Prediction of aerodynamic forces on bridge decks using artificial neural networks
The aeroelastic behaviour of slender structures, such as cable-stayed bridges with long spans, must be investigated as part of their design, as they can develop significant vibrations when subjected to atmospheric wind flow. The trend towards increasing the flexibility and reducing the mass of structures makes such problems more pronounced and the analysis more difficult, thereby increasing the need for accurate, robust and efficient predictive models.
Further information
Partitioned vortex methods for fluid-structure interaction in thin-walled flexible structures
The flow-solving algorithm developed at the department, based on the vortex particle method (VPM), has been extended for 2D and pseudo-3D FSI simulation of thin-walled systems. A structural solution algorithm based on a corotational finite element formulation is coupled with 2D-VPM to account for geometrically non-linear effects. This model is being further developed to carry out an aero-electromechanically coupled simulation for analysing the performance of various cantilever energy harvesters. In addition, the existing pseudo-3D multi-layer model is being extended to enable FSI simulation of thin-walled surface systems such as cantilevered roofs and cooling towers.
Development and evaluation of models in bridge aerodynamics
As part of this project, several key research areas are being considered, including: (i) a categorical modelling approach for the evaluation of aerodynamic models; (ii) a turbulent pseudo-3D vortex method for buffeting and flutter analyses of bridges; (iii) a synergistic comparison framework for CFD and semi-analytical models; (iv) time-series comparison metrics tailored to identify and quantify specific characteristics of the time-dependent aeroelastic response and forces; (v) a method for determining the complex form of aerodynamic admittance through the simulation of deterministic gusts.
Experimental methods for the validation of non-linear aeroelastic bridge models
In order to meet the growing need to account for non-linearity in slender bridge structures subjected to wind flows, new, modern models and methods are constantly being developed. However, these must be validated. The best way to achieve this is through the implementation of experimental procedures that can be used to test the proposed approaches.
Numerical models for the behaviour of fluid dampers in structural systems subjected to wind loads
Flexible, slender and tall structures, such as high-rise buildings, bridges or chimneys, are highly sensitive to dynamic wind loads, which can lead to damage or fatigue. For many of these structures, the provision of additional damping is the only reliable and practical method of controlling their dynamic behaviour. Owing to their low maintenance requirements, fluid dampers represent a field of research with great potential for practical application. This research aims to investigate the numerical coupling of fluid reservoirs in structures subjected to wind loads.
Non-linear interaction between motion-induced and gust-induced forces in long-span bridges
As span lengths increase and deck shapes become more complex, aerodynamic non-linearity becomes a critical issue in the design of long-span bridges. Conventional aerodynamic bridge models treat the forces induced by free-flow turbulence and deck motion as independent. These models do not take into account the two crucial aspects of interaction that are closely linked to aerodynamic non-linear behaviour: the effect of large-scale sinusoidal vertical gusts on the shear layer of a moving body, and the non-linear dependence of aerodynamic forces on the effective angle of attack. It is essential to investigate the problem of wind-induced vibrations in a turbulent environment in a synergistic manner in order to capture these effects.
Probabilistic quantification of the residual load-bearing capacity of infrastructure structures under extreme loads
Structures may be subjected to extreme loads throughout their service life, ranging from earthquakes and explosions to heavy impact loads caused by land, sea and air vehicles. The resulting requirements for the design and verification of structures generally exceed those arising from conventional use. In this context, consideration must be given to possible mechanisms that may positively or negatively influence the structural behaviour during an extreme load scenario.
Structural simulation and limit state analysis using mathematical optimisation strategies
With the aid of mathematical optimisation strategies, a wide variety of problems can be solved in the context of physically and/or geometrically non-linear structural analysis. In addition to the general non-linear simulation of structures subjected to load and constraint, which is also supported by a wide range of commercial and non-commercial finite element implementations, the use of mathematical optimisation enables the numerical solution of optimisation problems with a wide variety of objective functions.
Non-linear dynamic analysis of bridge piers subjected to ship impact
The ship’s impact behaviour is examined in detail using complex models, taking into account key factors such as the ship’s mass, impact velocity, impact angle, pier shape and pier size. To reduce computational costs, a non-linear mass-spring model (MSM) was developed as a substitute for the complex finite-element ship model.
Modern technologies and data analysis for assessing the condition of structures
Condition assessment is a key component in the safety and maintenance of structures. Modern technologies, such as image capture using unmanned aerial vehicles and photogrammetric reconstructions, can support the necessary inspection tasks. Automated analysis and processing of the captured data enables an effective assessment of the condition of the structures under investigation. This research focus centres on the use of geometric features to calibrate computational models, link structural information and identify mechanical properties.
Further information
Vibration-based monitoring for structural assessment
Within the framework of Structural Health Monitoring (SHM), vibration-based methods can be used to determine the current condition of a structure. This involves identifying changes in the structural condition through changes in dynamic properties. Measured structural responses can be used to calibrate numerical models and to determine the structural condition using system identification. Research in this field aims to identify structural properties and damage, taking into account the characteristics of the sensors used and the data acquisition system.
Further information
Drone-based inspection of structures
For infrastructure structures, as well as historic buildings, ensuring structural stability and fitness for purpose is the top priority. Structures are subject to various ageing and deterioration processes that can only be predicted with relative inaccuracy. Regular monitoring and assessment of the structure’s condition is therefore essential. As an alternative to traditional visual inspection, the use of small unmanned aircraft systems (UAS) has come to the fore. These systems, equipped with high-quality cameras, are suitable for a wide range of inspection tasks. Various research projects are investigating potential applications of UAS in the inspection of structures as part of a digital maintenance strategy.
Probabilistic physics-based models for machine learning to monitor structural condition
Appropriate structural condition monitoring systems support effective asset management strategies and reduce the life-cycle costs of structures. Nowadays, a wealth of heterogeneous data is generally available, as the number of embedded sensors monitoring our structures is rising exponentially.This project proposes physically grounded machine learning models that bridge the gap between purely traditional methods and artificial intelligence approaches. They offer an alternative for processing large volumes of data whilst retaining the technical knowledge represented by typical mathematical models.