Computers & Structures最新文献

An efficient method to address the three-dimensional modeling of the visco-elasto-plastic material behavior, specifically of bituminous conglomerates used in asphalt concrete production, is proposed. The method resorts to one of the most recent formulations for asphalt creep modeling, represented by the modified Huet-Sayegh fractional rheological model. The Grünwald-Letnikov representation of the fractional operator is adopted to treat the operator numerically in an efficient manner. Further, a coupling scheme between the creep model and elasto-plasticity is proposed by adopting the additive decomposition of the total strain tensor. This enables the numerical assessment of the mechanical behavior for bituminous materials under short- to long-term loading. In this context, both constant strain rate tests, and creep recovery tests are numerically simulated.

Numerical analyses are conducted at the meso-scale with the aim to evaluate the development of inelastic strains in the binder during creep, due to the local interaction between the different material components.

Life-cycle performance assessment of bridges is crucial for decisions pertaining to functionality, maintenance, and rehabilitation while accounting for inherent epistemic and aleatoric uncertainties stemming from noise or structural degradation. Since fatigue from repeated cyclic loads is a prominent source of performance degradation in bridges, a continuous and efficient method for structural monitoring is necessary. In fatigue assessment, engineers rely on strain response, which can be challenging to collect due to the labor-intensive and costly deployment of strain gauges that are not conveniently reusable. This paper proposes an indirect sensing approach that converts acceleration signals to strain signals, enabling a convenient and robust paradigm for a continuous, and accurate bridge fatigue assessment. A combination of convolutional neural networks and transformers are used in this work for estimating strain signals from acceleration measurements. The efficacy of the proposed framework is demonstrated through data collected from the Gene Hartzell Memorial Bridge in Pennsylvania, USA. Furthermore, physical insights have been drawn from the results that reinforce the rationale behind the proposed artificial neural network architecture. This novel framework for indirect sensing can be readily employed for strain estimation from acceleration measurements of the bridges, upon adequate training, which will contribute to bridge condition and life-cycle assessment.

Topology optimization (TO) is a pivotal technique for generative design of high-performance structures. Practical designs often face complex boundary conditions and require non-gradient optimizers for solving TO with thousands of design variables or more. This paper presents the Adaptive Deep Learning (ADL) which supports both gradient-based topology optimization (GTO) and non-gradient-based topology optimization (NGTO). The ADL roots in convolutional neural network to link material layouts with structural compliance. A small number of training data is generated dynamically based on the ADL’s prediction of the optimum. The ADL explores the region of interest in a probabilistic setup and evolves with increased data. The presented ADL has been evaluated on four cases including beam design, heat dissipation structure design, three-dimensional machine tool column design and heat transfer enhancement optimization. The ADL achieved 0.04 % to 4.08 % increasement of structural performance compared to GTO algorithm, and 0.88 % to 81.98 % increasement compared to NGTO algorithms.

The computational modeling of fractures in solids using damage mechanics faces challenge when dealing with complex crack topologies. One effective approach to address this challenge is by reformulating damage mechanics within a variational framework. In this paper, we present a novel variational damage model that incorporates a threshold value to prevent damage initiation at low energy levels. The proposed model defines fracture energy density ($\tilde{\varphi }$) and damage field (s) based on the energy density (ϕ), crack energy release rate ($G_c$), and crack length scale (ℓ). Specifically, if $\varphi \le \frac{G_c}{2\ell }$, then $\tilde{\varphi }=\varphi$ and $s=0$; otherwise, $\tilde{\varphi }=-\frac{G_c^2}{4{\ell }^2}\frac{1}{\varphi }+\frac{G_c}{\ell }$ and $s=1-\frac{G_c}{2\ell }\frac{1}{\varphi }$. Furthermore, we extend the model with a threshold value to a higher-order version. Utilizing this functional, we derive the governing equation for fractures that evolve automatically with ease. The formulation can be seamlessly integrated into conventional finite element methods for elastic solids with minimal modifications. The proposed formulation offers sharper crack interfaces compared to phase field methods using the same mesh density. We demonstrate the capabilities of our approach through representative numerical examples in both 2D and 3D, including static fracture problems, cohesive fractures, and dynamic fractures. The open-source code is available on GitHub via the link https://github.com/hl-ren/vdm.

Tensegrity structures obtained from the same connectivity patterns are said to belong to families. The Octahedron and X-Octahedron families are examples of these. In the literature, little attention has been paid to how the final geometries of the equilibrium forms of the members of both families are obtained. A compact formulation for controlling the equilibrium shapes of members of the Octahedron and X-Octahedron families is proposed in this article allowing the designer to get any geometry for the super-stable members of both families. Controlling the stability of folded forms is achieved by using the shape of the structure, and a detailed explanation of the formulation is provided here, as well as several examples that clarify the formulation. The geometrical control of the equilibrium shape is fundamental when applying it to tensegrity structures in an engineering context.

