Predicting the solar irradiance is the key to estimating the performance of the solar energy conversion system and ensuring the stable operation of the power grid. Solar irradiance is expressed by radiant flux density or power density (W/m²). For a solar power system, the effective solar energy can be expressed as the product of the solar irradiance (s) incident on the collector and the effective total area of the collector (W/m²xm²=W). Electric utilities operate their own power generation systems and charge customers based on the power consumption (kWh) within a certain period of time. The power generation of a solar power system is related to effective solar irradiance and other factors, such as specific system design performance and important environmental factors. Photovoltaic power plants show linear changes in the process of solar power generation; that is, during their operation, the overall conversion change is often less than 20%. On the other hand, due to the non-linear characteristics of thermal inertia and thermodynamics, this makes it challenging to establish a correlation between concentrating solar energy (CSP) power generation and direct radiation (DNI), at least in a short period of time. That’s it. There are several models for evaluating the performance of solar power systems.
The atmosphere acts as a medium in the process of solar radiation reaching the surface. The number and type of atmospheric components and the radiation properties related to wavelength determine the scattering, absorption and transmission of radiation in the atmosphere. As shown in Figure 1, cloud cover mainly affects the intensity and type of solar radiation used for energy conversion. In fact, most of the solar resource data used in the United States is not derived from the measured values of direct pyranometers and total pyranometers, but model estimates based on observations of the cloud cover on the ground and satellites (Wilcox, 2012). Solar radiation prediction is also highly dependent on cloud conditions within the prediction interval. The input information for solar prediction includes cloud type, height, relative movement and formation/dissipation area. In the process of solar energy prediction, the radiative transmission characteristics of clouds can be understood through detailed cloud information (for example, optical thickness, cloud liquid water content and cloud ice content, effective radius of cloud drops).
When the sky is clear and cloudless, there are also complex interactions between solar radiation and the “clear” atmosphere. For the effective solar radiation obtained by the solar collector, the factors that affect its spectral distribution are:
Types and quantities of atmospheric aerosols, total precipitation, ozone and other components. Figure 2 shows the predictable interannual changes in clear sky DNI measurements, which are affected by the earth’s orbit and periodic increases in atmospheric aerosol content. The data in this figure is continuously measured by a direct radiometer. As the highest DNI value in any hour of the month, it usually appears at noon when the sky is clear and cloudless.
Under clear sky and cloudless conditions, atmospheric aerosols cause forward scattering of solar radiation, thereby reducing the DNI value and increasing the DHI value. This redistribution of radiation near the surface of the solar disk is called circumsolar radiation. The intensity of solar radiation is of great significance to all concentrating solar power systems. The large amount of circumsolar radiation generated by atmospheric conditions can affect the shape of the sun or the DNI available for concentrating collectors (see Figure 3).
The spectral distribution of solar radiation on the ground is of great significance to solar power generation methods, especially the design and performance testing of photovoltaic devices. About 97% of the available radiation wavelengths in the solar spectrum are in the range of 290nm to 3000nm (Figure 4). The solar spectrum at the top of the atmosphere is quite constant, close to blackbody radiation at a temperature of 5520K. The atmosphere is equivalent to a continuously changing filter. By changing the relative content of DNI, DHI and GHI, different available radiation spectrum distributions can be obtained. There are fewer sources of spectroradiometric measurements (USDOE; NREL measurement and instrument data center). Taking the weather data under clear sky and cloudless and all-sky conditions as input values, the spectral distribution model of solar radiation can be established.
Under clear sky and cloudless conditions, we refer to the amount of DNI passing through the atmosphere to the surface as the atmospheric path length or relative atmospheric mass (AM). When the sun is somewhere directly above sea level, the length of the atmospheric path is 1.0 (for example, AM1.0). Figure 5 illustrates the dependence of AM on the position of the observation point (collector) relative to the sun. Since all locations and seasons do not meet the AM1.0 conditions, AM1.5 was determined as the clear sky standard solar spectrum when establishing the photovoltaic model (Figure 6).
The method of predicting solar irradiance must account for changes in the position of the sun and atmospheric characteristics, as well as the influence of these characteristics on the available solar irradiance. In the prediction interval, the basic methods of dynamic application of prediction are as follows. Firstly, the available radiation under clear sky conditions is evaluated through atmospheric composition measured by meteorology or remote sensing; secondly, the cloud cover is explained. Methods to obtain cloud scenes according to different prediction intervals include: surface observation, satellite observation, or numerical weather forecast evaluation. In addition, ground measurement devices can be installed at the power generation site to provide additional data for the input and verification of predictive models.
