In this paper, non-equilibrium multiscale molecular dynamics simulation is applied to investigate the transport properties (shear viscosity, diffusion coefficient and thermal conductivity) and structural properties of dense methane for Poiseuille flow within rough silicon nano-channel. In this simulation, the recently modified OPLS model is used to capture the micro-information of dense methane nanofluidics, as well as its coarse-grained model to solve the problem of wall-fluid interactions via the classical Lorentz-Berthelot mixing rule. In order to calculate the local shear viscosity in fluid layers accurately, the fitted local velocity profile method is introduced to determine the viscosity value by fitting the local velocity profile based on the Navier-Stokes equations. Additionally, the Green-Kubo formula is employed to compute the diffusion coefficient and thermal conductivity of methane nanofluidics. From the simulation, we conclude that atom number density, transport and structural properties of methane nanofluidics present significant variations near the rough nano-channel walls, and the obtained data in the central part of the nano-channel is close to experiment value of bulk system. Moreover, the streamlines are distorted near rough nano-channel walls. The results above are significant in understanding of the nanoscale hydrodynamics, and should be taken into account in fluid dynamics theory and the design of nanofluidic devices. (C) 2017 Elsevier Ltd. All rights reserved.