PowerPoint Presentation
Understanding and Improving the Seismic Behavior of Pile
Foundations in Soft SoilsBradley Fleming, Sri Sritharan, &
JinWei HuangIowa State UniversityKanthasamy Muraleetharan &
Gerald MillerOklahoma University
Dream it, Design it, Build it.
www.ccee.engineering.iastate.eduIowa State UniversityCivil,
Construction & Environmental EngineeringHello, my name is Brad
Fleming and I will be presenting the analytical portion of what was
just discussed. 1Check for other involvement (4 schools and 2
industry leaders?). This could be found in the media day
videos.Modeling TechniquesFinite Element (OpenSees)Detailed 3D
analysisUsed to understand complex interactions betweenpile and
improved soilimproved soil and unimproved soilp-y Analysis Method
(LPILE)Simple 2D analysis Attractive for engineers in
industryAccount for improved soil of limited width by applying
modification factors to p-y relationshipsIn our research we are
trying to understand and improve the behavior of pile foundations
in soft soil and translate our findings into a useful design
methodology.
To do this, we have constructed detailed finite element models
in the OpenSees and compared the results to field data to see how
well we can characterize the behavior of improved piles. This has
also helped us to understand the complex interactions between the
pile and soil as well as the improved soil and unimproved soil.
The p-y analysis method is very attractive for engineers in
industry and can easily be adopted into a design methodology.
However, this method only accounts for soil layers that have
constant properties to a infinite length in the horizontal
direction. We have found this to be useful for practical purposes
and can be applied to improved piles if the improvement is
sufficiently wide. For improved soil of limited width, we have
developed a simple method for modifying the p-y relationship to
account for this. LPILE is a program we use to apply modification
factors to known p-y curves for clay and analyze.2
OpenSees Finite Element ModelPile (forceBeamColumn)7.23 m total
length34 beam elements5.3 m embedded lengthNon-linear fiber section
(half of pile)
Soil Island (OpenSeesPL soil mesh generation)10.3 m long, 5.15 m
wide, and 7.62 m high3,450 nodes 2,492 soil elementsSoil Contact
Elements (BeamContact3D)Pilerm1m2slavePSoil Elements
(SSPbrick)ClaySandPressureIndependMultiYieldPressureDependMultiYieldGmax
= 3250 kN/m2Gmax = 1.0E+5 kN/m2c = 30.5 kN/m2 = 37 deg.sat =1.8
ton/m3sat =2.0 ton/m3ClaySandIn building the finite element model
we used a very useful tool called OpenSeesPL to generate the soil
island, pile elements, and assign elements material properties.
Only half the mesh was constructed due to symmetry.
The pile is made up of several beam elements having a cross
section made up of many fibers. The uniaxial characteristics of
each fiber is what makes the pile nonlinear if stressed beyond its
yield point. The stiffness and yield strength of the steel in our
piles was tested in the lab and applied to each fiber in the
cross-section.
Contact elements between the soil and pile were added to allow
for frictional slip, sticking, and separation. These elements and
respective material properties were developed by Pedro from the
University of Washington.
Soil elements were given uniform properties along its depth.
Initially, these were calibrated to give the behavior of the
unimproved pile in the field. Then improved soil was applied to the
elements surrounding the pile to see if the improved pile behavior
can be captured.3Pile Head Responses of Full-Scale Test and FEM
Unimproved PileImproved PileIn this slide, the left image is the
response of the unimproved pile and the right image is the response
of improved pile for both the field test and finite element model.
From this we see that the model did very well in capturing the
response of the improved pile. 4LPILE Model & Modified p-y
Curves
Now I would like to discuss the p-y analysis method.
In the traditional method the pile is made up of several beam
elements with nonlinear moment-curvature characteristics. The soil
is represented by a series of nonlinear springs that have a
resistance p to a lateral displacement y. The shape of the p-y
curve is defined by known models that can be found in various
publications. LPILE also has these models available in its library.
For improved soil of sufficient width, the behavior of the soil can
be characterized with a single spring. For improvement of limited
width, the proposed method is to take the p-y models for both
improved and unimproved soil and place them in series to obtain the
effects of each.
But how much soil improvement is needed to be considered
sufficiently long?5Effective LengthGuo and Lee (2001)Attenuation of
stresses in soil layer
This property we call effective length and can be found from
observing the stress attenuation along the radial distance away
from the pile. Guo and Lee developed a closed form solution for the
attenuation. In their equation, the stresses are non-zero out to
infinity. We found this to be impractical and, based on several
centrifuge tests, we determined an attenuation value of 5% of the
maximum stress to be sufficient for finding the effective length.
