Vibration-based damage identification in composite beams
View/ Open
Gu2023.pdf (11.35Mb)
Date
09/06/2023Item status
Restricted AccessEmbargo end date
09/06/2024Author
Gu, Yu
Metadata
Abstract
Composite structures are widely used in bridge and building construction; a typical
example is a bridge deck consisting of concrete slabs and steel beams, where shear
connectors are used to connect the concrete slabs and steel beams to form a
composite structure. The performance of a composite structure is well understood to
depend not only on the properties of the primary components, such as the slabs and
steel beams mentioned above but also, and importantly, on the properties and
condition of the shear connectors. Therefore, in a structural health monitoring and
damage identification process, it is imperative to distinguish the damages in the
primary components and in the shear connectors. However, in the existing literature
concerning damage assessment of composite structures, there is generally a lack of
differentiation between the damages in the two distinctive groups of constituent
entities, and oftentimes the damages are simply treated in terms of the gross flexural
stiffness with the use of an equivalent Euler-Bernoulli beam. This could lead not only
to a false identification of the actual severity of the damages but also misleading
results in case severe damage to shear connections occur.
In this thesis, a comprehensive study on the damage assessment of composite
beams, which to a large extent, represent the basic mechanics of composite structures,
is conducted. Firstly, the basic mechanics governing the overall transverse flexural
stiffness and the nominal sectional rigidity of a composite beam is investigated
analytically, and the essential differences between the component beam damage and
the shear connector damage on the distribution of the nominal flexural rigidity is
examined by numerical simulations. On this basis, the feasibility of differentiating the
two groups of damages from a damage identification process using vibration
information, namely the natural frequencies and mode shapes, is investigated by
means of a genetic algorithm (GA) - based finite element (FE) model updating.
Considering the limitation of the GA-based approach due to its global optimisation
nature, a neural network-based method is then developed for the identification of
damages in composite beams. In this development, the following sub-topics are
systematically studied, i) The sensitivity of the dynamic properties of composite beams
to the two groups of damages under different composite beam configurations; ii) The
normalisation of the shear connector stiffness and component beam flexural stiffness
and the corresponding utilisation in the neural network-based damage assessment
scheme, and iii) Identification of multiple damages in a composite beam using a neural
network-based damage assessment scheme. The wavelet packet node energy
(WPNE), which could allow for the use of a single or few limited measurement points,
is incorporated into the neural network-based damage identification scheme. It is
shown that the WPNE-incorporated scheme can be efficient for identifying damages
in composite beams. Finally, in conjunction with the analytical and numerical
simulation studies, a laboratory experimental programme has also been conducted,
and the feasibility of separating the shear connector damage and the beam section
damage in a physical measurement environment is verified by employing the
experimentally measured vibration data.
In conclusion, this thesis provides a comprehensive study on the damage
assessment of composite structures, with a particular focus on distinguishing
damages in the primary components and the shear connectors. The proposed
methods, including the GA-based finite element model updating and neural network-based damage identification scheme, are shown to be effective in identifying damages
in composite beams using vibration information. The laboratory experimental
programme verifies the feasibility of the proposed methods in a physical measurement
environment. The findings of this thesis will contribute to a better understanding of the
behaviour of composite structures and improve the accuracy of structural health
monitoring and damage assessment in practice.