Вы находитесь на странице: 1из 75

THE MECHANICS OF THE AXIAL

LOADING OF NONDISPLACEMENT
PILES IN SAND
Rodrigo Salgado

Third Bolivian International Conference on Deep Foundations, April 27, 2017


Acknowledgments
2

Prof. Monica Prezzi, Fei Han


National Science Foundation

INDOT and FHWA

Fugro Consultants and CTI Associates for their

support for COFFEE (Center for Offshore,


Foundation and Energy Engineering)
Outline
Fundamental Concepts and the Piling Engineering Problem
Realistic Modeling of Soil Response
Analysis
Challenges in computational simulations
Details
Results
Verification
Insights from Analyses
Base and shaft resistance of nondisplacement piles
Group interaction
Final Remarks
Fundamental Concepts
Pile installation
Pile axial load response controlling factors
Individual pile response in the context of groups/
rafts
Pile Installation
Wished in place (ideal nondisplacement pile)
Drilled/Bored (nondisplacement pile)
Drilled with minimal/no unloading (CFA piles)
Drilled displacement piles
Monotonically jacked piles
Piles jacked with multiple strokes "Probe Penetration Problem"
Driven piles
Piles in Different Contexts
One pile per column, one pile group per column or
piled raft are all different problems with respect to how
a single pile load-settlement response fits into the
overall load response
The soil profile (existence or not of a stiff bearing
layer) also affects how we use the individual pile load
response in design
It is highly desirable to have the capability to
accurately simulate individual pile response in any of
these contexts
Motivation for Theoretical Work
Develop capability to fully simulate the response of
piles to loading
Use theoretical (and experimental) simulations to
continue to develop pile design tools that work
accurately and reliably
Develop constitutive models and computational
approaches that can be applied to other problems in
which some of the same challenges appear
Direct Design

Soil mechanics? Unit resistance:


In situ measurement:

qc qb , qs
Limited data of good quality (lack of
load tests on instrumented piles in
well-characterized soil profiles loaded
to sufficiently large settlements

Still the only practical way to develop


design methods in structured soils
Purdue Pile Design Method
Fundamental (soil mechanics) basis for all equations
Validation through
field load testing
centrifuge testing
calibration chamber testing

Applicable to sand (and non-plastic silt) and clay


(and plastic silt)
Purdue Method Limit Unit Shaft Resistance qsL
qsL = K v0 tan c
c = critical-state friction angle in soil
K0 = coefficient of lateral earth pressure
at-rest
K 0.2 K 0 0.4
DR v0 [1] C1 = 0.7 for clean sand in general
=e C1 exp 1.3 0.2 ln qc = cone resistance
K0 100
p
A DR = relative density
'v0 = initial vertical effective stress
q 'h0 = initial horizontal effective stress
ln c 0.4947 0.1041c 0.841ln h 0
p p A 100% [2]
DR (%) = A

0.0264 0.0002c 0.0047 ln h 0
pA

[1] Salgado 2006, [2] Salgado et al. 2010, [3] Salgado & Prezzi 2007
Purdue Method Ultimate Unit base resistance qb,ult

qcb,avg = average cone resistance near


0.0066 DR
qb ,ult = 0.23e qcb , avg the pile base
DR = relative density

Note assumed independence from pile length or confining stress

Salgado et al. 2011


Bounding Surface Model
The inadequacy of traditional soil models
What is a bounding surface?
Bounding surface versus classical plasticity
Purdue Sand Model
Constitutive Model
Soil elements near the
pile experience various
loading conditions Shearing
Constitutive model must SS
accurately capture the PILE

soil behavior under all


of these loading
Extension
conditions TXE
Compression
TXC
Features Model Must Capture
Inelastic strains practically from outset
Peak effective stress ratio (bounding surface)
Critical state asymptote (critical state surface)
Correct volume change or pore pressure response (phase
transformation line or dilatancy surface)
A variety of other details, such as shape of critical state line
and dependence of critical state properties on stress path,
fabric evolution during deformation and other factors
Model Used in Simulations
A bounding surface model now in its third
generation was used in simulations
Key references:

