shaking table effects
Shaking Table Effects: Principles, Applications, and Practical Considerations
Shaking tables are mechanical devices used to simulate ground motion, primarily for geotechnical earthquake engineering, structural testing, and seismic performance evaluation. This article provides a direct overview of shaking table effects—how they generate controlled dynamic excitation, how scale models respond to such input, and what key parameters (frequency, amplitude, acceleration, and boundary conditions) influence test outcomes. We then compare shaking table testing with alternative dynamic testing methods, present a real-world case study from the University of California San Diego’s outdoor shake table, and conclude with frequently asked questions to clarify common misconceptions.
1. What a Shaking Table Does (and What It Cannot Do)
A shaking table applies a programmed time-history of displacement, velocity, or acceleration to a specimen mounted on its platform. The “effect” is twofold: first, the table imparts inertial forces into the test object; second, the object’s dynamic response (resonance, damping, nonlinear deformation, or failure) is observed under controlled, repeatable conditions. The most critical effect is that the table’s motion is not identical to free-field ground motion—it is filtered by the table’s own stiffness, actuator capacity, and control algorithm. For example, a uniaxial table can only reproduce one component of motion, while a 6-degree-of-freedom (6-DOF) table can simulate coupled horizontal, vertical, and rocking motions. The table’s payload mass and overturning moment capacity also limit the maximum acceleration that can be achieved without spurious table-specimen interaction.
2. Key Parameters That Define Shaking Table Effects
| Parameter | Definition | Typical Range (Civil Eng. Tests) | Effect on Specimen |
|---|---|---|---|
| Peak Ground Acceleration (PGA) | Maximum acceleration commanded at table surface | 0.1g – 1.5g (large tables) | Higher PGA increases inertial forces; may trigger nonlinear response or collapse |
| Frequency Content | Dominant frequencies in the input motion | 0.5 – 20 Hz (earthquake-like) | If input frequency matches specimen’s fundamental frequency, resonance amplifies displacement |
| Duration | Total shaking time, including strong shaking and decay | 10 – 60 seconds | Longer duration increases cumulative damage and pore pressure buildup in soils |
| Boundary Conditions | How specimen is fixed to table (bolted, free-standing, or with flexible base) | Fixed base vs. soil container | Fixed base overestimates structural stiffness; flexible containers (e.g., laminar shear box) simulate free-field soil behavior |
| Control Mode | Displacement, acceleration, or force feedback | Acceleration control is standard | Poor control can cause “table-specimen interaction” where the specimen’s motion alters the table’s output |
Table 1. Comparison of shaking table testing vs. other dynamic testing methods
| Method | Input Type | Scale | Best For | Limitations |
|---|---|---|---|---|
| Shaking Table | Real-time ground motion | 1:1 to 1:100 | Full-system interaction, soil-structure interaction, liquefaction | Size and payload limits; cost per test |
| Pseudodynamic (PsD) Test | Quasi-static displacement with numerical feedback | Large-scale substructures | Rate-independent behavior (e.g., steel frames) | Cannot capture rate-dependent effects (e.g., damping in rubber bearings) |
| Centrifuge Test | In-flight shaking on small model | 1:100 to 1:200 | Soil liquefaction, slope stability | Scaling laws for stiffness and damping are tricky; no full-scale validation |
| Cyclic Triaxial Test | Uniform sinusoidal loading on soil sample | Small sample (50–100 mm) | Soil stiffness and damping degradation | No multi-axial stress rotation; no structural interaction |
3. Real-World Case Study: UCSD Outdoor Shake Table (LHPOST)
The Large High-Performance Outdoor Shake Table (LHPOST) at the University of California, San Diego, is the largest in the United States. It is a single-axis (horizontal) table with a 7.6 m × 12.2 m platform, capable of shaking specimens up to 2,000 metric tons with a maximum displacement of ±0.75 m and a velocity of 1.6 m/s.
