Determination of atmospheric extinction for solar tower plants

Losses of the reflected direct normal irradiance (DNI) between the heliostat field and receiver in a solar tower plant are caused by atmospheric scattering and absorption by aerosols and water vapor in the atmospheric boundary layer and can vary significantly with site and time. The radiation losses can be larger in desert environments compared to the standard and constant atmospheric conditions which are usually considered in ray-tracing or plant optimization tools. The main challenge for solar tower plant project development is the limited site-specific information about atmospheric extinction for many sites which are interesting for concentrated solar power (CSP) projects. Moreover, this information also affects the plant operation strategy because atmospheric extinction influences the power which is concentrated at the receiver aperture.Pitman and Vant-Hull (1982) presented a transmittance model (P&V model) which enables the calculation of radiation loss in a tower plant due to extinction, but relies on on-site measurements of the scatter coefficient which is usually not available.Four different approaches to determine atmospheric extinction and respectively atmospheric transmittance with commercially available instrumentation are presented and evaluated in this work. The first two approaches are based on a correction method (ABC method) which uses radiative transfer simulations to transform measurements performed with the FS11 scatterometer of Vaisala and the LPV4 transmissometer of Optec to the broadband transmittance which is needed for CSP. The ABC method reduces the mean deviation of the transmittance at Plataforma Solar de Almería (PSA) between the FS11 and LPV4.Therefore, it is recommended to apply the ABC method for atmospheric transmittance measurements instead of using only raw measurements of the FS11 or LPV4 for solar resource assessment. A comparison of the P&V model with the ABC corrected FS11 and LPV4 measurements is performed. The result confirms the limitations of the P&V model and emphasizes the necessity of a careful choice of the source of the scattering coefficient.Furthermore, an approach is presented which derives atmospheric broadband transmittance from particle number measurements of the EDM164 particle counter of Grimm. The approach is mainly limited by the unknown inlet transmittance of the EDM164 as well as necessary assumptions about aerosol specifications.Moreover, another approach to develop a transmittance model of Sengupta and Wagner (2011), is tested for PSA and enhanced. The advantage of this approach is that it only requires data of a common meteorological measurement station including DNI measurements which is available for CSP resource assessment anyway. Costs for additional instrumentation which is necessary for other approaches can therefore be avoided. Within its uncertainty limits, this approach seems promising and will be tested also at other sites.The transmittance height profile at PSA can be considered as homogeneous in the first 100m above the surface according to two experiments performed at PSA. The horizontal derived transmittance at ground level can therefore be used representatively for the transmittance between the heliostats and the receiver at PSA.So far, only standard transmittance conditions with a temporal constant extinction coefficient are applied in plant yield calculations and field optimization tools. The annual plant yield for an exemplary solar tower plant at PSA is simulated in this work. The raytracing simulations show that overload dumping has to be considered in the discussion about the effect of atmospheric extinction in solar tower plants. Overload dumping takes place if some heliostats are defocused to avoid overload at the receiver or turbine. The annual plant yield of the exemplary solar tower plant is reduced by about 7.04% by atmospheric extinction at PSA and 1.56% if overload dumping due to the receiver and storage limitations is considered. The annual yield would be underestimated by about -0.38% (-0.16 %) if a standard clear transmittance model equation would be applied instead of the variable transmittance time series and -11.18% (-3.66 %) for a standard hazy model equation.To conclude, site- and time-dependent transmittance data sets should be considered by project developers during resource assessment, especially at arid sites where the annual plant yield can be reduced by several percent due to atmospheric extinction. At clear sites like PSA, standard assumptions might be accurate enough. To identify a clear site, the here presented extinction measurement systems based on DNI measurements can be applied. The system including FS11 data can be recommended for a detailed examination of atmospheric extinction also at hazy sites.The modeling of site- and time-dependent extinction in ray-tracing tools is now possible. Further, an evaluation tool is available to rate the accuracy of solar tower plant simulations concerning atmospheric extinction and one exemplary plant is evaluated for PSA