Optimizers... what are they good for?

[nwdiver]

I see a lot of significantly misinformed posts with regards to module level optimization. This topic frustrates me because it's a large source of wasted capital as far as solar installs go. By and large there is virtually no financial benefit to using optimizers. Even with the rapid shutdown requirement the additional cost of adding an optimizer is very unlikely to be recovered. Worse is that it adds another point of failure to the PV system. Here is a related thread: Solar Panel Shading

The only task optimizers are suited for is allowing panels in the same string to face different directions. That it. I've encountered very few installations where this task couldn't also be accomplished with a string inverter that had multiple independent MPP channels. Optimizers DO NOT prevent shaded panels from reducing the production of unshaded panels. Bypass diodes do that. Optimizers also generally do not increase the annual production of a panel. The variation from panel to panel is not significant enough that there's a tangible benefit to allowing each panel to operate at a separate MPP.

Here is the output from two systems so far in 2020. The top is a SMA string inverter and the bottom is a Solar Edge system with optimization on every panel. The SMA system was expanded at the same time the Solar Edge system was installed. We added 4 310w panels to the existing 15 295w panels. So not only is there no model level optimization in the SMA system but the strings are mixed. The last 2 panels in each string are 310s while the other panels are 295s. The production difference between the top and bottom rows on the SMA system is due to the system being oversized by 50% and SMA inverters clip one channel at a time so all clipping occurs on String A (this also allows to you calculate exactly how much production is lost oversizing by 50%... <5%)

So two systems. One with module level optimization and one without. The panels are from the same batch, installed at roughly the same time, in the same city. The only difference is an orientation of 150 vs 180 which should only yield a benefit of ~2% to the 180 facing panels.

 

[JeremyWhaling]

So a majority on this section of the forum are going to be buying or at least heavily consider Tesla energy, and don't have much flexibility in equipment (only number of panels). Under most conditions right now, unless you want to do something really custom, Tesla is the cheapest - optimizers or not. Tesla seems to be deploying mostly solar edge these days, although they use to use Delta in the past as well (with no optimizers).

I think Tesla is using optimizers not necessarily to guarantee or otherwise maximize production, but in flexibility of installations. They just need to throw up the panels and string the panels together. No need to worry about all the panels in a string facing the same way, or occasional shading. Looking at the layouts thread has really shown me how Tesla has a more "throw the panels anywhere" approach to this, and optimizers make it possible.

I don't get why they don't just go for microinverters to be honest.

 

[nwdiver]

I don't get why they don't just go for microinverters to be honest.

Why? Micros are even worse than optimizers from a cost vs performance perspective. They're also less flexible than optimizers.

I've seen a number of posts actively seeking optimizers in their proposals. My point is that's silly.

Unless something has changed I don't believe Tesla is using optimizers in all their installs. I know they were heavily using Delta inverters which are not optimized.

 

[woferry]

The only Tesla installs I've seen personally have all been Solar Roofs, and those are all Delta inverters (either Solivias or the newer Mx units). But it seems many (most? all?) of the Tesla panel installs use SolarEdge inverters with optimizers. The optimizers (or at least the ones that go with a 300W-ish panel) certainly wouldn't make sense on a Solar Roof since the tile wattage is much less. But I'm not sure why Tesla chooses to go with different brand inverters for the different types of installs instead of using the same thing (Delta) everywhere. As someone who loves data, I am jealous of the per-panel data available with either the microinverters or optimizers (I'm fighting to even get per-MPPT data from my Delta Solivias ).

 

[nwdiver]

As someone who loves data, I am jealous of the per-panel data available with either the microinverters or optimizers (I'm fighting to even get per-MPPT data from my Delta Solivias ).

You're definitely the exception. ~90% of my customers probably haven't looked at their system profiles in months. Most people check it often in the first thirty days until the novelty wears off then rarely look again. Module level data is cool... just not sure it's ~$30 per panel cool. There's also ways other than spending $60 per panel on optimizers... the SMA system I posted gets module level info from TS4-S modules that are half the cost of an optimizer.

 

[pdx_m3s]

Why? Micros are even worse than optimizers from a cost vs performance perspective. They're also less flexible than optimizers.

I've seen a number of posts actively seeking optimizers in their proposals. My point is that's silly.

Unless something has changed I don't believe Tesla is using optimizers in all their installs. I know they were heavily using Delta inverters which are not optimized.

My Tesla Solar setup is using optimizers per the technical design plans. 33 panels (2 strings - 16/17). All west-facing. Zero shading. SolarEdge 10kW. Perhaps optimizers are used because of the uneven string length?

 

[nwdiver]

My Tesla Solar setup is using optimizers per the technical design plans. 33 panels (2 strings - 16/17). All west-facing. Zero shading. SolarEdge 10kW. Perhaps optimizers are used because of the uneven string length?

Nope; The SMA system I posted also has uneven strings. Each string is on an independent MPP channel.

