Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Technical Solar PV question

This site may earn commission on affiliate links.

nwdiver

Well-Known Member
Feb 17, 2013
9,407
14,630
United States
This is a rather technical question... I'm helping a friend build a 15kW array tied to a 10kW inverter. The plan is to have 6 strings. 2 Strings will face S, 2 Strings will face SW and 2 strings will face W. The thought is that we can get ~10kW over a longer period of time without investing in a tracking system. Since voltage is dependent on temperature and current is based on insolation each string should have roughly the same voltage and since the strings are in parallel the different current levels will be irrelevant. Thoughts?
 
If you live in the US the best bet is the face South - this will give you the most hours of sunlight and production. The westerly panels who probably perform the worst depending on your location.

The objective here isn't most overall production... you are correct that South is best for that... TX pays $0.075/kWh exported but charges $0.12/kWh imported... so a kWh that is 'self-consumed' is worth >50% more than a kWh that is exported => producing MORE kWh in the evening is more valuable than producing more kWhs...
 
NREL's PV Watt calculator estimates hourly energy generation, and it allows one to tinker with all of the relevant variables, including azimuth. I'd suggest you play around with this application to tune the system your friend has in mind. My off-hand guess is that you'll end up with a narrower band of directions, e.g. 175°, 190° and 205°.

If you're going to be really sophisticated about this, you'd also want to look at the real-time price pattern at the location to optimize the economic value of the power, rather than maximizing the kWh. That might be too challenging, though, and subject to change as generation patterns evolve.
 
This is a rather technical question... I'm helping a friend build a 15kW array tied to a 10kW inverter. The plan is to have 6 strings. 2 Strings will face S, 2 Strings will face SW and 2 strings will face W. The thought is that we can get ~10kW over a longer period of time without investing in a tracking system. Since voltage is dependent on temperature and current is based on insolation each string should have roughly the same voltage and since the strings are in parallel the different current levels will be irrelevant. Thoughts?

While I expect you have looked into this, you will want to confirm that your inverter is capable of dealing with the excess power. Why aren't you putting in a larger inverter, to make use of the full amount of production?
 
While I expect you have looked into this, you will want to confirm that your inverter is capable of dealing with the excess power. Why aren't you putting in a larger inverter, to make use of the full amount of production?

One of the UL1741 standards is that inverters must be capable of handling 2x the power rating of the array... the inverter technically isn't even 'handling' the power, it simply moves the MP point making the array less efficient. So a 15kW array with max power voltage of 380v would simply operate at 440v with significantly reduced current.

More detail on array over-sizing here

There are several reasons an oversized array can be beneficial; 15kW AC is not allowed under TXUs program; Your cost per kWh can actually be lower with an oversized array since panels have gotten so cheap... you gain >15kWh/day with more panels and only lose <2kWh/day on average by having an undersized inverter. Plus kWs in excess of 10kW are almost always exported so they are 50% less valuable than the kWhs gained in the morning and evening with the addition of more panels.

My question is more geared towards any physical reason the S-SW-W array arrangement wouldn't work... I can't think of one. So long as I don't have any panels on the same string facing different directions it should work.
 
Are you trying to wing it? Solar Pathfinder will do the engineering for you. Might want to find someone NABCEP certified to model it for you and save yourself a lot of hassle and risk.

I wouldn't say I'm 'winging' it... I've built oversized arrays in the past... there's no risk to the inverter unless you put too many panels in series, in which case power is irrelevant... or you get a short.

I think that modules have been so expensive for so long that the emphasis on capturing every kWh is really ingrained in the industry. I know that SW is economically advantageous to S since the AC always runs in the afternoon but rarely in the morning. I also know that an oversize ratio of 1.3 is advantageous if modules are less than double the cost of the inverter per watt.

What I don't know is what kind of losses to expect from facing 6 strings in 3 slightly different directions tied to an inverter with 1 MPP tracker... slight I would think.
 
One of the UL1741 standards is that inverters must be capable of handling 2x the power rating of the array... the inverter technically isn't even 'handling' the power, it simply moves the MP point making the array less efficient. So a 15kW array with max power voltage of 380v would simply operate at 440v with significantly reduced current.