Existing studies on active tensegrity structure optimum design only focus on sizing and/or shape optimization i.e., the structural element topology does not change during the design process, which vastly limits the design space and further improvement of mass-saving performance. This study investigates the optimum design of active tensegrity structures through topology optimization, which has never been done to the best of the authors’ knowledge. Structural member topology and actuator layout are considered as binary design variables and their coupling relation is handled by auxiliary constraints. Member cross-sectional areas are treated as discrete design variables considering practical availability. Member prestress, actuator length changes, and other necessary auxiliary parameters are defined as continuous variables and designed simultaneously. Equilibrium conditions, member yielding, cable slackness, strut buckling, and the limitations on the nodal displacements as well as other design requirements are formulated as constraints. Linearization algorithm is proposed to transform the bilinear expressions in the objective and constraint functions to allow the problem to be solved to global optimum. Typical benchmark examples indicate that the topology-optimized active designs obtained through the proposed approach can further decrease the material consumption compared with sizing-optimized active tensegrity designs hence leading to more lightweight structures.

Traditional surrogate models always face the challenge of low accuracy when dealing with high-dimensional problems in structural optimization, this study aims to overcome this problem and proposes a vertex-based graph neural network (GNN) classification model. In contrast to conventional machine learning models that treat design variables as independent inputs, the proposed model develops a vertex-based graph representation to transform structural topological features and critical physical information into the graph data. According to a message passing mechanism based on the graph convolutional, it can extract the correlations among design variables and enhance its capability in handling high-dimensional structural optimization problems. Three truss examples, including a 10-bar with 10 variables, a 600-bar with 25 variables, and a 942-bar with 59 variables, are utilized to investigate the performance of the proposed surrogate model. The results demonstrate that the GNN-based surrogate model outperforms traditional machine learning approaches, particularly in the two high-dimensional problems, showcasing its superior ability to capture complex variable correlations and handle high-dimensional structural optimization tasks. Moreover, the proposed method significantly reduces the computational expenses by over 60% compared to conventional metaheuristic algorithms, while yielding optimal designs with comparable quality. These results demonstrate the efficiency and effectiveness of the GNN-based surrogate model in tackling complex, high-dimensional structural optimization problems.

Shape sensing, i.e. the reconstruction of the displacement field of a structure from discrete strain measurements, is becoming crucial for the development of a modern Structural Health Monitoring framework. Nevertheless, an obstacle to the affirmation of shape sensing as an efficient monitoring system for existing structures is represented by its requirement for a significant amount of sensors. Two shape sensing methods have proven to exhibit complementary characteristics in terms of accuracy and required sensors that make them suitable for different applications, the inverse Finite Element Method (iFEM) and the Modal Method (MM). In this work, the formulations of these two methods are coupled to obtain an accurate shape sensing approach that only requires a few strain sensors. In the proposed procedure, the MM is used to virtually expand the strains coming from a reduced number of strain measurement locations. The expanded set of strains is then used to perform the shape sensing with the iFEM. The proposed approach is numerically and experimentally tested on the displacement reconstruction of composite stiffened structures. The results of these analyses show that the formulation is able to strongly reduce the number of required sensors for the iFEM and achieve an extremely accurate displacement reconstruction.

In this work, a novel GPU-accelerated heterogeneous method for the automated multilevel substructuring method(HAMLS) is presented for dealing large finite element models in structural dynamics. Different parallel modes based on node, subtree, and eigenpair have been developed in the solution steps of AMLS to achieve a heterogeneous strategy. First, a new data management method is designed during the model transformation phase to eliminate the determinacy race in the parallel strategy of the separator tree. Considering the distribution characteristics of the nodes in the separator tree and the dependence of node tasks, a load balancing heterogeneous parallel strategy is designed to take full advantage of hosts and devices. By developing an adaptive batch processing program for solving eigenvectors during the back transformation phase, the overheads of launching kernels, as well as the GPU memory requirements, can be reduced by several orders of magnitude. Several numerical examples have been employed to validate the efficiency and practicality of the novel GPU-accelerated heterogeneous strategy. The results demonstrate that the computational efficiency of the novel strategy using one GPU can increase to 3.0x that of the original parallel AMLS method when 16 CPU threads are used.

Concrete with new features such as high strength and a high tension–compression ratio has been developed to enhance building safety and the defense structure capability, which also poses a challenge to classical constitutive models such as the Holmquist-Johnson-Cook (HJC) model.

This study proposes a flexible constitutive model that is suitable for concrete-like materials with varying strength and tension–compression ratios. Known as the three-invariant model, it features the explicit introduction of two mechanical characteristic parameters: the tension–compression ratio and the Lode angle. By strictly passing through (or closely approximating) six benchmark stress state points, the model effectively captures tension–compression anisotropy and yield behaviors across the entire range of hydrostatic pressure. To further extend the static model to dynamic conditions, a unified S-type strain rate equation is developed. This equation accounts for dynamic tension–compression anisotropy arising from the material’s intrinsic properties by considering the influence of hydrostatic pressure on strain rate effects. Experimental data from various rock and concrete specimens subjected to true triaxial stress states are compared with calculated data. The results confirm that the proposed model accurately reflects the yield strength and improves the predicted accuracy of structural responses under complex stress states.