Abstract
By the use of molecular dynamics and parallel algorithm on PC-cluster, we report a series of properties including velocity distribution, pressure distribution, stress tensor and molecular structure at equilibrium state for nano-scale contraction-expansion flow with different chain length.There are some interesting effects on chain length of contraction-expansion flow under the same driving force. With the increasing chain length, the density of local system gets more uniform. This is great different form the assumption that the density of fluid is incompressive. For the roughness of wall and the viscosity of fluid, the velocity profile is very special.Under the same driving force, the flow velocity and shear stress decreases with the chain length while the pressure and normal stress increases with the chain length. The end-to-end distance and orientation factor is large at the contraction flow. Because of the strong attraction of the wall, there is large end-to-end distance and orientation factor in the surface of wall at the expansion flow.The parallel efficiency changes with the computing data. For small computing data, the parallel method has lower efficiency. We have used the parallel algorithm coupling atomic decomposition and force decomposition parallel method successfully to simulate molecular dynamcis. With the lower number of parallel computer, the efficiency gets much better.