That's frustrating... ~$2k in top level electronics doing nothing but cooking on a roof You'd still need rapid shutdown that would cost ~$1k but that's still ~$1k cheaper than optimizers.

 

[arnolddeleon]

I think Tesla is using optimizers not necessarily to guarantee or otherwise maximize production, but in flexibility of installations. They just need to throw up the panels and string the panels together. No need to worry about all the panels in a string facing the same way, or occasional shading. Looking at the layouts thread has really shown me how Tesla has a more "throw the panels anywhere" approach to this, and optimizers make it possible.

I think this nails it. My SolarCity install from 2014 has a SolarEdge inverter/optimizers (it ended up different orientations). The optimizers along with their rail less installation process appears to give layout and design flexibility that saves them in install costs. The guys in the field made running changes to accommodate the situation as found in the field. My installation ended up with twist in that the city wouldn't sign off on the final layoff because of the then new setback/path requirements. In the end they had to remove some modules (refund me the difference).

I didn't even know about optimizers during this install, it was only afterwards that I learned more about them. I read I have to say I love the idea per module data (I don't have access to it and I haven't bothered Tesla for it). These days I'm with @nwdiver, if I had the roof space, I would spend my money on more modules instead of worrying about optimizers or upsizing my inverters to limit clipping.

 

[sleevemedia]

We would care if optimizers were a line item on an invoice. They aren't, you don't pay less if you remove them. That would matter if installers that do sell an itemized bill of materials could touch the price per watt we're buying, and they can't, even without optimizers.

It's commodity economics at this point, and that means there's going to be a federally acceptable amount of insect parts in your corn flakes.

 

[nwdiver]

We would care if optimizers were a line item on an invoice. They aren't, you don't pay less if you remove them. That would matter if installers that do sell an itemized bill of materials could touch the price per watt we're buying, and they can't, even without optimizers.

It's commodity economics at this point, and that means there's going to be a federally acceptable amount of insect parts in your corn flakes.

Except some people will choose the installer that uses optimizers over one that doesn't even if the one that does is more expensive and encourage others to ensure they get optimizers. Just because it's not a line item doesn't mean there isn't a choice... If you're justifying a higher priced install because they're using optimizers that's just silly.

Then there's the DIY crowd that reads this thread. The last thing a scrappy DIY'r needs is a large unnecessary expense. I've had to educate most of my recent customers on why optimizers are not beneficial.

 

[BrettS]

Yeah, I think there’s a lot of misinformation out there, and maybe even downright shady tactics from a lot of installers. Even going back more than a year ago when I first started considering going solar I got quotes from Tesla as well as a number of other installers and several installers specifically told me that their systems, using micro inverters would outperform Tesla’s system because Tesla used string inverters. I did my research at that time and came to the conclusion that the optimizers that Tesla used would provide similar benefits to the microinverters that they had quoted.

Frankly those tactics probably wound up backfiring in my case because when I did the research and saw that what they were saying wasn’t true it made me cross them off my list. But I’m sure that there are plenty of people who just take them at their word and don’t do any further research.

 

[sleevemedia]

I used EnergySage and beat the nearest quote by 20 cents and found out after the fact that Tesla uses optimizers. They don't tout the fact in their bids. A couple of the competing quotes used Enphase MPLE, but I didn't want micros on my roof and they were 40-90 cents over Tesla.

I watch a few of the DIY channels, learned a lot from Will.

 

[jrweiss98020]

My installer was very candid and informative. He told me performance would be similar between microinverters and optimizers, that optimizers would help due to intermittent shading by trees, that microinverters were a bit more expensive, and that future expansion would be easier with microinverters. I'm happy with the Enphase IQ7+ microinverters. YMMV.

 

[nwdiver]

My installer was very candid and informative. He told me performance would be similar between microinverters and optimizers, that optimizers would help due to intermittent shading by trees, that microinverters were a bit more expensive, and that future expansion would be easier with microinverters. I'm happy with the Enphase IQ7+ microinverters. YMMV.

.... yeah... optimizers and micros are the same schtick; They just cost ~$.10-$0.20/w more than a regular string inverter for no increase in performance. $ for nothing...

 

[wwhitney]

I think your point is that the cases where optimizers actually help are fairly rare and typically don't occur often enough for them to make up for their cost by increasing energy production. So to be fair to optimizers, what are the cases where they do help? My understanding is probably not complete, but I would think:

(1) If enough of the panels are fully shaded that enough of the bypass diodes would activate so that the string voltage would drop outside the operating window of the inverter. Then if there were optimizers, would the DC-DC conversion at the optimizers be enough to boost the voltage back up to keep the inverter operating? So the optimizers would effectively broaden the (unoptimized) string voltage operating window of the inverter?

(2) If the string has a mix panels that are partially shaded and unshaded, I would think a string inverter has to choose between operating at a string current that the partially shaded panels can handle, foregoing some power from the unshaded panels, or operating at the higher string current that would let the unshaded panel maximize energy production, but activates the bypass diodes on the partially shaded panels, foregoing their production. While with optimizers, each panel could operate at its own MPP.

