Anda di halaman 1dari 5

Air cylinder

100

50

100

100

ESP

uini = 100 kPa

200

300

TSP

Undrained

400

TSP

Drained

300

Undrained TC test
Vc = 300 kPa

Mean stress, p and p' (kPa)

200

ESP

150

Mean stress, p and p' (kPa)

u = 100 kPa

Drained TC test
Vc = 300 kPa

400

Detail:
1. DST is usually for examining the shear strength of soil.
2. Container can be split horizontally at the mid-height.
3. Vertical stress is loaded from the top or bottom platens.
4. Shear stress develops by giving the horizontal load to the half of the box that is
not fixed stationary.
5. Soil is obliged to shear along the horizontal shear zone.
6. The states of stress and strain cannot be determined completely. Only Vn and W
are known; therefore, it is impossible to construct a unique Mohrs circle of
stress.

Direct shear test

100

200

300

Stress paths

Cell pressure is measured by a pressure transducer connected to the cell water


(i.e., not the specimen) and axial force by a load cell

Operation:
1. The sealed specimen is enclosed by a chamber or cell, then the water is filled
into the cell;
2. Cell pressure supplies an equal uniform radial and vertical stress to specimen;
3. Axial force is applied by a piston passing through the top of the cell

Triaxial apparatus

Gear

Stress-controlled: the state of stress is changed (e.g., by air cylinder) and the
changes in the state of strain is observed.

Strain-controlled: the state of strain is changed (e.g., by gear) and the changes in
the state of stress is observed.

Control of loading

Deviator stress, q (kPa)

Fundamental Laboratory test

Deviator stress, q (kPa)

Fa is not equivalent to Va but the Va-Vr,


which is equal to deviator stress, q

A=

V V0 -V V0 1- V V0
=
=

L L0 -L L 0 1- L L 0
1- V V0
1- v
=A 0
=A 0

1- a
1- L L0

A 0 = V0 L 0

L=L0 -L

V=V0 -V

If the initial A, V and L is A0, V0 and L0, respectively:

As the specimen deforms, its dimensions change.


Therefore, the cross-sectional area change during a test.

Fa A = a - r

a = r + Fa A

0.0

-0.2

-0.4

-0.6

-0.8

-1.0

-1.2

0.0

0.2

0.4

0.6

0.8

1.0

1.2

100

10

12

14

16

0.2

0.0
0

, Dr (%)

Ticino

, Dr (%)

10

12

14

25 , 0.602 , 96.80
50 , 0.598 , 97.71
100 , 0.603 , 96.51
200 , 0.593 , 99.24

Vn (kPa) , e0

Ticino

25 , 0.602 , 96.80
50 , 0.598 , 97.71
100 , 0.603 , 96.51
200 , 0.593 , 99.24

Vn (kPa) , e0

Shear displacement, s (mm)

-0.2

-0.4

-0.6

-0.8

-1.0

-1.2

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

600

Ticino sand (Italy)

, Dr (%)

Hostun

25 , 0.635 , 96.62
50 , 0.638 , 95.92
100 , 0.631 , 97.56
200 , 0.627 , 98.43

Vn (kPa) , e0

500

400

590

Hostun sand (France)

, Dr (%)

Hostun

25 , 0.635 , 96.62
50 , 0.638 , 95.92
100 , 0.631 , 97.56
200 , 0.627 , 98.43

Vn (kPa) , e0

300

490

398

Normal stress, V and V' (kPa)

200

300

Total stresses: drained test

qmax =98 kPa;


V3 = Vc =300 kPa (constant); and
u = 232 kPa (at failure)
Thus,
V1 = 398 kPa;
V3 = 300 kPa
V1 = 166 kPa;
V3 = 68 kPa

Shear displacement, s (mm)

166

Effective stresses:
drained test

68

200

Total stresses:
undrained test

Effective stresses:
undrained test

Direct shear test result

-200

-100

100

200

qmax = 290 kPa;


V3 = Vc =300 kPa (constant); and
u = 100 kPa (constant)
Thus,
V1 = 590 kPa;
V3 = 300 kPa
V1 = 490 kPa;
V3 = 200 kPa

16

When qpeak exhibits or qmax is attained, specimen continues straining without further
changes of stress, volume (drained test) or pore pressure (undrained test)
Drained test:
Undrained test:

Failure in TC test

Triaxial apparatus

Drained test: the valve is opened and the pore pressure remains constant while the
water will be expelled from or sucked into soil specimen (measured by burette).
Because pore pressure remains constant, changes in total stress and changes in the
effective stress are identical (GV = GV+Gu (=0)).

