Two inverters or one for different orientations?

System · January 2009

We want to put in a 8 Kw system and I've been getting quotes from several local solar installers. Because of the size and our roof layout, it makes sense to split the panels between two roof areas facing two different orientations (one southern exposure, one western). I've been told alternatively that this is no problem for a system like a Sunnyboy inverter by one company, and another company proposes installing two Kaco inverters to control the separate strings. Any thoughts?

BB. · January 2009

Re: Two inverters or one for different orientations?

See this thread for White Paper.

Short answer--if series string "A" is on east roof, and series string "B" is on west roof (and string A=B, and no panels mixed between A&B), then one inverter is fine (1% or so loss maximum).

Decision just becomes--which is the most cost effective.

-Bill

Cariboocoot · January 2009

Re: Two inverters or one for different orientations?

I think we may be confusing 'inverter' with 'charge controller'.

Brock · January 2009

Re: Two inverters or one for different orientations?

I am assuming we are really talking about grid tied inverters. I would say you are going to need two grid tied inverters anyway since as far as I know Sunnyboy only makes a 7000w max unit. Unless you don't mind loosing that last 1000w of panels you have up there. So I would go with two units one on each roof section.

If you’re just wondering about overall system performance (assuming one inverter could handle the power) I would still go with two units because the two sections could be different voltages and amperage output, especially during the morning and afternoon. It would also be easier to wire two 4000w systems and inverters rather than try to run it all in to one 7000w unit. You wouldn't lose the whole system if something should happen and could monitor each section by itself as well.

BB. · January 2009

Re: Two inverters or one for different orientations?

Vmp is mostly set by the panel temperatures... Since they are both, roughly, in the same environment, the panel outputs are similar... As one set goes into shade--both its temperature and its output fall--so any "excess" losses are minimal (per the white paper and testing).

Regarding systems size--there is panel wattage and GT inverter wattage... For example, I have 3.5 kW of solar panels, but my system is rated something like 2,997 watts (and I would have no problem adding another 10% or so worth of panels as the max wattage is only hit a few times a year on my system--and since the controller simply limits max power--there is no damage or safety issues--within limits). Placing 8kW STC rated worth of panels on a 7 kW inverter should not be any problem (again, must follow the mfgr's installation requirements).

Regarding two smaller or one larger--for this system--it probably just is based on which ever is cheaper and easier to wire. Twice as many inverters means twice as many things to go wrong too. As long as the inverters are well proven designs--either should be acceptable (I don't know the background of the GT mfg... So, I am not suggesting anything one way or the other myself).

And, I would believe we are talking about GT Inverters here--For residential use, I have not seen any commonlly available 7+ kW solar charge controllers from the few Mfg. I have seen--There probably are some out there (solar charge controller, common maximum spec. would be around 80amps * 60 volts = 4,800 watts)

mradtke · January 2009

Re: Two inverters or one for different orientations?

Hi,

I am certainly not an expert on solar power, but my system has two differently oriented arrays, so I have some experience. The advice that you have gotten so far has been from folks far more knowledgeable than I am, but I didn't notice that anyone pointed out that differently oriented arrays will have their peak power at different times of day. With the two inverter solution, each inverter must handle the peak power of its corresponding array. However, with a single inverter, the combined power at any time will always be less than the peak power of both arrays since each array will reach its peak at a different time of day. You may find it a little daunting to sort through the calculations, but you may find that a single inverter may be sized small enough to win the cost comparison.

As far as one versus two and reliability, I think that it is a wash. With two, it is twice as likely to have a failure, but when an inverter fails, only 1/2 the system will be down.

Mike

System2 · January 2009

Re: Two inverters or one for different orientations?

Yes, we are talking about a system with an AC rating of around 6.8, so a SMA 7.0 inverter should be able to handle it. The peak power question was something that the company recommending two inverters had recommended, but the other company reps claim that the power lost is less than 1%. To complicate matters further, if we went with Sunpower panels we could fit enough panels on one area, so we wouldn't have to split the array between southern and western exposures. But, the Sunpower panels would work out to be about $2000 more. The installer representing Sunpower also claims that they have better performance in high heat conditions, and we do have some days in the 90-105 range. Is this just puffery, or do some panels or inverters perform better in high heat conditions?

solartek · January 2009

Re: Two inverters or one for different orientations?

