TL;DR
This paper introduces a novel numerical approach for Substructure Shake Table Testing (SSTT) that combines advanced finite element modeling with efficient model-based integration algorithms. This method allows for more realistic simulations of frame structures equipped with various dampers, overcoming limitations of previous simplified models and paving the way for more accurate and computationally efficient seismic performance evaluations.
What Problem Does This Paper Address?
Traditional shake table tests for evaluating seismic dampers often involve testing complete, scaled-down structures. While effective, this can be costly and time-consuming. Substructure Shake Table Testing (SSTT) offers a hybrid approach where one part of the system (e.g., the damper) is tested experimentally, while the other (e.g., the primary structure) is simulated numerically. However, previous numerical models for the analytical substructure in SSTT have often been simplified (e.g., single-degree-of-freedom or shear-type structures) to reduce computational cost. This simplification can lead to unrealistic results, especially for complex structures. This paper addresses this by proposing a more sophisticated finite element (FE) model for the analytical substructure and integrating it with efficient model-based integration algorithms to achieve both accuracy and computational speed.
Methods
The core of this research lies in the development and application of a numerical framework for virtual SSTT. Key methodological aspects include:
- Finite Element Modeling: A four-story steel frame is modeled using stiffness-based beam-column elements with fiber sections. This approach allows for a more detailed representation of the structural behavior, including P-Δ effects, compared to simpler models.
- Model-Based Integration Algorithms (GCR): The study employs a family of dual-explicit model-based integration algorithms, specifically the Generalized CR (GCR) algorithms. These algorithms are highlighted for their unconditional stability and explicit displacement formulation, which are crucial for efficient and accurate simulations of dynamic systems, especially when dealing with nonlinearities.
- Virtual SSTT Simulation: The paper simulates SSTT by treating the frame as the analytical substructure and various dampers (TMD, TLD, PD, PTMD) as experimental substructures. The numerical simulation then mimics the interaction between the tested damper and the simulated frame under seismic excitation.
- Parameter Studies: The research systematically investigates the influence of several critical parameters on SSTT results:
- Auxiliary Mass Ratio: How the mass of the damper relative to the structure affects performance.
- Integration Parameters (K₁, K₂): The impact of specific coefficients within the GCR algorithms on numerical accuracy.
- Time Step (Δt): The trade-off between computational efficiency and accuracy with varying time step sizes.
- Time Delay (τ): The effect of delays inherent in shake table operations on the simulation results.
Datasets
The study utilizes the 1940 El Centro NS ground motion as the seismic input for the simulations. No new benchmark datasets were introduced; instead, the focus was on simulating the behavior of a four-story steel frame with different damper configurations.
Key Insights and Ideas
- TLD Superiority: Among the tested dampers, the Tuned Liquid Damper (TLD) consistently demonstrated the best performance in reducing seismic responses for the specific structure and ground motion studied. Its effectiveness also increased with a higher auxiliary mass ratio.
- Realistic FE Models are Crucial: The use of fiber-sectioned beam-column elements for the analytical substructure provides a more realistic representation of the frame's behavior compared to simplified SDOF or shear models, which is essential for accurate SSTT.
- GCR Algorithms Offer Efficiency and Accuracy: The GCR algorithms proved to be highly effective, providing accurate results even with relatively large time steps. Their explicit nature eliminates the need for iterations, significantly saving computational time compared to implicit methods like the Newmark CAA algorithm.
- Time Step Flexibility: The GCR algorithms showed robustness with larger time steps, particularly for TMDs and TLDs, suggesting their suitability for real-time applications where computational efficiency is paramount. However, for particle-based dampers (PDs and PTMDs), larger time steps led to increased errors.
- Time Delay is a Significant Factor: Time delays, characteristic of experimental shake table setups, negatively impact SSTT results, introducing errors that increase with the delay magnitude. This highlights the need for delay compensation techniques in future experimental SSTT.
Key Results
- Damper Performance: The TLD showed the most significant reduction in seismic responses, with its effectiveness increasing with auxiliary mass ratio. TMDs offered positive but less pronounced effects, while PDs and PTMDs showed mixed results depending on the mass ratio and response metric.
- Integration Parameter Insignificance: For the specific structure and time step used (0.001 s), the choice of integration parameters (K₁, K₂) within the GCR algorithms had a negligible impact on the SSTT results, indicating the robustness of the GCR family for this application.
- Time Step Impact: While larger time steps generally increased errors, the GCR algorithms maintained acceptable accuracy for TMDs and TLDs even with a time step of 0.01 s. PDs and PTMDs, however, showed substantial error increases at larger time steps (e.g., 0.02 s).
- Time Delay Impact: Errors increased with time delay, with PDs exhibiting the largest errors. TLDs and PTMDs were less sensitive to time delays compared to TMDs and PDs.
Limitations and Future Work
- Generalizability: The authors explicitly state that the conclusions drawn are specific to the particular structure, ground motion, and damper parameters studied and may not be generalizable to other scenarios.
- PD/PTMD Sensitivity: Particle-based dampers (PDs and PTMDs) showed higher sensitivity to larger time steps and time delays, suggesting that their accurate simulation in SSTT might require more refined numerical schemes or careful parameter selection.
- Future Work: The paper suggests that for scenarios with larger time steps and stiffer structures, the influence of integration parameters might become significant. Additionally, for significant time delays, delay compensation techniques will be necessary to mitigate adverse effects. Further research could explore the application of these advanced SSTT methods to a wider range of structures and seismic inputs.
Summary updated