Semi-plenary lectures

SPL01 Title: 2D and 2.5D Responses of long underground tunnels to moving train loads: a comparison study
Lecturer: Prof. Yeong-Bin Yang, Taiwan University, China
A comparative study is conducted for the responses of soil-tunnel systems to moving train loads using the 2D and 2.5D finite/infinite element approaches, considering the effects of train speed, rail roughness and floating slab. Focus is placed on the wheel-rail interaction forces in the presence of rail roughness. The following are the major findings of this paper: (1) For all the cases studied, the 2D soil response is always higher than the 2.5D response. (2) The 2D result (with plane strain condition) is the limit of the 2.5D analysis with infinite train speed for smooth rails. (3) The 2D frequency response function (FRF) is contributed by frequencies of the whole range, being less sensitive to variation in roughness frequencies, while the 2.5D FRF is affected seriously by the frequencies of rail roughness. (4) With the floating slab tracks, the velocity and acceleration predicted of the soil are largely reduced for frequencies above the threshold using both approaches. But for frequencies below the threshold, the 2D approach shows higher amplified response. In short, the 2D approach saves tremendous computation time, as the system matrices is relatively smaller. But the 2.5D approach is more realistic, since it can account for various factors of the half space, including rail roughness and wave transmission along the tunnel axis.
SPL02 Title: Multi-scale and –physics tsunami disaster simulation for disaster prevention and mitigation
Lecturer: Prof. Mitsuteru Asai, Kyushu University, Japan
On March 11, 2011, the huge tsunami caused by the great east Japan earthquake devastated many infrastructures in pacific coast of north eastern Japan. Particularly, the damage of outflow of bridge girders caused a traffic disorder and these collapse behaviours led to delay of recovery after the disaster. After 2011 tsunami, disaster prevention and mitigation techniques are actively developing in coastal infrastructures and establishing prediction method for tsunami disaster is one of the severe issues toward the next millennium tsunami.
In our study, a multi -level and –physics tsunami simulator based on the Smoothed Particle Hydrodynamics (SPH) Method has been developed. The concept is summarized in Fig.1. The last level (Level-2) can treat multi-physics problem shown in Fig.2, and the simulator can predict bridge wash out accidents during tsunami by using one of the modules of multi-physics simulation. Recently, SPH is widely used in field of fluid and solid dynamics, and a stabilized ISPH has been developing to treat the coupling behaviour among structure-fluid-soil mechanics. Each level can be connected by a proposed virtual wave maker. The performance and efficiency of our developed multi-scale and –physics tsunami disaster simulation tool is given by a couple of numerical examples.
SPL03 Title: Stochastic process of earthquake motion phase and its inherent features
Lecturer: Prof. Tadanobu Sato, Southeast University, China
Decomposing the earthquake motion phase into the linear delay and the fluctuation parts, we investigate the stochastic characteristics of the phase difference in the fluctuation part. The probability density function of the group delay time, which is approximated by the quotient of phase difference with respect to its discrete circular frequency interval, is expressed by a unique stable distribution function for any arbitrary circular frequency intervals. Because the variance of the stable distribution function cannot be defined it is analytically derived that the group delay time, as well as the phase difference, are discontinues functions with respect to the circular frequency. The earthquake motion phase, therefore, is a continuous but un-differentiable function with respect to the circular frequency. We propose a new type of stochastic process being able to represent these stochastic characteristics of phase difference by the use of Lubesgue-Stieltjes type integral equation. In which the Kernel plays a role to realize the self-affine and auto correlation natures of phase difference and the integration function represents the main stochastic characteristics of earthquake motion phase. The proposed stochastic process does not obey to the central limit theorem which is the essential concept in the modern stochastic process. Because of this characteristic the proposed stochastic process is an innovative one. Comparison of probability distribution functions of several numerically simulated phase differences with those of observed earthquake motion phase differences results in the efficiency of the newly proposed stochastic process to simulate a realistic earthquake motion phase.
SPL04 Title: Ratchetting and Ratchetting-fatigue Interaction of Metallic Materials: Experiments and Modeling
Lecturer: Prof. Guozheng Kang, Southwest Jiaotong University, China
A series of uniaxial stress-controlled cyclic tests were first conducted for three kinds of metallic materials with different cyclic softening/hardening features in order to reveal their ratchetting and ratchetting-fatigue interactions, and the dependences of the ratchetting strain and fatigue life on the applied mean stress and stress amplitude, as well as stress ratio were discussed. Then, based on the obtained experimental results including the determined rules of damage evolution, a damage-coupled unified visco-plastic constitutive model was established to describe the whole-life ratchetting of the metallic materials and predict the fatigue life of them by introducing a suitable failure criterion. Finally, the developed constitutive model was verified by comparing the predictions with corresponding experimental results, and the comparison showed that the predictions were in good agreement with the corresponding experimental ones.
