Inverter size

Status
Not open for further replies.

wwhitney

Senior Member
Location
Berkeley, CA
Occupation
Retired
If you start with a given DC size and a DC/AC ratio of 1, and lower the inverter size, you will lose increasing losses to clipping. There will be a point at which the inverter produces less energy from the last module in the array than it clips. That's the equivalent point of diminishing returns going in that direction.
I don't see that as a particularly meaningful point in terms of tradeoffs. Panel size is arbitrary.

Remember the question is whether "clipping losses ... are ... made up for ... by increased production".
I took that as a purely electrical statement. E.g. that depending on the inverter efficiency curve, it's possible that having, say, 0.01% annual clipping losses would give you a greater total annual production, as improved efficiency in the shoulders would yield more than 0.01% annual improved production. Not sure if that's really possible, obviously it's not if inverter efficiency is constant across all input power levels below nameplate.

Now the economics are a different story. My posts that haven't mentioned dollars or economics explicitly were statements without consideration of cost.

Cheers, Wayne
 

jaggedben

Senior Member
Location
Northern California
Occupation
Solar and Energy Storage Installer
I don't see that as a particularly meaningful point in terms of tradeoffs. Panel size is arbitrary.
I'm not sure what meaning your attaching to any of this. I thought it was just a mathematical question. Also fwiw panel size is arbitrary to the general theory but meaningful when designing an actual system.

I took that as a purely electrical statement. E.g. that depending on the inverter efficiency curve, it's possible that having, say, 0.01% annual clipping losses would give you a greater total annual production, as improved efficiency in the shoulders would yield more than 0.01% annual improved production. Not sure if that's really possible, obviously it's not if inverter efficiency is constant across all input power levels below nameplate.
Yes inverter efficiency has next to nothing to do with this, and it seems like you missed the meaning of ggunn's post #10 which was the one that started this side discussion. When ggunn said 'power curve' I'm pretty sure he meant the curve of an array's (or panel's) power output over the course of the day as it varies with isolation.
 
Last edited by a moderator:

wwhitney

Senior Member
Location
Berkeley, CA
Occupation
Retired
Yes inverter efficiency has next to nothing to do with this, and it seems like you missed the meaning of ggunn's post #10 which was the one that started this side discussion. When ggunn said 'power curve' I'm pretty sure he meant the curve of an array's (or panel's) power output over the course of the day as it varies with isolation.
I understand that meaning of "power curve". The statement that seems to have led to this kerfuffle was "And clipping losses in the middle of the day are sometimes made up for, or more, by increased production on the shoulders of the power curve."

So I took "made up for" to mean either (a) you lose some kWh at midday by not having a larger inverter, but a larger inverter would lose you some kWh in the shoulders (due to the efficiency curve), so upsizing the inverter would not increase total kWhs or (b) something similar, but where the accounting is $ based depending on TOU rates, rather than valuing all kWhs the same.

But I guess now the meaning was "the clipping losses in the middle of the day (due to not spending $ on extra inverter power) are offset by the increased production during the shoulder period (due to instead spending those $ on extra panels)." Which is absolutely correct and is just another way of putting my original point today. I guess ggunn's rephrasing was not spelled out enough for me today, sorry.

Here's an old but still relevant white paper from Enphase on the matter:
That's not going to work for others, but I don't think it's necessary anymore.

Cheers, Wayne
 
Last edited by a moderator:

wwhitney

Senior Member
Location
Berkeley, CA
Occupation
Retired
And clipping losses in the middle of the day are sometimes made up for, or more, by increased production on the shoulders of the power curve.
Assuming I have the correct meaning of the above now, let me say that I agree, although I expect that for a properly economically designed system, "sometimes" would be "almost always". The almost is due to the granularity of DC size and inverter size, if they were continuously selectable, it would be always.

