Combining Solar Panel Strings facing different directions - Virtual Tracking

richardimorse

When combining strings with a different azimuth in an off grid system, lets say NNE facing with North Facing and NNW facing (Southern Hemisphere), you catch the morning and evening sun better, this is called "Virtual Tracking" where fixed panels can start charging the batteries sooner and more effectively in the morning and can manage the load for longer without draining the batteries in the evening, thus enabling your battery operating window to be shortened.

Now my question is regarding the electrical characteristics of this practice, if you were to combine the panel strings facing in different directions in the same sub-array, combiner and charge controller, 1-combiner = 1 charge controller for off grid.

One assumes that each panel string (i.e. the ones facing different directions) produces a different current output at the same voltage, if this is the case, I guess their is no problem (although at the very start and end of the day, the voltage might not be the same which could cancel out the benefit of doing virtual tracking in the first place).

I guess that if the panels facing different directions (Azimuth) were grouped together on the same combiner and charge controller (i.e. different directions use a different combiner and charge controller pair), there would be no problem and EVERYTHING is telling me that this is the only best practice way to implement virtual tracking to avoid imbalance and any electrical back-feeding between the panel strings.

Lets assume three ±45Voc panels are in each string and three strings are combined into one 150V MPPT charge controller, e.g. Outback FlexMax 80, what is going to happen electrically if we mix strings with panels facing different directions, i.e. one facing North, one facing NNE and facing NNW, in the morning the NNE panel string is going to wake up the charge controller before the N and NNW panel are electrically active and the CC will then perform MPPT calculations using the NNE string only

Then their is the issue of potential voltage imbalance during the early morning hours between three strings on the same combiner to consider while the N and NNW facing strings get their electrical activity ramped up from zero to the Vmppt level.

Estragon

I think what will happen is in the morning, the E facing array will warm in the sun and voltage will drop to ~80% of STC Vmp.  With ambient indrect light, the W array will rise to ~STC Voc, depending mainly on ambient temp.  The two arrays have quite different MPPs, so a single controller won't likely find an optimal max power point.  Exact behavior would dependin part on the specific programming (settable on some, such as Midnite classics).

In other words, separate controllers for each array needed for optimal virtual tracking IMHO.

mcgivor

The modules need very little light to produce voltage but require more direct sunlight to actually produce current, having differently oriented sub arrays on a single controller would allow a longer harvesting period than a single array, of the same module count facing in the same direction, albeit at a lower overall current value. So it really comes down to the maximum amount of current  the batteries can absorb. It's generally better to spread the charging over a longer period of time, because the  net result would be that the batteries would have to support loads,  unassisted for a shorter period of time. This  would have the adverse effect on opportunity loads however, one size fits all doesn't apply.

 Having two controllers, each with its separately oriented array would probably perform marginally better than the single controller arrangement, but the added cost has to be considered. The  amount of headroom, PV voltage versus battery nominal voltage, would be a key factor in the equation. Say for example PV voltage is is 140V with a battery of 48V nominal, the headroom  would still allow the MPPT controller to utilize the available input voltage  to down convert  to the required nominal voltage, in a dual array arrangement.

The question probably has numerous answers which may differ dramatically, opinions may vary, but in the end they're just opinions, no more no less. 

Dave Angelini

Anyone can have an opinion. It is pretty much a no brainer to me to do this the right way. This forum is really great for penny pinching.
Not so great sometimes for giving advice based on evidence because so few actually have done the testing or live offgrid all of the time.

In practice what happens when you do not have separate mppt is the mppt will be at the wrong voltage or hunt. On a day a cloudy day with clouds moving is the best way to test this. 

The other point is to just do it because if you are a technical person, you know it is the right way. 

I do alot of testing here because it is what I do for my business. I have run testing for Sanyo and Sunpower. I use 3 separate 2kw azimuth tracked arrays.

The results show me that it is worth doing the right way. It does cost more, I do care if it gets my battery full on a pretty bad day.

Estragon

It doesn't even have to cost much more.  You could use two MN Kid CCs instead of one Classic, for example, for about the same total cost.

richardimorse

 Estragon said:

I think what will happen is in the morning, the E facing array will warm in the sun and voltage will drop to ~80% of STC Vmp.  With ambient indrect light, the W array will rise to ~STC Voc, depending mainly on ambient temp.  The two arrays have quite different MPPs, so a single controller won't likely find an optimal max power point.  Exact behavior would dependin part on the specific programming (settable on some, such as Midnite classics).

In other words, separate controllers for each array needed for optimal virtual tracking IMHO.

This is what I am planning to do.  The current system uses one array of 2100Wp and one array of 2625 Wp on two separate charge controllers, this is producing less power than I would have anticipated only 17.9kWh per day and part of this may be due to the fact that the installer (7 years ago) mixed the NNE, N and NNW facing panels on the same charge controllers.

