Reliability of Solar Panels

[pleppik]

NREL recently published a paper which analyzes the three-year reliability of 1.7GW of solar installations all across the U.S. Here's the link. My apologies if this is old; the paper was presented at a conference in September, but I only just saw it today.

Some interesting takeaways:

• 85% of all solar installations (including those with known defects or problems) produced at least 90% of the expected annual output.
• Installations in hotter climates degraded more than installations in more temperate regions, though the authors don't know if that's because of inverter degradation, panel degradation, or dust.
• The most common hardware problem was with the inverter (affecting 0.5% of systems), with underperforming modules affecting 0.1%.
• "Unauthorized shutdown" (i.e. some yahoo pulls the breaker) wasn't a common problem, but it happened enough to put in its own category.
• Hail damage was extremely rare, affecting 0.05% of systems.
• Underperformance due to storms was more likely to be due to power outages than direct damage to the solar panels.

[Drastyn]

Interesting! Thanks

[rollug]

And is there any tests about the realiability after 10 years or so ? Because most of tchem have a warranty for 10-15 years.

[Mike90250]

Well made factory PV panels have a design life of 20 - 25 years with 10% loss of power. Sometimes there are defective panels, but not very often,

The INVERTERS have about a 10 year lifetime for the high power electronics inside them.

[peakbagger]

Many of the original panels that have been installed for 20 plus years are still putting out power. Barring manufacturing defects the life spans predicted by accelerated aging tests appear to have been conservative. The classic example are the Carrizo solar panels originally misapplied in a concentrating solar power plant. They were overheated to the point where they turned brown. After the plant was shut down, the panels were salvaged and resold to home solar enthusiasts. Despite the 10 years of abuse (1984 to 1994) there are still quad lams out there that produce power.

Inverters on the other hand have a far less durable reputation. In order to compete many company's cut corners and many solar installers overfed them so they were running 100% full out much of the time. I have a 12 year old grid tied inverter installed in a cool basement with less than rated input that I hope will keep running as finding a replacement is going to be a challenge. Many inverters also were installed in less than desirable locations like outdoors on the sunny side of a house or were lacking ventilation. Early Enphase microinverters apparently had a high failure rate.

[veritass]

Many of the original panels that have been installed for 20 plus years are still putting out power. Barring manufacturing defects the life spans predicted by accelerated aging tests appear to have been conservative. The classic example are the Carrizo solar panels originally misapplied in a concentrating solar power plant. They were overheated to the point where they turned brown. After the plant was shut down, the panels were salvaged and resold to home solar enthusiasts. Despite the 10 years of abuse (1984 to 1994) there are still quad lams out there that produce power.

Inverters on the other hand have a far less durable reputation. In order to compete many company's cut corners and many solar installers overfed them so they were running 100% full out much of the time. I have a 12 year old grid tied inverter installed in a cool basement with less than rated input that I hope will keep running as finding a replacement is going to be a challenge. Many inverters also were installed in less than desirable locations like outdoors on the sunny side of a house or were lacking ventilation. Early Enphase microinverters apparently had a high failure rate.

Inverters should get better/cheaper with time. Any guess how much cheaper an inverter will be in 5 years?

[foo1bar]

Inverters should get better/cheaper with time. Any guess how much cheaper an inverter will be in 5 years?

My guess is not much in non-adjusted dollars. I think they are getting to the point where getting cheaper is more difficult. So I'm guessing maybe 5-15% cheaper - but my guess is inflation for 5 years will be 8-11% (based on 2005-2010 and 2010-2015 inflation)

These aren't like a computer chip - adding more transistors for the same cost isn't going to make it cheaper - it's more like a microwave oven or other appliance which is about the same price as they were 5 years ago.

[veritass]

Inverters should get better/cheaper with time. Any guess how much cheaper an inverter will be in 5 years?

You are too pessimistic. One analysis is predicting a 20-25% price decline by 2020.


" The global market for solar Photovoltaic (PV) inverters will decline in value from an estimated $5.7 billion in 2014 to approximately $5.2 billion by 2020, due to the rapidly decreasing cost of inverters, according to research and consulting firm GlobalData.
The company’s latest report* states that while the annual addition of solar PV systems is increasing globally every year, the falling cost of inverters continues to reduce the market size, with prices expected to decrease by 20–25% in the forecast period to reach $0.13 per Watt (W) by 2020.
A solar inverter converts direct current electricity generated from solar PV panels into alternating current to facilitate both grid connectivity and use with appliances.
Prasad Tanikella, GlobalData’s Senior Analyst covering Power, says: “The average PV inverter price was $0.48 per W in 2010, following a sharp decline of around 60% over the preceding four years. The price will continue to drop as major tariff cuts impact installations and demand for PV inverters by 2020.
“Manufacturing costs of PV inverters will also fall with improving economies of scale. The expansion plans of large manufacturers, such as SMA Solar Technology, Power-One, KACO New Energy and Refusol, will increase their supplies and saturate the market.” "

[foo1bar]

You are too pessimistic. One analysis is predicting a 20-25% price decline by 2020.

