The average wavelength in diffuse solar irradiance is shorter (more blue) than that of direct sunlight. Modules of (poly)-crystaline silicon have a broad spectral response (see figure), and the the influence of the spectrum is not so large. Thin film amorphous silicon solar cells have narrow spectral response peaking at short wavelengths. The efficiency of a-si modules is therefore better under diffuse conditions than under clear sky conditions.

However, there are many types of thin film solar cells. CIGS for example have a broad spectral response and therefore their efficiency is not so sensitive to the spectral distribution of the irradiance.

Also the advantage of a-Si under diffuse conditions is often exaggerated by manufacturers. A few years ago I was involved in a German project to compare several solar module types under North German Climatological conditions (FLINS, http://www.flins-projekt.de). It turned out that no significant difference between the performance of the modules could be found: The uncertainty in the module STC power as specified by the manufacturers , was larger than the differences in annual yield measured in the field.

The average wavelength in diffuse solar irradiance is shorter (more blue) than that of direct sunlight. Modules of (poly)-crystaline silicon have a broad spectral response (see figure), and the the influence of the spectrum is not so large. Thin film amorphous silicon solar cells have narrow spectral response peaking at short wavelengths. The efficiency of a-si modules is therefore better under diffuse conditions than under clear sky conditions.

However, there are many types of thin film solar cells. CIGS for example have a broad spectral response and therefore their efficiency is not so sensitive to the spectral distribution of the irradiance.

Also the advantage of a-Si under diffuse conditions is often exaggerated by manufacturers. A few years ago I was involved in a German project to compare several solar module types under North German Climatological conditions (FLINS, http://www.flins-projekt.de). It turned out that no significant difference between the performance of the modules could be found: The uncertainty in the module STC power as specified by the manufacturers , was larger than the differences in annual yield measured in the field.

The FLINS website is in German only, unfortunately.

Anyway, it is important to consider that STC conditions refer to an AM1.5G spectrum under clear conditions. As correctly stated, the diffuse spectrum is blue-rich and thus more favorable to a-Si than other types of cells. However, clouds act as a relatively neutral grey filter, so that the spectral distribution under clouds is similar to the standard spectrum for a same airmass.

The actual periods when the actual spectrum favors a-Si most (at least from a purely spectral standpoint, since there might be other effects) is therefore in early morning and late afternoon during clear summer days. Assuming a fixed array at mid-latitudes and tilted at latitude or more, the cells are then exposed to diffuse radiation from the blue sky only. These periods are relatively short and have low irradiance, which might explain Jethro's results.

Chris, I cannot agree with you on the effect of clouds, they do tend to cause a blueshift. See for example Fig. 5 in  Behrendt  et al. 2013, Solar Energy, 2013; 98:100–110.

  (https://www.researchgate.net/publication/259094161_Solar_spectral_irradiance_derived_from_satellite_data_A_tool_to_improve_thin_film_PV_performance_estimations?ev=prf_pub)

The thing with the FLINS project is that there are several effects that happen simulltaneously. The modules each have a different low irradiance behaviour, a different spectral effect, and  different temperature behaviour.  Wit the limited monitoring equipment in that small project it is difficult to seperate these effect.  In the current PVKLIMA project we will be able to look more into detail.

@Anton: From a technical point of view you are correct at stating: ' "Performing well at low irradiance" is another way of saying "performing not-so-well at high irradiance". '

However, from an economic perspective things can look differently. Most potential PV-system owners look at the system cost per Wp, but what counts in the end is the cost per kWh. If you can get more kWhs per Wp for the same Wp price modules with a poor high irradiance performance might be the best choice.

Furthermore, at high PV penetration levels  the power produced during low irradiance has more value than power produced at high irradiance (when everybody else is also throwing their PV power onto the spot market). In an unsubsided situation it would therefore make sense to (economically) optimise PV systems for low irradiance circumstances.

Article Solar spectral irradiance derived from satellite data: A too...

Jethro,

The Behrendt study is very interesting, BUT assumes that APE is a fundamental and "invariant" property of PV cells. I found out this is not the case, see http://spie.org/Publications/Proceedings/Paper/10.1117/12.826071

As I mentioned, the transmittance of clouds is roughly spectrally neutral. This is demonstrated in the attached figure, based on simulations with the libRadtran code. Of course there might be cases when some blus-shifting occurs, but this should remain a second-order effect anyway.

Betts, T R, R Gottschalg, and D G Infield. “Spectral Irradiance Correction for PV System Calculations.” In Proceedings of the 19th European PVSEC, Paris, edited by W Hoffmann, J.-L. Bal, H Ossenbrink, W. Palz, and P Helm, 3:2533–36, 2004.

Hallo Chris,

Your paper sounds interesting, I will order it at our library. Especially the difference between ice and water clouds has relevance to my current work.

I wonder how we can look at the same results at interpret it so differently: For me the Figure that you posted is consistent with our results: During  cloudy conditions the transmittance for longer wavelengths is reduced more than for shorter wavelengths, i.e.  the average wavelength becomes shorter, ergo a blue-shift. The larger the COD the stronger this effect.  This is the same result as found in Fig. 5 of the Behrendt et al article.

Regarding the average photon energy (APE), I think you misread the article there. (There are probably some germanisms that make the article less clear  ;-)  ). The APE is just a convenient way to summarise the spectral distribution in one number. Just as the spectral distribution, it varies in space and time.

The APE says nothing about  solar cells. However, it can be used as an input to a simple  model for the spectral dependence of the solar cell efficiency. See for example:

(downloadable from the EUPVSEC site)

I must confess, i do not have a intuition yet on how much clouds contribute to the overall spectral effect on PV system yields. But thanks to this discussion I have gotten some ideas that I will try to incorporate in my upcoming research.

Jethro,

Thanks for your input. Unfortunately, the paper you cite is old and does not seems to be available from the EUPVSEC website...

https://www.researchgate.net/publication/237353389_Spectral_Irradiance_Correction_for_PV_System_Yield_Calculations

Conference Paper Spectral Irradiance Correction for PV System Yield Calculations

Jethro, Thanks for the link but paper not available from RG. I finally found it from the EUPVSEC repository. I still think the whole idea of using APE as a "universal" way of characterizing spectral effects needs to be verified.

You mean something like this?:

https://www.researchgate.net/publication/245170092_Uniqueness_verification_of_solar_spectrum_index_of_average_photon_energy_for_evaluating_outdoor_performance_of_photovoltaic_modules

Article Uniqueness verification of solar spectrum index of average p...

That study was interesting, but limited to a single site. Moreover, the atmospheric conditions might be relatively stable there. My question is whether APE is a site-specific, or truly universal tool...

i had done my Mtech research work on the performance analysis of Thin Film technology so you can see the work on the Researchgate page

Jethro,

About the APE limitations, please see this new study:

https://www.researchgate.net/publication/315838711_Is_the_average_photon_energy_a_unique_characteristic_of_the_spectral_distribution_of_global_irradiance?_iepl%5BviewId%5D=pkYt3dAndOLzoALnNG7WeAU7&_iepl%5BprofilePublicationItemVariant%5D=default&_iepl%5Bcontexts%5D%5B0%5D=prfpi&_iepl%5BtargetEntityId%5D=PB%3A315838711&_iepl%5BinteractionType%5D=publicationTitle

Article Is the average photon energy a unique characteristic of the ...

The spectral distribution of diffuse irradiance is different than direct normal irradiance.  This diffuse spectral distribution also depends on a cloudiness factor and maybe an air mass factor.

There are severalother factors that also come into play so there is no one answer that fits all circumstances.