Explore the vast range of titles available.

Seismic analysis of structural systems often relies on numerical methods, such as the finite element method, for dynamic response prediction. Establishing a high-fidelity numerical model becomes challenging when structural parameters are unknown, and the increasing complexity of numerical models requires prohibitively heavy computation, especially for large engineering problems exhibiting nonlinear hysteretic behaviors. Data-driven black-box modeling methods describe the input–output relationship without relying on precise physical assumptions, but this leads to difficulty in learning interpretable latent spaces of dynamic systems. In recent years, the modeling approach of incorporating physics into machine learning for dynamical systems has emerged as a mainstream trend. In this study, a novel state space model-based recurrent neural network (RNN) embedded with physics is proposed, which integrates gated recurrent unit (GRU) networks and the Runge–Kutta method as its framework, named the RKGRU network. A progressive training strategy with three stages was adopted during the network training process. Three examples are presented to illustrate the capability of the RK4GRU networks to model different structural behaviors, including a linear numerical system with three degrees of freedom (three-DOF), a nonlinear single degree of freedom (SDOF) numerical system with Bouc–Wen hysteresis and an actual overpass with on-site monitoring data. Compared to state-of-the-art physics-enhanced modeling networks, the RKGRU networks exhibit superior performance, requiring fewer training samples and achieving higher prediction accuracy for long trajectories. Additionally, inspired by the transition equations of the state space vectors, the recurrent weight-sharing neural networks (NNs) embed the physical mechanisms of variable equality as hard constraints. The weight-sharing NNs can be flexibly designed based on prior physics knowledge of different structures.

Structural response reconstruction (SRR) and prediction modeling is an area of growing interest in earthquake engineering research. Within the post-earthquake environment, SRR models are useful for estimating seismic demands in uninstrumented buildings using the response measurements from those that are equipped with strong motion sensors. This paper introduces a dual model that uses kriging combined with the extreme gradient boosting (XGBoost) algorithm to reconstruct seismic response demands in buildings during real earthquakes. The model is constructed with data that consists of responses from 207 buildings and 35 earthquakes. Kriging is first used to predict peak ground accelerations (PGAs) at the location of the uninstrumented buildings, using the measured PGAs from instrumented building sites with similar features (e.g., event magnitude, source-to-site distance, and location). The XGBoost algorithm is then used, with PGA as one of several features, to reconstruct the maximum (over the building height) peak (over the response history) story drift ratio and peak floor acceleration in the uninstrumented buildings. The residuals are then examined to assess overall model performance.

To address the stochastic, nonlinear, and strongly coupled characteristics of multivariate long-term structural response in bridge health monitoring, this study proposes the Informer-SEnet prediction model. The model integrates a Squeeze-and-Excitation (SE) channel attention mechanism into the Informer framework, enabling adaptive recalibration of channel importance to suppress redundant information and enhance key structural response features. A sliding-window strategy is used to construct the datasets, and extensive comparative experiments and ablation studies are conducted on one public bridge-monitoring dataset and two long-term monitoring datasets from real bridges. In the best case, the proposed model achieves improvements of up to 54.67% in MAE, 52.39% in RMSE, and 7.73% in R2. Ablation analysis confirms that the SE module substantially strengthens channel-wise feature representation, while the sparse attention and distillation mechanisms are essential for capturing long-range dependencies and improving computational efficiency. Their combined effect yields the optimal predictive performance. Five-fold cross-validation further evaluates the model’s generalization capability. The results show that Informer-SEnet exhibits smaller fluctuations across folds compared with baseline models, demonstrating higher stability and robustness and confirming the reliability of the proposed approach. The improvement in prediction accuracy enables more precise characterization of the structural response evolution under environmental and operational loads, thereby providing a more reliable basis for anomaly detection and early damage warning, and reducing the risk of false alarms and missed detections. The findings offer an efficient and robust deep learning solution to support bridge structural safety assessment and intelligent maintenance decision-making.