Therefore, if the 5% attenuation value is located within the
improved zone, the p-y relationship for the soil at that layer can
be represented with a single spring. If the 5% value is outside the
improved zone, the effective length is much larger because of the
distribution of stresses in the soft soil and the effective
stiffness of the soil is smaller. 6R Equivalent rigidity (analogous
to AE for axially loaded member)Leff - Length of uniform soil layer
ki Equivalent stiffness of the p-y curveS - Stiffness of
springEquivalent Rigidity
Improved SoilUnimproved Soil
To find the effective stiffness of the combined soil, we took a
simple approach to combine the equivalent stiffnesses or K of the
p-y curves for both improved and unimproved soil. Since K
represents the stiffness of an infinitely long soil layer, we
multiply it by the effective length found from Guo and Lees
equation to get an equivalent rigidity. This is analogous to AE for
axially loaded members. Applying this rigidity to the length of
improved soil or Li and unimproved soil or Lu will give the
stiffness of an equivalent spring. Then, the effective stiffness of
the combined springs can be calculated for two springs in
series.
Li is easy to determine but Li is not because it requires the
effective length for the combined soil.7Combined Properties
Cont.
Therefore, we created a simplifying assumption to eliminate Lu.
Pictured on this slide is the attenuation for fully improved, fully
unimproved, and an attenuation for the combined soil. From this it
is easy to see that Lu approaches zero as Li increases and Lu
approaches the effective length for unimproved soil as Li
decreases. This led to the development of the linear relationship
shown above. After subsituting into the equation for two springs in
series, we came up with an equation for the effective stiffness of
the p-y curve for the combined soil.8p-y Modification Factors
JinWei Huang (2011)We use this stiffness to interpolate the
values for the full curve and calculate the modification factors to
be inserted into the LPILE analysis tool. In this case omega p and
omega y are used to reduce the soil resistance and increase the
soil displacement of the improved p-y curve.9Global Response
Comparisons (Centrifuge)JinWei Huang (2011)This slide shows the
camparison of measured load-displacement response envelopes with
the LPILE computed load-displacement responses using the newly
developed p-y curve modification factors for test #1.the calculated
elastic lateral stiffness for piles from centrifuge test #1 were
slightly higher than that obtained from the experimental data;the
calculated lateral load resistance for centrifuge test #1 increased
from 100 kN to 420 kN/m, which is in close agreement with the
increase from 80 kN to 420 kN/m (94.4 kips) from the experimental
data;
10Global Response Comparisons (Centrifuge)JinWei Huang
(2011)Calculated Elastic stiffness is in excellent agreement with
the average value of the push and pull experienmental data.The
calculated lateral resistance at max. target disp. increased from
55 kN to 125 kN which is really close to the increase form 55 kN to
118 kN from the experiement data.11Global Response Comparisons
(Centrifuge)JinWei Huang (2011)Calculated Elastic stiffness is in
excellent agreement with the average value of the push and pull
experimental data, however, no significant increase in elastic
stiffness can be observed for these two piles although the soil
improvement dimensions increased from 13D13D9D to 17D17D12D.
12Field & LPILE Global Responses
JinWei Huang (2011)The calculated elastic lateral stiffness
increased by 480% from 759 kN/m to 4396 kN/m, which agrees well to
the 420% increase obtained from the experimental data.Then
calculated lateral resistance in the inelastic region for
unimproved pile agrees well with the test data while the lateral
resistance in the inelastic region for the improved pile is about
20% lower than the measured values.It should also be noted that the
pile in the improved ground failed at the first cycle of loading at
lateral displacement of 8 in., primarily due to local buckling at
the side wall just above the ground surface, & fractured the
pile as a result of low cycle fatigue. 13Field & LPILE Test
Local Responses
JinWei Huang (2011)Calculated max. moment location is about 1.5
m below the ground for pile with no ground improved, & at the
ground surface for pile with ground improvement. Which agrees very
well with the experimental data. The moment decreased to zero at a
depth of 1.4 m below the ground surface in both calculated &
measured moment profiles for the improved pile, which indicates
that the effective depth of the ground improvement is a lot less
the actual improvement depth used in the test.14ConclusionsBoth
LPILE and OpenSees closely resembles centrifuge and field
behaviorOpenSees is an effective analysis tool but requires
specialized knowledge and involves high computation costsLPILE is
an attractive tool for engineers and has flexibility to modify p-y
curvesThe proposed method for modifying p-y curves does well in
characterizing the behavior of piles in improved soil of limited
width15