The finite element method based on fluid-structure interaction is used to systematically study the inertial migration of double vesicles in microtubule flow with a two-dimensional model. The results show that the equilibrium position of inertial migration of two circular vesicles with initial symmetry is always symmetric about the center of the channel, and with the increase of Reynolds number, the equilibrium position will be closer and closer to the center of the channel. Secondly, for the double vesicle system composed of circular vesicles and elliptic vesicles, when the initial positions of circular vesicles and elliptic vesicles are located on both sides of the channel, the equilibrium position of circular vesicle inertial migration is almost constant with the increase of Reynolds number, but the elliptic vesicles shift to the center of the channel and across the center to the other side of the channel, and finally move slowly to the wall with the increase of Reynolds number, and at Re≥500, the radial displacement of elliptic vesicle reaches the maximum value. When the circular and elliptical vesicles are located on the same side of the channel, the final equilibrium position is closer to the wall of the channel with the increase of Reynolds number, regardless of whether the elliptic vesicles are anterior or posterior. The present study elucidates the physical mechanism behind the vesicles based on their forces, and the related results can facilitate the application of inertial microfluidics in the precise separation and manipulation of vesicles.

Pseudopotential model of lattice Boltzmann method is partially non-local for pseudopotential calculation with coupling of lattices, which leads to synchronization of threads in parallel implementation process. Besides, it uses a large number of registers and much time of data access operations when access global memory in calculation process. They lead to low computational efficiency. In this paper, a multi-relaxation-time(MRT) 3D pseudopotential model with D3Q15 lattice is adopted as an example to investigate performance of parallel computing based on GPU. To address limitation of parallel computing of pseudo-potential model, efficiency of reading and writing of global memory is improved by using merge access method. To improve efficiency of grids retrieving data which are in boundary of lattice, a "Directional Transfer" algorithm is proposed. The role of computing resource configuration is investigated with different sizes of block, and optimal resource configuration scheme is obtained.