Valli,

The problem with answering your question is that there is not a solar system design tool available that can handle this type of design: parallel connected strings of panels into a single MPPT grid tie inverter where the orientation and tilt angle of the strings is different. So the best anybody can probably do is model it as two different systems with separate inverters, model it a one large system with all strings in the southern orientation, model it as one large system with all strings in the western orientation and then make an "educated" guess.

As to the question of inverter sizing, if you look at detailed plots of inverter efficiency versus % of rated operating power, most inverters have maximums in the 50% to 75% of their rated power range. I personally tend to design systems which put the Array power in those ranges. If oversizing the array, the real issue to watch out for is the maximum DC input short circuit current that the inverter can handle. This will set the maximum size of the array the inverter can handle.

Finally, if you look at the datasheets for Sunpower panels versus other crystal Si panel manufacturers, the Sunpower panels look like they have a smaller temperature dependence on their power rating (-0.38%/deg C) then most other crystal Si panels. So I would expect their temperature performance to be better.

HTH,

Scott.

niel · January 2009

Re: Two inverters or one for different orientations?

i would think the only possible power loss would come from one pv string dissipating power into the other if there's no blocking diodes to prevent this power backwash. it should not amount to much as even the lesser of the pv strings will be outputting power making any dissipated power from the other string from that differential voltage small. it's not like one string is in complete darkness and the other in full sun, but it has some comparison as when a battery might backfeed some of its power through to a pv if the pv is in darkness and no means of stopping it is in place noting that many pvs have a builtin blocking diode. only 1 such diode would be needed for each series string so it may even benefit to take out all of the blocking diodes of a string except 1. do not confuse the blocking diode with any bypass diodes as these must be left in place due to shading possibilities.

BB. · January 2009

Re: Two inverters or one for different orientations?

Actually, it is not difficult to model an east/west/more set of strings with one inverter.

The PV Calculator Tool on the final output page includes an option to download all of the data hour by hour (Output Hourly Performance Data):

"PVWATTS: Hourly PV Performance Data"
"City:","SAN_FRANCISCO"
"State:","CA"
"Lat (deg N):", 37.62
"Long (deg W):", 122.38
"Elev (m): ", 5
"Array Type:", "Fixed Tilt"
"Array Tilt (deg):", 37.6
"Array Azimuth (deg):", 180.0
"DC Rating (kW):", 1.0
"DC to AC Derate Factor:", 0.770
"AC Rating (kW):", 0.8

"Year", "Month", "Day", "Hour", "AC Power (W)"
1966, 1, 1, 01:00, 0
1966, 1, 1, 02:00, 0
1966, 1, 1, 03:00, 0
1966, 1, 1, 04:00, 0
1966, 1, 1, 05:00, 0
1966, 1, 1, 06:00, 0
1966, 1, 1, 07:00, 0
1966, 1, 1, 08:00, 0
1966, 1, 1, 09:00, 211
1966, 1, 1, 10:00, 409
1966, 1, 1, 11:00, 561
1966, 1, 1, 12:00, 640
1966, 1, 1, 13:00, 647
1966, 1, 1, 14:00, 590
1966, 1, 1, 15:00, 464
1966, 1, 1, 16:00, 303
1966, 1, 1, 17:00, 51
1966, 1, 1, 18:00, 0
1966, 1, 1, 19:00, 0
1966, 1, 1, 20:00, 0
1966, 1, 1, 21:00, 0
1966, 1, 1, 22:00, 0
1966, 1, 1, 23:00, 0
1966, 1, 1, 24:00, 0
1966, 1, 2, 01:00, 0
1966, 1, 2, 02:00, 0
1966, 1, 2, 03:00, 0
1966, 1, 2, 04:00, 0
1966, 1, 2, 05:00, 0
1966, 1, 2, 06:00, 0
1966, 1, 2, 07:00, 0
1966, 1, 2, 08:00, 0
1966, 1, 2, 09:00, 62
1966, 1, 2, 10:00, 379
1966, 1, 2, 11:00, 390

I generate output for the individual strings, then put them all together in one spread sheet. Sort on Column 1, 2, and 3 (in that order), then add the first and second entry (date, time same, last column is Watt*Hour for that time).