SPL05 Title: Simulation of Extreme Events with Finite Element Material Point Method
Lecturer: Prof. Xiong Zhang, Tsinghua University, China
Extreme events, such as hypervelocity impact, shock and explosion, metal forming, slope failure, and liquid sloshing, are highly nonlinear problems. In these kinds of problems, materials are severely distorted, fragmented, melted, or even vaporized. Both purely Lagrangian and Eulerian methods suffer some difficulties so that it is desirable to develop new approaches to better tackle these challenging problems. The material point method (MPM) is a fully Lagrangian particle method which utilizes the advantages of both Lagrangian and Eulerian methods. Although the MPM has demonstrated obvious advantages in tackling extreme deformation problems, its accuracy and efficiency are lower than that of the FEM for small and moderate deformation problems.
This talk will present our different schemes to combine the MPM with FEM to fully take respective advantages of these two methods for different extreme events. These methods have been implemented in our three-dimensional explicit parallel MPM simulation software, MPM3D®, whose applications in hypervelocity impact, perforation, explosion, slope failure, metal cutting, and fluid-structure interaction will also be introduced.
SPL06 Title: A novel cohesive model for simulating delamination propagations in composite laminated structures
Lecturer: Prof. N. Hu, Chongqing University, China
Usually, when using comparatively coarse cohesive elements to simulate the interface damage propagations, such as delamination propagation, there are two frequently occurring phenomena, the first is the numerical instability caused by a well-known elastic snap-back instability, which occurs just after the stress reaches the peak strength of the interface; the second is the error of the peak load in the load-displacement curve. Especially for those interfaces with high strength and high initial stiffness, these two problems become more obvious. In this paper, we propose a new cohesive model to stably simulate delamination propagations in composite laminates under transverse quasi-static or impact loads when using comparatively coarse cohesive elements for reducing the computational cost. In this model, in the front of the original softening zone located at the delamination tip, we set up another pre-softening zone. In this pre-softening zone, with the increase of effective relative displacements, the initial stiffness of cohesive elements is gradually reduced as the interface strength decreases. However, the onset displacement for starting the real softening process is not changed in this model. The critical energy release rate of materials for determining the final displacement of complete decohesion is not changed too. In this cohesive model, the lower limit of the interface strength and stiffness can be theoretically defined according to the mesh size. This cohesive model is implemented in the explicit time integration scheme for evaluating the delamination propagations in composite laminates. A DCB problem is employed to analyze the characteristics of the present cohesive model, and it is found that this new cohesive model can effectively remove the numerical instability and errors in the peak loads. Also, a stress-based criterion is adopted for judging some other in-plane damages, such as matrix cracks, fiber breakage etc. Then, this numerical simulation method is extended into other more complex laminated structures, such as CFRP laminated plates under transverse quasi-static or impact loads. The corresponding experiments are done and the results are used to illustrate the validity of the present method.
SPL07 Title: Doing Topology Optimization Explicitly and Geometrically Based on Moving Morphable Components (MMC) approach-A New Paradigm
Lecturer: Prof. Xu Guo, Dalian University of Technology, Dalian, China
Structural topology optimization, which aims at placing available material within a prescribed design domain appropriately in order to achieve optimized structural performances, has received considerable research attention. From geometry representation point of view, most of the existing topology optimization methods are developed within the pixel or node point-based solution framework. Although remarkable achievements have been made by this approach, there are still some challenging issues need further explorations. Firstly, the pixel-based geometry/topology representations is not quite consistent with that in modern Computer-Aided-Design (CAD) modeling systems. Secondly, since no geometry information is embedded in the pixel-based topology optimization approaches explicitly, it is difficult to give a precise control of the structural feature sizes (i.e., minimum/maximum length scale, minimum curvature), which is usually very important from manufacturing considerations. Finally, since the element-wise material distribution is utilized to represent the structural topology, the computational efforts involved in pixel-based topology optimization approaches are relatively large especially when three-dimensional problems are considered.
In the present lecture, I intend to demonstrate how to do topology optimization in an explicit and geometrical way. To this end, a new computational framework for structural topology optimization based on the concept of moving morphable components is proposed. Unlike in the traditional solution frameworks, where topology optimization is achieved by eliminating unnecessary materials from the design domain or evolving the structural boundaries, optimal structural topology is obtained by optimizing the layout of morphable structural components in the proposed approach. One of the advantages of the proposed approach, which may have great potential in engineering applications, is that it can integrate the size, shape, and topology optimization in CAD modeling systems seamlessly. The approach can combine both the advantages of explicit and implicit geometry descriptions for topology optimization. It also has the great potential to reduce the computational burden associated with topology optimization. Some representative examples are presented to illustrate the effectiveness of the proposed approach.