Undrained test: the valve is closed and the soil mass and volume are constant ('Vw =
0) but pore pressure will change (measured by pressure transducer).

To control the drainage during a test,


specimen must be sealed with membrane.
In addition, usually: 1) pressure transducer;
2) valve; and 3) burette are connected in
series.

We control:
1. Effective stress by controlling both the
total stress and pore pressure; and
2. Volume of the specimen by controlling
the volume of water expelled from or
sucked into the specimen.

Control of pore pressure and drainage

Modelling of soil behaviours.

Also, monitoring for the states of stress and strain and the pore pressure are
performed continuously during the loading.

Soil specimen is placed in a loading machine and the state of stress is changed.

Laboratory soil tests are performed:


1. to classify soils into groups before performing more detailed tests;
2. to examine the permeability;
3. to examine the compressibility by consolidation test
4. to examine the mechanical behaviour, in particular, the strength and deformation
during loading.

Shear stress, W and W' (kPa)

Vz
Vy

Wzy

Wxz

Wzx
Wyx
Wyz

Vy

Back
pressure

TE does not necessarily means that tensile stresses


exist in the specimen.

Typically, specimen will be isotropic consolidated to Va =


Vr = Vc. Then, the loading piston is moved to compress
or to pull the specimen. Hence, we get:
1. Triaxial compression (TC): Fa is compressive and Va
> Vr; therefore, Va = V1 and Vr = V2 = V3.
2. Triaxial extension (TE): Fa is tensile and Va < Vr;
therefore, Va = V3 and Vr = V1 = V2.

Therefore, the radial or horizontal deformation can be


calculated from axial deformation and volume change.

Volumetric change is measured by a burette.

Axial deformation is measured by the movement of the


loading piston by using LVDT or dial gauge and so on.

Cambridge type
Detail:
1. One pair of platens are allowed to rotate.
2. Horizontal direct strain, Hb = Hc = 0.
3. For NGI type, only Vn and W on horizontal plane are known; therefore, the state
of stress in soil cannot be determined uniquely.
4. For Cambridge type, Vn and W for all faces are known; so the state of stress.
Mohrs circle of stress can be constructed.
5. In both types, Hc = 0 while axial and shear strains are calculated from the
displacements and rotations of the platens. Thus, the state of strain is
completely defined.

NGI type

Simple shear apparatus

Detail:
1. Specimen is a disc (e.g., I60 x 20 mm), contained within a rigid metal ring and
loaded from the top via a rigid plate.
2. State of strain is 1D with Hr = Hb = Hc = 0.
3. Axial stress is principal stress.
4. Top and bottom platens consist of porous stone discs; therefore, pore water flow
is also in the axial direction (1D).
5. Pressure in porous stone disc is zero but pore pressure within the body of soil
specimen.

Oedometer apparatus

Connect to valve,
pressure transducer
and soil specimen

Loading
piston

LVDT

Triaxial apparatus

Specimen boundaries are principal planes of stress and of strain and they are coincided.

Vx

Wxy

Vz

However, when near to a slope or a structure, there will


be a completely general state of stress

Below the level ground, the axisymmetry is relevant: i.e.,


Vx = Vy and Wxy = Wyz = Wzx = 0.

Classification of soil tests

Vx = Vy; and
Wxy = Wyz = Wzx = 0.

Vx

Level ground

Stress in field

Wzy
Wxy
Wxz

Wyx
Wyz

Vy

Retrieve
Wxz

Wzx
Wyz

Wyx

Laboratory

Vx

Wxy

Wzy

Vz

Vy

Hvol
(%)

10

Compression

Drainage

30

No significant
change in Hvol

u = 100 kPa

Drained TC:
V3 = 100 kPa

40

In standard oedometer tests, Vr cannot be


measured and tests are usually performed
in a stress-controlled scheme.

a = -L L 0

a = Fa A

Stress and strain calculation:

q=Fa /A

Deviator stress, q:

1- V V0
1- v
A=A 0
=A 0

1- a
1- L L0

Instantaneous A:

a = -L L 0

vol =- V V0 = Vw V0

Volumetric and axial strains:

V0 =A 0 L 0 =88.4610-6 m3

A 0 = 2 4 =1.13410-3 m 2

Initial specimens cross-sectional area


and volume

Cell pressure: Vc = 300 kPa;


