Real-Time Simulation And Analysis Of DAM-Break Flood Based On GPU-CA Model

The simulation approach based on cellular automata, having the benefits such as convenient boundary handling, simpler arithmetic, better support for spatial analysis, has become a powerful tool and hot research topic in the field of dam-break flood simulation and analysis. However, the present simulations of dam-break wave and cellular automation are usually performed by the CPU (Central Processing Unit) computation, leading to relatively low computational efficiency. Especially, the system’s response speed and interaction efficiency will be severely affected when high resolution DEM data are used to simulate and analyze the dam-break flood flouting progress. In order to improve the computational efficiency of the numerical calculation of dam-break flood process and enable real-time simulation and risk assessments, technologies like mass parallel computing platform CUDA (Compute Unified Device Architecture), virtual geographic environment and geographic information system were employed in this work to research the dam-break flood model based on GPU-CA (Cellular Automata based on Graphic Processing Unit). The interaction methods to simulate and assess the risks of dam-break flood routing were designed based on virtual geographic environment platform, which is based on the existing dam-break flood models and risk assessment methods. The main work in this paper is consisted of the following aspects.1) Cellular automaton model for dam-break flood process based on GPU-CA design. Based on the natural similarity between the parallel computing features of cellular automata and the parallel computing architecture of the CUDA, a dam-break flood computing model based on GPU-CA was proposed. Key technologies, such as the computational frame, data mapping, data race and protocol computation, calculation flow and the program optimization method are discussed in detail.2) Dam-break flood simulation and risk assessment based on virtual geographical environment platform. According to the concept of virtual geographical environment, the cellular automaton model of dam-break flood was integrated and a virtual geographical environment platform framework was designed for the simulation of dam-break flood routing and risk assessment. Furthermore, the interaction method was also devised for the simulation of dam-break flood routing and risk assessment.3) Prototype system developments and case experiment analysis. The Microsoft Visual C++2012, CUDA5.5, OpenSceneGraph and osgEarth were employed to develop a prototype system to realize real-time simulation and analysis of Dam-break flood routing, and the Xiaojiaqiao barrier lake was selected for carrying out an experiment. The experimental results showed that the dam-break flood computing model based on GPU-CA could greatly improve the computing efficiency and ensuring the validity of the simulation results. Moreover, the model supports real-time simulation analysis and improves the risk assessment of the dam-break flood routing greatly.