7

Many manufacturers of photovoltaic solar panels offer some sort of warranty that is valid for 15 to 25 years. Many people suggest that after that period of time, a replacement of the panels is necessary

For example:

The lifetime of solar panels (photovoltaic panels) on average ranges between 20-25 years – different solar panel companies may give different guaranties for the performance and operation of their solar panels.

Also this:

PV panels can last up to 25 years or more, some with a maximum efficiency loss of 18% only, even after 20 years of operation.

Both suggesting that there is about a 20 year lifetime and then you have to replace your solar panels.

I am skeptical that something that works 20 years would, all of a sudden, stop working and require replacement. I expect that a non-mechanical device has little wear-out. That's why there is a relatively long warranty period. For that reason, it is not reasonable to expect much performance loss over time. I expect that even after 20 years, a solar panel will still generate at least 80% of its initial performance which really makes me wonder about why a replacement should be necessary.

Oddthinking
  • 140,378
  • 46
  • 548
  • 638
humanityANDpeace
  • 179
  • 5
  • I have "struck-out" your notability link, because nothing on that page suggests that the lifetime is limited strictly to a fixed time. A better notability reference would be desirable. Especially, you have claimed that the replacement is due to sudden failure, rather than gradual efficiency loss. – Oddthinking Nov 16 '13 at 12:37
  • @Oddthinking I am honestly still struggeling with the `lifetime` of an PV-module here. I am sorry if you are unsatisfied with my references. Indeed it is that my **skeptic question** was with regard to people claiming that. When asked they are themselves always without a reference, which actually induced me being skeptic. I will try to find some references and if so will edit the post here. – humanityANDpeace Nov 16 '13 at 13:34
  • 1
    Sorry, I was unclear. [Welcome to Skeptics!](http://meta.skeptics.stackexchange.com/questions/1505/welcome-to-new-users). I am not looking for a quality reference that has evidence of an answer. I am looking for a "notability" reference - it is an example of someone making the claim. There are a number of reasons we ask for those references - but partly to make sure that our answers addresses a real claim, not just a misunderstanding. At the moment, we don't know if anyone actual thinks that PV cells fail after 20 years. – Oddthinking Nov 16 '13 at 13:58
  • 1
    @Oddthinking thank you then for the information. I will as suggested strive to provide such a reference. Thank you. – humanityANDpeace Nov 16 '13 at 14:00
  • Perhaps I have misunderstood. I thought people were saying "PV cells almost always survive until the warranty is over, then die suddenly" and you were skeptical that they fail in a short window after a long period. But people are saying "They last for two or more decades, on average.", without giving a small window, and you seem to be saying "If they last 20 years, they should last forever." – Oddthinking Nov 16 '13 at 17:41

1 Answers1

5

The life time of photo voltaic (PV) cells is fundamental in assessing the ecological impact and greenhouse gas emissions of solar power. As such, it has been studied in thousands of papers, some of which report on measurements spanning 20 years.

The problem with the panels is that their efficiency drops a certain amount per year. After a while, the panel fails -- there are studies around the various failure modes as well. This yearly degradation has been measured to be between 0.5% and 1% (depending on study and PV type), however failure is due to cumulative damage and is catastrophic (e.g. a panel at a certain point short circuits).

So obviously you need to replace a PV panel only when it short-circuits or shunts and stops working, or when it's not cost-efficient anymore due to a grossly degraded efficiency, however if you have many -- such as in a power plant -- what matters is the number of average years: some panels will break sooner, some later and basically only the average affects your bottom line.

A history of degradation rates using field tests reported in the literature during the last 40 years has been summarized. Nearly 2000 degradation rates, measured on individual modules or entire systems, have been assembled from the literature and show a mean degradation rate of 0·8%/year and a median value of 0·5%/year. The majority, 78% of all data, reported a degradation rate of <1%/year.

A very good meta-analysis of ~2,000 studies has been performed by the US National Renewable Energy Laboratory.

What they found is that recent PV systems are predicted to last over 25 years (maybe 30 or so), the older one are a litlle less reliable. The figure that follows is takend from the NREL study linked above and gives an idea of typical measured life times of panels in studies varying by the year the study was published.

