# Hyung-Suk Shin Pedro University of Washington Steven L. Kramer

## Presentation on theme: "Hyung-Suk Shin Pedro University of Washington Steven L. Kramer"— Presentation transcript:

Numerical Seismic Soil-Pile-Structure Interaction Analysis Using OpenSees
Hyung-Suk Shin Pedro University of Washington Steven L. Kramer Mahadevan Ilankatharan Bruce L. UCdavis

Outline Centrifuge test 1 Centrifuge test 2 Conclusion
Bent & Bridge Model Experimental and numerical Results Centrifuge test 2 Oriented Bent Model Conclusion First, I will briefly describe our project. Then I will explain how to model the soil-pile-structure interaction of a bridge bent. Then I will move to the results from experimental and numerical simulation. Conclusion.

Centrifuge Test 1 1 / 52 scale
Again, the test was performed in 1/52 scale. Notice the geometry of soil is not horizontal.

Centrifuge Test 1 Shaking direction
A picture of the centrifuge test set-up.

Centrifuge Test 1 Shaking direction
This is another shot showing a plane view.

Soil - Nevada sand Dry soil Dr = 80 % Unit weight = 16.29 kN/m3
ρd = 1.66 Mg/m3 Soil, we used Nevada san. It is all dry and the relative density is 80%.

Pile We used aluminum tube for pile.
In order to mimic the same pile cross section properties as those in shaking table, it was necessary to cover the aluminum pile with plastic wrap.

Input motion Northridge 1994
The calculated OpenSees motion at the equivalent fix depth was slightly different than the measured centrifuge motion. Because of that, we had to slightly modified the base motion. This is motion we scaled to 4 different horizontal acceleration. Northridge 1994

Interface spring model in OpenSees
Interaction model p-y spring t-z spring q-z spring Soil model in OpenSees Pressure dependent multi yield elasto-plastic material model (by Elgamal and Yang) Pile model in OpenSees Nonlinear fiber beam column element Interface spring model in OpenSees 1-D nonlinear springs (by Boulanger) y p pult (Reese 1974) y50 (API 1993) When we have an earthquake, the shear wave is propagated and a free field motion is generated. The free field motion gives a kinematic force to the pile and superstructure has an inertial force. Soil-pile-structure interaction occurs near the pile. We call it “Near-field”. The zone which is not affected by the interaction is called “Far-field”. So in our pile modeling, the freefield is modeled with a soil column. The near-field is modeled with interface springs as well-known “p-y”, “t-z”, and “q-z” springs. For soil, we used “Pressure dependent multi yield elasto-plastic material model. We use nonlinear 1-D spring models defined in terms of “pult” and “y50”.

Bent Model & Simplification
We model a two-pile bent using a single pile constrained at pile top and connected it to a single soil column through interface springs. In the second simplification, we used two soil column and two sets of interface springs with original shape of structure. In this simplification, we can capture “rocking motion of structure”. This is the third simplification using a continuous soil.

Experimental & Numerical Results (Single bridge bent)
We also checked the soil motion at around 2.5 times of pile diameter. This shows the comparison of acceleration of time history. These are spectra.

Experimental & Numerical Results (Single bridge bent)
Let me show you some results. We finished all the simulation before the test. However, as I mentioned the centrifuge base motions were slightly modified, the simulation should be repeated. We changed only input motion in the repeated simulation. So we called it as “Class-A prime prediction”. All the results I show in this presentation is based on the class-A-prime prediction. Let me first show the result of a single bent. We measured horizontal acceleration here at bent top in this direction and compared it with theses simulation results. The blue one represents CFG results and the red one represents OpenSees results. OpenSees can capture the response well. The response spectrum is shown.