Cheers, Wayne
 

wwhitney

Senior Member
Location
Berkeley, CA
Occupation
Retired
Assuming I have the correct meaning of the above now, let me say that I agree, although I expect that for a properly economically designed system, "sometimes" would be "almost always". The almost is due to the granularity of DC size and inverter size, if they were continuously selectable, it would be always.
Sorry, the above is not quite right. For an economically optimal system, the extra energy in the shoulders would always match or exceed (in whatever metric is appropriate) the clipping losses. As otherwise, the other system would be a better choice. [Why I originally assumed post #10 was not about a comparison of two equal priced systems.]

Cheers, Wayne
 
But I guess now the meaning was "the clipping losses in the middle of the day (due to not spending $ on extra inverter power) are offset by the increased production during the shoulder period (due to instead spending those $ on extra panels)." Which is absolutely correct and is just another way of putting my original point today. I guess ggunn's rephrasing was not spelled out enough for me today, sorry.




Cheers, Wayne
FWIW Wayne, I also questioned that statement gunny made, and almost called it out as being wrong, so you are not alone. I guess I also interpreted it with different conditions and "different starting point". I agree with your quoted part above as being accurate.

The way it usually works for me, is I have a fixed DC size, and must choose an inverter. As the inverter size is lowered, clipping will start and energy production will decrease (of course).. the question becomes are the savings in a smaller inverter large enough that it is worth the hit on production.
 

winnie

Senior Member
Location
Springfield, MA, USA
Occupation
Electric motor research
Just considering the system I am looking at for my home, the limiting factor is not DC array KW, but rather array square feet.

I suppose like any optimizing problem, the direction that things go to hit 'optimum' can change radically depending upon initial constraints.

IMHO the most we can say to the OP is 'an inverter that clips is often a perfectly fine choice but the devil is in the details'.

-Jon
 

ggunn

PE (Electrical), NABCEP certified
Location
Austin, TX, USA
Occupation
Electrical Engineer - Photovoltaic Systems
...the devil is in the details'.
So true. We once had a customer who looked at his egauge monitoring every day; he called us one day when he saw a little flat spot on the top of his power production curve and thought it meant that he needed to replace his inverter with a larger one. We looked at a year's worth of his production and ran the numbers, we showed him that if he upsized the inverter it would take him something like 50 years to break even.
 

wwhitney

Senior Member
Location
Berkeley, CA
Occupation
Retired
IMHO the most we can say to the OP is 'an inverter that clips is often a perfectly fine choice but the devil is in the details'.
I would go a bit farther than that: since inverters aren't free, the economically optimal DC/AC ratio will always be greater than 1.0. Now with the granularity imposed by discrete inverter and panel sizes, the achievable ratio closest to economically optimal may be less than 1.0.

In other words, if the design methodology is pick the number of panels, then pick the inverter, a good starting point for the optimization is to choose the inverter that gives you a DC/AC ratio of 1.0 to 1.3.

Cheers, Wayne
 

ggunn

PE (Electrical), NABCEP certified
Location
Austin, TX, USA
Occupation
Electrical Engineer - Photovoltaic Systems
I would go a bit farther than that: since inverters aren't free, the economically optimal DC/AC ratio will always be greater than 1.0. Now with the granularity imposed by discrete inverter and panel sizes, the achievable ratio closest to economically optimal may be less than 1.0.
PV modules deployed in the field rarely if ever experience STC and many of them never even come close. We occasionally design systems with DC:AC < 1.0, but it's never because of potential clipping losses. Sometimes due to roof area and available inverter(s) it just works out that way.
 

pv_n00b

Senior Member
Location
CA, USA
So true. We once had a customer who looked at his egauge monitoring every day; he called us one day when he saw a little flat spot on the top of his power production curve and thought it meant that he needed to replace his inverter with a larger one. We looked at a year's worth of his production and ran the numbers, we showed him that if he upsized the inverter it would take him something like 50 years to break even.
I remember when talking about allowing clipping was blasphemy. You did not lose a single electron those expensive modules produced. Now it's like, yeah just throw up a few extra modules to make sure and call it a day. :)
 

winnie

Senior Member
Location
Springfield, MA, USA
Occupation
Electric motor research
It totally makes sense.

