Collaborative Research:

Centrifuge Modeling for

Soil-Pile-Bridge Interaction

Bruce Kutter, Professor

Mahadevan Ilankatharan, Graduate student
University of California, Davis

NEES 4th Annual Meeting

June 21-23, 2006 Washington DC

 Main Points

¾ Scope of project
¾ Scope of centrifuge test program
¾ Collaborative design – column length issue
¾ Comparison of centrifuge and shaking table tests
¾ Component tests vs system tests
¾ Comparison of simulations and experiments
¾ Data archiving
 Prototype Structure

Data UW, UCB, UCD

Computational Outreach
Kansas
Models
SJSU
Collaboration in Centrifuge Tests Design
UW

Ground motion, Many hours of

numerical simulations
- Video conferences

- Conference phone calls

Dimensions &
details of UCD - Face to Face meetings
structural models Soil properties

- Email

UNR Data archiving UT

NEESit
NEES Geotechnical Centrifuge at Davis
 41.1

Centrifuge Model:
X X

Shaking BENT 1 BENT 2 BENT 3

direction
93.6 85.8

Y
SINGLE
PILE

First BENT 4 BENT 5

Centrifuge
Test Series PLAN VIEW
(MIL01)
2.6 3.9
6.4

27.8

ELEVATION AT X-X
Features of centrifuge models:
2.6 3.9
6.4

¾ Centrifuge g level : 52 g
¾ Soil : dry Nevada Sand (Dr = 80 %)
¾ Piles: strain gauged aluminum tube
¾ Ground motion :
¾ Realistic ground motions selected by UW
¾ Frequency sweeps
¾ Scaled amplitude in successive events
¾ 20 different superstructures
¾ Two pile-bents, 30° 60°

¾ Varied orientation relative to shaking

¾ Varied mass of bent
BENT B BENT C
¾ Varied clear height of pile
¾ Single piles
¾ Two span segment of bridge
¾ Dates of testing: December 2004 – January 2006
 UCD Centrifuge test:

1/52 scale
aluminum tube piles

UNR shake table test:

- 1/4 scale reinforced concrete

columns
What should be the shaking table column height to achieve
- fixed support on shake tables
similar natural frequencies, and moment and shear distribution?
Equivalent depth of fixity

We chose to model
the column stiffness
H clear, pile so that natural
H col, shake frequencies would be
table
the same in
Lf centrifuge and shake
table.
Shake table
column Equivalent depth of
fixity would be
L f : Equivalent depth different if you want
of fixity Centrifuge Pile to model the column
(Chai, 2002)
capacity.

H clear, pile= H col, shake table - L f

Instrumentation:

Free field soil

Strain gage instrumentation

Bent cap
 Comparison of centrifuge & 1-g shake table test results:
Deck accelerations
0 .4 (i) S h o rt b e n t

0 .2

0 .0

-0 .2

-0 .4
D e c k m o tio n @ C e n tr ifu g e te s t
Deck acceleration (g)

0 .4 (ii) T a l l b e n t D e c k m o tio n @ S h a k in g t a b le te s t

0 .2

0 .0

-0 .2

-0 .4

0 .4 (iii) M e d i u m b e n t 2.6 3.9

6.4
0 .2

0 .0

-0 .2

-0 .4
S T M
Medium amplitude shake: Peak base acc = 0.25g (in
0 5 10 15 20 25
centrifuge test)
T im e (p ro to ty p e se c o n d s)
 Comparison of Centrifuge and Shake Table Results for 0.25 g shake

Free field motion @ 2.5 m depth @ Centrifuge test

2 Shaking table base motion

1 (i) Short bent (ii) Tall bent (iii) Medium bent

ARS (g)

0
Deck motion @ Centrifuge test
2 Deck motion @ Shaking table test

(iv) Short bent (v) Tall bent (vi) Medium bent

1

0
0 1 2 30 1 2 30 1 2 3
Period (s)
Shake may
damping tablebemotions
greaterare a bit different damping
for for each shake table (possible
two modes (translation +due to radiation
torsion) in centrifuge
are closer together in shake table test
specimen-table interaction)
than centrifuge test (distorted span length in centrifuge test).
Component vs System behavior: Single bent configuration

2.6 3.9
6.4

S T M
Component vs System Behavior: Bridge bent configuration

2.6 3.9
6.4

S T M
Component vs System Behavior:
-8

Single bent
Short bent-cap motions Bridge bent

Single bent -4
1.5
Bridge bent

Depth from ground surface (m)

1.0
0

(i) Short Bent (Hc/D=2.2)

0.5

4
0.0
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Period (s)

Spectral acceleration of bent cap 8

motion is much less when it is
connected to the deck. But, (i) Short Bent (Hc/D=2.2)
bending moments are only 12
slightly smaller Pile bending
-2000 moments0@ maximum
-1000 1000
bent-cap displacement
Component vs System Behavior:
Single bent
Bridge bent
Medium bent-cap motions
0.0
Single bent
1.5 Bridge bent

1.0

(iii) Medium Bent (Hc/D=3.3)

0.5

0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Period (s)

Spectral acceleration of bent cap

motion is similar when it is
(iii) Medium Bent (Hc/D=3.3)
connected to the deck. But
bending moments are much -2000 -1000 0 1000

greater! Pile bending moments @ maximum

bent-cap displacement
Shin, Ilankatharan, Arduino, Kutter, and Kramer
(8 NCEE, 2006)

Excellent
P-y and shear
comparisons
beam
between
analyses
OpenSEES
using
and centrifuge
OpenSEES are
results!
verified Some
for
analyses done
piles in dry
during
sand. testing
Data archives:
-MIL data is the most completely documented experiment in NEEScentral
-Have already heard about this data in NEESit report later
-Recent interactions with NEESit have been productive! Project

Experiments

Trials

Data

DAQs

-Unprocessed data
-Converted data
-Corrected data
-Derived data
Concluding Remarks:

¾ Complementary experiments on multiple NEES sites

requires collaboration amongst multiple experts. Cross-
disciplinary training may lead to more holistic soil-structure
system designs.

¾ Direct comparison of results from different types of

facilities is valuable because it can clearly expose flaws
that we might otherwise ignore, e.g.,
¾ importance of distorting bent spacing,
¾ specimen-actuator interaction

¾ Extension of element behavior to system behavior via

numerical analysis cannot be taken for granted and must be
tested. Major NEES facilities enable testing response of
multiple component systems (e.g., multiple span bridge
decks).
 Concluding Remarks:
¾ Data from three series of highly
instrumented centrifuge tests and
approximately fifty shaking eventsis
archived and available through
NEEScentral

¾ Results from OpenSEES analyses

were able to accurately predict the
experimental results.

¾ A UW graduate student, Hyung-Suk Lee

spent about a month helping Lanka perform each experiment.
¾ His assistance with the experiment helped him understand and
confidently use the test data for his numerical simulations
¾ Pre-test analyses helped us design the specimens
¾ Analyses during the test helped us figure out how hard and how many
times to shake the specimens and what to look for in the data.
Acknowledgements:

¾ Visiting Scholar: Tetsuya Sasaki, PWRI, Japan

¾ Student at UNR: Nathan Johnson

¾ Students at Austin: Puneet Agarwal, and Asli Kurtulus

¾ Faculty: Arduino, Kramer, Wilson, Jeremic, and Wood

¾ IT advice Roger Clermont and Shannon Whitmore (NEESit)

¾ Centrifuge Technicians: Chad Justice, Tom Coker, and Tom

Kohnke