Case: 2011 Full-Scale Reinforced Concrete Bridge Column Test
A 7.6 m tall, 1.2 m diameter circular bridge column was mounted on the table. The input motion was the 1994 Northridge earthquake record (Rinaldi Station), scaled to 0.8g PGA. The table successfully reproduced the near-fault pulse, and the column exhibited flexural-shear failure after 3 cycles of large displacement. Key observation: the table’s control system compensated for the column’s nonlinear stiffness changes, but the measured base shear was 12% lower than the analytical prediction due to table-specimen interaction—a well-documented effect that engineers must account for when interpreting results.
This test demonstrated that shaking tables can reproduce realistic failure modes, but also highlighted that the table’s own compliance (the platform bends slightly under heavy payloads) introduces additional flexibility that is not present in a fixed-base numerical model.
4. Practical Effects to Watch For in Shaking Table Tests
- Table-Specimen Interaction (TSI): When the specimen is heavy and stiff, its dynamic response can feed back into the table’s control loop, causing the actual table acceleration to deviate from the command. This is more pronounced in uniaxial tables with limited force capacity.
- Boundary Effects in Soil Tests: If soil is placed in a rigid box on the table, waves reflect off the box walls, creating artificial stress states. Laminar shear boxes (stacked rings) reduce this effect but do not eliminate it.
- Scaling Distortion: For reduced-scale models, gravitational acceleration cannot be scaled. This leads to incorrect stress-dependent behavior (e.g., soil friction angle changes with confining pressure). Centrifuge shaking tables solve this by spinning the model to increase effective gravity, but they are limited in payload and frequency range.
- Acceleration vs. Displacement Control: Most earthquake records are acceleration time-histories. However, at low frequencies (<1 Hz), displacement control is more stable. Switching control modes mid-test can introduce spurious transients.
5. Frequently Asked Questions (FAQ)
Q1: Can a shaking table reproduce a real earthquake exactly?
No. A table can only reproduce the recorded motion at a specific location, and even then, it is filtered by the table’s hydraulic system and platform stiffness. High-frequency content (above ~20 Hz) is usually lost, and near-fault velocity pulses are difficult to reproduce on small tables due to displacement limits.
Q2: Why do we use scale models if full-scale tests are possible?
Full-scale tests are extremely expensive and limited to simple structures (e.g., single columns or small buildings). Scale models allow parametric studies (changing mass, stiffness, damping) at lower cost. However, scaling laws require careful attention: for example, if length is scaled by 1/10, time must be scaled by 1/√10 to maintain dynamic similarity, which is not always practical.
Q3: What is the difference between a “uniaxial” and a “6-DOF” shaking table?
Uniaxial tables move in one horizontal direction only. 6-DOF tables can move in three translations (x, y, z) and three rotations (roll, pitch, yaw). The latter is necessary for testing asymmetric structures or simulating multi-directional ground motion, but they are far more complex to control and calibrate.
Q4: How do engineers avoid table-specimen interaction?
They use robust control algorithms (e.g., acceleration feedback with adaptive feedforward compensation), limit the specimen mass to less than 10% of the table’s payload capacity, and perform a “bare table” calibration run before each test. In some cases, they use a “reaction wall” to absorb the specimen’s reaction forces, but this is not possible for all test setups..jpg)
Q5: Is a shaking table test always more reliable than a numerical simulation?
Not necessarily. Shaking tables have physical limitations (boundary effects, scaling distortion, actuator noise). Numerical models can simulate any input motion and any boundary condition, but they rely on constitutive models that may be inaccurate for complex materials (e.g., liquefiable soils). The best practice is to use shaking table tests to validate numerical models, then use the validated model for parametric studies.
Conclusion
Shaking table effects are not simply “applying an earthquake to a specimen.” They involve a complex interplay between the table’s mechanical limits, the specimen’s dynamic properties, and the control system’s ability to track a desired input. Understanding these effects is essential for interpreting test data correctly and for designing experiments that yield meaningful, transferable results. The UCSD case study illustrates both the power and the pitfalls of this testing method. For any engineer planning a shaking table test, a pre-test calibration, a careful selection of boundary conditions, and a clear acknowledgment of the table’s limitations are non-negotiable steps.