R

Undrained Soil Behavior

(and modeling)
Gioacchino (Cino) Viggiani
Laboratoire 3SR (Sols, Solides, Structures - Risques), Grenoble, France

many of the slides were originally created by:

Helmut Schweiger (Technical University of Graz, Austria)
Ronald Brinkgreve (Plaxis bv / Delft University of Technology)
Antonio Gens (Technical University of Catalunya, Barcelona)

Course “Computational Geotechnics” 1/37

 contents R

• drained / undrained (conditions and analysis)

• drained / undrained soil behavior
– typical results from drained and undrained triaxial tests
– Skempton's parameters A and B
– strength parameters
– what is the critical case: drained or undrained?
• modeling undrained behavior with Plaxis
– three methods
– effective stress analysis: how does it actually work
• undrained shear strength
– undrained behavior with Mohr-Coulomb Model
– undrained behavior with Hardening Soil Model
– influence of dilatancy
• summary

Course “Computational Geotechnics” 2/37

 drained / undrained (conditions and analysis) R

in undrained conditions, no water movement takes place and,

therefore, excess pore pressures are built up
Du  0, Ds  Ds'

in drained conditions, no excess pore pressures are built up

Du = 0, Ds = Ds'

• drained analysis appropriate when

– permeability is high
– rate of loading is low
– short term behavior is not of interest for problem considered

• undrained analysis appropriate when

– permeability is low and rate of loading is high
– short term behavior has to be assessed

Course “Computational Geotechnics” 3/37

 an example (deep excavations) R

suggestion by Vermeer & Meier (1998)

T < 0.10 (U < 10%)  undrained analysis

T > 0.40 (U > 70%)  drained analysis

k = permeability
Eoed = stiffness in 1-d compression
k E oed w = unit weight of water
T t D = drainage length
γw D 2
t = construction time
T = dimensionless time factor
U = degree of consolidation

Course “Computational Geotechnics” 4/37

 triaxial test (NC soils) – drained / undrained R

drained undrained

Course “Computational Geotechnics” 5/37

 triaxial test (OC soils) – drained / undrained R

drained undrained

Course “Computational Geotechnics” 6/37

 stress paths in undrained triaxial test – NC / OC R

Course “Computational Geotechnics” 7/37

 Skempton’s parameters A and B R

Skempton 1954: Dpw  B Ds 3  ADs1  Ds 3 

- fully saturated soil

- no inflow / outflow of pore water
- bulk modulus of soil grains is considered to be very high
- isotropic linear elastic material behavior (Hooke´s law)

D vol , skeleton  D vol , pore water

Dp' K' 
E´
D vol , skeleton  31  2´ 
K'
n Dp w
D vol , pore water 
Kw

Course “Computational Geotechnics” 8/37

 Skempton’s parameters A and B R

in standard triaxial compression: Ds 1 ; Ds 2  Ds 3

D vol , skeleton  D vol , pore water

Ds1  2Ds 3  3Dpw K w

Dp w  
3K' n

 
D pw 
1
 D s 3 
1
D s 1  D s 3 
1
nK '  3 
Kw

Dpw  B Ds 3  ADs1  Ds 3 

1
B 1
1
nK ' (always) A only for linear isotropric elasticity
Kw 3

Course “Computational Geotechnics” 9/37

 Skempton’s parameters A and B R

a few notes on parameters A and B:

– for Kw large compared to K', parameter B ~ 1.0

(corresponds to Dpw = Dp  Dp' = 0)
– small amount of air (lack of saturation) reduces parameter
B significantly
– "parameter" A is NOT a soil parameter. It depends on stress
path: for a given model, it cannot be determined a priori
but is a result of the model behavior for the stress path
followed (e.g. A=1/3 for linear isotropic elasticity)

Course “Computational Geotechnics” 10/37

 strength parameters R

Mohr-Coulomb parameters in terms of effective stress (real soil behaviour)


  c  s  tan  


c s

s 1  s 3  s 1  s 3 c   c 
   sin   ; 
t  s   sin  
2  2 tan     tan   

however …

Course “Computational Geotechnics” 11/37

 strength parameters R

MC parameters in terms of total stresses (only undrained conditions!)

   c  s  tan   total stresses

Cu
 s 1  s 3  s1  s 3 
 2 
 F
 2 
 F   cu s , s

-Cu
Effective stresses

 Soil behaves as if it was purely cohesive (zero friction)

 cu ( su ) : undrained shear strength
 cu only changes if drainage occurs (no change if undrained conditions prevail)