Back pressure: u = 100 kPa
Specimen: I38 x 78 mm

CVE661 Advanced Soil Mechanics

290

290

265

200

100

q
(kPa)

Compression

Excess pore water pressure 'u


Effective stress Va = Va - 'u

Total stress, Va

1.579

1.360

1.228

1.175

1.152

1.134

A
(m2 x 10-3)

No significant
change in q

35.00

24.50

15.00

7.50

2.50

Axial strain, Ha (%)

20

9.50

9.50

Ha
(%)
0.00

Oedometer apparatus

10

100

200

8.4

300

8.4

-27.3

7.99

7.07

-11.7
-19.11

458

325
394

4.21

3.72

-5.85

0.99

0.88

-1.95

0.00

235

0
115

'Vw
(mm3 x 103)

'L
(mm)

Fa
(N)

Drained TC test

In this class of soil tests, we usually know only the normal stress and the
corresponding shear stress on the boundary or cross-section of the specimen
while the principal stress value and its direction are needed to be calculated or
unknown.

Specimen boundaries are usually not the principal planes of stress nor strain.

Classification of soil tests

Generally, laboratory tests are performed under limited conditions of boundary.

However, it is extremely difficult.

The apparatus should be able to impose a completely general state of stress and
to change the state of stress so that principal planes may rotate during loading.

Ideally, soil specimen from the field should be transferred to a laboratory


apparatus without disturbance, load or unload the stresses.

Field

Vx

Wzx

Vz

Simulation of general state of stress in the field

Deviator stress, q = V'1-V'3 (kPa)


Volumetric strain, Hvol (%)

If flexible diaphragm was used, no shear stresses will be


transmitted to the soil; therefore, the in-contact plane is
the principal plane in the soil and the principal stress is
approximately uniform across the contact area. However,
direct strain measurement is difficult.

If there are local load transducers installed with the rigid


platen, variation of stress across the contact area can be
measured.

Normal and shear forces are applied to the platen and


they are transmitted to the soil specimen.
If the contact stresses are uniform, their values can be
directly calculated from the applied loads and the area of
soil that is in contact. And, the corresponding strains are
obtained by measuring the movements of platens.
By definition, if there is no shear stress between soil and
platen, normal stresses are the principal stresses.

-19.5

136
148

25

18

12

5.5

2.5

Ha
(%)

1000

20

15

10

LL= 43.6 %
PL= 23.6 %
PI= 20.0 %

Axial stress, Va (kPa)

10

Strain-controlled

25

Axial stress, Va (kPa)

100

Stress-controlled

LL= 43.6 %
PL= 23.6 %
PI= 20.0 %

100

Strain rate =
0.05 %/min

wc= 42 %

-1

-1

Terminology:
Consolidation = time-dependent process of volume change in soil as pore water
is expelled.
Compression = the relationship between the volume of the soil and the effective
stress, which is a relationship independent of time.

Volume changes in soils are due to exclusively changes in the effective stress;
therefore, if the volume of a soil changes, the effective stress must also change.
The effective stress changes with time as the soil volume changes due to the
flow of water from the pore towards the boundaries.
As the flow of water also depends on the hydraulic gradient and the rate of flow,
they affect on the changes of effective stress.
Variation of soil volume with time will be governed by complex interactions
between effective and total stresses, pore pressure, seepage, and
compressibility.

1000

Constant-rate-of-strain test

Fujinomori

p'c

Cc*

10

wc= 42 %

Standard consolidation test

30

q=Fa /A

Deviator stress, q:

A=A 0 1- L L0 =A 0 1- a

Instantaneous A:

a = -L L 0

vol =0

Volumetric and axial strains:

A 0 = 2 4 =1.13410-3 m 2

Initial specimens cross-sectional area

Cell pressure: Vc = 300 kPa;


Initial pore pressure: u = 100 kPa
Specimen: I38 x 78 mm

Fujinomori

Cs*

Introduction

25

20

15

10

u = 100 kPa

No significant
change in u

20

98

98

96

80

50

q
(kPa)

Drained TC:
V3 = 100 kPa

No significant
change in q

1.512

1.383

1.289

1.200

1.163

1.134

A
(m2 x 10-3)

Axial strain, Ha (%)

10

232

232

224

200

Oedometer test result

100

200

0
300

100

-9.36
-14.04

124

100

0
-1.95
-4.29

165

u
(kPa)

'L
(mm)
0
58
96

Fa
(N)

Undrained TC test

Detail:
1. Height is usually twice the diameter;
2. Top and bottom ends are rigid and assumed to be smooth or made lubricated;
3. Specimen is enclosed with rubber membrane and sealed to the top and bottom
ends by O-rings or rubber bands;
4. Rubber membrane acts as flexible boundary and as a seal to separate pore
pressure and total stress (i.e., effective stress);
5. Specimen is assumed to remain a right cylinder; so, the vertical sides of
specimen are principal planes of stress and strain.

