Extreme Shading!

[Outtasight]

Now heres a question... Given that the energy that isn't being taken out of the array presumably ends up as heat, how much and for how long will that heat gain reduce the efficiency of the panels when the shading is removed?

The bypass diodes prevent the cells from heating up when shaded (bypassing the power around the shaded bank of cells) and being in the shade, the Sun isn't heating them either.  The bypass diodes get warm but they're in the junction box so don't heat the cells.  They don't get very hot though so aren't a factor.

One unknown factor is that of mechanical longevity of the modules when subjected to partial, but hard cyclic shading.  A small hard shadow "walking" across a bunch of modules every day for 1/3rd of the year will cause a walking expansion stress in the glass and other components (maybe most significantly the solder joints between cells).  The effect is a slow one but repeated hundreds of times over 25 years.  It could cause a mechanical stress failure in the module.  Normally a module is designed to expand and contract with uniform sunlight on it (or uniform shading from clouds).  I'm not sure if they've modelled or designed for expansion waves crossing the surface, causing some portions of the surface to contract while others are expanding.

My cheapo plastic framed TopRay 15W panels suffered multiple failures of the module terminal wiring due to expansion cycling (or maybe just wind stress). It caused a very weak copper pad, badly fixed to the glass substrate, to crack and detach, making the whole panel go open circuit.  I've had at least two panels fail in this way (only one confirmed by "autopsy", that was out of warranty and I repaired it).  The failures occurred within the first 6-18 months of operation!  I flogged all the others at a car boot sale before they went bad too.

[billi]

Madness on the other-hand that  PV panels have to be in series for on Grid ideas  = dramatically affected by shading

Is it not possible to produce higher voltage modules (thin film do by the way ) that one could parallel all panels ? To reduce shading and the need off several inverters or MPPT trackers ?

Or why is the industry not producing an Inverter that works with say 44 Volt Modules or 88 volt (as a pair ) to feed the grid , so all panels on their own could be paralleled  ?


Billi

[StBarnabas]

Billi
I agree it is crazy. I have 936 cells all in series (26 panels x 36 cells per panel). When I was setting out to design the StB system I was hoping to use about 4 separate strings but soon discovered that even with two strings the inverters I researched would not even fire up most of the time. Having 936 cells in series is asking for trouble. Fortunately my own shading is caused by trees (all of which I now own) so pollarding is on the agenda..

[Outtasight]

Dunno why the industry still persists with central inverters either...

Enecsys http://www.enecsys.com, a UK based (US owned) company, is a maker of module integrated micro inverters.  A paper they wrote (take figures with a pinch of salt) talks about the relative merits of string based high Voltage DC central inverters and module based AC inverters.

http://www.enecsys.com/downloads/MicroInverterEconomics.pdf

200W micro inverter product sheet (they have a range that goes up to 280W).  G83 certified, but they're silent on MCS approval
http://www.enecsys.com/downloads/datasheet_SMI-200-G83.pdf

Looks promising.  They work at Voltages suitable for single "24V" modules (that usually have a max power Voltage of 35V) or pairs of "12V" panels in series, or most of the higher power (>220W) grid tied modules that are in existence now (must be >27Vmp).

Initial capital outlay is estimated to be 10% higher than the cost of a comparable central inverter but other costs are lower due to no DC wiring, the attendant DC disconnect switch.  They also claim a 20 year warranty on the inverters (SMA et al only offer 5-10 years).  Seems they did this by coming up with a design that eliminates the use of electrolytic capacitors (the most unreliable type).

Add to that the much higher shade tolerance (actually each panel can face a different direction too, so E-W systems are possible) and it looks like a winner (if they can get it MCS approved).  You string each PV module in parallel together on the AC line.  They communicate with a monitor using data over the AC line.

Each inverter is independent, has its own MPP tracker and if one module is shaded, the others carry on producing at optimal power.  The electronics in the inverters are less stressed, as they're working at 27-45VDC, not 600VDC.  Also means the system is much more expandable.  You can add one PV module at a time and you never hit an inverter capacity ceiling (as you would with a central inverter).

One European research group is looking at the opposite though... Very high Voltage thin film modules (600V) that allow a transformer-less high voltage three phase inverter to be used.  Not aimed at the current generation of PV modules, that only work at 17-36V per module (or 60V for amorphous). More for the next generation of building integrated PV.

http://www.pvmips.org/over.html

[billt]

I'm in the process of getting quotes for a system and have looked at the Enecys inverters. I've decided on a conventional inverter as the array will not be shaded at all so the USP doesn't apply, the roof is very difficult to access, so maintenance or replacement of a faulty inverter would be difficult and expensive. A standard inverter can be put in the garage; very easy access. A good conventional inverter is slightly more efficient.