The sum would represent the inverter output, accurate to 1% (I would assume that the calculated output would be, typically at most, 1% high--and that would be during the evening/morning (when panel temperature differences are the greatest).

The other limitation would be the max amount of solar panels allowed by the UL/NRTL certification--The GT inverters (at least the leading brands) regulate their maximum power output themselves--so having some excess panels should not be a problem (bright cool clear day that happens a few times a year for me). Too many problems (X times rated) may get into issues of too much current available, Isc (and a waste of money). Again, check the installation manuals.

Regarding the blocking diode--I would not bother. For my 3kW GT system, it is already pretty dark (no direct sun/sun is setting--just getting "sky glow" and cloud reflections) and the panels are still outputing pretty close to Vmp--and the inverter keeps faulting because it cannot get enough current (~190-250 volts and 0.2 amps for my system) to even start converting. If the sun is already down, and my inverter is already faulting due to insufficient sun, I don't care if one of the panels is, possibly, leaking back a few hundred milliamps).

Adding diodes is another point of failure, mounting issue, insulating issue, regulatory safety issue, and a heat sink / dissipation issue. For Grid Tied / Utility Interactive Inverters--it just creates more problems for a non-problem in the first place.

-Bill

Just to answer a question that confused me at first... the "1966" year in the first column is really the date that represents the "average solar" day over the ~20 years of monitoring that is used to calculate the "average" output for your Solar System.

solartek · January 2009

Re: Two inverters or one for different orientations?

BB. wrote: »

Actually, it is not difficult to model an east/west/more set of strings with one inverter.

The PV Calculator Tool on the final output page includes an option to download all of the data hour by hour (Output Hourly Performance Data):
I generate output for the individual strings, then put them all together in one spread sheet. Sort on Column 1, 2, and 3 (in that order), then add the first and second entry (date, time same, last column is Watt*Hour for that time).

The sum would represent the inverter output, accurate to 1% (I would assume that the calculated output would be, typically at most, 1% high--and that would be during the evening/morning (when panel temperature differences are the greatest).

<snip>

Like I said, most people make an educated guess.

I'm not sure that I would assume the total would be accurate to within 1%. It is possible that we would need to use a smaller DC/AC derate factor for each of the strings to account for the fact the the maximum power point for each of the strings will be different because they are in different orientations (and maybe tilt) and who knows what the MPPT algorithm in the different inverters choose to do in this case. Maybe Solar Guppy knows, but I don't.

And this is where I think the power loss comes from, not from power in one string being dissipated in another, but from each string not running at their maximum power point potential.

BB. · January 2009

Re: Two inverters or one for different orientations?

From the studies in the Fronuis white paper:

Although the current (and hence, the power) is significantly lower from the string with the less-direct sunlight, the voltage is not as adversely affected. Even indirect sunlight will allow a PV module (or string) to achieve substantially high voltages.

As a result, tracking the strongest string allows the FRONIUS IG to optimize the total power out of this type of array as well. Since the MPP voltages of the string in direct sunlight and in less direct sunlight are largely similar, there is once again a very small difference in the net impact on power output.

Studies show that the percentage of total power lost as a result of operating on the MPP of the directly-lit string but off the MPP voltage of the weaker string is approximately 1% for crystalline silicon module. This effect would be even smaller for thin-film materials since the maximum power point is not as well-defined, meaning the difference in power at different operating points is even smaller than for crystalline silicon.
...snip...
As can be seen from the above charts, the larger difference in output is the difference in orientations itself – not the difference in operating voltage. This low difference is the result of the string that is in indirect sunlight naturally outputting substantially less power, so the total power loss achieved by operating near but not exactly on the less-directly luminated sting’s MPP is a relatively trivial value. The optimally-oriented string is still outputting 1103W, but as the sun becomes increasingly indirect (or the roof face is turned farther from optimum), the non-optimum string outputs less and less. But since the Vmpp is still extremely close between the two strings, there is very little difference (< 1%) between the best this system could do if the strings were each operating at their own optimum Vmpps and what it can do when operating at the optimum string’s Vmpp only.

