Refine
Year of publication
- 2026 (1)
Document Type
- Doctoral Thesis (1)
Language
- English (1)
Has Fulltext
- yes (1)
Institute
- Fachbereich 4 (1)
Fiber melt spinning is an essential manufacturing process for the creation of polymer fibers, which are a key component in the textile industry. Due to the fact that industrial spinning devices are often constructed for the large-scale production of polymer fibers, an experimental variation of setup parameters for creating fibers with the desired properties leads to high expenses. Hence, there is a need for numerical simulations of the considered process. However, in order to use computer-aided design, every material law, including the polymer’s temperature and strain-rate-dependent elongational viscosity law, must be specified. Due to the high strain rates in fiber melt spinning, a direct measurement of elongational viscosities is currently infeasible. As a consequence, the identification of a suitable viscosity law is challenging and hinders the predictive capability of numerical process simulations.
This thesis investigates the identification of the polymer’s elongational viscosity law based on diameter measurements along the spin-line for varying process parameters. Mathematically, this leads to a nonlinear optimization problem in a function space subject to ordinary differential equations prescribed with boundary conditions. For the solution of this problem, we propose a two-step procedure. First, we introduce a fiber model based on a boundary value problem of ordinary differential equations, as well as a numerical framework for the simulation of generalized Newtonian materials.
The key idea of the first step is to identify triples of temperatures, strain rates, and the corresponding viscosities for each set of diameter measurements. To achieve this, we propose and investigate methods for determining the polymer’s elongational viscosity along the spin-line, using the proposed fiber model. We analyze the resulting initial and boundary value problems as well as the formulated optimization problems with respect to existence and uniqueness of solutions and supply our analytical findings with academic numerical examples.
In the second step, symbolic regression is applied to the in step one generated dataset to reconstruct an approximate viscosity law. The two-step solution strategy is evaluated using artificially created measurement data of four different Newtonian and generalized Newtonian materials. For every material, we compare the identified viscosity law against the actual law and furthermore compare the associated diameter profiles with the reference solutions. Overall, the results indicate that the proposed approach yields elongational viscosity laws which enable accurate numerical simulations of fiber diameters in the considered melt spinning processes.