How much production am I losing?

[DrSmile]

I'm over paneled by what I think is a pretty standard 1.2 ratio with 13.2KW of panels feeding a Solaredge 11.4 KW inverter. The current Coronavirus emergency has screwed up my usage for the year, so I probably won't be able to add more panels despite needing more because we now have a second Tesla. To make things worse I'm the second to last house on the utility line so I am regularly below the spec voltage resulting in the inverter only pushing out 11.2KW most of the time.

My thought is to see if I can somewhat increase production by replacing the single inverter with 2 say 7KW inverters. My question is how do I calculate how much production I am losing? I know I clip 9 months out of the year with my current setup, but obviously this varies by day and sky conditions. But even for a single day like today (see graph) where I clipped for just over three hours, how do I calculate the loss as accurately as possible? I realize it may not be a lot but it may be significant enough to make a difference for me.

 

[miimura]

Just a crazy eyeballed estimate of the missing top of that plot, I think you're missing about 1.5kWh which is only 1.75% of the daily production on that day. Of course, other days may be different, but a cool Spring or Fall day is probably the worst clipping since the higher temperature of a Summer day will reduce the panel efficiency and therefore the peak output. I don't think it's worthwhile to do anything about it.

 

[MXLRplus]

What you can do sometimes, is to aim some panels more to the western sky, this reduces the peak but retains much of the kWh of production and can even increase production when you are truncated.

Note: I have done the exact same thing but with a SE20kW inverter and paired optimizers. I flattened my curve when the utility moved the TOU window to later in the day to recoup some of my $ losses.

 

[MXLRplus]

Here's what it looks like after I 'shaped' the peak out of the curve, but I do have some tree issues:

 

[DrSmile]

What you can do sometimes, is to aim some panels more to the western sky, this reduces the peak but retains much of the kWh of production and can even increase production when you are truncated.

Note: I have done the exact same thing but with a SE20kW inverter and paired optimizers. I flattened my curve when the utility moved the TOU window to later in the day to recoup some of my $ losses.

That's some great information thank you very much! I hadn't considered moving some panels but I do have an empty Western facing roof so maybe moving 6 panels there may make sense. It will obviously lower the amount before clipping in the morning but hopefully make up for it in the afternoon.

I did answer my own question regarding actual loss on this day by consulting my Calculus book, and assuming a parabolic curve the area under the curve is about 2/3 the area of a rectangle, so assuming ~13KW peak production, that's 2/3 of 1.8 x 3, or 3.6KWh. I would guess that moving the panels will lose another 1/3 of that gain, so I think it's probably not worth it for 2.4Kwh on a good day.

 

[MXLRplus]

That's some great information thank you very much! I hadn't considered moving some panels but I do have an empty Western facing roof so maybe moving 6 panels there may make sense. It will obviously lower the amount before clipping in the morning but hopefully make up for it in the afternoon.

I did answer my own question regarding actual loss on this day by consulting my Calculus book, and assuming a parabolic curve the area under the curve is about 2/3 the area of a rectangle, so assuming ~13KW peak production, that's 2/3 of 1.8 x 3, or 3.6KWh. I would guess that moving the panels will lose another 1/3 of that gain, so I think it's probably not worth it for 2.4Kwh on a good day.

If you are in a Time Of Use tariff plan, not all kWh's are created equal. In my case, kWh after 4 pm are almost twice as much $ as before 4 pm. What they did to many solar producers in my state was to move the high $ window from 12 noon to 4 pm. So moving the panels reduced my losses. At higher latitudes in the summer, it becomes even more important since the sun is still high at 4 pm.

 

[VT_EE]

I only see clipping on a few days in April or May. The rest of the year it’s fine. The amount of lost energy is negligible for the short period it happens. I certainly wouldn’t do anything about it.

 

[charlesj]

Here's what it looks like after I 'shaped' the peak out of the curve, but I do have some tree issues:

View attachment 529917

Yep, fix one, get bitten by another

 

[charlesj]

That's some great information thank you very much! I hadn't considered moving some panels but I do have an empty Western facing roof so maybe moving 6 panels there may make sense. It will obviously lower the amount before clipping in the morning but hopefully make up for it in the afternoon.

I did answer my own question regarding actual loss on this day by consulting my Calculus book, and assuming a parabolic curve the area under the curve is about 2/3 the area of a rectangle, so assuming ~13KW peak production, that's 2/3 of 1.8 x 3, or 3.6KWh. I would guess that moving the panels will lose another 1/3 of that gain, so I think it's probably not worth it for 2.4Kwh on a good day.

If you have an idea how many days you have had with that much clipping, over may years it may be worth it.

 

[charlesj]

Here's what it looks like after I 'shaped' the peak out of the curve, but I do have some tree issues:

View attachment 529917

Boy, that is a very nice bell curve, pretty even on both sides.

 

[Shygar]

I'd rather have that then my shorter window of generation. You are making more energy in the outside of the curve than you are losing at the top.

 

[Vostok]

It's good to know I'm not the only person who does this kind of thing.

I have stored all 5-minute PW2 data since 1 July last year in a spreadsheet. One thing I did was use that data to generate an 'ideal' production curve for every month. I wrote a VB macro to pull out every production data point for a given 5-minute period in a given month, finding the largest one, and taking that as the most solar that can be produced in that 5 minute period for that month (it is very unlikely that any given 5-minute period would be clouded over every single day for an entire month). Then repeat for the next 5-minute period etc. to produce an "ideal" diurnal production curve for that month.

The data's a bit noisy because Tesla stores data points to only one decimal place and of course solar production changes a fair bit over a month so the methodology is imperfect. But the results were interesting nonetheless. I don't have a full year of data yet, but here's how many minutes of clipping I get per day with "perfect" production:

• July: 100
• August: 135
• September: 170
• October: 160
  
• November: 100
  
• December: (unreliable data due to extensive and persistent bushfire smoke most of the month, I estimate ~100)
  
• January: 115
  
• February: 145
  
• March: 175

Looks like a lot - nearly 3 hours in March. But of course not every day is perfectly sunny for the entire day. In fact few are. Here's how many days each month I generated 90% or more of the "ideal" generation for that month:

• July: 11
  
• August: 7
  
• September: 4
  
• October: 4
  
• November: 5
  
• December: 2
  
• January: 2
  
• February: 0
  
• March: 2

Note I am in Australia. December and January were wiped out with bushfires. February was wiped out with rain. The closest a day's production has come to "ideal" is 95.7%. The worst was 2.9%.

Conclusion: I am not very concerned about the clipping. It doesn't happen often enough.

 

[DrSmile]

I had trouble with my grid voltage causing the system to go down all Summer last year, but even given that I have determined that I clip similar to my initial graph even in the Summer, so basically every sunny day for 6 months. I clip for 1.5 months on either side of that but to a much lesser extent, I estimate a total of about 70 days worth of clipping 3.6KWh, so that totals about 250 Kwh. That's enough to drive my model 3 1000 miles, but still only 1.5% of my annual production. In NJ that means I'm giving up 1/4 SREC every year for 10 years, so about $550 dollars there, plus at 20 cents per KW delivered that's another $500 over 10 years (not betting the inverter lasts longer than that anyways). That's not enough loss to justify changing anything major, so I'm probably going to leave everything as is.