Ageing infrastructure leads to the need for a proper assessment and final decisions considering its state. In the case of prestressed concrete structures, knowledge of the residual state of prestressing is the crucial factor. Therefore, reliable diagnostic techniques for determining the residual value of the prestressing force are needed. This information is subsequently used in the process of the quantification of the load-carrying capacity and remaining service life of prestressed concrete structures. The presented paper introduces an evaluation of a monolithic 62-year-old prestressed concrete bridge, which was built in 1959. The assessment was carried out as a result of concerns after exposure of the anchorage area of the bridge, which was executed during the construction of the new system of anti-flood barriers in the town of Banska Bystrica in central Slovakia. Therefore, the diagnostic survey and subsequent determination of the residual prestressing force included the application of the saw-cut method, the structural response method, and the Barkhausen noise technique. Finally, the experimental program supported by numerical analysis provided information about the actual state of prestressing in the bridge. Results of performed analysis suggested that the state of prestressing of the bridge in question does not significantly differ from the expected level of prestressing after 62 years of service. Subsequently, obtained conclusions enabled the determination of the load-carrying capacity for future use in the form of a pedestrian bridge.

Estimating the losses in the immediate aftermath of an earthquake is a key component of seismic response. Seismic rapid-loss estimates provide first responders with a prediction of where and what to prepare for. Improving the precision of quick loss estimates requires an estimate of how a buildings in the affected zone may have reacted to an event. Structural response prediction models are a novel approach to estimating building response from the observed displacement of instrumented buildings. Current SRPMs are built on relatively small databases but offer potential for expansion. There exists no robust building-specific database which could facilitate the construction of these models. As a reaction to this gap, this study applies, abstractly and concretely, the OGC SensorThings data model to building seismograph records. The harmonized records form part of a proposed abstract and concrete Structural Response Prediction Model to make estimates of building-response on other un-instrumented buildings. The utility of a abstracted observation data-model and pipeline is shown, with the potential for unifying existing data-sources. The work shall show that the OGC SensorThings integrates generally well, with some limitations, with the requirements of seismic observation record keeping.

Structural response reconstruction is an important technique for structural health monitoring. However, aiming at the problem of discretization error in the discrete state space model used in structural response reconstruction, a novel structural response reconstruction method based on a continuous-discrete state space model is proposed, which can avoid the discretization error and improve the reconstruction performance. Firstly, the structure is modelled with a continuous-discrete state space model and the process noise covariance matrix is transformed into the process noise gain matrix. Secondly, square root cubature Kalman filter is applied for response reconstruction and the Gaussian nested implicit Runge-Kutta method is used for state recursion. Then the ill-conditioned matrix inverse problem is solved in the posterior estimation with the Tikhonov regularization method. Finally, the proposed method is validated through numerical simulation of a two-dimension truss and response test of an overhanging beam. The results show that the proposed method can achieve high reconstruction accuracy even when the noise level is 30 %, demonstrating its effectiveness and robustness in practical engineering applications.

To understand sandwich panel behaviour under fire, expensive full-scale tests, or potentially more efficient fire-structure simulations can be carried out. However, these simulations have only been demonstrated to work for specific applications, either on the global scale (a fire on a simple panel) or on the small scale (a temperature load on a single screw connection), often loaded by a standard fire curve. In this paper, the quality of simulations for combined situations is investigated, i.e. a furnace fire on a set of panels including details and connections. First two existing tests are introduced, a sandwich panel façade test and a studs bolt test, followed by the presentation of their basic fire-structure simulations. In general, the heat transfer analyses agree well with the tests, whereas the structural response analyses need investigation: For the first test, out-of-plane deflections are overestimated at the beginning of the test. A parameter study indicates that this is most likely due to adhesive decomposition, resulting in face delamination and related instabilities. For the second test, the basic simulation does not show any failure, whereas the test failed by vertical bearing. However, with a two-scale model the ultimate load is estimated, and increasing vertical displacements and the onset of vertical bearing are predicted. It is concluded that future tests should include more simulation-relevant measurements. Also, global-scale models need to include features specific to the structure to be simulated, only known after tests and basic simulations, and connections may be decisive for global-scale behaviour, which can be incorporated by a two-scale model. Finally, the tests exhibited complex behaviour across different scales, and modifications and improvements of the simulations increased their fidelity. Therefore fire-structure simulations should always be verified with tests and compared with basic simulations, and modifications in the simulation models should be anticipated.