1G: Loukidis and Salgado (2009)


2G: Woo and Salgado (2015)
3G in progress
Undrained Triaxial Compression
LS model (2009) WS model (2015) This study

DR = 26%
DR = 47%

DR = 39%

Dry-deposited Toyoura sand with p'0 = 400 kPa and various relative densities DR
Gray lines with white dots: experimental results (Yoshimine, 1995-1999)
Lines: simulation results
Undrained Triaxial Compression
LS model (2009) WS model (2015) This study

DR = 39%
DR = 47%

DR = 26%

Dry-deposited Toyoura sand with p'0 = 400 kPa and various relative densities DR
Gray lines with white dots: experimental results (Yoshimine, 1995-1999)
Lines: simulation results
Drained Triaxial Compression
LS model (2009) WS model (2015) This study

DR=81%, p'0=400kPa

DR=42%, p'0=400kPa

DR=81%, p'0=200kPa

DR=40%, p'0=100kPa

Air-pluviated Toyoura sand with various p'0 and relative densities DR


Gray lines with white dots: experimental results (Fukushima and Tatsuoka, 1984)
Lines: simulation results
Drained Triaxial Compression
LS model (2009) WS model (2015) This study

DR=42%, p'0=400kPa

DR=40%,
DR=81%, p'0=100kPa
p'0=400kPa

DR=81%, p'0=200kPa

Air-pluviated Toyoura sand with various p'0 and relative densities DR


Gray lines with white dots: experimental results (Fukushima and Tatsuoka, 1984)
Lines: simulation results
Undrained Triaxial Extension
LS model (2009) WS model (2015) This study

DR = 54%

DR = 46%

DR = 31%

Dry-deposited Toyoura sand with p'0 = 400 kPa and various relative densities DR
Gray lines with white dots: experimental results (Yoshimine, 1995-1999)
Lines: simulation results
Undrained Triaxial Extension
LS model (2009) WS model (2015) This study

DR = 54%

DR = 46%

DR = 31%

Dry-deposited Toyoura sand with p'0 = 400 kPa and various relative densities DR
Gray lines with white dots: experimental results (Yoshimine, 1995-1999)
Lines: simulation results
Undrained Simple Shear
LS model (2009) WS model (2015) This study

DR = 35%

DR = 29%

DR = 24%

Dry-deposited Toyoura sand with p'0 = 300 kPa and various relative densities DR
Gray lines with white dots: experimental results (Yoshimine, 1995-1999)
Lines: simulation results
Undrained Simple Shear
r1 r1 r1
LS model (2009) WS model (2015) This study

Critical-state surface
Critical-state surface Critical-state surface

DR = 24, 29 and 35%

Normalized plane

Dry-deposited Toyoura sand with p'0 = 300 kPa and various relative densities DR
Gray lines with white dots: experimental results (Yoshimine, 1995-1999)
Lines: simulation results
Undrained Simple Shear
LS model (2009) WS model (2015) This study

'1

Dry-deposited Toyoura sand with p'0 = 300 kPa and various relative densities DR
Gray lines with white dots: experimental results (Yoshimine, 1995-1999)
Lines: simulation results
Direct Simple Shear
LS model (2009) WS model (2015) This study

DR = 73%

DR = 61%
DR = 49%

Air-pluviated Toyoura sand with '11=98.1kPa, '22='33=44.15kPa and various relative densities DR
Gray lines with white dots: experimental results (Pradhan et al., 1988)
Lines: simulation results
Direct Simple Shear
LS model (2009) WS model (2015) This study

DR = 73%

DR = 61%

DR = 49%

Air-pluviated Toyoura sand with '11=98.1kPa, '22='33=44.15kPa and various relative densities DR
Gray lines with white dots: experimental results (Pradhan et al., 1988)
Lines: simulation results
Direct Simple Shear
t 1 3
=
s 1 + 3