And, I am being a bit factitious in saying 1% accuracy... In reality, unless you have the type of equipment that Solar Guppy owns--I would bet that a +/- 10% is about as close as anyone can measure/predict solar vs panel output on our home systems.

-Bill

niel · January 2009

Re: Two inverters or one for different orientations?

"Regarding the blocking diode--I would not bother. For my 3kW GT system, it is already pretty dark (no direct sun/sun is setting--just getting "sky glow" and cloud reflections) and the panels are still outputing pretty close to Vmp--and the inverter keeps faulting because it cannot get enough current (~190-250 volts and 0.2 amps for my system) to even start converting. If the sun is already down, and my inverter is already faulting due to insufficient sun, I don't care if one of the panels is, possibly, leaking back a few hundred milliamps).

Adding diodes is another point of failure, mounting issue, insulating issue, regulatory safety issue, and a heat sink / dissipation issue. For Grid Tied / Utility Interactive Inverters--it just creates more problems for a non-problem in the first place."

bb,
i think i am being slightly misconstrued here when i discussed the blocking diodes. my point was that that is the only losses i see that should be possible as i agree that it should be looked at as 2 separate systems added together as i believe that is how the mppt will also see it, as a combination. maybe sg or bob can correct me if i'm wrong on this. my thought is that say one string was below the threshold of operation and the other just above it that the lower string will not drag the good string down to stop operations, but rather you'll see the output operation of only the lit well enough string of pvs without any added benefit from the other(excepting backfed power due to no blocking diode). the other will contribute when the voltage from it is high enough to be at the point of operation by the inverter to be added to the output of the stronger pv string. if there's no blocking diode it can stop the operation point from being reached sooner or maintained longer because some of it will be dissipated in the other sting even though it's small.

i might also add that many pvs have provision for these diodes. the newer ones with sealed j boxes may or may not have one as one can't be sure as this probably varies with pvs, but it could be sealed in permanently by the factory if it does have one.

the losses from backfed power is real, but would be small during daylight hours of charging and during the night there aren't any batteries to worry about so if one wishes to have 1 blocking diode it wouldn't hurt anything, but many pvs have builtin blocking diodes on every pv and it could be slightly advatangeous to take all out excepting one in each string if one wants to, even though that loss would also be small in the overall scheme of things.
i am seeing this just like pvs feed mppt controllers that go to batteries as the batteries are only the load to the controller as the grid is the load to the inverter. would i suffer a loss if one pv was darkened and the other in noonday sun? only that which was not fed to the controller from the darkened pv is lost as no degradation in output from the illuminated pv should be present. i believe it's all additive without algorithm degredation unless somebody better qualified can say otherwise.

System2 · January 2009

Re: Two inverters or one for different orientations?

Hi all first post.

I am based in the U.K and use this software which allows you to add many different strings at different orientations.

http://www.solardesign.co.uk/pvsol.htm#pvsol

solartek wrote: »

Valli,

The problem with answering your question is that there is not a solar system design tool available that can handle this type of design: parallel connected strings of panels into a single MPPT grid tie inverter where the orientation and tilt angle of the strings is different. So the best anybody can probably do is model it as two different systems with separate inverters, model it a one large system with all strings in the southern orientation, model it as one large system with all strings in the western orientation and then make an "educated" guess.

As to the question of inverter sizing, if you look at detailed plots of inverter efficiency versus % of rated operating power, most inverters have maximums in the 50% to 75% of their rated power range. I personally tend to design systems which put the Array power in those ranges. If oversizing the array, the real issue to watch out for is the maximum DC input short circuit current that the inverter can handle. This will set the maximum size of the array the inverter can handle.

Finally, if you look at the datasheets for Sunpower panels versus other crystal Si panel manufacturers, the Sunpower panels look like they have a smaller temperature dependence on their power rating (-0.38%/deg C) then most other crystal Si panels. So I would expect their temperature performance to be better.

HTH,

Scott.

solartek · January 2009

Re: Two inverters or one for different orientations?

I thought the PVSOL software only handled multiple strings with different orientations connected to inverters having multiple, independent MPPT inputs. Are you saying it can simulate multiple strings with different orientations connected in parallel to a single MPPT input inverter?

Scott.

Two inverters or one for different orientations?

Comments

Categories