I am decommissioning 525Wp (1 string) to make 4200Wp (24 * 175Wp) on a single charge controller (4200 - 7 years at 1% degradation = 3900Wp) and I plan to re-orient all strings in the existing array so that all of it faces North.

I am also installing two new arrays at 3900 Wp (12 * 325Wp for each array), I plan to install the two new arrays facing NNE and NNW, keeping everything clean and balanced with the same panel azimuth and zenith angle and panel type on 3 separate controllers with 3900 Wp power on each charge controller.

What I am wondering is what is going on at the moment electrically with the existing panel strings mixed on the same charge controllers and facing different directions, the panel diodes are probably protecting them from back-feeding voltage from other panel strings (when their is a string voltage imbalance) but I wonder what the other electrical implications are.

Speculation
The worst case scenario could be if the combiner and mppt algorithm cause one string of panels on the combiner to operate at 100% while the others remain fairly idle during most of the day and that the panel string operating at 100% changes (i.e. tracks the sun) during the day.

Mppt tracking only happens at panel power up by default unless configured otherwise on the charge controller.

Dave Angelini

"Mppt tracking only happens at panel power up by default unless configured otherwise on the charge controller."

Most decent CC's continuously track and adjust voltage based on the requirements of the battery/loads.

Estragon

Doesn't make much sense to me for any mppt CC to only track at power up.  Surely as the panels warm up, and the sun angle changes, the MPP changes too?  

mcgivor

What was the anticipated Kwh figure?  If the original configuration produced 17.9Kwh from a total of 4725W, calculating using a derating of 75 %, 3543W it would mean an equivalent of full power for  ~5 hours spread across the day. 

This was with both sub arrays on a common controller, it would be interesting to see what the total harvest would be with two controllers, to get those figures. Despite all the theoretical predictions regarding MPPT tracking etcetera  it would appear to be working reletivaly well.

To achieve an accurate comparison two identical systems in terms of array, one with a single controller the other with two run similtaniously, recording and compiling data over a given period of time wIll reveal definitively the better arrangement. Short of this everything is hypothetical, but I do believe that hypothetically, the advantage would favor the two controller setup.
Edit. There is some discussion on another forum with regards to virtual tracking using a single MPPT device some may find interesting 
https://forums.energymatters.com.au/solar-wind-gear/topic5064.html

Estragon

Interesting discussion (though long and a bit wandering, not that we ever do that )

In that case virtual tracking on a single mppt may have performed reasonably well because the arrays were apparently straight east and west at steep tilts, and the results were for near summer solstace.  In effect, the array opposite the sun would be near totally shaded.  At low latitudes, such an arrangement might work reasonably well.  The mppt would likely find a good mpp for the unshaded array, and ignore the shaded one.

I'm not sure it would work as well with nne and nnw orientations OP was considering.  Experimentation might be wise/required.  If at low latitude, more of an ene/wnw or straight e/w may work better?

mountainman

Estragon I think the op is in the southern hemisphere. When i was in Virginia late Sept the lower sun angle behind trees mountain etc. I was only  getting 3 hrs 12:30 to 3:30 sun where my panels were. So I moved half the array 2 100 watt panels 2series 2 parrelell  to a location where the sun was 10:30 to 1:30. When the sun was on just 2 panels 11 amps. then from  12:30 to 1:30 sun  on all 4 panels 22 amps then back to 11 amps at 1:30 to 3:30. This was on a single mppt. Seemed to work for me. 

jonr

> any electrical back-feeding between the panel strings

You don't need to worry about this.

See below for data on virtual tracking working well to extend time (with a single controller):

https://forums.energymatters.com.au/solar-wind-gear/topic5064.html

IMO, extra watts in the morning is important to get lead acid batteries through the bulk charging stage.  Falling off in the evening is less of an issue (absorb doesn't need so many watts).

Estragon

@mountainman - that sort of fits with what I was thinking, if two panels were pretty deeply shaded while two were in good sun.  Basically the mppt ignoring the shaded panels.

Where I was thinking there could be an issue is if both pairs were producing, but with different mpp.  This might be the case with arrays just east or west of south (or north, for OP).  

solardad

@Estragon

Not sure I follow why you think there would be an issue if there were two MPPT controllers, one for each array pointed in the opposite direction (assuming a master/slave setup would address differences in the arrays).  Were you thinking the controllers were not linked together? 

Estragon

Two controllers for two arrays is fine - my system is done that way.  The potential issue was if a single controller was trying to find max power point on two producing arrays in different orientations.  

It looks like it should work okay if one array is producing little/no current as the controller seems to find something close to mpp on the producing array.  What I'm unsure of is the effect of having two arrays, both with enough light to produce meaningful current, but with different mpp (temperature differences, etc).  

jonr

Specifically, the combined system IV curve can have a local maxima that isn't the same as the global maxima and some poor algorithms might select the local maxima.     But I see reports of it working fine and none for it not working right in the case of east/west parallel arrays (not to be confused with series connections of partially shaded panels - that's quite different).

solardad

@Estragon got it. Thanks for confirming 

Dave Angelini

The arrays harvest need good static mpv and dynamic. There are plenty of cases where one controller fails to do this right. The article below was from 2010 when we were field testing the first 600v mppt. The shade tolerant feature was pretty amazing and it worked well in moving clouds.