Course “Computational Geotechnics” 12/37

 what is the critical case: drained or undrained? R

ESP
Loading

Unloading
s, s’

note that for soils in general:

factor of safety against failure is lower for short term (undrained)
conditions for loading problems (e.g. embankment)
factor of safety against failure is lower for long term (drained)
conditions for unloading problems (e.g. excavations)
however … all depends on the sign of Dpwp, which depends on soil and SP

Course “Computational Geotechnics” 13/37

 contents R

• drained / undrained (conditions and analysis)

• drained / undrained soil behavior
– typical results from drained and undrained triaxial tests
– Skempton's parameters A and B
– strength parameters
– what is the critical case: drained or undrained?
• modeling undrained behavior with Plaxis
– three methods
– effective stress analysis: how does it actually work
• undrained shear strength
– undrained behavior with Mohr-Coulomb Model
– undrained behavior with Hardening Soil Model
– influence of dilatancy
• summary

Course “Computational Geotechnics” 15/37

 modeling undrained behavior with PLAXIS R

Undrained A (analysis in terms of effective stresses):

effective strength parameters c', ', '
effective stiffness parameters E50', '

Undrained B (analysis in terms of effective stresses):

total strength parameters c = cu,  = 0,  = 0
effective stiffness parameters E50', '

Undrained C (analysis in terms of total stresses):

total strength parameters c = cu,  = 0,  = 0
total stiffness parameters Eu, u = 0.495

Course “Computational Geotechnics” 16/37

 FE modeling of undrained behavior (method A) R

• both changes in s' and u are considered

• constitutive equations are formulated in terms of s'

principle of effective stress  Ds  Ds ' Ds f

with Ds f  Dpw Dpw Dpw 0 0 0
T

since the strains are the same in each phase,

Ds  D D
Ds '  D' D D  D' D f
Ds f  D f D pore fluid stiffness, related to the
bulk modulus of pore fluid Kf
13 03 Kf
D f  Ke Ke 
03 03 n

Course “Computational Geotechnics” 17/37

 FE modeling of undrained behavior (method A) R

all the above (which is valid for any soil for which the principle
of effective stress applies) can be easily combined with the FEM

• instead of specifying the components of D, specify D'‚ and Ke

D  D' D f then same as in the drained case

• when calculating stresses,
Ds f  K e D v
Ds  Ds ' Ds f
Ds '  D' D
a value must be set for Ke

the pore-fluid is assigned a bulk modulus that is substantially

larger than that of the soil skeleton (which ensures that during
undrained loading the volumetric strains are very small)

Course “Computational Geotechnics” 18/37

 what PLAXIS does R

PLAXIS automatically adds stiffness of water when undrained

material type is chosen using the following approximation:

Kw Eu 2 G1   u 
K total  K'   
n 31  2 u  31  2 u 

E' 1   u 
K total  assuming u = 0.495
31  2 u  1  '

Notes:
• this procedure gives reasonable B-values only for ' < 0.35 !
• real value of Kw/n ~ 1•106 kPa (for n = 0.5)
• in Version 8 B-value can be entered explicitely for undrained materials

Course “Computational Geotechnics” 19/37

 two numerical examples R

Example 1:

E' = 3 000 kPa, ' = 0.3, u = 0.495

 K' = 2 500 kPa, Ktotal = 115 000 kPa  Kw/n = 112 500 kPa
1
B = 0.978 fair value for saturated soil
nK '
1
Kw

Example 2:

E' = 3 000 kPa, ' = 0.45, u = 0.495

 K' = 10 000 kPa, Ktotal = 103 103 kPa  Kw/n = 93 103 kPa

B = 0.903 poor value for saturated soil

Course “Computational Geotechnics” 20/37

 FE modeling of undrained behavior (method A) R

uf
u
TSP
cu ESP

s, s’

– single set of parameters in terms of effective stress (consistent)

– realistic prediction of pore pressures (if model is appropriate)
– the undrained analysis can be followed by a consolidation analysis
– Cu is a consequence of the model, not an input parameter!!