Triaxial apparatus

Boundary condition

Deviator stress, q = V'1-V'3 (kPa)


Pore pressure, u (kPa)

Introduction

Axial strain, Ha (%)

CVE661 Advanced Soil Mechanics

Stress ratio, WV

Normal displacement, d (mm)

Stress ratio, WV


Normal displacement, d (mm)

Axial strain, Ha (%)

V0

Stage I

u0

V0

Stage II

u0+'u

Stage III

'Vt

Stage IV

uf=u0

Stage IV:
Total stress = V0+'V
Pore pressure = u0
Effective stress = V0+'V
Volume = V0+'V
Settlement = 'Uf

t0

'Vt

u0+'ut

V=-Vw =-

V0+'V

Time

V0+'V
V0+'V

Detail:
The resistance to flow of water due to
the valve represents the resistance to
flow of water past the soil grains. So, the
valve opening degree models the
permeability of a soil.
The stiffness of the spring represents
soil compressibility but unnecessarily
linear or reversible.
The piston has unit area: F = V = total
stress
If water pressure = u, the force in the
spring = (V-u) = V = effective stress.
If the piston settles by 'U, the volume
within the cylinder changes by 'V, we
obtain:

From curve fitting, we get: O = 0.2. Subsequently, 1 = 3.27

At the start of test, Vv= 30 kPa, u= 0, Q= 2.67, thickness z = 20 mm.


Find: K0, O and 10
Solution: When u= 0, the sample is in equilibrium and total and effective stresses
are equal. K 0 = ch cv = h v =0.5
2
p is obtained from Vv as: pc= cv
3
For 1D: = z z z= 2.67 20=0.134mm -1

Mathematic on 1D compression

K 0 =1-sinc

K0 for NC soil can be approximately determined by


the Jakys equation as:

K0 = coefficient of earth pressure at rest, varying


with overconsolidation ratio

ch =K 0 cv

During 1D compression, Vh will vary with Vv as:

1D test is substantially easier to perform than


isotropic compression test.
As sediments are deposited to form soils, they are
compressed by the weight of successive layers of
sediment.
Because of symmetry, there will be no horizontal
strain in the ground and the state of stress will
correspond to the state of stress in 1D compression.

One-dimensional compression

If isotropic compression tests were performed by repeatedly unloading and


reloading on clay sample, the results can be represented by a single line AC
together with a family of unloading-reloading lines: e.g., D1B1 and D2B2

Isotropic compression of clay

Stage IV:
After a very long time, the hydraulic
gradient and the flow of water through the
valve will be negligible and the model is
once again in equilibrium with zero
excess pore pressure.

Volume

V0

Stress

Compression and consolidation

Compression and consolidation

V0

V0

Stage II

u0+'u

t0

Stage III

'Vt

'Vt

u0+'ut

Stage IV

uf=u0

V0+'V

Time

V0+'V
V0+'V

Detail:
The level of water in the standpipe
(h0+'h) measures the pore pressure
(u0+'u) at the mid-height of the cylinder.
If the valve is open, the system is in
equilibrium ('h=0) and u=u0. This is
known as steady-state pore pressure
and will not change with time.
If 'h is above or below its equilibrium
position, there will be an excess pore
pressure 'u=Jw'h.
If there is 'u and the valve is open,
water will flow and 'u will diminish;
therefore; 'u will change with time.
After an infinite time, 'u will become
zero, the system is once again in
equilibrium ('h=0), the final pore
pressure is the steady-state one.

pcm
pc

p reminds that Rp is the ratio of mean normal stress


Rp always equal to or more than 1.0

qc=cv 1-K 0

cvm
cv

pc
pc

m v =/cv

=-

Clearly, if O is constant, mv cannot be


constant either. Therefore, vol =m v cv
is valid only for the relatively small
increments of stress for which the NC
line may be assumed to be linear.

