Get started
Work

Dynamica van structurele lijmverbindingen. Experimentele karakterisatie en modelgebaseerde parameteridentificatie

Year: 2024
Type: dissertation
Abstract: In recent years, automotive, aerospace and manufacturing sectors have been witnessing a clear trend towards the adoption of lightweight and more sustainable materials. This shift is driven by the need to meet higher requirements in terms of energy efficiency, performance and sustainability in mechatronic systems. As materials evolve, there is a corresponding need to adapt the structural joints that hold these systems together. In this context, there is an increasing use of structural adhesive joints. These joints offer a better stress distribution compared to traditional mechanical fasteners, such as rivets and bolts. During their lifetime, mechatronic systems are both source and receiver of dynamic mechanical loads. These loads manifest as noise and vibrations that can lead to discomfort to the human users, negative impact on the functioning of the system, and fatigue damage which, in extreme cases, can lead to catastrophic structural failure. To account for, anticipate and improve the structural dynamic behavior of mechatronic systems, vibrations are typically considered during the design phase via prototype testing and numerical analysis. From previous scientific studies, it is well known that mechanical fasteners have an impact on the structural dynamic behavior of systems. What is less clear is how the adhesive joints influence the system vibrations. More in detail, two questions emerge: what is the magnitude of their influence, and how accurately can numerical models predict this influence? The availability of accurate models will not only support the design of structural adhesive joints, but also enable the model-based monitoring of these joints during operations. This work aims to address these questions above by combining finite element modeling with experimental modal analysis. Firstly, the influence of adhesive bonds on system vibration is estimated in terms of modal features and frequency response functions using finite element modeling, in which the adhesive is treated with dedicated solid elements. These models are also used to compute the sensitivity of the eigenfrequencies to the design parameters of the joint, such as the dimensions and material properties of the adhesive layer. Secondly, the accuracy of these models is verified with reference to experimental modal analysis, which is based on vibration tests with hammer excitation and accelerometer measurements. The sensors location is defined to maximize the sensitivity to the adhesive stiffness and minimize the impact of location errors. As compared to other joints, such as bolts, using an adhesive bond only has a limited influence on mode shapes but can lead up to a 6-fold increase of the modal damping values and 10% increase of the eigenfrequencies. The validity and robustness of the experimental characterization procedure against external disturbances is assessed. Sensors intrusiveness, inaccurate substrates material properties and air radiation damping are quantified and found negligible with respect to the adhesive influences. Finally, the accuracy of the considered numerical models is enhanced through vibration-based identification of the most sensitive adhesive model parameters, which is performed on plate-to-plate joints. Different levels of model complexity are considered, depending on the adhesive type. For stiff adhesives, such as epoxies and acrylics, linear-elastic models are used and the adhesive Young's modulus and loss factor are identified with a framework based on a surrogate parametric modal model. Novel algorithms allow the correct interpolation between the samples of the surrogate model, solving the computational issues related to mode veering, mode crossing, and sign inversion. For softer adhesives, visco-elastic models are considered instead. In this latter case, the four parameters governing the fractional derivative formulation are identified with a framework based on a parametric reduced order model. The improvement in model accuracy is consistent for the model of the plate-to-plate joints, for all considered adhesives. When the parameters identified on the plate-to-plate joint are used to estimate the dynamic response of another geometry, then the accuracy is only enhanced in the most sensitive cases, particularly when the adhesive is significantly more flexible than the substrates. This limit in the scalability of the identification framework is attributed to the geometrical variability of the adhesive joints arising from the bonding process. Through the presented experimental observations, numerical analysis and identification frameworks, this work improves the understanding on the influences of adhesive joints on system dynamics as well as the accuracy of numerical models to predict these. Besides supporting the design of structural adhesive joints in lightweight sustainable mechatronic systems, these accurate models contribute as essential building blocks to the 'digital twins' of real mechatronic systems, adding value at every phase in their lifetime.
Source: Lirias
Language: English
Open Access status: green
OpenAlex ID: W7018557580
Cites:
Cited by:
Related to: