Understanding the rheological properties of fresh cement slurries is essential to maintain optimal pumpability, achieve dependable zonal isolation, and preserve long-term well integrity in oil and gas cementing operations and the 3D printing cement and concrete industry. However, accurately and efficiently modeling the rheological behavior of cement slurries remains challenging due to the complex fluid properties of fresh cement slurries, which exhibit non-Newtonian and thixotropic behavior. Traditional numerical solvers typically require mesh generation and intensive computation, making them less practical for data-scarce, high-dimensional problems. In this study, a physics-informed neural network (PINN)-based framework is developed to solve the governing equations of steady-state cement slurry flow in a tilted channel. The slurry is modeled as a non-Newtonian fluid with viscosity dependent on both the shear rate and particle volume fraction. The PINN-based approach incorporates physical laws into the loss function, offering mesh-free solutions with strong generalization ability. The results show that PINNs accurately capture the trend of velocity and volume fraction profiles under varying material and flow parameters. Compared to conventional solvers, the PINN solution offers a more efficient and flexible alternative for modeling complex rheological behavior in data-limited scenarios. These findings demonstrate the potential of PINNs as a robust tool for cement slurry rheological modeling, particularly in scenarios where traditional solvers are impractical. Future work will focus on enhancing model precision through hybrid learning strategies that incorporate labeled data, potentially enabling real-time predictive modeling for field applications.
Composite reinforcements are attracting attention in the reinforced concrete (RC) field for their high corrosion resistance, low thermal conductivity, and low electromagnetic interference behavior. However, compared to metallic reinforcements, composites are less ductile and may lead to brittle failure. Three-point flexural tests provide information on the mechanical behavior of metal- and composite-reinforced concrete beams with distinct crack patterns. The structural conditions and failure mechanisms can be defined based on stress change and crack propagation. This study employs finite element analysis (FEA) to simulate the mechanical responses of composite- and metal-reinforced concrete beans under three-point flexural tests and predict the crack propagation in the beams. Machine learning-based algorithms are trained using FEA data to assess the spatial–temporal conditions of the RC beams. The findings indicate that composite rebars provide better reinforcement than metallic rebars in terms of stress fields (30.27% less stress in composite rebars) and crack propagation (fewer cracks in composite RC beams), with the initiation of shear cracks and maximum von Mises stress in rebars being correlated. The findings highlight the effectiveness of the Random Forest Regression (RFR) algorithm () in assessing RC beam conditions under flexural loads, offering insights for efficient industry applications.
Efficient deployment of distributed fiber optic sensors (DFOS) is critical for capturing complex strain behavior in structural health monitoring, especially in scenarios involving non-uniform deformation such as buckling. This study presents an intelligent design framework that integrates experimental testing, finite element analysis (FEA), and advanced data-driven techniques to improve strain field reconstruction and guide sensor deployment. A buckling test on a steel plate instrumented with DFOS was conducted to generate high-resolution strain data, which was used to validate a corresponding FEA model. Four interpolation methods, including bilinear, cubic, inverse distance weighting, and radial basis function (RBF), were selected as benchmarks across various sensor counts. Results indicate that the RBF method provides the most accurate reconstruction performance, particularly under sparse sensor configurations. Moreover, multiple placement strategies were investigated to enhance sensor layout efficiency, including weighted K -means clustering, high-gradient hotspot selection, and density-based spatial clustering of applications with noise. An integrated strategy combining these methods was developed to achieve balanced spatial coverage while capturing key deformation zones. The findings demonstrate that data-informed deployment strategies can significantly improve monitoring accuracy while reducing sensor redundancy. Finally, several deployment methods for the DFOS were evaluated by six metrics, offering quantitative insights into trade-offs between sensing accuracy and deployment feasibility. While the case study is based on a specific buckling scenario, the proposed methodology is generalizable to other structural conditions. Specifically, for large-scale infrastructure, the framework can be employed to capture strains at critical nodes and to reconstruct full-field strain distributions for digital-twin-based infrastructure monitoring. Overall, this framework provides a flexible and scalable reference for optimizing DFOS installation in high-strain-gradient environments.
This paper presents a novel framework for reconstructing structural buckling deformation from strain distribution. The reconstruction was based on UNet, a convolutional neural network (CNN) that inputs the strain field of the buckling structure to predict the deformation. Two neural network architectures UNet and UNet++ were trained to reconstruct buckling deformation. A new knowledge distillation approach was designed to transfer features of layers from the larger teacher model (UNet++) to the smaller student model (UNet). This approach can improve the accuracy of the student model without increasing model size. To improve knowledge distillation, we replaced uniform weights for feature transformation with adaptive weights. The developed method was validated on a mixed strain-deformation dataset from the finite element analysis and distributed strain measurement, which provided a real-world implication with diverse information. The trained UNet and UNet++ achieved Mean Absolute Percentage Errors (MAPE) of 2.90% and 1.52%, respectively. According to the training results, the best UNet model trained with the proposed knowledge distillation method achieved a MAPE of 1.71%, demonstrating a 41.01% accuracy improvement. A parametric study was conducted to investigate the effect of transferring weights in the proposed framework. In addition, the effect of the framework for deformation reconstruction under varying conditions was evaluated, which indicated a general improvement ability. This study provides a powerful tool to advance the capability of identifying structural buckling by leveraging CNN, smart sensing, numerical modeling, and knowledge distillation, which contributes to health monitoring and anomaly detection of structures.