Carbon nanotube sheets or films, also known as "buckypaper", have been proposed for use in actuating, structural and filtration systems, based in part on their unique and robust mechanical properties. Computational modelling of such a fibrous nanostructure is hindered by both the random arrangement of the constituent elements, as well as the time and length scales accessible to atomisticlevel molecular dynamics modelling. Here we present a novel in silico assembly procedure based on a coarse-grain model of carbon nanotubes, used here to attain a representative mesoscopic buckypaper model that circumvents a need for probabilistic approaches. By variation in assembly parameters, including the initial nanotube density and ratio of nanotube type (single-and double-walled), the porosity of the resulting buckypaper can be varied threefold, from approximately 0.3 to 0.9. Further, through simulation of nanoindentation, the Young's modulus is shown to be tunable through manipulation of nanotube type and density over a range of approximately 0.2 to 3.1 GPa, in good agreement with experimental findings of the modulus of assembled carbon nanotube films. In addition to carbon nanotubes, the coarse-grain model and assembly process can be adapted for other fibrous nanostructures such as electrospun polymeric composites, high performance nonwoven ballistic materials, or fibrous protein aggregates, facilitating the development and characterization of novel nanomaterials and composites as well as the analysis of biological materials such as protein fiber films and bulk structures.