PV Lifespan

Sklivvz
  • 78,578
  • 29
  • 321
  • 428
  • Can you summarize this more clearly, please: is the problem that efficiency degrades slightly, or that units fail entirely, or both? Why does efficiency degrade, and is there a graph of how much typical efficiency degrades over years? And another graph of expected complete-failure rate over years? Where does your "Figure 4" come from, and what does it show (it's not clear to me how to read it): what does "Degradation Rate Publication" mean on the vertical axis, and why does it show a cluster of "degradation rate publications" in the first 5 or 10 exposure-lengths-years? – ChrisW Nov 16 '13 at 12:34
  • Where a device has maintenance costs (e.g. cleaning, rental of space, opportunity costs of limited roof-space, etc.), the degradation of efficiency may dictate a replacement earlier than the total failure date. I speculate that may be why people may be saying it will need replacing after 25 years, even if it hasn't failed outright. – Oddthinking Nov 16 '13 at 12:49
  • @ChrisW as I said there are multiple failure modes. Degradation occurs slowly but failure happens catastrophically. If you are interested in the details, you can look at the paper I've linked. I'll try to clarify. – Sklivvz Nov 16 '13 at 13:14
  • 1
    @Oddthinking that could be, however it wouldn't impact warranty lengths, which are clearly correlated to degradation rate as per Fig. 4. – Sklivvz Nov 16 '13 at 13:16
  • Thank you for your answer Sklivvz! I understand the replacement by short-circuit (that is valid). Can you enlighten me how one could understand the QUOTE or when it's not cost-efficient anymore due to a grossly degraded efficiency QUOTEEND ? If for instance I have no cost for the land and sunshine is free, what would be a good idea to think I would risk becoming cost-in-efficient ? How much would in your assumption be this " grossly degraded efficiency"? – humanityANDpeace Nov 16 '13 at 13:30
  • @humanityANDpeace there are two cases: maintenance costs become too high in relation to energy produced -- even periodic electrical checks have a cost -- and opportunity costs -- you may want to upgrade because the efficiency improvement of having a new unit makes it worthwhile and you can't simply add a new unit because of space, etc. – Sklivvz Nov 16 '13 at 13:40
  • @Sklivvz again thanks. Reading the linked resources I quote "Typically, a 20% decline is considered a ailure, but there is no consensus on the definition of failure, because a high-efficiency module degraded by 50% may still have a higher efficiency than a non - degraded module from a less efficient technology.". Assuming a very pessimistic yearly rate of degratation of 2%, to reach a accumulated loss of 50% efficiency would be 34 years. The real struggle here is to know about the `catastrophic failure` situation. The gradual degradation seems "manageable". – humanityANDpeace Nov 16 '13 at 13:56
  • @humanityANDpeace see http://www.nrel.gov/docs/fy11osti/47755.pdf where this is studied extensively. – Sklivvz Nov 16 '13 at 13:59
  • To answer my own question, IMO figure 4 shows that most studies re degradation rate are published within a few short years of installation, and only very few studies are published after the warantee expires. It does *not* show "measured lifetimes", it shows publication/study times, and warantee periods. Any other figure from Ibid would have been more interesting (more worth quoting) in your answer. – ChrisW Nov 16 '13 at 15:42
  • @ChrisW I think you are misunderstanding the graph. It shows the measured lifespan and the warranty period (as comparison). Measured lifespan increased over the years because product quality increased. – Sklivvz Nov 16 '13 at 16:07
  • So you're saying that in most cases the measured lifespan is much shorter than the warantee period?? IMO it's saying that in 2010 for example there were several publications measuring degradation, and that many of these measurements/publications were taken/made only a few years after the product was installed. – ChrisW Nov 16 '13 at 16:13
  • @ChrisW yes, that is not surprising because the published measurements are about all available panels, whereas the warranty refers to a *specific* model of panel, which may or may not be mesured as part of the set (or could be an outlier). – Sklivvz Nov 16 '13 at 17:22
  • @ChrisW and on top of that, the warranty period covers technical failure only, economic failure (thus, the panel still working but no longer being economical to keep in service) may well come much sooner, and isn't covered by the warranty. So say warranty would kick in if the panel were to short out after 25 years, but if because of environmental factors it degrades to where it is no longer economical to run it despite still producing some power after 15 you will retire it and warranty does not cover that (and that's the usual failure mode). – jwenting Nov 21 '13 at 06:57
  • 1
    @Sklivvz the figure in the answer doesn't represent what you are saying. You say "gives an idea of typical measured life times of panels", but the y-aixs is merely showing the lengths of the studies by publication date. The length of the studies have no relationship to the lifetimes of the panels themselves. – DavePhD May 27 '16 at 12:23