Should we point them away from the sun? -A study in PV spectral tracking in a Scandinavian climate

Abstract

Conventional photovoltaic solar tracking is commonly done by aligning the surface normal
of a PV module with the direction from which direct solar irradiation is coming from. While
this tracking design has proven to perform well in sunny climates, solar tracking is altogether
less common in northern climates where the direct solar irradiance is weaker. In this study a
radiative transfer model is used to approximate the wavelength distribution of incident solar
radiation reaching Visby on the island of Gotland in Sweden during 2011 to investigate if the
site specific irradiation of a northern climate in combination with different PV materials
optical properties can result in circumstances where there are more productive things to track
than the direct solar irradiation. The results of the study indicates that the performance of a
tracking system to some extent depends on the choice of PV cell material used in the system
and that the spectral response of different materials often make them achieve optimum
productivity at slightly different tilt/azimuth combinations. The study concludes that a
conventional tracking system in Visby during 2011 in general would generate approximately
10 kWh/m² less electricity compared to a theoretically optimal tracking system.

Popular Abstract

Conventional photovoltaic solar tracking is commonly done by aligning the surface normal of a PV module with the direction from which direct solar irradiation is coming from. While this tracking design has proven to perform well in sunny climates, solar tracking is altogether less common in northern climates where the direct solar irradiance is weaker. In this study a radiative transfer model is used to approximate the wavelength distribution of incident solar radiation reaching Visby on the island of Gotland in Sweden during 2011 to investigate if the site specific spectral irradiance of a northern climate in combination with different PV materials optical properties can result in circumstances where there are more productive things to track than the direct solar irradiation.

When the performance of photovoltaic cells is determined, under so called Standard Test Conditions, the incident solar energy is assumed to have a spectral distribution regulated by a standardized AM1.5 spectra. AM1.5 stands for Air Mass 1.5 and is supposed to represent a situation where sunlight has to travel through 1.5 times as much atmosphere compared to the shortest possible atmospheric path at sea level (AM1). Expressed as solar elevation angle AM1.5 corresponds to roughly 42 degrees above the horizon. During large parts of the year in Scandinavia the solar elevation angle is often far lower than 42° and consequently the sunlight has to travel through far more atmosphere during these times than what is assumed in the Standard Test Conditions. This additional atmospheric length causes radiation attenuation that lowers the overall broadband irradiance reaching a surface on the earth, but it does not necessarily have to diminish it completely evenly across the electromagnetic spectrum. Different atmospheric constituents all have their own characteristic locations, their absorption bands, in the spectra where they cause radiation attenuation. Depending on how long the atmospheric length is and what chemicals are in the atmosphere, the solar irradiations spectral distribution will differ from what was assumed when the cell was performance rated. Therefore, simulating the annual energy output from cells without considering the site specific wavelength distribution of incident solar energy will to some extent yield an unrealistic forecast.

In this study the solar spectra during every minute of a year was simulated using a radiative transfer model together with a numerical model of the atmosphere and its annual chemical variations. The spectrally resolved irradiation was then used to calculate a unique Spectral Mismatch Factor for every minute, tilt and azimuth for eight different semiconducting materials to determine how appropriate the site specific solar energy at different orientations is for conversion to electric current. Since different PV materials have different spectral responses, the light reaching a surface oriented towards one direction could very well be more suitable for one PV material than for another.

In this study the following eight PV materials were evaluated; Amorphous silicon, Cadmium Telluride, Gallium Arsenide, Gallium Indium Phosphide, Monocrystalline Silicon, Multicrystalline Silicon, Copper Indium Gallium Selenide with Zinc Oxide and Inorganic and Nanostructured Photovoltaics.

The calculated Spectral Mismatch Factor was then combined with real minute-by-minute measurements of the broadband irradiance reaching the site during the year and a weighted irradiance was calculated for every minute, tilt, azimuth and material type. The weighted irradiance, which is the product of Spectral Mismatch Factor and broadband irradiance, should provide a more realistic estimation of how appropriate the incident radiation is for electricity generation since it considers the underlying wavelength distribution of the solar energy.

The results of the study shows that the materials Gallium Indium Phosphide and Amorphous silicon, due to their limited range in spectral response, could be spectrally unfavoured by the Scandinavian solar energy during parts of the year when the solar elevation angle is low. The results also indicates that for every material and every day of the simulated year, there are time steps were there are, at least in theory, better things to track other than the sun. During sunny days the optimum tracking orientation is however generally very close if not exactly the same as the orientation of direct solar irradiation regardless of material. The study concludes that if a tracking system were to be built that could identify the point at which maximum weighted irradiance occurs instead of orienting itself towards the sun, the annual electricity production could be increased by around 10 kWh/m².

• author: Stefansson, Petter
• supervisor: Elisabeth Kjellsson
• organization: Division of Energy and Building Design
• course: AEBM01 20151
• year: 2015
• type: H2 - Master's Degree (Two Years)
• subject: Technology and Engineering
• keywords: PV spectral response., radiative transfer model, spectral mismatch factor, spectral irradiance, Solar tracking
• language: English
• id: 7761721
• date added to LUP: 2016-05-02 09:39:24
• date last changed: 2016-05-02 09:39:24