During the transient process of load rejection, the hydraulic pressure applied to the pump-turbine and plant concrete changes dramatically and induces high dynamic stress on the spiral case. The preloading spiral case has been widely used in large-scale pumped-storage power stations due to its excellent load-bearing capacity. However, studies on the impact of preloading pressure on the structural response during load rejection are still few in number. In this paper, 3D flow domain and structural models of a prototype pump-turbine are designed to analyze the hydraulic characteristics and flow-induced dynamic behavior of the preloading steel spiral case under different preloading pressures during load rejection. The results show that the asymmetric design of the logarithmic spiral lines ensures an axially symmetric potential flow within the spiral case domain with uniform pressure distribution. Higher preloading pressure provides larger preloading clearance, leading to greater flow-induced deformation and stress, with their maximum values located at the mandoor and the inner edge, respectively. The combined effect of the asymmetrical shape, internal hydraulic pressure and unbalanced hydraulic force leads to an asymmetrical preloading clearance distribution, resulting in an asymmetrical distribution along the axial direction but a symmetrical characteristic near the waistline of the structural response. Stress variations at sections and between sections share similar characteristics during load rejection. It follows the same trend as the hydraulic pressure under lower preloading pressures, while there is a delayed peak of stress due to the delayed contact phenomenon when the preloading pressure reaches the maximum static head. The conclusions provide scientific guidance for optimizing the preloading pressure selection and structural design for the stable operation of units.

In the era of Industry 4.0, researchers in various fields have paid special attention to digital twin technology, which can realize real-time mapping between virtual and physical space. In this paper, a new general framework for response prediction of composite structures based on digital twins is proposed. The tensile testing process of standard samples of carbon fiber-reinforced composites (CFRCs) is used as the twinning object. Moreover, the development of a digital twin and composite structural response prediction based on the generic framework is demonstrated. First, standard CFRC tensile samples are prepared, and relevant raw data are acquired. Subsequently, the microscopic parameters of the standard CFRC tensile samples are obtained by scanning electron microscopy. Geometric measurements are performed to determine the macroscopic parameters, which, together with the material properties of carbon fibers and matrix, are used as the input parameters of a multi-scale virtual physical model (MVPM). The MVPM is used to simulate the actual tensile process using the multi-scale finite element method (FEM). Then, the real-time measurement data from the physical space are transferred to the virtual space through sensors. At the same time, the computationally time-consuming MVPM is downscaled to meet the real-time requirements for the online deployment of the digital twins. In this paper, the backpropagation (BP) neural network model is used to train the input and output parameter data of the MVPM to obtain a reduced-order model (ROM). In addition, to improve the prediction accuracy of the structural response of the digital twin, three model update strategies (MUS) of the ROM are proposed: 1) MUS 1 is based on the ROM, adding the tested sample historical data for the training model update strategy; 2) MUS 2 is based on the ROM 1, adding the measured real-time data of the current sample for training and updating to obtain the ROM 2; 3) MUS 3 is based on the predicted structural response data of ROM 2. Combined with the real-time measured data of the current sample, a higher-order fitting real-time correction is performed to obtain ROM 3. Finally, the tensile process of five CFRC standard samples is demonstrated based on the structural response prediction of the digital twin. The strain response prediction and contour visualization of the whole sample is achieved with limited strain gauge data. By comparison, MUS 2 has higher prediction accuracy than MUS 1 after adding the real-time measured data of the current sample. The prediction errors of MUS 1 and MUS 2 at the later stages of the stretching process are within 10%, with the minimum error of MUS 1 being 15.73% and that of MUS 2 being 3.36%. With the correction of high-order fitting, MUS 3 can achieve a stable prediction error of 20% or less in future moments, and the error can be reduced to less than 5%, reaching a minimum error of 0.44% at the critical tail section near tensile failure.

Besides the design freedom offered by additive manufacturing, another asset lies within its potential to accelerate product development processes by rapid fabrication of functional prototypes. The premise to fully exploit this benefit for lightweight design is the accurate structural response prediction prior to part production. However, the peculiar material behavior, characterized by anisotropy, thickness dependency and scatter, still constitutes a major challenge. Hence, a modeling approach for finite element analysis that accounts for this inhomogeneous behavior is developed by example of laser-sintered short-fiber-reinforced polyamide 12. Orthotropic and thickness-dependent Young’s moduli and Poisson’s ratios were determined via quasi-static tensile tests. Thereof, material models were generated and implemented in a property mapping routine for finite element models. Additionally, a framework for stochastic finite element analysis was set up for the consideration of scatter in material properties. For validation, thin-walled parts on sub-component level were fabricated and tested in quasi-static three-point bending experiments. Elastic parameters showed considerable anisotropy, thickness dependency and scatter. A comparison of the predicted forces with experimentally evaluated reaction forces disclosed substantially improved accuracy when utilizing the novel inhomogeneous approach instead of conventional homogeneous approaches. Furthermore, the variability observed in the structural response of loaded parts could be reproduced by the stochastic simulations.