(3) Others?

Cheers, Wayne

 

[nwdiver]

I think your point is that the cases where optimizers actually help are fairly rare and typically don't occur often enough for them to make up for their cost by increasing energy production. So to be fair to optimizers, what are the cases where they do help? My understanding is probably not complete, but I would think:

(1) If enough of the panels are fully shaded that enough of the bypass diodes would activate so that the string voltage would drop outside the operating window of the inverter. Then if there were optimizers, would the DC-DC conversion at the optimizers be enough to boost the voltage back up to keep the inverter operating? So the optimizers would effectively broaden the (unoptimized) string voltage operating window of the inverter?

(2) If the string has a mix panels that are partially shaded and unshaded, I would think a string inverter has to choose between operating at a string current that the partially shaded panels can handle, foregoing some power from the unshaded panels, or operating at the higher string current that would let the unshaded panel maximize energy production, but activates the bypass diodes on the partially shaded panels, foregoing their production. While with optimizers, each panel could operate at its own MPP.

(3) Others?

Cheers, Wayne

I mentioned it;

The only task optimizers are suited for is allowing panels in the same string to face different directions. That it.

In most shading scenarios there's no benefit to optimizers; If one cell group of a panel is unshaded then the other two cell group are bypassed whether there's an optimizer or not. In situations where all cell groups are shaded then it's essentially blood from a stone.

The chimney is beginning to shade these panels. Yes, with a string inverter the output would be 0w instead of 40w. IMHO the cost of an optimizer isn't worth the ~40w being harvested in the limited amount of time the panel is in full shade. The 65w panel is one full cell group so you'd get 65w from a string inverter too.

Optimizers also experience their own situation similar to string inverters in which too many panels are in shade to produce sufficient voltage to be in the MPP band. With optimizers this is called 'voltage blocking'; Essentially instead of varying current with insolation as solar modules do naturally they vary the voltage. This is the 'magic' that allows them to produce different levels of power in the same string. But optimizers cap out at 60v. If you have a string of 10 panels the output of that string MUST be ~350vdc with solar edge. If 6 are in shade by 4 are in full sun those 4 cannot meet 350v because each is limited to 60v. So those 4 panels will be handicapped by the 6 that are shaded.

This is a less extreme real world example; More power is available in the morning but cannot be produced because the optimizer is 'voltage blocked'

 

[niveknow]

Optimizer are now being installed by Tesla by default at least for me. I saw them listed in my equipment list one per panel as a line item. Guess this may be a new standard which is great because I didn't have to request. Can't recall where on this forum I read, but is there a way to see performance at the individual panel and optimizer level? I recall someone digging into this and realizing that one of their panels as in fact not operating and thus needed to be replaced.

 

[nwdiver]

Optimizer are now being installed by Tesla by default at least for me. I saw them listed in my equipment list one per panel as a line item. Guess this may be a new standard which is great because I didn't have to request. Can't recall where on this forum I read, but is there a way to see performance at the individual panel and optimizer level? I recall someone digging into this and realizing that one of their panels as in fact not operating and thus needed to be replaced.

Hypothetically speaking... if optimizers were not included, how much would you be willing to pay to have module level monitoring?

 

[niveknow]

Would prefer to pay zero =). Module level monitoring wouldn't be needed a regular basis because most of us wouldn't care for the grauluar data, but would be an occasional checking to make sure they all work as expected. Imagine running a system for years and finding out that you had a couple panels that weren't working as designed and no way of knowing when/where it fail to address the problem. You would have missed out on production. How does a system know ALL the panels are working as designed?

 

[nwdiver]

Would prefer to pay zero =). Module level monitoring wouldn't be needed a regular basis because most of us wouldn't care for the grauluar data, but would be an occasional checking to make sure they all work as expected. Imagine running a system for years and finding out that you had a couple panels that weren't working as designed and no way of knowing when/where it fail to address the problem. You would have missed out on production. How does a system know ALL the panels are working as designed?

That's kinda my point with this thread. You may not see it on your invoice but it costs ~$60 per panel for optimization + monitoring or ~$30 just for monitoring. That adds up quick. So for a 10kW system that's ~$1800 for optimization or ~$900 just for monitoring. And for all that added expense there's really no added value.....

Panel failure is exceedingly rare. In >10 years I've never seen a broken panel that was working when installed aside from an obvious mechanical failure like broken glass from stress or an impact but even in that case the panel was still working. Panel failures can be spotted from a decrease in production then the exact panel can be narrowed to the string. IMHO $900 on module level monitoring isn't worth the <1% chance there will be a panel failure in the life of the system OR the ~20 minutes of additional troubleshooting required to find a bad panel absent module level monitoring.

Adding optimization also makes a failure FAR more likely. While I've never seen a panel fail I have seen several optimizers fail.