PV array strings

[zzdodge]

We are adding 68 425 W panels on a barn roof at a rural farm residence which has existing solar. I would like to do four strings of 17 modules (Q Peak 425W) but it appears that will overvoltage the SMA 7.0 inverters. So I am considering splitting the 17 modules in to 8 and 9 MPP strings per inverter. Unfortunately the barn is about 150 feet from the existing solar combiner, so I will have longer runs. Any other strategies which we might consider to reduce the wire runs? The module is good for about 49 volts and just over 10 amps. Since the roof is 4/12 we will not have direct insolation most of the time, so the current will be lower.

 

[Little Bill]

@zzdodge
What is your occupation?

 

[zzdodge]

@zzdodge
What is your occupation?

Engineer

 

[Little Bill]

Engineer

Thank you for letting us know. Most occupation fields were lost when we transitioned to this new software.

 

[jaggedben]

We are adding 68 425 W panels on a barn roof at a rural farm residence which has existing solar. I would like to do four strings of 17 modules (Q Peak 425W) but it appears that will overvoltage the SMA 7.0 inverters. So I am considering splitting the 17 modules in to 8 and 9 MPP strings per inverter. Unfortunately the barn is about 150 feet from the existing solar combiner, so I will have longer runs. Any other strategies which we might consider to reduce the wire runs? The module is good for about 49 volts and just over 10 amps. Since the roof is 4/12 we will not have direct insolation most of the time, so the current will be lower.

As I understand your question you are looking for ideas to avoid running 8 DC conductors instead of 4 between the array and the inverters. My first instinct is to tell you that running 8 DC conductors may actually be the best design. Taking advantage of the inverters dual MPPTs will probably result in enough extra production over the life of the system to offset the extra cost of wire. We're talking small wires here, extremely marginal compared to the cost of digging the trench I presume you need to dig. Your post is a little light on enough details to understand whether there are other relevant considerations. Here are a couple additional thoughts with no particular endorsement of the ideas as being advantageous.

1) You could possibly pair two strings of 9 on each of two inverters and two strings of 8 on each of the other two inverters. This avoids string mismatch and reduces your conductors from the array to 4. But even strings with the same number of panels will end up with mismatch that using the dual MPPTs would mitigate. At a certain point the time it takes to truly figure out the correct answer is more expensive than the amount of money riding on the choice.
2) Maybe you could put the new inverters at the barn and run AC to the existing combiner. Probably a terrible idea compared to the other options since your wires will be bigger and your voltage drop will be worse at lower voltages.

Finally, just a comment that at this scale of system the code cares nothing whatsoever about the direction the panels face and such. You need to design for max voltage at record low temperature and size conductors for 156% of Isc or whatever the exact wording of 690 requires in your code cycle. Until you get to 5 megawatts, you could install the panels upside down in a cave and the code would still require that.

 

[zzdodge]

Jaggedben,

Thank you for your considered reply. To set a few things straight. We are talking 68 panels, which is 4 * 17 modules.

Is it known if running two strings of 8 panels (about 49 Vdc each) is favorable over running a string of 9 and a string of 8 to a given SMA 7.0 inverter?
Our experience with SMA is that they provide a data sheet but little beyond that data sheet.

To clarify, the inverter will handle a max of 18 amps, so producing more (on the rare instances of optimal insolation and temperature) will merely result in clipping of the output of the inverter, and SMA has indicated this is not a problem. The max module current output is about 11 A which is well within the 18 A short circuit current rating of the inverter.

10 ga Cu wire is planned, so current limits are not a factor. However, 8 pair of 10ga will be needed for the array.

Finally, this is a simple installation in that the roof faces due south, and there are no obstructions or shading.

Any direction on the preference or efficiency of MPP string configuration is welcome.

 

[wwhitney]

I hope I'm not intruding, but your post has made me curious about a couple aspects of string inverter design:

1) If V_oc,STC = 49V, what is your design extreme minimum temperature and what is the temperature coefficient of V_oc? For example, one random 425W module I looked had a temperature coefficient of -0.286%/C for V_oc. So if your extreme minimum temperature is -15 C, that's a 40C delta from STC, meaning that V_oc would rise by 11.44%, to 54.6V. Since the SMA inverters are limited to 600V, that would mean your maximum string size would be 11 panels (which would be 600.6V in this example, but I assume that's close enough.)

2) From the discussion of 4 strings vs 8 strings, I infer that you are currently considering (4) 7 kW SMA inverters, or 28 kW_AC for panels rated at 68*425 = 28.9 kW_DC. That would be an AC/DC ratio of 1.03, which I understand is lower than the cost-effective optimum. [Unless you are interested in maximizing total power produced for a fixed roof area, without regard to the marginal cost of your marginal watt.] So have you considered dropping two panels to have 66 panels in 6 strings of 11, and just using (3) 7.7 kW SMA inverters? That assumes your computation as in (1) above allows 11 panel strings. That would give you an AC/DC ratio of 1.21.

Or if you want to stick with 68 panels, you could use 4 strings of 8 and 4 strings of 9, for string DC powers of 3.4 and 3.8 kW respectively. Then you could put one of each on a 6 kW SMA inverter, for a DC/AC ratio of 1.2 Or use two different inverter sizes, either 5 kW or 6 kW for 2 strings of 8, and then either 6 kW or 7 kW for 2 strings of 9.

Cheers, Wayne

 

[jaggedben]

Jaggedben,

Thank you for your considered reply. To set a few things straight. We are talking 68 panels, which is 4 * 17 modules.

Is it known if running two strings of 8 panels (about 49 Vdc each) is favorable over running a string of 9 and a string of 8 to a given SMA 7.0 inverter?
Our experience with SMA is that they provide a data sheet but little beyond that data sheet.

