Three modeling methodologies based on mathematical, empirical and artificial neural network based on radial basis function have been compared for their ability to predict fabric properties. It has been observed that artificial neural network model produced the least error as well as minimum range of error as compared to the other modeling methodologies.

View full abstract

Fiber suspension flow and fiber orientation through a parallel plate channel were numerically simulated for mono- and bi-disperse fiber suspensions in order to study the mechanism of the evolution of fiber orientation distribution. In the present simulations, fiber-fiber interaction was not taken into account.
For suspensions including small aspect-ratio fibers, flip-over or oscillation phenomenon of the orientation ellipsoid was predicted. For the r_a = 10 + 40 (bi-dispersion including fibers with aspect-ratio r_a = 10 and r_a = 40) and r_a=10 + 100 suspensions, fibers with r_a = 40 and r_a = 100 almost completely aligned along the channel wall, however these fibers had little influence on the evolution of the orientation ellipsoids for the bi-disperse suspensions because the number of fibers was much lower when the same fiber parameters were assumed for both small and large aspect-ratio fibers. In contrast fibers with r_a = 20 could affect the evolution of the orientation ellipsoids for the bi-dispersion with r_a = 10 + 20. It could be found that the phenomenon taking place, flip-over or oscillation, depends on the initial orientation distribution of fibers, the number of fibers of each aspect-ratio for bi-dispersion and the flow kinematics of fiber suspension.

View full abstract