ADS Spallation Target Assisted Visualization System Based On GPU-CUDA

Today nuclear energy has become an outstanding energy. While making great contribution to the society, it also brings amounts of challenges. One of the key problems is how to deal with the huge amounts of nuclear waste. Accelerator Driven Sub-critical System, also named as ADS evolution system, is a very promsing scheme of handling nuclear waste. It can decrease the emission ability of nuclear waste significantly. The energy produced during the process improves the fuel efficiency as well. In 2011, Chinese academy of sciences established strategic pilot program, and one of its ultimate goal is to complete an industrial-grade ADS system.In recent years, both computer software and hardware are developing fastly. Computer can simulate the real world visually. CPU, cooperating with GPU, can save the experiment time as well as money greatly. In this paper, we do some research from the perspective of computer science, using Qt to develop an ADS spallation target assisted visualization system and using GPU-CUDA to accelerate the process of physical simulation. The system mainly includes three modules: modeling the spallation target, granular flow transportation simulation and high-energy Monte-Carlo simulation, and visualizing theirs related results. The modeule of modeling the spallation target uses Open CasCade geometric library to construct and operate geometric model, and uses STEP standard to translate the modeling result to a special format which can be recognized by MCNP. This module can also provide support for granular flow transportation simulation. The module of granular flow transportation simulation uses DEM method to simulate the process of granular flow transportation. It also uses GPU-CUDA and cluster technology to realize parallel computation. The key issue is to handle the process of collision detection between particles as well as particels and vessel walls. After the calculation, we can pick out the meaningful information and using OpenGL to render it. Through contrast experiment, we find that using different size of Block and Cell in the simulation can achieve different performance improvement. After using CUDA to parallel the code, the speed of simulation increases several times. The module of High-energy Monte-Carlo simulation involves large amount of knowledge of nuclear physics, so in this paper, we do not do much theoretical research, our main job is to integrate the code provided by The Chinese academy of sciences institute of modern physics to the system. We also design and implement a parameter input card to help researchers collecting the parameters conveniently. After the calculation, we save the results and using Open GL to realize visualization.After developing the three modules mentioned above respectively, we use Qt to complete integration. By means of visualization to assist the research of spallation target can release resaerchers from tedious work which has nothing to do with the analysis of the data itself, and thus saving a lot of time as well as providing support for designing the ADS spallation target.