Roof Mounted 11.25 kW System Review & Questions

[J.P.M.]

well another mistake is using the wrong location, so say they put it in southern CA by mistake by copying an existing design..
Seen that too.

Anecdotally, when I was working, I'd see mental lapses such as that. After I got to be the boss and eliminated radios/earphones, I still saw them, but the number of such errors seemed to decrease, but vigilance and calc/spec./drawing checking still went on. Go figure.

[ButchDeal]

I appreciate what you write. In this case however, running PVWatts without any shading ought to show about or within a couple/few % of the same output as another model running with no shading, provided both produce what are deemed reliable results - whatever that means. I'd think shading on another model would produce less output than the same model run without shading.

Anyway, if I was the OP, I'd get a sunpath diagram of site shading and then, if ambitious, get familiar w/ SAM (PVWatts on steroids), include the shading from that model and see what it has to say.

well another mistake is using the wrong location, so say they put it in southern CA by mistake by copying an existing design..
Seen that too.

[J.P.M.]

Oh we have seen other quotes that are all over the place. Common to just set up the house without any shadows, sometimes they set it up as if it is facing due south, leave off roof obstructions. Most common is to leave off neighboring buildings which sometimes cause pretty good shadows.
All of these short cuts in the modeling will cause over estimate in production

I appreciate what you write. In this case however, running PVWatts without any shading ought to show about or within a couple/few % of the same output as another model running with no shading, provided both produce what are deemed reliable results - whatever that means. I'd think shading on another model would produce less output than the same model run without shading.

Anyway, if I was the OP, I'd get a sunpath diagram of site shading and then, if ambitious, get familiar w/ SAM (PVWatts on steroids), include the shading from that model and see what it has to say.

[ButchDeal]

Thank you.

I appreciate folks cut corners but I've not seen too many solar vendors that over estimate output. If more did there would probably be fewer oversized and more smaller but more cost effective systems around.

Maybe the folks who pay that designer cut corners and bought cheap/inexperienced help.

I certainly agree with the GIGO principle perhaps being at work in this case.

Oh we have seen other quotes that are all over the place. Common to just set up the house without any shadows, sometimes they set it up as if it is facing due south, leave off roof obstructions. Most common is to leave off neighboring buildings which sometimes cause pretty good shadows.
All of these short cuts in the modeling will cause over estimate in production

[J.P.M.]

The application is free to use. As you know installers often cut corners and/or are not well trained in modeling and often make poor models.
no application is that does a model is going to give accurate results given poor inputs.

personally we have found excellent results with Aurorasolar and have tried many modeling tools, but it is not cheap.

Thank you.

I appreciate folks cut corners but I've not seen too many solar vendors that over estimate output. If more did there would probably be fewer oversized and more smaller but more cost effective systems around.

Maybe the folks who pay that designer cut corners and bought cheap/inexperienced help.

I certainly agree with the GIGO principle perhaps being at work in this case.

[ButchDeal]

Thank you for the response.

So, if the OP wanted to duplicate the SolarEdge modeling work, any thoughts on how he might go about it ?

The application is free to use. As you know installers often cut corners and/or are not well trained in modeling and often make poor models.
no application is that does a model is going to give accurate results given poor inputs.

personally we have found excellent results with Aurorasolar and have tried many modeling tools, but it is not cheap.

[J.P.M.]

Hard to say without duplicating the Solaredge design to insure it was laid out correctly and in the right locaton.

Thank you for the response.

So, if the OP wanted to duplicate the SolarEdge modeling work, any thoughts on how he might go about it ?

[ButchDeal]

I'm having a hard time seeing how SolarEdge gets that much annual output (16.27mWh/yr.) at that location.

The OP used PVWatts' model's standard defaults for the PVWatts inputs including a system loss value of 14.06%, 3% of which of which, according the help screens is for shading loss.

I reran the OP's input, changing the panels to premium, the system loss parameter to 10 % and the DC/AC ratio to 1.125.

Since the ground coverage ratio is somewhat poorly defined but generally meant (it would seem) to apply to arrays with rows of panels arranged in sawtooth fashion and perhaps, but not mandated to be ground mounted, and rereading the NREL stuff on power optimizers and shading, I used a ground coverage ratio of 1.0.

With those change the PVWatts model output increased from 13,792 kWh/yr. to 14,743 kWh/yr. Using the PVWatts default ground coverage of 0.4 does not change the PVWatts model output for my runs.

I'm not sure how the SolarEdge designer got 16.27 MWh/yr. annual output. 1.45 MWh/yr annual output per STC kW seems kind of high for VT, even if no shading is present. That designer may have more info about site shading than I do, but w/the PVWatts value of 3%, which I effectively more than eliminated by reducing the system loss parameter to 10% from 14 % (and which, if optimistic would tend to err in the direction of any optimizer enabled performance restoration of some shading loss), I don't see where SolarEdge can restore performance by 10 % from what's already an unshaded modeled output that's done in a way that's been a reliably accurate representation of system output in the past.