Therefore:

From:

cv cv = pc pc

For 1D normal consolidation, K0 is a constant; therefore

=-m v cv

Alternatively:

Kaolin

For 1D compression, an increment of volumetric strain GHvol by a change of effective


vertical stress GVv is related by:
vol =m v cv

Coefficient of volume compressibility, mv

R0 is not generally equal to Rp

R0 =

For 1D compression, overconsolidation ratio, R0 is:

=-H

vol =- V V =- H H

Since the Hh=0 and Hvol = Hv, therefore:

For 1D compression, q and p are:


1
pc= cv 1+2K 0
3

One-dimensional compression

A soil is normally consolidated if its present state of stress has never been
exceeded.

Rp =

Overconsolidation ratio for isotropic compression:

Soil that is in equilibrium with 'u=0 and


its state lies on AC line is normally
consolidated (NC) and AC line is the
normal consolidation line (NCL).
If soil is in equilibrium with 'u=0 but its
state lies on a line such as BD, this soil
is overconsolidated (OC) and lines such
as BD are swelling lines.
The location of swelling line can be
defined by the maximum previous stress
pm corresponding to the point B.

Consolidation = Stage III (time-dependent)


Compression = Stages IV and I (equilibrium where 'u=0)

If we apply many increments of total stress, both loading and unloading,


allowing the excess pore pressures to dissipate completely before the next
increment applied, we may plot the volume against the effective stress for all
points where equilibrium is reached and the excess pore pressure is zero.

Stage I

u0

Overconsolidation

Volume

V0

Stress

Compression and consolidation

Compression and consolidation

V0

Stage I

u0

V0

Stage II

u0+'u

Stage III

'Vt
Stage IV

uf=u0

V0+'V

Time

V0+'V
V0+'V

vol =-

V Vw
=
V
V

The volumetric strain is:

p= c
pc= c -u

q=qc=0

The state of stress in isotropic compression is:

Stage I:
Total stress = V0
Pore pressure = u0
Effective stress = V0 = V0-u0
Volume = V0
Settlement = U0 = 0

t0

'Vt

u0+'ut

Along AB, the sample is 1D normally consolidated and K0 has a constant value.
For NC clay, K0 is typically near 0.7.
Along BD, the sample is 1D overconsolidated and K0 varies.
For large value of R0, K0 may exceed 1.0; therefore, Vh > Vv.

CS | 2.303

As the K0 is not constant during overconsolidation, we will have no simple


relationship between N and CS; however, if we assume that K0 is constant:

CC =2.303

As the K0 is constant during normally consolidation, we get:

and cv cv = pc pc

pcd
dpc
-=
From:

for overconsolidation

for normal consolidation

c d
d
=2.303 v
d lncv
dcv

de
d log10 cv
-CC =2.303

-CS =

de
d log10 cv

Next:

Swelling index:

Compression index: -CC =

In current geotechnical engineering practice, the 1D test result is typically plotted


as e vs. log10 Vv

Compression index and swelling index

Kaolin

One-dimensional compression

If NC soil was isotropically loaded from


A, the state will follow line AC.
Its state may be moved to the left of the
AC line by unloading along a swelling
line: e.g., BD.
However, it is not possible to move the
state of the soil to the right of the AC
line.
The line AC represents a boundary
between possible states to the left and
impossible states to the right.

Possible states of isotropic compression

pc = p-u 0

We could apply a constant back pressure u0 at the burette so as to saturate the


sample by dissolving any air and gas present in the pore water. In this case, the
equilibrium effective stresses before loading and at the end of consolidation would
be given by:

Isotropic compression test

Stage I:
The model is in equilibrium with the valve
open

Volume

V0

Stress

Compression and consolidation

V0

Stage I

u0

V0

Stage II

u0+'u

Stage III

'Vt
Stage IV

uf=u0

V0+'V
V0+'V

At small times, the hydraulic gradient


is relatively large; therefore, a large
value of slope.
After a very long time, the hydraulic
gradient should reach zero. Thus,
there should be no flow of water from
the soil and the volume change
should stop.

V0+'V

Time

Stage II:
Total stress = V0+'V
Pore pressure = u0+'u =u0+'V
Effective stress = V = V0
Volume = V0
Settlement = U0 = 0

t0

'Vt

u0+'ut

CVE661 Advanced Soil Mechanics

take ln

simplify

The position of N-line is not unique,


depending on the pm. If we defined QN as
Q of OC soil at p = 1 kPa on the swelling
line that corresponds to pm on the AC
line:
= -lnpc

Shear testing

The slopes and positions of the lines


depend on the particular clay.

