Research on damage characterization of SCC and rock composite specimens based on CPU and GPU heterogeneous code acceleration

The interface between self-compacting concrete (SCC) and rock has significant effects on the strength, damage, and crack growth of rock-filled concrete (RFC). In this paper, the strength, failure characteristics, and damage mechanism of SCC-rock composite specimens with different interface inclination angles under uniaxial compression are studied by physical tests and numerical simulations. The 3D finite element solver is developed by using CPU-GPU heterogeneous computing, which greatly improves computing efficiency and can be applied to large-scale mesh model calculation with tens of millions of degrees of freedom. The GPU-accelerated solver and the Realistic Failure Process Analysis (RFPA) theory are utilized to simulate the uniaxial compression of 3D mesoscopic stochastic mechanical models of three-phase SCC-rock composite specimens at different interface inclination angles. The results show that the compressive strength, peak strain, and accumulated acoustic emission (AE) energy of the composites decrease first and then increase with the increase of interface inclination angle, and the composites with different interface inclination angles show different failure characteristics. The progressive failure process of the composite specimen and its mechanical behavior are reproduced by numerical simulation. The research results provide an important reference for RFC engineering design and research.