Generating electricity via solar thermal power is a new technology of applying clean and renewable power source. It concentrates the solar power through heliostat field and coverts luminous power into thermal power for generating electricity. In the-sis, we propose an efficient, precise, and adaptable simulation system for large-scale heliostat field. In the system, we first give the sunshape mode, pre-compute the sha-dows casted on the mirrors, and optimize the heliostat microfacet reflection model. Monte Carlo ray tracing is adopted to simulate luminous power transfer from solar to the concentrator. As a result, luminous power transfer efficiency of heliostat field is obtained. The system can be implemented on both CPU and GPU for the sake of adapting various computer hardwares. In order to exploit parallel performance capacity of algorithms on both CPU and GPU, a multithread SIMD feature is employed on CPU and the algorithms are optimized via CUDA on GPU. Implementation results show that the system is adapted to massive data and computing in the heliostat simulation. It can efficiently compute solar power gathering efficiency. It will be applied to the he-liostat field layout optimization in the future. |