Data from 1D compression tests on


different clays may usually be idealised
to similar straight lines.

d
pcd
=
d lnpc dpc
For O-line, if 1 is the Q at p = 1 kPa, we
obtain:
=-lnpc

-=

Along BD:

One-dimensional compression of clay


Kaolin

d
pcd
-=
=
d lnpc dpc

Along AC:

Mathematic on isotropic compression

Secondary compression is the time-dependent deformation under a constant


effective stress in soil.

However, additional small volume changes occur slowly as time passes; these are
due to soil creep and the process is often known as secondary compression.

Secondary compression

Stage II:
Total stress increases by 'V while the
valve is closed. As a result, 'Vw = 0 and
'V = 'U = 0. The piston and spring have
not moved and compressed. Therefore,
'V = 0. So, 'u = 'V.

Volume

V0

Stress

Compression and consolidation

V0

V0

Stage I

u0

Stage II

u0+'u

V = 1+ 0.4092.65 67.7 =0.031cm -3

qc 3 1-K 0
=
pc 1+2K 0

=N 0 -N lnpc

(for swelling line)


During NC, K0 is constant; thus, q/p is
constant along NC line. As a result, there
will be a family of O-lines each
corresponding to a particular value of q/p.

Soil is usually subjected to not only the normal stress but also the shear stress.
Therefore, we need to study the response of soils to combinations of shear and
normal stresses.
The feature of shear testing apparatus is that soil specimens can be subjected
to known and controllable combinations of shear and normal stresses while any
strains and distortion are observed.

Introduction

For 1D compression:

For isotropic compression: qc pc=0

Similar to isotropic compression, we can


define:
(for NC line)
= 0 -lnpc

If data from 1D and isotropic compressions


are superimposed in Q:ln p space, the
corresponding slopes during normally
consolidation and overconsolidation are
very similar.
Approximately, we can assume that both
NC lines are parallel with slope O and all
swelling lines are parallel with slope N.
The respective NC and swelling lines for
1D compression is always below the ones
for isotropic compression.

1D and isotropic compressions of clay

From curve fitting, we get: O = 0.2 and N = 0.05. Subsequently, 1= 3.32

Then, we get:

V0+'V

Time

V0+'V
V0+'V

Simplify

Stage IV

At the end of test when Vc= 60 kPa, V = 67.7 cm3 and w = 40.9 %.
And, Gs = 2.65. Find: O, N and 1
V =1 Vs =const
Solution:
=1+e=1+G s w

Mathematic on isotropic compression

The slopes of the lines and their


positions will be different for different
soils.

Stage III

'Vt

uf=u0

Stage III:
Total stress = V0+'V
Pore pressure = u0+'ut
Effective stress = V0+'Vt
Volume = V0+'Vt
Settlement = 'Ut

t0

'Vt

u0+'ut

Take ln

If the small loop in the unloading and


reloading cycle BoDoB is neglected, Qlog p relationships can be idealised by
two straight lines.

Kaolin

Isotropic compression of clay

Stage III:
The valve is open at t0, water begins to
flow and the piston sinks. At time t, the
settlement is 'Ut and the volume is
V0+'Vt. As the piston sinks, the spring
compresses; therefore, the effective
stress has increased to V0+'Vt and the
pore pressure has reduced to u0+'ut

Volume

V0

Stress

Compression and consolidation

Undrained plane

Drained tests

pcf =exp - 0
qcf =exp - 0

0 = f

NC clay is isotropically compressed to A


on NCL and then undrained TC test to
failure at B on CSL.
Because dQ = 0, test path remains on a
constant Q plane (i.e., plane ACDE),
undrained plane through A.
If the initial state and test condition are
known, we can locate a point on CSL at
which sample will fail.
Suppose that on NCL, we have p0 and
Q0 and we would like to find qf, pf and Qf
on CSL by standard undrained TC test:

Detail:
Each specimen is isotropically
compressed to p0 = a, 2a and 3a.
The Hvol Ha relations are similar.
q Ha curves are all of the same shape.
Specimen that is compressed to higher
p0 gives higher q at failure.
q Ha curves may be normalised by
plotting q/p0 Ha relation.

Normalised

Shape of q-Ha curve is similar to the one of undrained test on NC clay.


Shape of 'u-Ha curve is similar to Hvol-Ha curve of drained test on OC clay.
ESP goes to the right of TSP as 'u is negative

34.5

Undrained TC on OC clay (remoulded Weald clay)

827

Uses:
Determination of stress at failure on a
given plane or on pre-existing failure
surfaces.