Local void ratio in shear band


measured by -rays or x-rays

Air-pluviated Leighton Buzzard sand (type B) with various 'n and relative densities DR
Dots: experimental results (Roscoe, 1970; Stroud, 1971)
Lines: simulation results
Simple Model?
5,000
4,500
4,000
3,500

Load (kN)
3,000
single-elastic
vs. 2,500
2,000
1,500
1,000
500
0
0.00% 10.00%
Settlement at pile top (%)

DR = 80%
Purdue Sand Model 1G Linear Elastic Model
Simple Model?
400
350
300

Load (kN)
250
MC-total
vs. 200 MC-shaft
150 MC-base

100
50
0
0.00% 5.00% 10.00%
Settlement at pile top (%)

DR = 80%
Purdue Sand Model 1G Mohr-Coulomb Model
Challenges in Computational Simulations
Interface Response (Contact Formulation)
Mesh Distortion
Constitutive Model Integration
Crushing
Complexities in Pile Installation/Loading
Drainage
Fully drained loading in clean sand
Fully undrained loading in pure clay
Pile size versus particle size
Continuum (solid) mechanics applicable if B / D50 large enough (required ratio
different for base and shaft response)
Very large shear strains and rotations near the pile-soil interfaces (mesh
distortion, model integration)
Very large induced mean stresses around the pile base (crushing)
Complex interface response (different relative roughness leads to different
interaction with the soil)
FE Simulation Steps
Meshing
Geostatic step
K0= 0.4 for dense sand, and 0.45 for loose sand
Same unit weight is applied to pile and sand
Loading step
Vertical velocity is applied at the pile head
The velocity is smoothly increased from 0 to 2 mm/s in the
first second
Mesh Configuration for the Three-Dimensional FE Analysis

Elements next to the 10 m


pile should take the B/2 ts

size of the shear band


thickness: about 5D50 Pile Sand
10 m

10 m
Rate (inertial) Effect
The constitutive model is rate-
700
independent
600
Rate (inertia) effects is examined by
using three different loading speeds 500

at the pile head: 6 cm/s, 1cm/s and 400 6cm/s

Load (kN)
2mm/s 300 1cm/s
2mm/s
Oscillations are observed in the load- 200
settlement response when 6cm/s is B = 0.3 m, L = 10 m
100 Ottawa sand, DR =
used; 1cm/s and 2mm/s produce 80%
0
almost the same load-settlement 0% 2% 4% 6% 8% 10%
curves, without oscillations Settlement at pile top(%)
Shaft Resistance

35
Surface Roughness
Definition of the maximum roughness Rt and the normalized roughness Rn

Peak

Valley
Rt
Rn =
D50

Low Rn Rough
Rn 1
Shear Band Formation for Different Pile Surface Roughnesses

Smooth Medium rough Rough


(Rn = 0.03) (Rn = 0.26) (Rn = 1.14)
Spectrum of Possible Interface Response

Sliding dominates; soil Soil gets engaged, a


Shearing occurs inside
engagement is limited; (thinner) shear band
the soil possible to
no shear band forms in forms in the soil, but some
model, up to a point,
the soil; needs to be sliding occurs and needs
using "perfect
modeled through an to be modeled through
contact" (a single mesh
appropriate contact an appropriate contact
for pile and soil)
formulation formulation

=/50
0 1
Shear Band Development (Rough Interface)
Contact Modeling
Traditional contact formulation, with Coulomb
friction along the interface, does not capture the
complexities of the response of any of the three
types of interface response
In this presentation, our focus is exclusively on

rough contact, which implicitly captures the


interface response in a realistic manner
Shear Band Thickness
Shear Band Development

Pile Sand
Unit Shaft Resistance
The profiles of the unit shaft Unit shaft resistance (kPa)
0 10 20 30 40 50 60 70 80 90 100
resistance along depth at several 0
relative settlement levels 1
2
Ottawa sand
3
L = 10 m 4

Depth (m)
B = 0.3 m 5
DR = 80% 6
7
8
9
10
w=0.005B w=0.01B
w=0.017B w=0.03B
w=0.1B
Stress in Shear Band Element vs. Pile Head
Settlement
120.00