The company’s latest report* states that while the annual addition of solar PV systems is increasing globally every year, the falling cost of inverters continues to reduce the market size, with prices expected to decrease by 20-–25% in the forecast period to reach $0.13 per Watt (W) by 2020.

Maybe I am being a little pessimistic - but better that than being overly optimistic like what you're quoting.

I believe $.13/W in 2020 would be dropping ~50% from now.
I just don't expect inverters costing $1300 for a 10kW inverter in 5 years.
$2500 instead of $2800? Sure.

[veritass]

Maybe I am being a little pessimistic - but better that than being overly optimistic like what you're quoting.

I believe $.13/W in 2020 would be dropping ~50% from now.
I just don't expect inverters costing $1300 for a 10kW inverter in 5 years.
$2500 instead of $2800? Sure.

Why do you think I am being overly optimistic? Do you have another other sources for the projection of the price of inverters by 2020?

[foo1bar]

Why do you think I am being overly optimistic? Do you have another other sources for the projection of the price of inverters by 2020?

Why?
Because:
I don't think there's a large profit margin on inverters (too much competition for that)
I don't see manufacturing improvements reducing the price by >50%
I don't see economy of scale providing >50% price reduction either (even if combined with the above)

https://www.db.com/cr/en/docs/Deutsche-Bank-report-Make-way-for-the-Sun.pdf

page 41 has a nice graph of cost of solar that's broken down by inverter/module/BoS.
You can see it doesn't show price of the inverter being cut in half from 2015 to 2020. It shows it as very close to the same.

Anyhow YOU asked "Any guess how much cheaper an inverter will be in 5 years?"
I gave you my guess - if you don't like it, that's fine. Your response was to write/quote something where the numbers don't even add up. (And not provide what that source was you were quoting)

Really it doesn't matter - if an inverter costs $2800 or $1300 in 5 years doesn't really change anything today.

[veritass]

Why?
Because:
I don't think there's a large profit margin on inverters (too much competition for that)
I don't see manufacturing improvements reducing the price by >50%
I don't see economy of scale providing >50% price reduction either (even if combined with the above)

https://www.db.com/cr/en/docs/Deutsche-Bank-report-Make-way-for-the-Sun.pdf

page 41 has a nice graph of cost of solar that's broken down by inverter/module/BoS.
You can see it doesn't show price of the inverter being cut in half from 2015 to 2020. It shows it as very close to the same.

Anyhow YOU asked "Any guess how much cheaper an inverter will be in 5 years?"
I gave you my guess - if you don't like it, that's fine. Your response was to write/quote something where the numbers don't even add up. (And not provide what that source was you were quoting)

Really it doesn't matter - if an inverter costs $2800 or $1300 in 5 years doesn't really change anything today.

Thanks for the report. I didn't see a numerical prediction for invert costs in the report.

"Our US solar team in its latest report Crossing the Chasm by Vishal Shah, published on 2 March 2015, expects a ~30-40% reduction in costs over the next 4-5 years as current solar module costs continue to decline, increasing panel efficiencies and leading to a decline in the balance of system costs due to scale and increasing competition. Historically, polysilicon has accounted for major cost reduction; however, in the coming years major reductions are expected to come from other system costs. Efficiency improvements are expected to further reduce the cost of cells and modules. Apart from this, balance-of-plant costs, such as inverters, mounting structures and cables, could fall faster than modules. The solar team expects EPC cost reduction to come from economies of scale. Other costs could decline with solar gaining acceptance and the process becoming more standardized."

[J.P.M.]

Thanks for the report. I didn't see a numerical prediction for invert costs in the report.

"Our US solar team in its latest report Crossing the Chasm by Vishal Shah, published on 2 March 2015, expects a ~30-40% reduction in costs over the next 4-5 years as current solar module costs continue to decline, increasing panel efficiencies and leading to a decline in the balance of system costs due to scale and increasing competition. Historically, polysilicon has accounted for major cost reduction; however, in the coming years major reductions are expected to come from other system costs. Efficiency improvements are expected to further reduce the cost of cells and modules. Apart from this, balance-of-plant costs, such as inverters, mounting structures and cables, could fall faster than modules. The solar team expects EPC cost reduction to come from economies of scale. Other costs could decline with solar gaining acceptance and the process becoming more standardized."

I always believe everything I read, especially when it agrees with and reinforces what I've already convinced myself to be reality and the truth.

[veritass]

I always believe everything I read, especially when it agrees with and reinforces what I've already convinced myself to be reality and the truth.

Solar prices have been steadily declining since the 1970's. Solar prices have declined 50% in the last five years. To think that a 30-40% price drop by 2020 is unreasonable goes against a long established trend and the majority of forecasts of analysts and companies in the industry You offer no evidence or detailed analysis for why you think prices won't decline. For someone who allegedly wants to see more solar installed you would think you would welcome a price decline. Just look at the forecasts for solar costs by Solar City and First solar and many analysts in the industry. Please show me evidence that backs up you assertion that solar prices won't decline much.

[sensij]

You may find in this attachment some of the projections of system and component costs used by parties of CPUC's NEM 2.0 proceeding (see Attachment B page 6, pdf page 13 on my display).