More detail on array over-sizing here

There are several reasons an oversized array can be beneficial; 15kW AC is not allowed under TXUs program; Your cost per kWh can actually be lower with an oversized array since panels have gotten so cheap... you gain >15kWh/day with more panels and only lose <2kWh/day on average by having an undersized inverter. Plus kWs in excess of 10kW are almost always exported so they are 50% less valuable than the kWhs gained in the morning and evening with the addition of more panels.

My question is more geared towards any physical reason the S-SW-W array arrangement wouldn't work... I can't think of one. So long as I don't have any panels on the same string facing different directions it should work.

Thanks for the background and reference to the paper. Our system is a bit more than 20% overpowered, and I had been told that we were approaching the limits of the inverter to deal with the excess power (but that was five years ago and the technology is no doubt continuing to evolve).
 
So I did a little exercise using PV Watts for Houston (using IAH weather). I used that to calculate the output of a 5 kW array pointing from 160 to 270 degrees in 10-degree intervals. I was a bit lazy and used a common tilt at each direction, 27°.

This exercise showed that there is remarkably little to be gained by using multiple angles. The plot below compares the output of 3 groups of 5 kW arrays over the course of the day (average over the year).
The blue line shows the output when all three are set at 180°.
The green line shows the output when all three are set at 230°.
The red line shows the output when one set is at 170°, another at 220°, and the third at 270°.
Array Diversity Annual.png

It looks to me that the desired flattening of output isn't happening as much as one's intuition would suggest, and if you want to time-shift the output, it's more effective to rotate the entire array.
 
Last edited:
@Robert.Boston - So orienting an entire uniplane panel at 270* is pretty much the same as 180* assuming you wanted to 'evening shift' your generation for personal reasons? Also, putting your same test panels on single post twist 'n tilt standards would do what to daily output?

@nwdiver - so keeping strings on the same plane is still a good rule of thumb as it avoids various issues?

Trying to understand some of the basics here.
--
 
nwdiver,
The System Advisor Model on the NREL site will give you the answer. The program allows you to input the modules, strings, inverters and azimuths of each string. If you have your hourly kW consumption data from your utility, you may upload that as well. SAM also has a rate plans database where you can specify your utility and rate plan. If you take the time to learn the program, it will give you an extremely detailed financial analysis of your proposed system. It's also easy to save different models and do a what if with different azimuths/string combinations.
 
@Robert.Boston - So orienting an entire uniplane panel at 270* is pretty much the same as 180* assuming you wanted to 'evening shift' your generation for personal reasons? Also, putting your same test panels on single post twist 'n tilt standards would do what to daily output?
Trying to understand some of the basics here.
This chart might help demonstrate what's going on. It plots the average daily output for all 12 azimuths I modeled (e.g. A160 -> Azimuth 160°). You can see that anything off 180° lowers output overall but shifts generation (some).
Array Spectrum.png

Note that I've only modeled 160° to 270°, so we're not exploiting the morning quite as aggressively as the evening.

If the solar array tracked the sun, you could get the outer envelope of these curves (plus something similar in the morning). At this location, that gives about 16% more power between noon and sunset. Given the cost of such a tracking system (both upfront and repairs), my guess is that it would be cheaper to put 20% more panels up and use static mounts.
 
Personally I would go for two axis tracking with less panels. Unless you have full sun at the peak of the day that day's production is severely impacted. Having the panels on trackers can save your day if the cloud cover is only present in the morning or the afternoon. Most estimates for effects of trackers don't account for weather pattern issues. I can't even count the number of days where I could have made a lot more power had the clouds cleared a few hours prior when the sun would be at peak angle. Trackers keep the panels at peak angle all day long. Cut the number of panels to be in line with your available inverter capacity and spend that money on a twin axis tracking system.
 
My understanding of the PV Watt model is that it uses actual weather data to infer the insolation level. The charts I've posted are for Houston, but it would be straightforward to calculate them for other locations.

I'm also making the (big) assumption that all kWhs are equal. That assumption might not hold in all cases, depending on retail rate tariffs, etc.
 