To clarify, the inverter will handle a max of 18 amps, so producing more (on the rare instances of optimal insolation and temperature) will merely result in clipping of the output of the inverter, and SMA has indicated this is not a problem. The max module current output is about 11 A which is well within the 18 A short circuit current rating of the inverter.

10 ga Cu wire is planned, so current limits are not a factor. However, 8 pair of 10ga will be needed for the array.

Finally, this is a simple installation in that the roof faces due south, and there are no obstructions or shading.

Any direction on the preference or efficiency of MPP string configuration is welcome.

With those clarifications, I'd wager that the better design is to parallel two matched strings (8 or 9 panels) at the array for each inverter. As you say, with all the panels facing the same direction, you'll get significant clipping in the middle of the day if you use both MPPTs.

Note that your sentence I've highlighted can't be answered for the general case. For example, if the strings were facing east and west at step angles and the current was peaking at different times, you might never get close to the 18A input. Whereas if you paralled two such strings to one MPPT, the different irradiance on the strings would reduce their total output most of the time,even with identical strings. But you're not dealing with those considerations here.

Two paralleled strings don't require overcurrent protection, you can just splice them together in a junction box somewhere, so no significant added cost there. The fact that you reduce the wires in your conduit running to the inverters is gravy.

Finally, SMA does have an online design tool. So they do offer more than the data sheet. You might give it a shot.

 

[zzdodge]

I hope I'm not intruding, but your post has made me curious about a couple aspects of string inverter design:

1) If V_oc,STC = 49V, what is your design extreme minimum temperature and what is the temperature coefficient of V_oc? For example, one random 425W module I looked had a temperature coefficient of -0.286%/C for V_oc. So if your extreme minimum temperature is -15 C, that's a 40C delta from STC, meaning that V_oc would rise by 11.44%, to 54.6V. Since the SMA inverters are limited to 600V, that would mean your maximum string size would be 11 panels (which would be 600.6V in this example, but I assume that's close enough.)

2) From the discussion of 4 strings vs 8 strings, I infer that you are currently considering (4) 7 kW SMA inverters, or 28 kW_AC for panels rated at 68*425 = 28.9 kW_DC. That would be an AC/DC ratio of 1.03, which I understand is lower than the cost-effective optimum. [Unless you are interested in maximizing total power produced for a fixed roof area, without regard to the marginal cost of your marginal watt.] So have you considered dropping two panels to have 66 panels in 6 strings of 11, and just using (3) 7.7 kW SMA inverters? That assumes your computation as in (1) above allows 11 panel strings. That would give you an AC/DC ratio of 1.21.

Or if you want to stick with 68 panels, you could use 4 strings of 8 and 4 strings of 9, for string DC powers of 3.4 and 3.8 kW respectively. Then you could put one of each on a 6 kW SMA inverter, for a DC/AC ratio of 1.2 Or use two different inverter sizes, either 5 kW or 6 kW for 2 strings of 8, and then either 6 kW or 7 kW for 2 strings of 9.

Cheers, Wayne

Wayne,

The desire is to cover the available roof area for this project, which provides for the 68 modules. The price savings for a 6kW SMA inverter as quoted by potential suppliers amounted to $65, so the max savings would be $130. Given the advantages of having just one model of inverter, it was decided to stick with (4) 7kW inverters.

On other panels at the same site, albeit ones which are at 45 degrees rather than 18 degrees, the snow tends to melt from top to bottom, hence the predilection of having one inverter per horizontal row. Silly, but easy to observe.

Thanks for your thoughts.

 

[zzdodge]

With those clarifications, I'd wager that the better design is to parallel two matched strings (8 or 9 panels) at the array for each inverter. As you say, with all the panels facing the same direction, you'll get significant clipping in the middle of the day if you use both MPPTs.

Note that your sentence I've highlighted can't be answered for the general case. For example, if the strings were facing east and west at step angles and the current was peaking at different times, you might never get close to the 18A input. Whereas if you paralled two such strings to one MPPT, the different irradiance on the strings would reduce their total output most of the time,even with identical strings. But you're not dealing with those considerations here.

<omitted>

The subject modules Isc is 10.78, so I think I will not parallel strings, and just use an individual MPPT for each. Granted there will be some limiting during ideal conditions. But the relative cost of adding a few panels to max out the roof area is rather small.

 

[wwhitney]

OK, but using 6 kW inverters instead of 7 kW inverters would give you a DC/AC ratio of 1.2, rather than 1.03. And while you might only save $260 in inverter costs, there would typically be savings in things that need to get sized at 125% of inverter output current.

PVWatts for Buffalo, NY, with 5% system losses, 18 deg tilt, 180 deg azimuth, and 97% inverter efficiency says the annual generation difference is 0.9 kWh/kW_DC, or 26 kWh/year for your array size. I don't think $260 is worth an extra 26 kWh/year.

Admittedly I'm not sure what to use for system losses, I think 5% is conservatively low for this comparison. If I put in 0% for system losses (implausible), then the annual generation difference jumps to 4.7 kWh/kW_DC, or 136 kWh/year. Depending on the value of electricity and the discount rate, that could be on either side of break even. Which is an unattainable best case.

Cheers, Wayne

 

[electrofelon]

Everytime I have looked at running strings back a significant distance, I have found it cheaper and easier to just run an AC feeder out and put the inverters at the array (unless of course the inverters can't go there for some reason). Assuming AL feeder and copper PV wire, the feeder will be cheaper (even considering the slight voltage advantage of the DC). AL PV wire is available FWIW.

One other thought is to use fronius inverters which will do 1KV strings, maybe that will change the calculus a bit (although it didn't for me the several times I ran the numbers).

Last year I do an 800 foot and a 600 foot run and neither could I justify running the strings back.