Numerical studies of extrudate swell and distortion of polymer melts

This thesis reviews recent developments in calculating molecular weight distributions (MWD) from polymer viscoelastic properties. Uniqueness in MWD prediction is obtained by using an analytical relation between the relaxation spectrum and molecular weight. This scheme gives comparable results with others methods of determining MWD from rheological data without introducing arbitrariness. Accurate predictions of MWD confirm a completeness of frequency range in oscillatory shear experimental data. In turn, the wholeness of relaxation spectra as substantiated by MWD predictions, bolster the level of confidence when using constitutive models based on these spectra. This is particularly important when using these models in simulations of complex flow phenomena.; PolyFlow, a software package based on the finite element methods was employed to simulate the extrudate swell for polybutadiene of various molecular weight (M_w) and molecular weight distribution (MWD). We calculated the relaxation spectra for the different samples and then inserted the spectra into a standard viscoelastic constitutive model used in the numerical simulations. We elucidated the importance of using the full range of relaxation spectrum rather than a short range around a typical shear rate for the accuracy of the numerical predictions. We found extrudate swell ratio (ESR) to be strongly dependent on MWD and stress conditions at the die exit.; We also used PolyFlow to study sharkskin phenomena. A stick-slip mechanism is used as the basis for the simulations. We use a superposition of stress relaxation/stress growth and a periodic change in extrudate swell governed by the die exit stress level to depict sharkskin.; Three relatively monodisperse polybutadienes were used in this study. The simulated sharkskin time period was found to be in good agreement with experimental findings. We found that the simulated pictures of sharkskin are similar for all three molecular weight samples.; A comparison between simulated sharkskin and experiments show qualitative resemblance. The main problems preventing us from quantitatively reproducing sharkskin amplitude lie in our limitations in depicting stress singularity, limitations in mesh design refinement and the constitutive model employed. In spite of these limitations, the qualitative agreement between simulation results and experimental data is good.