The modules have gotten cheaper and cheaper. The 'balance of system' is significantly more expensive than the modules.

Hell north facing modules now make sense in some situations.

Jon
 

pv_n00b

Senior Member
Location
CA, USA
Walking around and seeing an old residential install with the modules racked up off the roof to point due south and at the exact angle needed to maximize production makes me think about the old days when array design was more than how many modules can we fit on the roof.
 

inforaj

Member
Location
Chiago
Occupation
Member
You probably know this, but wherever you are connecting three strings to an MPPT channel, all three must be individually fused at 15A.
You mean each MPPT will be 15A X 3 = 45A. The SunPower had a wire connector 2 to 1 fused 15A.

HARNESS, DC STRING, 2-TO-1, FUSED, 15A, POSITIVE POLARITY,
HARNESS, DC STRING, 2-TO-1, FUSED, 15A, NEGATIVE POLARITY,
HARNESS, DC STRING, 1-TO-1, FUSED, 15A,
 

ggunn

PE (Electrical), NABCEP certified
Location
Austin, TX, USA
Occupation
Electrical Engineer - Photovoltaic Systems
You mean each MPPT will be 15A X 3 = 45A. The SunPower had a wire connector 2 to 1 fused 15A.

HARNESS, DC STRING, 2-TO-1, FUSED, 15A, POSITIVE POLARITY,
HARNESS, DC STRING, 2-TO-1, FUSED, 15A, NEGATIVE POLARITY,
HARNESS, DC STRING, 1-TO-1, FUSED, 15A,
I'm not sure what that means, but if you are connecting three strings to an MPPT channel, all three strings need to be fused even if there are two strings on one input and one on the other. If you are under the 2020 NEC you do not have to fuse both the + and- conductors anymore, but I don't remember when that change came in.
 
I'm not sure what that means, but if you are connecting three strings to an MPPT channel, all three strings need to be fused even if there are two strings on one input and one on the other. If you are under the 2020 NEC you do not have to fuse both the + and- conductors anymore, but I don't remember when that change came in.
I am confused, maybe just a typo but you can clarify gunny?
 

ggunn

PE (Electrical), NABCEP certified
Location
Austin, TX, USA
Occupation
Electrical Engineer - Photovoltaic Systems
I am confused, maybe just a typo but you can clarify gunny?
Sure. On the SMA Core 1 inverters there are 6 MPPT channels and each channel has two inputs which are unfused and wired in parallel. If you are using high voltage low current PV modules, you may not be able to put enough modules in a string for 12 strings to fill the inverter capacity. The Isc may be low enough that you can put three strings on a single input, though, but if you do that you need to fuse all three strings whether you connect them all to one input or two strings to one input and one string to the other.
 
Sure. On the SMA Core 1 inverters there are 6 MPPT channels and each channel has two inputs which are unfused and wired in parallel. If you are using high voltage low current PV modules, you may not be able to put enough modules in a string for 12 strings to fill the inverter capacity. The Isc may be low enough that you can put three strings on a single input, though, but if you do that you need to fuse all three strings whether you connect them all to one input or two strings to one input and one string to the other.
Ok so by "input" you just mean a physical termination that is common with other "inputs", all on the same MPPT?
 

inforaj

Member
Location
Chiago
Occupation
Member
Sure. On the SMA Core 1 inverters there are 6 MPPT channels and each channel has two inputs which are unfused and wired in parallel. If you are using high voltage low current PV modules, you may not be able to put enough modules in a string for 12 strings to fill the inverter capacity. The Isc may be low enough that you can put three strings on a single input, though, but if you do that you need to fuse all three strings whether you connect them all to one input or two strings to one input and one string to the other.
1644268833572.png
HARNESS, DC STRING, 2-TO-1, FUSED, 15A, POSITIVE POLARITY,
HARNESS, DC STRING, 2-TO-1, FUSED, 15A, NEGATIVE POLARITY,
HARNESS, DC STRING, 1-TO-1, FUSED, 15A,

These are the standard available.
 
Status
Not open for further replies.
Top