Efficient ray-tracing with real weather data

New approaches for the computation of the sampling points for an annual simulation of a solar tower power plant are presented. The annual sun-path in azimuth and elevation and in ecliptic longitude and the hour angle are considered. Real measured weather data is considered in the computation of these sampling points. INTRODUCTION The simulation of solar tower power plants becomes increasingly important. It allows to assess the expected annual energy yield and to optimize the planned power plant configuration before construction. With this gain of information, the produced energy can be increased, and the costs can be reduced. For a heliostat field layout optimization, the underlying simulation model should be accurate and fast. As optical simulation model, the convolution method, as in e.g. UHC, Delsol and HFLCAL, or recently more and more the Monte-Carlo ray-tracing method, as in e.g. SolTrace, Tonatiuh and STRAL [1,2] are used. The main influences on runtime are the spatial (number of integration points or –rays, respectively) and temporal (number of time points) discretization. A higher discretization leads to a higher accuracy, but also to a higher runtime. For the annual simulation, usually weather data from clear sky models is used, e.g. the clear atmosphere model from Hottel [3] or the meteorological radiation model (MRM) [4]. This data shows a symmetric behavior: a day is symmetric before and after noon and a year is symmetric before and after June 21st. This information is used to strongly reduce the number of temporal sample points [5]. FIGURE 1. Plot for the DNI distribution over a whole year with the MRM model (a) and the measured weather data (b) for Mumbai from EnergyPlus. But for industrial performance computations, real measured weather data (e.g. from a TMY file) should be used. For this case, the symmetric approach fails, compare with Fig. 1. To consider non-symmetric weather data for an annual simulation the brute-force method would be to simulate all 8760 hours of a year (weather data is usually provided in hourly data). But of course, this method is computationally too costly. (a) (b) With a smart choice of the sampling points, the number of simulations can strongly be reduced, while maintaining accuracy. In this paper, different temporal integration approaches are presented and discussed. In the following, two main concepts are presented: the temporal and the angular integration. TEMPORAL INTEGRATION The annual energy production Eyear of the solar tower power plant can be computed with the sum over all days d and the integral of the daily power production, Eyear =%&' P(t, d) dt -./-01 -./23-0 4 578 9: ;(<=>(?)) @AB