Course “Computational Geotechnics” 21/37

 modeling undrained behavior with PLAXIS R

Undrained A (analysis in terms of effective stresses):

effective strength parameters c', ', '
effective stiffness parameters E50', '

Undrained B (analysis in terms of effective stresses):

total strength parameters c = cu,  = 0,  = 0
effective stiffness parameters E50', '

Undrained C (analysis in terms of total stresses):

total strength parameters c = cu,  = 0,  = 0
total stiffness parameters Eu, u = 0.495

Course “Computational Geotechnics” 22/37

 FE modeling of undrained behavior (method C) R

• analysis in terms of total stress

• u does not change

• constitutive equations are formulated in terms of s

total strength parameters c = cu,  = 0,  = 0

total stiffness parameters Eu, u = 0.495

Ds  D D

Course “Computational Geotechnics” 23/37

 FE modeling of undrained behavior (method C) R

TSP=ESP
cu
s, s’

– parameters in terms of total stress

– no prediction of pore pressures (only total stresses are obtained)
– the undrained analysis can not be followed by a consolidation analysis
– Cu is an input parameter!!

Course “Computational Geotechnics” 24/37

 modeling undrained behavior with PLAXIS R

Undrained A (analysis in terms of effective stresses):

effective strength parameters c', ', '
effective stiffness parameters E50', '

Undrained B (analysis in terms of effective stresses):

total strength parameters c = cu,  = 0,  = 0
effective stiffness parameters E50', '

Undrained C (analysis in terms of total stresses):

total strength parameters c = cu,  = 0,  = 0
total stiffness parameters Eu, u = 0.495

Course “Computational Geotechnics” 25/37

 FE modeling of undrained behavior (method B) R

• analysis in terms of effective stress

• u changes
• constitutive equations are formulated in terms of s’ (but
strength in total stresses!)

total strength parameters c = cu,  = 0,  = 0

effective stiffness parameters E50', '

Ds  D D
D  D' D f
Resulting undrained stiffness parameters
3 E
Eu  ;  u  0.495
2 1  

Course “Computational Geotechnics” 26/37

 FE modeling of undrained behavior (method B) R

ESP TSP
cu
s, s’

– parameters in terms of total stress and effective stress

– prediction of pore pressures (generally unrealistic)
– the undrained analysis should not be followed by a consolidation
analysis (pore pressures unrealistic)
– Cu is an input parameter!!

Course “Computational Geotechnics” 27/37

 contents R

• drained / undrained (conditions and analysis)

• drained / undrained soil behavior
– typical results from drained and undrained triaxial tests
– Skempton's parameters A and B
– strength parameters
– What is the critical case: drained or undrained?
• modeling undrained behavior with Plaxis
– three methods
– effective stress analysis: how does it actually work
• undrained shear strength
– undrained behavior with Mohr-Coulomb Model
– undrained behavior with Hardening Soil Model
– influence of dilatancy
• summary

Course “Computational Geotechnics” 28/37

 undrained shear strength from the Mohr Coulomb model R

“method A” (c’, f’ input parameters)

tana’ = sinf’

s1  s3  Effective Stress

2 Path, ESP B’

cu
a’ = c’cosf’

A’

s' 1 s' 3  , s1  s 3 

2 2
1
cu  c 'cos f ' s 'sin f '  c 'cos f ' s vo  s ho
  sin f '
2
cu c 'cos f ' 1
  1  K 0  sin f '
s 'v 0 s 'v 0 2

Course “Computational Geotechnics” 29/37

 undrained shear strength from advanced models R

‘Method A’
 Although it is possible, in some cases, to obtain an analytical
expression for cu, it is advisable to perform a numerical “laboratory”
test to check the value of undrained shear strength actually supplied
by the model
 It is important to perform the numerical “laboratory” test under the
same condition as in the analysis
 Plane strain, triaxial, simple shear
 Correct initial stresses
 Compression, extension, simple shear
 Not all cu values are achievable with a particular model

Course “Computational Geotechnics” 30/37

 Soft soil model R

Parameters
c '  0.1 kPa f '  23º K 0NC  1  sin f '  0.609  ur  0.15  *  0.11  *  0.0275

Desired cu/sv’=0.22
140.00
140.00

120.00
120.00

100.00 100.00
1-3 (kPa)

80.00

1-3 (kPa)
80.00

60.00 60.00

40.00
40.00

20.00
20.00

0.00
0 20 40 60 80 100 120 140 160 0.00
0.00 20.00 40.00 60.00 80.00 100.00 120.00 140.00 160.00
p' (kPa)
p' (kPa)

