Coupled computational fluid dynamics and multifilament fiber-spinning model



A fundamental fiber-spinning model (proposed by Doufas and colleagues) is coupled to a three-dimensional Navier–Stokes computational fluid dynamics (CFD) code, where source terms for momentum and energy transfer between the fibers and the surrounding quench air are treated implicitly. The three-dimensional equations are solved for the airflow using a preconditioning technique in a structured multiblock framework. Scalable parallelism is achieved by assigning an arbitrary number of grid zones to a predetermined number of processors. Individual fibers are also divided equally among the available processors, allowing for thousands of fibers to be solved in only a few hours in realistic industrial-scale spinning processes. Model results are shown to agree with results obtained using Doufas et al.'s stand-alone fiber-spinning model in the limit of a small number of fibers. The coupled model is used to compute the temperature and air flow around individual fibers on a generic three-dimensional fiber-spinning application that contains multiple rows of fibers. It is found that, for large fiber bundles (such as 72 fibers) a significant variation in fiber cooling and tensile stresses exists across the bundle, which would result in significant variation in fiber tensile properties within the bundle. It is proposed that the cooling air velocity profiles across a multifilament system that exhibits stress-induced crystallization effects can be extracted from results obtained using the new coupled CFD/fiber-spinning model and can then be used in improved calculations (using a stand-alone fiber model) of the heat-transfer and air-drag coefficients within the fiber bundle. © 2006 American Institute of Chemical Engineers AIChE J 2007