Experimental & Numerical Results (Single bridge bent)
From double integration of the vertical accelerations on top of bent, we got the vertical rotation angle of the bent. OpenSees can capture well the rocking motion. 2 soil column model

Experimental & Numerical Results (Single bridge bent)
Maximum bending moment Ground surface Now, let me show the maximum bending moment. Blue dots are from centrifuge test. Red solid lines are from OpenSees simulation. When we used the 1-soil column model in numerical simulation, OpenSees gave a slightly higher value. The 2-soil column model give the better results. When we used the continuous soil model, we got a weird and asymmetric results. I guess this is from the additional local soil deformation around pile due to interaction force. (a) 1 soil (b) 2 soil

Two-Span Bridge Model & Simplification
equalDOF Bridge deck This one is after bridge deck connection. For bridge model simplification, we used the first simplification model. We applied the same base motion. Soil profile is different. Each bent structure is connected with an adjacent structure through bridge deck.

Experimental & Numerical Results (Two-Span Bridge)
Let me the result of bridge by focusing on the shortest bent here. The horizontal acceleration at the bent top in this direction is measured and compared. OpenSees gave a good results, again.

Experimental & Numerical Results (Two-Span Bridge)
This slide shows the maximum bending moment of three bent for 0.25g and 0.6g event. Again, the blue dot is for CFG and the red one is for OpenSees. From left, short bent, long bent, and medium bent. This one is for maximum bending moment. We also looked at the time history at around 2.5 times of pile diameter. Overall, OpenSees can capture well the maximum bending moment as well as the time history. 0.40g (a) short bent (b) long bent (c) medium bent

Experimental & Numerical Results (Two-Span Bridge)
Increasing intensity I summarized here about the maximum bending moment and its corresponding depth of three different bents over four different motions. Left plot is for CFG and the right one is for OpenSees. The left data is for the short bent, and the middle one is for the longest bent. Let me quickly show you how I get the maximum bending moment and its depth. For example, we got the maximum value here and its depth like this. The shortest one has higher values than other longer one. And the maximum bending moment occurs at a deeper place in short bent. It is important to know the maximum bending moment increases with event intensity and it is different at each bent. A similar pattern is observed with maximum bending moment depth. It is not exactly 2.5m for all bents. short bent long bent medium bent short bent long bent medium bent (a) centrifuge (b) OpenSees

Centrifuge Test 2 1 / 52 scale
Again, the test was performed in 1/52 scale. Notice the geometry of soil is not horizontal.

Centrifuge Test 2 Again, the test was performed in 1/52 scale.
Notice the geometry of soil is not horizontal.

Oriented Bent Model & Simplification
Decompose base motion axbase Calculate freefield motions based on decomposed base motions Apply absolute freefield motions to a series of p-y springs ay ax We model a two-pile bent using a single pile constrained at pile top and connected it to a single soil column through interface springs. In the second simplification, we used two soil column and two sets of interface springs with original shape of structure. In this simplification, we can capture “rocking motion of structure”. This is the third simplification using a continuous soil.

3D Bent Structure & Soil axbase* sin() axbase* cos()

Experimental & Numerical Results (free field motion)

Experimental & Numerical Results (Oriented Bents – 0o, 30o, 60o, 90o)
ax ay1 60o 90o y x z

Experimental & Numerical Results (Oriented Bents – 0o, 30o, 60o, 90o)
ax y 60o 90o x z

Experimental & Numerical Results (Oriented Bents – 0o, 30o, 60o, 90o)
ay1 60o y 90o x z

Experimental & Numerical Results (Oriented Bents – 0o, 30o, 60o, 90o)
-z zz xx yy y x

Experimental & Numerical Results (Oriented Bents – 0o, 30o, 60o, 90o)

Future work Increasing diameter Increasing stiffness

Conclusion Traditional p-y curves work well for the dynamic soil-pile-structure interaction analysis of dry sand. Vertical rocking motion of the bridge bent should be included in the analysis to capture the correct structural response. Simplified coupled bridge models using one soil column for each bent provides a good estimation. The maximum bending moment of the individual bent in a full bridge occurs at different depths and changes with input motion intensities. Simplified angled bridge models using decomposed free field motions capture well the primary structural response.