100.00
Stress on shaft (kPa)

80.00

60.00 Shear
stress
p'
40.00
B = 0.3 m
20.00 L = 10 m
Ottawa sand
DR = 80%
0.00
0% 2% 4% 6% 8% 10% 0= 100
Settlement at pile top (%B) kPa
K0 = 0.4

Unit Shaft Resistance
K
0 1 2 3
0
1
2
Profiles of K values along depth at several rela 3
tive settlement levels (L = 10 m, B = 0.3 m, DR 4

Depth (m)
= 80%) 5
6
7
8
9
10

Back calculated Gauss point


Build Up of Lateral Stress during Shearing along Pile Shaft

Results from calibration chamber tests


at 70 kPa vertical effective stress
K Estimated using FE Analyses

rough: no sliding along interface


Values of from axisymmetric FEAs vs. Centrifuge Test Data

Data from [1, 2]; a = 80g


Toyoura sand: D50 = 0.22 mm
Model pipe pile
Aluminum alloy, E = 70 GPa
Pile diameter B = 10 mm
Pile length L = 245 mm
pile surface relative roughness Rn = 0.45 (by
mechanical turning)
Shear band thickness ts [3] = 5D50 = 1.1 mm

[1] Fioravante, V. (2002). On the shaft friction modelling of non-displacement piles in sand. Soils and foundations, Japanese Geotechnical Society, 42(2), 2333.
[2] Colombi, A. (2005). Physical modeling of an isolated pile in coarse grained soils. University of Ferrara.
[3] Tehrani, F. S., Han, F., Salgado, R., Prezzi, M., Tovar, R. D., and Castro, A. G. (2016). Effect of surface roughness on the shaft resistance of non-displacement piles
embedded in sand. Gotechnique, 2, 115.
Base Resistance

50
Mesh at the pile base
The trial meshes used near the pile base: (a) square pattern elements with the size of 4 cm and (b)
square pattern elements with the size of 3 mm

Pile Pile

4 cm 3 mm

(a) (b)
Mesh density at the pile base
The effect of the mesh density
near the pile base on the
normalized base resistance qb/qc
at different levels of relative
settlement s/B
FE simulation
pile pile

Vertical movement Horizontal movement


DR= 87%
DIC
DIC = Digital Image Correlation
Comparison of successive frames to track the same

"point" and calculate the displacement of that


"point
The point is, in imaging terms, a group of pixels; in

mechanics terms, it is an REV


Observations within the Domain
Traditional testing (calibration chamber, centrifuge
or field testing): measurement of loads on moving
boundaries (pile base, cone tip, pile shaft)
DIC and similar techniques enable us to check not

only boundary loads/displacement but


displacement and strain within the domain of the
BVP
DIC Observations
( B = 31.75 mm, L = 490 mm, surcharge = 50 kPa, D = 87% )
R

Vertical movement Horizontal movement


FE simulation
pile pile

Vertical movement Horizontal movement


DR= 87%
Quantitative Comparison
2 2.5
1.8 FE

Displacement in soil (mm)


Displacement in soil (mm)

1.6 DIC 2
1.4 verAcal
1.2 1.5
1 verAcal
0.8 1
0.6 horizontal
0.4 0.5
0.2 FE
0 0 DIC
0 1 2 3 4 0 1 2 3 4
Pile head displacement (mm) Pile head displacement (mm)
Delayed Base Resistance Development

Unit base resistance (MPa)


2

0
0 1 2 3 4
Preinstallation process in chamber Pile head displacement (mm)
generates base-soil contact that
requires some base movement for
the resistance to kick in (B = 31.75 mm)
Soil response near pile base: shear strain and void ratio

Pile base Pile base

Ottawa sand
DR = 50%
=10 m
(e0 = 0.63) B = 0.3 m
s = 0.1B

Soil response near pile base: mean stress and shear stress

Pile base
Pile base

Ottawa sand
DR = 50%
=10 m
B = 0.3 m
s = 0.1B

Soil response near pile base: plastic strain and plastic dissipation

Ottawa sand
DR = 50%
=10 m
B = 0.3 m
s = 0.1B

qb vs. Settlement

4000 DR 4000 DR
3500 80% 3500 80%
Unit base resistance (kPa)

Unit base resistance (kPa)