Difficulties:
Complete state of stress in soil is
unknown.
Even Vn = FN/A and W = FS/A are
assumed uniform on the failure plane,
we get only single point in a Mohr stress
diagram.
Strains cannot be observed correctly as
the deformation is not uniform
throughout the sample but the sample
deforms largely in some region such as
ABCD.
It is therefore not suitable to the
interrelation between stress and strain.

The Vc = constant during shearing by


compressing the piston.
The 'u changes during shearing while the
volume is constant ('Hvol=0)
Total stress path also has slope= 3 as 'Va
increases while 'Vr=0.
If 'u is known during shearing, the effective
stress path can be drawn by following p = p-u.
The Q-p relation in an undrained test is simple,
being horizontal as 'Q=0.

Standard undrained TC test

Direct shear test

Vr,Hr

VT,HT

Advantages:
Axial, radial, and tangential stresses are
the principal stresses.
Usually, VT=Vr and HT=Hr.
Stress: Vr=VT=Vc and Va = Vc+(Fa/A).
Radial strain: Hr = (Hvol-Ha)/2.
Effective stress: Va=Va-u and Vr=Vr-u.

Drained plane

Drained tests

CVE661 Advanced Soil Mechanics

Constant p path:
q is increased while p is kept constant.
This can be done by reducing Vc at a
half of the rate that the axial stress, Va
is increased: 'Vr =-'Va/2; 'p=0.

Triaxial extension:
The Va is reduced while Vr and u are
kept constant; therefore, Va < Vr at
failure.
q = Va-Vr becomes negative and the
stress path is directed downwards with
slope dq/dp = 3.
This test is more specifically called
extension unloading as p is reduced.

f =-ln 3pc0 3-

pcf = 3pc0 3-

qcf = 3pc0 3-

qcf =pcf

qcf =3 pcf -pc0

NC clay is isotropically compressed to A


on NCL and then drained TC test to
failure at B on CSL.
Test path must lie on plane that is
parallel to Q-axis, have a projection in q
: p space with a straight line of slope 3,
drained plane through A.
If the initial state and test condition are
known, we can locate a point on CSL at
which sample will fail.
Suppose that on NCL, we have p0 and
Q0 and we would like to find qf, pf and Qf
on CSL by standard drained TC test:

Detail:
All test paths are straight in q : p space,
with the slope dq/dp = 3 from the
different initial p0 values.
Failures (i.e., dV = 0 or dHvol = 0 when
shear) occur at values of q and p which
define a straight line in q : p space.
The test paths in Q : p space are
curves, starting on the NCL and then go
to lower right as p increases and Q
decreases during drained shear.
Similar to undrained case, failure points
B1-B3 in Q : p space define a smooth
curve that has similar shape to the NCL.

Critical state line


and
Roscoe surface

Special drained triaxial test

Basic shearing modes:


TC: Fa>0; so, Va>Vr and therefore Va=V1 and Vr=V2=V3
TE: Fa<0; so, Va<Vr and therefore Va=V3 and Vr=V1=V2

Triaxial test

2
1 -3
3

s =

2
a - r
3

1
ca +2cr
3

Detail:
Lines of failure points on Q : p spaces
are similar for drained and undrained
tests.
Test data on NC Weald clay by Parry
(1960) shows that, at failure states of
drained and undrained TC tests, data
points define a single straight line in q :
p space and a single curve in Q : p
space.
This unique line or curve respectively
seen in q : p and Q : p spaces is called
Critical State Line (CSL).
Irrespective of different stress paths, the
failure will occur once the stress state
reach the CSL.

For different values of p0, there will be different drained or undrained planes.
The undrained test paths are helpful in defining the surface, for each test path
traces out a section of the surface at constant Q.
Families of undrained and drained tests on NC samples define the same 3D
surface in q : p : Q space?

The Roscoe surface

Critical state line

We need to unify the observed shearing behaviour of clay.


The observed behaviour for different loading histories or stress paths, undrained
and drained conditions should be able to unified.
We will firstly discuss only to NC clay.

Introduction

The Ha at failure is relatively large (> 20 %).


The Hvol is also relatively large (compressed).

1.632

207

By these definitions, q is positive for


TC but q is negative for TE

vol = a +2 r

s =

pc=

qc= ca -cr

To distinguish TC and TE

Drained TC on NC clay (remoulded Weald clay)

vol = 1 +2 3

1
1c +2c3
3

pc=

qc= 1c -c3

In terms of principal stresses and strains

Basic parameters:

Triaxial test

=-lnpc

Reminder, for NCL, we get:

=-lnpc

(Q p becomes linear when taking ln on p and


parallel to NCL; therefore, the same slope O can
be assumed.)