Despite the claims of high reliability I have my doubts about 18-20 modules working at high temperature and relatively high power running for 25 years without faults developing.

[Outtasight]

Maybe access is something that is important.

In considering the probability of a failure, there is a lot to be said for having one of something. 

If a central inverter has a probability of 1/1000 units developing a fault in 25 years of service, then you have a 1/1000 chance of you getting the one that dies and needs replacing inside of 25 years.

If you have 20 micro inverters which are twice as reliable as a central inverter, then each one presents a 1/2000 risk of a failure in the 25 years.  But because you've got 20 of them, the chance of any one out of the 20 developing a fault is 20/2000, or 1/200...  Not so good.  For the micro inverter system to be more reliable overall (no failures in 25 years), each individual inverter would have to be 20 times as reliable as one central inverter.  At 1/20,000 failures, if you have 20 of them, the resulting probability of a failure would be 20/20,000 or 1/1000.  Obviously, the more micro inverters you have, the worse the failure rate of the system will be (a system failure being a failure of ANY one of its components).

Enecsys argued that a central inverter is a single point of failure and that having n inverters means that if a micro inverter fails you lose 1/n functionality in the system.  That's true, but it still means going up on to the roof to replace it, rather than the loft, garage, or utility room to do the job. 

The micro inverter system is more resilient, in that it can suffer some component failures without the "service" failing (where the "service" is providing some electricity reliably)...  This is common design practice in IT systems, where multiple identical servers are used in a parallel cluster, rather than one super-computer.  One member of the cluster can fail and the service (web shopping transactions, or whatever) can continue (at reduced capacity) until the failed server is repaired.  It is the service that is more reliable, as a result of the extra complexity of having more servers.  Having more servers actually makes the likelihood of something going wrong more likely to happen, but it lessens the impact of it happening.

The same applies to the Enecsys system.  One module failing does not mean you have no electricity, just less than expected.  If a system was specified as being "a minimum 3000Wp capacity, guaranteed for 25 years", an installer could build in some redundancy.  If n modules with integrated inverters will deliver the 3000Wp capacity, but there is a significant risk of a failure within 25 years, the installer may decide that it is cheaper to install n+1 modules.  That way, one could fail in 25 years and he wouldn't be called out to repair it (as the contract was still fulfilled, to deliver a minimum 3000Wp capacity for 25 years).  The extra module may cost him £600 to install, but if it avoids a cost to him of at least £600 in a call-out to replace a failed module on a hard to access roof, then it would be economically viable.  Of course, he could just install n modules and hope they all work for 25 years...  But as you'll have now guessed, the more installations he does, the higher the total rate of failure amongst all of them will be... So what he saves today on not installing a redundant module in each job, he may lose in the future on wasted trips to repair failed modules under his service warranty and the cost of damage to his business reputation (from selling services that do not reliably meet the terms of the contract).

Or he may have long since run away with the money and tells the customer "Nuffing to do wiv me guv'... Talk to Mastervolt, it's their problem.".  Heard that one recently 

[rhys]

In the US a company Clarian appears about to launch a "plug in" micro inverter system.
Clearly woundn't meet our regs but micro inverters and PV's as plug in consumer appliances does have some attractions.
Not sure about the  turbine graphic htough!!

[kenilworthian]

To follow up my initial post about the effects of shading here is the result of splitting my string of 12 panels into 2 strings of 6 - one upper the other lower. Gone is the Diehl Ako 2100 now replaced with two SMA 1200s.

As anticipated the upper string is earning its keep, whilst the the lower is struggling.

Whothy of note is the drop in the upper string output at around 10:45 not recovering until around 14:45.  I suspect the tinest shadow from my neighbours TV antenna pole just visible in the 12:00 photo. Actually at this time it is aligned with a panel joint and the output is seen to jump.

Thus if you have any doubt that shading might affect your proposed installation know that if you don't have a shadow-free roof - it will.

My next action is to establish whether there may be damage being caused to my panels.  I would appreciate any discussion and technical evidence on the subject.

Meanwhile another photo-montage for your edification

[StBarnabas]

Ken
many thanks for this. Another excellent illustration of just how important shading can be.
Sean