Polymer fiber processes, such as high-speed spinning of nylon and PET, are highly complex involving a complicated interplay between an evolving internal molecular microstructure, macroscopic transport phenomena, such as fluid mechanics and heat transfer, and also non-equilibrium thermodynamics and kinetics affecting nucleation and subsequent crystal growth. All of the above processes are important in determining the final product's semi-crystalline morphology which is primarily responsible for its mechanical properties. Our approach to attack this problem is a hierarchical one: A macroscopic (continuum) model is developed based on the Hamiltonian formalism of non-equilibrium thermodynamics for flowing systems [Beris and Edwards, Oxford University Press, Oxford, 1994]. This approach allows us to follow the dynamic evolution of several macroscopic (continuum) variables involving both kinematic and structural parameters. To implement this approach successfully, an accurate modeling of the (extended) free energy (Hamiltonian) of the system under consideration and the dissipation therein is necessary. While the later is, at the moment, phenomenological, we are developing a first principles approach for the former based on a microscopic modeling of chain conformations using a lattice model. Lattice models have been used extensively before for the analysis of chain conformations in both purely amorphous and semi-crystalline polymers. We have reinterpreted some of the earlier lattice models by systematically deriving the relevant statistics of polymer chains and by outlining the a priori approximations in those models which are necessary to arrive at closed-form expressions. Although we do make use of such earlier work, we have also extended it through a computer-aided analysis. This analysis has enabled us to generate from first principles free energy surfaces for a system consisting of polymer chains, represented as multiple self- and mutually-avoiding random walks, on a 2-D fully populated semi-crystalline lattice. It is shown that the numerical results for dense semi-crystalline systems can be fitted with low order polynomials that provide closed-form approximations for the configurational entropy in terms of non-equilibrium structural parameters, such as the orientation and stretching of the polymer chains. Finally, chain statistics for bulk amorphous polymers have been validated against their theoretical predictions.