The ability to deploy a tracker would be best suited for what you are trying to do. I have two trackers in the yard and 14 panels on the roof.
The tracker configuration produces anywhere from 30% to 40% at any given time. The panels can also be reoriented for the summer and winter months as well.
I have two inverters both are SunnyBoy SMA products. 1 2500 and 1 3500 14 Sharp 200 watt and 12 Sharp 230 watt panels on the trackers.
I also have a small 4 battery bank with charge controller and small inverter which feeds power back into the system at night.
Hardware is pretty inexpensive these days you can't loose no matter which route you take
 
My understanding of the PV Watt model is that it uses actual weather data to infer the insolation level. The charts I've posted are for Houston, but it would be straightforward to calculate them for other locations.

I'm also making the (big) assumption that all kWhs are equal. That assumption might not hold in all cases, depending on retail rate tariffs, etc.

That's why I invested the time to model my system with System Advisor Model System Advisor Model (SAM) |

I already have 18kW of PV installed which has eliminated my on peak use and I wanted to evaluate the payback by installing enough solar to cover all of my usage. PV Watts was giving me unrealistic, overly optimistic results because I already have some solar installed and I am on a time of use plan. Since I have eliminated peak use with solar and demand shifting, my actual cost per kWh is 8.5 cents. PV Watts model was showing savings more than double my actual utility bills.

When I ran a SAM model with enough PV to cover all of my usage, it did yield a reasonable payback of 8 years and LCOE of less than 8c/kWh. Regrettably, my service entrance is insufficient to cover that much additional PV. If I factor in the additional expense of installing larger service entrance equipment, the return goes tilt. If I install just as much PV as allowed by my service entrance or about 9 kW, my utility bills would increase because I would have to change to a standard rate plan instead of TOU. Otherwise, I would over-produce during peak hours and receive a buy back of less than 3c/kWh for the excess at year end. I don't like the idea of selling kWh's to APS for 3c/kWh and APS in turn selling these KWh's to my neighbor for 25c/kWh. Alternately, I could simply increase my on peak usage to match my production.
 
Thank you very much Robert... that was extremely helpful... I'm now leaning towards a ~13kW array facing 230 degrees.

I tried to use PV Watts but instead of a graph it spit out a bunch of numbers... what did you use to make the graph?

PV Watts does have some issues... it doesn't appear to be 'clipping' production at 10kW even after selecting a 10kW inverter under the advanced options tab. Come 'on NREL we NEED you.
 
Thank you very much Robert... that was extremely helpful... I'm now leaning towards a ~13kW array facing 230 degrees.

I tried to use PV Watts but instead of a graph it spit out a bunch of numbers... what did you use to make the graph?

PV Watts does have some issues... it doesn't appear to be 'clipping' production at 10kW even after selecting a 10kW inverter under the advanced options tab. Come 'on NREL we NEED you.
There's an option to dump hourly numbers into a .CSV file. I did so and used Excel to manipulate the data.
 
Solar Is Cheaper Than Electricity From the Grid in 42 of 50 Largest U.S. Cities

As this recent report points out, we are at the tipping point where a wholesale shift to solar power is now cost competitive. See: http://nccleantech.ncsu.edu/wp-cont...-Ranking-Solars-Value-to-Customers_FINAL1.pdf

Most Americans are unaware of the true financial value of solar today. Seen by many as a technological luxury, solar energy is not seriously considered as an option by most homeowners in the U.S. However, our analysis shows that, in 46 of America’s 50 largest cities, a fully-financed, typically-sized solar PV system is a better investment than the stock market, and in 42 of these cities, the same system already costs less than energy from a residential customer’s local utility. Of the single-family homeowners in America’s 50 largest cities, we estimate that 9.1 million already live in a city where solar costs less than their current utility rates if they bought a PV system outright – and nearly 21 million (93% of all estimated single-family homeowners in those cities) do if low-cost financing is available.

So why aren’t more Americans investing in solar? There is a clear information gap, and with this report, we intend to open the eyes of average homeowners by showing that solar can generate both significant monthly savings and long-term investment value, and not infrequently, cost less than energy from some of America’s largest electric utilities. What’s more, it is our hope that people will come away realizing that solar is now not just an option for the rich, but a real opportunity for anyone looking to take greater control over their monthly utility bills and make a long-term, relatively low-risk investment.