Special Mechanical and Civil Engineering Seminar

In recent years radical urban development is observed while at the same time a large part of existing infrastructure is reaching the end of its design lifecycle. Developed societies are therefore now met with the urgent need for sustainable and resilient systems. This task is by far not a straightforward one as all structural systems are inherently characterized by uncertainty due to a number of factors including, lack of knowledge of the system, limited information on input loads, potential nonlinear behavior and ageing effects. Monitoring of infrastructure can be a valuable source of information for evaluating structural integrity, durability and reliability throughout its lifecycle, ensuring optimal maintenance planning, safe operation, early warning and subsequent localization of damage in face of extreme events.

 In this presentation, computational approaches to tackle these issues will be addressed. Developments in the state-of-the-art identification algorithms for behavior that lies beyond the usual assumption of the linear range will be presented, including the Unscented Kalman Filter and Particle Filter methods. Next, it will be demonstrated how reduced order models (surrogates or "metamodels") can efficiently simulate nonlinear systems, while lowering computational time and quantifying uncertainty especially for loads that are random in nature, such as earthquake or strong wind events. Finally, in addressing the issue of rapid simulation whilst maintaining accuracy for highly complex systems, such as the widely used composite materials, a multiscale methodology will be described. This method employs a mapping for embedding the required micro-scale information of the various material layers into a coarser grid of elements. A generic hysteretic FE formulation will be demonstrated leading to minimal re-evaluation of the system in the case of nonlinear dynamic analysis. Once again, by extending beyond linearity, a robust computational method is derived pertinent to a wide range of fields that explore novel materials, spanning from construction and material sciences to bio-engineering.

 In a world that is largely dependent on the power of information, knowledge in the form of data is continuously stocked from various domains such as transportation, infrastructure and even social networks. This talk presents building blocks of an integrated computational approach directed towards successfully interpreting the large arrays of collected data for predicting response to future events such as earthquakes and efficiently managing extended networks of infrastructure, and no longer isolated units. This framework can then form part of a grand scheme of "smart" information grids where seemingly unconnected information such as structural integrity and human behavior are fused to design reaction scenaria in face of extreme events (earthquakes, typhoons).