SPL08 Title: Development of a Citywide Real-time Landslide Warning System in Busan, Korea
Lecturer: Prof. Seung-Rae Lee, Korea Advanced Institute of Science and Technology, Korea
Rainfall-induced landslides have been one of major disasters in Korea where 70% of territory is covered by mountainous regions. As magnitude and frequency of extreme rainfall events has increased due to the global climate change, the number of landslide has significantly enlarged that results in casualties and property damages. In order to mitigate the landslide risk and to provide with an effective tool for public officials to manage the landside disasters, a citywide real-time landslide warning system has been developed by taking into account for situations in Busan, the second largest metropolitan city in Korea, as an application target area. The system provides with warning information based on five-alert levels that are classified as Normal, Attention, Watch, Alert, and Emergency. The warning level is determined by applying several thresholds developed by statistical, and physically-based as well as direct measurement-based approaches using forecasted/observed rainfall data or data obtained from a ground monitoring system. As a first step, the target area at ‘normal’ state is upgraded to an ‘attention’ state if the statistical thresholds are exceeded. Subsequently, the area delineated in the previous step can be updated to the next warning levels (watch and alert) by applying an infiltration-slope stability analysis when the safety factor of slope failure is less than 1.3 and 1.0, respectively. Finally, the ‘emergency’ state is determined by applying a debris-flow mobilization criterion and the subsequent potential debris-flow risk at specific local-scale areas are evaluated by conducting runout analyses. In order to validate the system applicability, landslide historical data and previous rainfall data during rainy seasons from 2009 to 2016 have been used. The system performance demonstrates a good agreement with the past landslide events. The developed system, therefore, will serve as a powerful tool to decision makers for landslide disaster preparedness.
SPL09 Title: Uncertainty Quantification and Propagation in Materials and Structures
Lecturer: Prof. C. F. Li, Swansea University, United Kingdom
The presentation aims to provide a balanced overview for computational techniques in uncertainty quantification and propagation related to materials and structures. These mainly include heterogeneous material modelling, structural reliability analysis, and stochastic finite element methods. High-level summaries and comparison studies will be provided for each of these interrelated topics, to provide an unbiased review of the latest computational technologies available in the research communities. Besides the review, case studies in material, structural, and oil & gas applications will be included. By no means will the presentation be exhaustive, but it is aimed to help answering the following questions relevant to computational engineering and science for safety and environmental problems:
1. What types for uncertainties are encountered and how they can be classified into tangible groups?
2.What computational methods are available to tackle these uncertainty related challenges?
3.What are the advantages and disadvantages for the various existing methods?
4.For solutions to practical problems, what has been done, what needs to be done, and where the most significant impacts are likely to appear.
SPL10 Title: Nonlinear Time-History Analysis of A City: Methodology and Application
Lecturer: Prof. Xinzheng Lu, Tsinghua University, China
Modern cities are becoming integrated systems that consist of a high density of population and buildings. Once they are hit by earthquakes, the damage or collapse of buildings will result in huge economic losses and casualties. The simulation of potential damage of buildings in an entire city region can help to gain a better understanding of the consequences of an earthquake, so as to reduce the human and economic losses of a city. Because many modern cities in China as well as in other countries are lack of sufficient historical earthquake records to establish the empirical models for seismic damage prediction, the nonlinear time-history analysis (THA), which strictly follows the fundamentals of structural dynamics, is used in this study to predict the seismic response of tens of thousands of buildings in a city. For the three key problems (i.e. modelling, computing and visualization) in the nonlinear THA of a city, a high-fidelity modelling method, a high-performance computing approach and a realistic visualization technique are proposed, respectively. With respect to the modelling problem, the multi-degree-of-freedom (MDOF) shear model for multi-story buildings and the MDOF flexural-shear model for tall buildings are proposed. Simultaneously, their parameter determination methods are proposed based on widely accessible urban building information. For the computing problem, a parallel computing method based on GPU (graphics processing unit) for regional building seismic damage simulation is proposed, leading to an improvement of almost 40 times in efficiency. In regard to the visualization problem, a 3D-GIS (geographic information system) data generation method using the existing 3D urban polygonal model and a realistic visualization method are proposed for displaying the simulated urban earthquake disaster scenario. The above proposed methods have been successfully used for the cities with tens of thousands of buildings. Accurate, efficient, and realistic simulations for regional building seismic damage have been achieved with these methods. In addition to the building seismic damage, the secondary disasters such as fire following earthquake and falling debris due to earthquake can be simulated based on the THA results of buildings. The outcome of this study provides important technology for regional building seismic damage simulation, which can improve the ability of urban disaster prevention and mitigation.