Aging Monte Carlo RMS parameters

[Dong]

Hello,

When I was doing the PVsyst simulation and inputting the parameters in the aging tool, I read the help manual and got to know that PVsyst has supposed Gaussian distribution with Imp and Vmp RMS dispersion value. According to the help manual, by considering the 2*"RMS dispersion", it is applicable to 95% of the modules. 

It is well known that 2 sigmas in the Gaussian distribution equals to 95% of the hits. Does it mean that "RMS dispersion" of Imp and Vmp can be treated as the "sigma" for Normal distribution?

Looking forward to your kind reply!

BR

[André Mermoud]

Yes you are right. The term is not well  chosen.

Here what is named RMS is indeed the standard deviation (sigma) of the gaussian distribution. You have statistically 68% of modules within the interval -Sigma to + sigma.

[Dong]

Thanks Prof. Andre!

Since most of module supplier can warrant the maximum degradation of 0.4% for every TOPCon module, it can predict that the degradation of all modules can be controlled within the span of 0-0.4% after 1 year operation, 0-0.8% after 2 years operation, and so forth. Therefore, inside the Guassian distribution, I would say the inrerval of "-2*RMS to 2*RMS" shall equals to the warranty span I mentioned above. 

Because module supplier (I'm from LONGi) doesn't have the sufficient data to indicate the "Average degradation factor", "Imp RMS dispersion" and "Vmp RMS dispersion". Can I make the hypothesis that:

1. "Average degradation factor"="1/2 of TOPCon yearly warranty"="0.2%",

2. "Average degradation factor"+2*"Imp/Vmp RMS dispersion"="0.4%", so that "Imp/Vmp RMS dispersion"="0.1%"

Please correct me if my understanding is wrong, or my hypothesis is not reasonable.

BR,

Dong

[André Mermoud]

No, these parameters IMP RMSDispersion and Vmp RMSdispersion      don't describe an average degradation along the year. 

They are used for the evaluation of the evolving mismatch between modules along the years, due to the fact that the modules are not degraded in the same way.

For a given year, each PV module is affected by a Imp/Vmp deviation, randomly chosen according to these parameters (Gaussian distribution). We calculate the corresponding mismatch loss for this year, which is an addiitional contribution to the average degradation loss. The modules are then further randomly degraded for the next year.

 

 

[Dong]

Sorry it was my bad, I didn't make my idea clear. 

Indeed RMS dispersion describes how wide the span is, where modules will randomly degrade the Imp or Vmp by means of Gaussian distribution.

Therefore, I made the proposal that the range of "-2*RMS to 2*RMS" (as known as "-2*sigma to 2*sigma") is quite sufficient to represent 95% of modules. So that the total distribution range can be controlled as "4*RMS*operation years" wide.

Since module manufacturer can make the power warranty of 0.4%/year for any of the module, the degradation only will be happened inside the range of "0-0.4%*operation years" (e.g. 0-0.4% for 1st year, 0-0.8% for 2nd years).

So that If I make the proposal that 95% "Guassian distribution range" equals to "module warranty range", I can draw the conclusion that the "4*RMS*operation years" equals to "0.4%*operation years". So that "RMS"="0.1%/year".

Hope the following table can explain the logic I mentioned above.

Thanks for your time in advance!

BR,

Dong

[André Mermoud]

First observation: the usual warranty for any module is usually about 15-20 % loss after 25 years, i.e. a max. degradation  of 0.6-0.8%/year. I have never seen datasheets with a warranty of 90% after 25 years. 

Now in the random choice of the PV modules degradation when calculating the Monte-Carlo for the mismatch evolution, we should indeed limit the lower value to 0.8% * no. of year. This is indeed done as in our Monte-Carlo calculation the dispersion is limited to 2 * sigma, i.e. 2 * 0.4% with the default sigma values.