Triaxial (compression) Triaxial (extension)

cu/sv’=0.279 cu/sv’=0.214

Course “Computational Geotechnics” 31/37

 Soft soil model R

Parameters
c '  0.1 kPa f '  23º K 0NC  1  sin f '  0.609  ur  0.15  *  0.11  *  0.0275

Desired cu/sv’=0.22
140.00

120.00

100.00
1-3 (kPa)

80.00

60.00

40.00

20.00

0.00
0 20 40 60 80 100 120 140 160
p' (kPa)

Plane strain (compression) Plane strain (extension)

cu/sv’=0.279 cu/sv’=0.277

Course “Computational Geotechnics” 32/37

 influence of dilatancy on undrained shear strength R

if we set   0 then, negative volumetric plastic deformations

occur at failure:

D v  D ve  D vp (elastic-plastic behavior)

D v  0 (undrained conditions)

D vp  0  D ve  0  Dp '  K D ve  0
At failure: Dq  M Dp '  Dq  0

result: unlimited increase of q, i.e. infinite strength!!

Course “Computational Geotechnics” 33/37

 influence of constitutive model and dilatancy R

UNDRAINED TRIAXIAL TEST

1 vs q
simulation of undrained triaxial compression test – MC / HS model - q vs 1

300

275

250

225

200

175
q [kN/m ]
2

150

125

100

75
MC non dil
50 MC dil
HS_1 non dil
25 HS_1 dil

0
0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00

1 [%]

Course “Computational Geotechnics” 34/37

 influence of constitutive model and dilatancy R

UNDRAINED TRIAXIAL TEST

simulation of undrained triaxial compression
p' vs q test – MC / HS model - q vs p'

300

275 MC non dil

MC dil
250 HS_1 non dil
HS_1 dil
225 total stress path
200

175
q [kN/m ]
2

150

125

100

75

50

25

0
0.00 25.00 50.00 75.00 100.00 125.00 150.00 175.00 200.00 225.00 250.00

p' [kN/m2]

Course “Computational Geotechnics” 35/37

 summary R

• FEM analysis of undrained conditions can be performed in

effective stresses and with effective stiffness and strength
parameters (Method A)

• undrained shear strength is a result of the constitutive model

• care must be taken with the choice of the value for dilatancy
angle

• Methods B and C provide alternative ways to analyze undrained

problems but:
– the constitutive models do not represent the true soil
behavior
– useful only in stability problems in undrained conditions

Course “Computational Geotechnics” 36/37

 R

Course “Computational Geotechnics” 37/37

 Nicoll Highway Collapse, Singapore R

24/3/04

Course “Computational Geotechnics” 38/37

 Nicoll Highway Collapse, Singapore R

Course “Computational Geotechnics” 39/37

 Nicoll Highway Collapse R

15:33, 20 April 2004

Course “Computational Geotechnics” 40/37
 Nicoll Highway Collapse R

Course “Computational Geotechnics” 41/37

 Nicoll Highway Collapse R

• Undrained stability problem. Method A and Mohr Coulomb

constituive model used for design analysis

cu A

cu real
s, s’

Course “Computational Geotechnics” 42/37

 Nicoll Highway Collapse R

0.21s'v0 Section M3
Piezocone Data 100
100 NKT = 14
Profile: ABH-32
Best Estimate Line Test
Cu (Mohr-Coulomb)
AC-3 95 Cu (Geotechnical Design Parameter Table)
AC-2
Upper Marine Clay

3007
3008

90
90
Reduced Level, RL (m)

85

Elevation, RL (m)
F2
Method A
80
80
Lower Marine Clay

75
Original
Design Used in
FE Analyses
70
70

65

F2
60
OA
60 0 20 40 60 80 100
0 20 40 60 80 100
Undrained Shear Strength, s u (kPa) Shear strength (kPa)

Overestimation of shear strength: 43%-62.1%

Course “Computational Geotechnics” 43/37
 Nicoll Highway Collapse R

Method A Method B
Underestimation of wall displacements (about a factor of 2)
Course “Computational Geotechnics” 44/37
 Nicoll Highway Collapse R

Method A Method B
Underestimation of bending moment (about a factor of 2)
Course “Computational Geotechnics” 45/37
 Nicoll Highway Collapse R

Method A
(no collapse)

Predictions
(backanalysis)

Method B
(collapse)

Course “Computational Geotechnics” 46/37