Computational Fluid Dynamics of Polymer Flow Induced Crystallisation

This thesis models flow induced crystallisation (FIC) in polymers using the Rolie-DoublePoly (RDP) model [1] in combination with the polySTRAND model [2] to create a computationally efficient method for modelling FIC. The RDP model incorporates interactions
between different chain species in a polydisperse melt. The polySTRAND model is a
computationally efficient FIC model that provides a molecular basis for the formation of
crystals in a polymer melt.
The RDP and polySTRAND models are validated against published results and are
implemented in OpenFOAM [3] using the RheoTool library [4]. We then simulate an
idealised bimodal polymer blend in a channel to demonstrate the successful implementation of the RDP model. We then take inspiration from the experimental work done by
Scelsi et al [5] and simulate flow through a contraction expansion geometry that has the
same dimensions as in their experiments. We then compared the flow induced crystallisation effects in the flow where only one chain species contributes to the acceleration of
crystallisation and compared our results to the Scelsi experiments.
The study of the contraction expansion flow is then extended to include multiple chain
species having an effect on the acceleration of crystallisation in flow. This required three
extensions to the previous chapter. We show how a general N mode polydisperse RDP
model can be implemented in OpenFOAM. We then construct a procedure for choosing
relaxation times in the RDP model to fit with rheological data. Finally we modify the
polySTRAND model calculation to account for contributions from multiple chain species
to FIC. With all of this in place we are able to make a more meaningful comparison to
the experimental investigation of Scelsi et al [5].
We further extended our framework to include the effects of temperature on the rheology and the crystallisation dynamics in the system. We need to make modifications to
our framework to account for how temperature affects the growth rate of crystals, relaxation times, viscosity and the energy barrier to crystallisation. We first show how cooling
a polymer melt after a shear pulse can affect the FIC properties in a channel geometry,
following a methodology that is commonly used in the literature [6]. Finally we reconsider
our contraction expansion geometry and our polydisperse melt and simulate flow through
our geometry when the walls are either cold or hot relative to the melt. We then compare
these results to the isothermal polydisperse simulations that we conducted earlier. We
show that by controlling the temperature distribution it is possible to change the spatial
distribution of crystallites.