Estragon

@Dave Angelini

Interesting stuff... well worth the read.

BB.

Thank you Dave for the article... I have added it to the FAQ thread:

https://forum.solar-electric.com/discussion/4426/working-thread-for-solar-beginner-post-faq/p3

-Bill

Dave Angelini

You both are welcome! Seasons greetings and here to with lot's of snow in the southern Sierra. Bone dry in October.

richardimorse

jonr said:

> any electrical back-feeding between the panel strings

You don't need to worry about this.

See below for data on virtual tracking working well to extend time (with a single controller):

https://forums.energymatters.com.au/solar-wind-gear/topic5064.html

IMO, extra watts in the morning is important to get lead acid batteries through the bulk charging stage.  Falling off in the evening is less of an issue (absorb doesn't need so many watts).

It is a good point, the charge controller will only provide a maintenance charge to the batteries and not a charge to support the loads, I am thinking of what inverters do when supporting load passthru from a generator and not what charge controllers do when the batteries are fully charged and are in Absorb/Float.  

The charge controllers are not coming out of absorb/float during the evening period in any case once the batteries are charged.  So would I be correct in saying for virtual tracking the morning period is MUCH more important than the evening period.

- In the morning Batteries start charging in BULK and supply morning peak demand period and first 3.5 hours more effectively.
- Batteries reach full charge during the day and so evening period only provides a maintenance charge.

Dave Angelini

Each offgrid application has its own specific requirements. Some want to run cooling into the evening. Some want to pump water as long as possible. Many want to minimize the need for a genset and want the winter arrays all pointing at the sun path.

The benefit of electrical tracking over virtual is it is all done automatically.

richardimorse

The maintenance cost of electrical tracking is much higher than expected over the lifetime of the system (no of years)
Nowhere near the 20 year lifetime of the solar panels or the 5 to 10 year lifetime of the batteries, even simple things like the materials used requires intense scrutiny as many of these systems use steel which will rust badly as opposed to aluminium and as it has moving parts needs a lot of TLC compared with the rest of the solar system, also wind is a major problem, as costs of panels comes down I don't see that tracking systems are required unless you are space limited.

Dave Angelini

Maybe what you say is true in your part of the world but there are hundreds of electrical tracked utility projects in the this country in the 1000's+ of megawatts. I pretty much disagree with everything in your post so I will just let it go. 

richardimorse

Utility projects are not exactly maintenance free like your average off grid system aspires to be, but then again off grid systems should also have regular maintenance as well, in remote areas people seem to install off grid solar as fit and forget, until something goes wrong, they don't even bother changing the filters in the generator and then wonder why the oil starts leaking.

Personally, I wouldn't like to be responsible for tracker maintenance after seeing what a Lorentz Etatrack system designed for solar parks looks like after 6 years in a coastal installation with salt water in the air, I am sure they function well with inland locations, then again my car wouldn't do too well if parked at the beach every day for 6 years without a wash down.

Dave Angelini

I use Array Technology and have most of my career. Getting around 40 years now in Solar and have seen it all.

 The days of virtual tracking because the panels were costly and there were very few trackers, the days when the tracker became the way for max power over the length of day, to now where the overall cost makes economic sense.

It it alot of work to squirt a grease gun twice a year and some of the actuators need an LPS lube ever 5 years or so. Arrays can be ordered in stainless, galvanized, or painted. Array does this for residential and commercial tracking.

jonr

Using PVWatts, I get a 15-20% improvement in December kWh with a two axis tracker (vs both axis fixed).   Also depends on the fixed angle you pick.  Pretty steep if you are optimizing for December.  Sometimes even vertical for snow shedding.

Dave Angelini

A two axis tracker does not need an angle choice Jon. A single axis does rodger?

The improvement for Offgrid is not really the power increase as we can't use most of that when we are full.

For Offgrid it is much more about the lengthening of the power day or when the clouds clear 2 hours before the end of the solar day.
Two hours of clear skies at the start of the solar day and then clouds fill in and we are full or close.
Two hours of good solar in the middle of the day not such a big deal.

It can save the use of gensets or minimize their use. You do have to have the solar power in large quantity though and often a large fixed array and a single axis tracker work very well together offgrid. I do that all of the time for my clients.

littleharbor2

If you have, say, a home based workshop you can take advantage of opportunity loads all afternoon.

My neighbor in Baja has a Freon powered dual axis tracker. It's been working for 20 years in the, sometimes, very salty air. He has always made liberal use of the grease gun and it still works as smooth as silk.

Dave Angelini

Even on the shortest day in the hemisphere!littleharbor2   At our place it is battery free power after 10 am except when it snows