50% 50%
3000 3000
2500 2500
2000 2000
1500 1500
1000 1000
500 500
0 0
0% 2% 4% 6% 8% 10% 0% 2% 4% 6% 8% 10%
Relative settlement s/B at pile head Relative settlement s/B at pile Base
(a) (b)
qb/qc vs. settlement
Normalized unit base resistance qb/qc mobilized in sands with different relative densities vs relative
settlement s/B: (a) at the pile head and (b) at the pile base

crushing?

(a) (b)
*qc is determined by Salgado, R., and Prezzi, M. (2007). Computation of Cavity Expansion Pressure and Penetration Resistance
in Sands. International Journal of Geomechanics, 7(4), 251265.
Crushing below the pile base
Unit Base Resistance versus Settlement from
Axisymmetric FE Analyses and Centrifuge Tests
Centrifuge test data from [1]; a = 80g 18
16
Toyoura sand: D50 = 0.22 mm, DR = 90%

Unit base resistance (MPa)



14
Model pipe pile 12
Aluminum alloy, E* = 21 GPa 10

Pile diameter B = 10 mm 8
Centrifuge
6 test (Rn=0.45)
Pile length L = 245 mm Centrifuge
4 test (Rn=0.06)
pile surface relative roughness Rn = 0.45 (by FE simulation
2
mechanical turning)
0
A shear band thickness ts = 5D50 = 1.1 mm 0% 10% 20% 30% 40% 50%
Pile head settlement
was assumed to be the mesh size

E* = equivalent elastic modulus after correction for the model piles cross-sectional area
[1] Fioravante, V. (2002). On the shaft friction modelling of non-displacement piles in sand. Soils and foundations, Japanese Geotechnical Society, 42(2), 2333.
Resistance in Layered Soil Profile
-2B
-B
-0.5B 0B B 2B

Relative settlement s/B = 10%


Summary
Goal
Building a scientific (soil mechanics) basis for
development of pile design methods
Analysis need to be realistic
Challenges to Realistic Simulations
Realistic constitutive model
Interface (contact) modeling
Large deformations, rotations (with implications in terms of
mesh distortion if installation/penetration is modeled and
model integration)
Crushing (without modeling crushing, resistances are
overestimated for denser sands)
Insights from Analyses Single Pile
Considerable build up in K on shaft of pile as it is
loaded due to dilative response along the shaft
Crushing in denser sand may reduce base resistance by
10-20%
From the point of view of the mechanics of the problem,
embedment of a pile in a bearing layer does not
increase base resistance in a measurable way
Thank you!

Muchas gracias!
VUMAT
Update stress and user-defined state variables from
strain increment/deformation gradient at each Gauss
point
Integration algorithm:
Semi-implicit backward-Euler (cutting-plane) algorithm
adapted with sub-incrementation and error control [1,2]
Stress error tolerance = 10-5
Yield function error tolerance = 10-9

[1] Abbo, a J. (1997). Finite element algorithms for elastoplasticity and consolidation. University of Newcastle.
[2] Loukidis, D., and Salgado, R. (2008). Analysis of the shaft resistance of non-displacement piles in sand. Gotechnique, 58(4), 283296.
Shear Band

Undeformed
element

Pile
Deformed element

Horizontal
expansion
(Dilation)
Crushing of sand particles
Crushing happens when sand is subjected to high mean
stress level p'
For Toyoura sand
Threshold mean stress for crushing: pc' = 5.85 MPa[1]
Maximum mean stress below the pile base p' = 6.7 MPa >
5.85 MPa
Crushing is not considered in the constitutive model

[1] Yao, Y.-P., Yamamoto, H., and Wang, N.-D. (2008). Constitutive Model Considering Sand Crushing. Soils and Foundations, 48(4), 603608.

Вам также может понравиться