You are here

Warning message

Attention! This event has already passed.

Identification of the permeability values of fiber reinforcements of composite materials by inverse methods

Monday, 24 August, 2009 - 17:00
Campus: Brussels Humanities, Sciences & Engineering campus
Gerd Morren
phd defence

Liquid Composite Moulding (LCM), is the state-of-the-art technology
for producing textile reinforced composite parts. In LCM, liquid resin is
injected into a mould holding the reinforcement material. Once this
material is impregnated and the resin is cured, the finished component
can be de-moulded. Virtual prototyping software has been developed to
assist the engineer in correctly designing the mould. However, for
accurate simulations it is absolutely necessary to have reliable input data,
of which the key parameter is the permeability of the reinforcement.
Measurement of the permeability is not yet standardized, and many
different set-ups have been proposed. Moreover, the measurements are
very sensitive to various factors and hence prone to error.

The thesis describes an inverse method for permeability identification.
The proposed method is a so-called mixed numerical/experimental
technique (MNET) for material property identification. The
experimental part is represented by the PIERS (Permeability
Identification using Electrical Resistance Sensors) 2D central injection
rig. The heart of this set-up is a solidly supported steel mould that holds
120 DC-resistance based sensors. The reinforcement is placed inside the
mould and is centrally injected with fluid which triggers the sensors on
arrival as it propagates through the reinforcement. The numerical
component of the technique involves adjusting the permeability
parameters in a customized finite element model until it satisfactorily
simulates the PIERS experiment.

With a single experiment, this fast and robust MNET allows to
determine all components of the in-plane permeability tensor.
Furthermore, since the finite element simulation accounts for the entire
PIERS experiment, all sensor data can be used in the permeability
calculation. In contrast, the typical analytical methods are valid only until
the injected fluid reaches an edge of the reinforcement and subsequently
reached sensors can not be used in the calculation. It is shown that, for
reinforcements that are not approximately isotropic, the finite element
simulation significantly improves the precision of the parameter
identification compared to the formerly used analytical method.

The thesis also presents a textile-like solid specimen with anisotropic
permeability that is produced with a stereolithography technique. It is
designed as a reference for calibration and comparison of permeability
measurement set-ups and for validation of numerical permeability
computation software. Unlike real textiles, the permeability properties of
such reference specimens do not vary from test to test. When used for
benchmarking, any discrepancy between different measurements on this
specimen must be attributed to the set-up and data processing.

In the final section, the reference specimen and the proposed MNET are
put to the test. The first experimental measurements of the permeability
of the reference specimens are presented and discussed. It is shown that
an excellent repeatability of the experiments is obtained. Since the
reference specimen allows isolating the variability arising from the
measurement technique, a clear assessment of the MNET is obtained.
Moreover, statistics from the series of measurements corroborate that
the proposed FE-based method, for processing the PIERS measurement
data, performs significantly better than the common analytically based
method, which was used before. Finally, the experimental results are
shown to be in good agreement with the values predicted using
numerical permeability computation software.