation and assume it's a valid parameter. Something called a "T" distribution might have some validity if you're interested, Another way to view variation in a qualitative sense might be to look at each array's average panel output divided by each corresponding array's range of panel outputs (highest output - lowest output) as a measure of panel/panel variation over an array. For the 13 panel array that # is ~ 0.10. For the 10 panel array, that number is ~ 0.20.
I'd go out and look at the arrays during various times of the day and keep track of any shading. And, FWIW, keep track of hourly total output and compare it to the PVWatts hourly output on clear days and look for large differences in clear day hourly output which might be an indication of shading.
Apologies for the late response. Something unexpected potted up.
Good luck.
Comment