Mixing orientations on grid-tie inverter

theenergyguy

Does anyone have any experience or know of any research done wherein a single grid-tie inverter is installed with one string facing east and the other west? Yes, conventional wisdom says "no!" Understood.

I've only done it once myself about 3 years ago for a customer who was (and fortunately still is!) a longtime friend and he said the actual performance was better than he thought it would be (he's got monitoring and knows how to read it).

Yes, Microinverters would be an obvious solution but at greater cost and, yes, two smaller inverters would also be obvious but, again, at greater cost. So I'm looking for any real experience anyone else may have had - or studies anyone else may have seen) wherein this sort of arrangement has been quantified in some way.

RCinFLA

Re: Mixing orientations on grid-tie inverter

Not a good idea. MPPT will go for non-optimal point.

There are some GT inverters that have two separate inputs that individually track MPPT for two different arrays.

There is alway two inverters solution.

stephendv

Re: Mixing orientations on grid-tie inverter

Fronius have done some studies on identical strings at different orientations, through 1 inverter: http://www.fronius.com/cps/rde/xbcr/SID-7644B1A7-A3C03366/fronius_usa/PDF_fronius_ig_reaction_to_non_optimal_conditions.pdf

See the section "Similar Strings at Dissimilar Orientations". In a nutshell, as long as both strings have the same nominal voltage the losses through a single inverter will be minimal.

"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."

Solar Guppy

Re: Mixing orientations on grid-tie inverter

The Fronius paper is BS, its some pencil pusher writing a white paper, not actual measured results.

The issue is the VMP of the east and west strings will be different, not due to irradiance but due to the temperature differences. Panels being in direct sun can easliy have a 25-30C temperature difference than ones significantly off angle, same for the roof temperatures the panels are mounted over.

Will it work ?, sure, will it be optimal, no! and unless your going to hook up power analyzers, temperature probes and other gear to be able to compute the true VMP, you will never know what your losing ... and I have done all of the above as part of my work.

solarix

Re: Mixing orientations on grid-tie inverter

The small added cost of dual inverters, microinverters, or a distributed inverter (SolarEdge) will far better than the loss of power caused by mixed orientations.

stephendv

Re: Mixing orientations on grid-tie inverter

Solar Guppy wrote: »

The issue is the VMP of the east and west strings will be different, not due to irradiance but due to the temperature differences. Panels being in direct sun can easliy have a 25-30C temperature difference than ones significantly off angle, same for the roof temperatures the panels are mounted over.

Ah temperature, interesting! Can the MPPT algorithm make a difference here? E.g. SMA have been vaunting their Optitrack algorithm (http://www.sma.de/en/products/knowledge-base/made-for-shade.html) which is supposed to minimize the partial shading issue.

drees

Re: Mixing orientations on grid-tie inverter

Sounds like what we need to do is to get 2 setups and compare outputs:

1. One array angled East, one West on one inverter sharing MPPT tracking
2. One array angled East, one West on two separate inverters or an inverter with separate MPPT tracking per array

Plot the voltage, current and power output of each over some period of time and compare.

The best setup will be the one that gives you the most power per dollar.

Solar Guppy - sounds like you may have the results of such a test already?

theenergyguy

Re: Mixing orientations on grid-tie inverter

So far no one has quantified the effect of a single (2 string) inverter with measurements I've seen (I think Solar Guppy's right - the white paper he refers to doesn't appear to explicitly say the mixed orientation results are from measurements), which is only way to really answer the question " are alternatives worth it". Two smaller inverters are a whole lot more, microinverters are more competitive but a few percent lower than most string inverters to start with (Enphase's new D380 unit is helping to reduce microinverter cost over string inverter).

I've heard both sides on 1 & 2 above (one side saying losses are minimal, the other side saying they're significant). Seems like the only way to tell conclusively is to monitor different inverter setups with the same (1/2 east, 1/2 west) array, such as:

1) Fronius or SMA with one string east & one string west
2) Microinverters
3) Two smaller inverters

Corrections for irradiance & temperature may be necessary if the same array was used in each case on different days. Otherwise you'd need 3 seperate but perfectly matched arrays to be really accurate.

Solar Guppy

Re: Mixing orientations on grid-tie inverter

Yes I have done this type of testing and the issue is the variables involved become to site or vendor specific to give hard numbers, at least to publish a paper about.

I have the equipment to know exactly what the vmp should be which allows me to know why a device ( charge controller or GT inverter ) is mis-tracking ... this is a fundamental part of my work

If you assume a device that's between the solar and loads is perfect in tracking one could model dis-similar vmps, but its not a linear function, below the vmp point power drops somewhat linear, above its logarithmic.

The vmp difference of East vs West using a 25C temperature difference will be around 10% which could mean anywhere from 10-30% power harvest loss on one of the arrays, depending on how good the mppt tracking is,

This might translate into a 5-15% combined loss, again depending on the Mppt, Temperature delta.

The point I'd make is the Fronius white paper is obviously written from someone who has no actual lab/field experience with photovoltaic's, probably written by a sales "engineer" to support the use of that companies product.

As others have posted, proper designed system would use one of the following:

• An Inverter for each array
• An Inverter that has independent inputs
• An Inverter for each panel

The Fronius claim of 1% loss is silly as using two arrays, in the same orientation won't have that type of efficiency due to mismatches in per-panel vmps and field install variable ( wiring , mounting , temp variations from inside to mid of an array ect )

Dave Angelini

Re: Mixing orientations on grid-tie inverter

"above VMP it is a log funtion" Thanks SG ! I was wondered about that when I would watch loads come on in float. You would see the CC track the array voltage down for the load. It did not seem like much voltage for an array when the load went from 100 watts to 1500 watts. Log function makes sense!

drees

Re: Mixing orientations on grid-tie inverter

Solar Guppy wrote: »

This might translate into a 5-15% combined loss, again depending on the Mppt, Temperature delta.

Thanks - so for planning purposes let's assume 15% worst case scenario and you are building a 5kW array using components from NAWS for a total of $19k.

To make up the difference, you'd have to swap out the $3.4k 5.1kW inverter for 2 2.5kW inverters at a cost of $4k or about $600 more. The extra labor/wiring will likely cost at least a couple hundred more, too., so let's call it $1000.

For $1100 you could buy 2 more of the Kyocera 205w panels bumping up the power from 4920W to 5330W or about 8%.

So if we are assuming 15% loss from the separate arrays, it's clear that you will be better off using 2 inverters purely on a dollar/watt ratio but if the loss is not as bad as the worst case, it will be pretty close.

My comparison is very simplistic - you'd want to run the numbers more carefully for sure - but at least at first glance it appears to be more cost effective to split the arrays.