A few% +/- maybe, but 10 % ?

Hard to say without duplicating the Solaredge design to insure it was laid out correctly and in the right locaton..

[J.P.M.]

Because you left the defaults.

https://www.solaredge.com/us/node/8195

and you likely should also set the module type premium

I'm having a hard time seeing how SolarEdge gets that much annual output (16.27mWh/yr.) at that location.

The OP used PVWatts' model's standard defaults for the PVWatts inputs including a system loss value of 14.06%, 3% of which of which, according the help screens is for shading loss.

I reran the OP's input, changing the panels to premium, the system loss parameter to 10 % and the DC/AC ratio to 1.125.

Since the ground coverage ratio is somewhat poorly defined but generally meant (it would seem) to apply to arrays with rows of panels arranged in sawtooth fashion and perhaps, but not mandated to be ground mounted, and rereading the NREL stuff on power optimizers and shading, I used a ground coverage ratio of 1.0.

With those change the PVWatts model output increased from 13,792 kWh/yr. to 14,743 kWh/yr. Using the PVWatts default ground coverage of 0.4 does not change the PVWatts model output for my runs.

I'm not sure how the SolarEdge designer got 16.27 MWh/yr. annual output. 1.45 MWh/yr annual output per STC kW seems kind of high for VT, even if no shading is present. That designer may have more info about site shading than I do, but w/the PVWatts value of 3%, which I effectively more than eliminated by reducing the system loss parameter to 10% from 14 % (and which, if optimistic would tend to err in the direction of any optimizer enabled performance restoration of some shading loss), I don't see where SolarEdge can restore performance by 10 % from what's already an unshaded modeled output that's done in a way that's been a reliably accurate representation of system output in the past.

A few% +/- maybe, but 10 % ?

[billboe]

thanks for the followup emartin!

another "tidbit" question...

can i use liquid-tight non metallic conduit on the outside of my house? it would run between the the solar meter, the disconnect, and the main panel. of course, i could use PVC but flexible would be more flexible...

if i go the PVC route, is schedule 40 sufficient? or do i need sch 80?


thanks all!

p.s. renvu just confirmed my order... no shipment details yet...

[emartin00]

I was thinking they made rain tight elbows, but maybe they don't?
I guess you would just go with a rain tight threaded connector to a threaded elbow or LB

https://www.homedepot.com/p/1-in-Electrical-Metallic-Tube-EMT-Rain-Tight-Connectors-62511/100548279

The Blue color is typically rain tight or rated for wet locations, which is what you are looking for.
You'll also want to use thread sealant on your threaded connections to keep water out.

The connector in your image below is not rain tight. I found it at HD and it says in doors or in concrete.

[billboe]

(1) You'll want to use a rain-tight pull elbow or LB on the conduit so you can go horizontal. Then enter into the J-box on the side with a rain-tight fitting. Nothing too complicated.

(2) I would stick with 2 strings. Having 2 strings at the inverter allows easier trouble shooting if you ever have issues in the future.

(3) Update your inputs in accordance with what SE recommends in the link Butch provided. That will probably get them closer.

thanks emartin!

re: (1), yeah, i was thinking a rain-tight elbow on the EMT but i'm not sure which kind of fitting to use (especially one that is rain tight). something like this (see below) may work if i went directly into the j-box but it isn't clear that it is rain-tight.

Halex EMT 90.jpg

an LB may be good but they are either threaded or have a set screw. for threaded, would i need a compression to threaded connector? and i can't imagine the setscrew version being rain-tight!

you wouldn't have specific fittings in mind, would you? THANKS!

for (2), i hadn't thought of the troubleshooting advantage of keeping the strings separate. thanks!

finally (3), i did play with the inputs some but it still more than 10% difference... really, i'm not worried about it at this point! the panels are ordered!


thanks again!!!

[ButchDeal]

thanks butch! so i guess that means SE designer should be more accurate...??? i presume SE doesn't subtract anything for shading or snow (etc.)...?

it would on that part of the simulation.

SolarEdge would definitely subtract for shading. Optimizers are not going to negate shadows but they will get as much as possible with shadows.

[emartin00]

(1) You'll want to use a rain-tight pull elbow or LB on the conduit so you can go horizontal. Then enter into the J-box on the side with a rain-tight fitting. Nothing too complicated.

(2) I would stick with 2 strings. Having 2 strings at the inverter allows easier trouble shooting if you ever have issues in the future.

(3) Update your inputs in accordance with what SE recommends in the link Butch provided. That will probably get them closer.

[billboe]

Because you left the defaults.

https://www.solaredge.com/us/node/8195

and you likely should also set the module type premium

thanks butch! so i guess that means SE designer should be more accurate...??? i presume SE doesn't subtract anything for shading or snow (etc.)...?