Modeling Progressive High-Cycle Fatigue in Polymer Composites

Abstract

The lack of reliable computer methods for simulating the mechanical response
of fiber-reinforced polymer composites is slowing down the industrial uptake of
innovative designs and novel manufacturing procedures. Polymer composites
are advanced engineering materials, whose microstructure gives rise to a complex
interaction and competition between various failure processes on different
space and time scales. Consequently, developing accurate, yet efficient and robust
simulation tools for predicting the mechanical performance of composites is not
straightforward.

Over the past few decades, significant progress has been made towards simulating
fatigue in polymer composites. Several modeling approaches have been
proposed for propagating initial cracks in elementary specimens. However, the
development of reliable failure simulation tools for cases often encountered in practice,
where cracks initiate and propagate, plastic deformations interact with fatigue
damage and processing effects play a role, is still a challenge.

In this dissertation, new numerical methods are developed for simulating high-cycle
fatigue in polymer composites. Firstly, a framework for simulating inter- and
intra-laminar damage in multidirectional laminates under cyclic loads is presented,
while taking into account the effects of thermal residual stresses as a result of
curing. Secondly, the framework is extended to simulate failure in overmolded
thermoplastic composites and is used to provide insights into the effect of processing
on the mechanical performance.

In Chapter 2, a high-cycle fatigue cohesive zone model covering initiation and
propagation is embedded in an XFEM framework for simulating the interactions
between transverse matrix cracks and delaminations in multidirectional laminates.
For efficient and robust progressive failure simulations, the fatigue cohesive zone
model is enhanced with a fully implicit time integration scheme. Furthermore, the
model is consistently linearized to achieve fast convergence. With the implicit time
integration, adaptive time stepping based on global convergence rates is enabled,
further increasing efficiency of the simulations.

To account for varying local stress ratios as a result of thermal residual stresses,
the framework is extended in Chapter 3 with an adaptive cycle jump strategy where
local stress ratios are computed during the simulation. The accuracy of the framework
is demonstrated through validation against experiments on multidirectional
laminates with two different stacking sequences. With the XFEM description for
transverse cracking, it is possible to simulate a transition from diffusive damage to
localized failure.

To make the step towards failure predictions of overmolded thermoplastic composites,
a mesoscopic constitutive model for continuous fiber-reinforced thermoplastic
composites is developed in Chapter 4. It is shown that the mesoscopic constitutive
model captures the homogenized (transversely isotropic) response of a detailed
micromodel with isotropic viscoplastic matrix and elastic fibers under various stress
states. For efficient high-cycle fatigue analyses, a two-scale time-homogenized version
is derived in Chapter 5, avoiding the need to simulate every cycle individually,
thus drastically increasing computational efficiency of simulations.

The final framework (Chapter 5) combines cohesive fatigue with time-homogenized
viscoplasticity and is applied to the simulation of a T-section cut
from an overmolded rib-stiffened panel. The effects of processing are taken into
account by assuming different healing profiles along the overmolded interface.
Furthermore, an engineering approach is proposed to obtain a processing-induced
geometry and mesh by applying artificial boundary conditions. The simulation
outcomes demonstrate that the effects of viscoplasticity and the process-induced
mesoscopic geometry of the laminate influence the mechanical response. The effect
becomes more pronounced with improved bond quality at the overmolded
interface.

The developed numerical methods enable efficient and robust fatigue simulations
of polymer composite components under a large number of cycles with
interacting and competing failure processes, plastic deformations and processing
effects. These tools contribute to more reliable virtual testing and facilitate reaching
the full potential of polymer composites.