CSL on Q : ln p space:

qc=pc

CSL on q : p space:

Detail:
Each specimen is isotropically
compressed to p = a, 2a and 3a.
If p is on the NCL for a Q value, it is
denoted by pe (equivalent pressure).
Specimen that is compressed to higher
p gives higher q at failure.
However, shape of q Ha curves for
different initial p values are similar.
We can normalise q by the
corresponding pe value obtained at the
start of shearing, located on NCL; then
plot q/pe Ha relation.
Three q-Ha relations collapse into a
single q/pe Ha relation.

If NC samples are subjected to the same effective stresses differently by


undrained and drained TC tests, the specific volume, Q, is the same?
That is, at point B, Q by drained TC test along path ABC is the same as the one
by undrained TC test along path DBE?
If yes, undrained and drained tests define the same 3D surface in q : p : Q
space.
However, we need a large number of test.

The Roscoe surface

Critical state line

Normalised

Undrained tests

The Ha at failure is relatively large (> 15 %).


Because undrained, relatively large 'u is generated.
The generated 'u reduces p in the soil.

1.632

207

Therefore:
The Q is more appropriate than Hvol when
correlating the behaviour of soils having
widely different initial conditions as the
initial volume of a sample is significant
only for that sample alone.

If plotted by volumetric strain, Hvol:


The key feature that all samples
reached the same Q after large shear
displacements cannot be observed.
This is a lost simply due to that we
choose Hvol to plot instead of Q.

If plotted by specific volume, Q:


Samples with high initial Q compressed
while the ones with low initial Q dilated
during shear.
Importantly, all samples reached the
same Q after large shear displacements.

Undrained TC on NC clay (remoulded Weald clay)

SS test on
randomlypacked
steel balls

Uses of volumetric strain

The Roscoe surface

CSL in q : p : Q space

Undrained tests

A series of drained TC tests were


performed starting at different p0
values. By knowing Q at different stages,
we can construct contours of constant Q
in q : p space.
The undrained test paths in q : p space
themselves are the contours of constant
Q.
The two sets of contour are of the same
shape and consistent with each other?
If yes, undrained and drained tests
define the same 3D surface in q : p : Q
space.

As CSL is a function of q, p and Q, it is


sometime better to plot in 3D q : p : Q
space.
As q = 0 during isotropic compression
on NC soil, NSL is on the floor.
CSL rises as p increase and Q
decreases.

Detail:
Shapes of different stress paths are
similar.
Therefore, all stress paths could be
collapsed into a single q/pe p/pe
relation.
In Q p plot, for undrained condition,
the paths move from NCL to the left at
constant Q values.
Points at failure (e.g., B1-B3) define a
straight line in the q : p space and a
curve that is similar to NCL in the Q : p
space.

A clear peak exhibited at Ha = 8 %.


OC clay contracts initially but then dilates strongly. At the end of test, Hvol = -3 %.
Stress path rises to qmax then drops to the residual value.

34.5

827

The Vc = constant during shearing by


compressing the piston.
The 'u=0 during shearing, dq/dp =
dq/dp = 3
If the volume change during shearing is
known, we can plot Q-p relation.
Stress path is completely fixed in q:p
space with slope = 3.
However, path is not fixed in Q:p space
and depends on magnitude of the
volume change.

Drained TC on OC clay (remoulded Weald clay)

Standard drained TC test

B1,C1

CSL

A1

(pe)1

D2

D2

A2

Normalise

Henkel (1960) plotted contours of


constant water content in Va : 2 Vr
space for both undrained and drained
tests on Weald clay.
The two sets of contour are of the same
shape and consistent with each other?
If yes, undrained and drained tests
define the same 3D surface in q : p : Q
space.

pce =exp -
qc=gexp - +hpc

depends only on current Q and soil constants

We get:

-h pcf =gexp
+lnpcf

-
g= -h exp

Hvorslev surface intersects CSL at qf, pf


and Qf where:
qcf =pcf ; f =-lnpcf

We get:

From NCL:

From figure: qc pce =g+h pc pce


qc=gpce +hpc

At failure, both drained and undrained samples are moving towards the CSL at
rates which are related to the distance of samples from the CSL.
This above is applied to both OC and NC samples, even though the position of
CSL is established from the data of NC samples alone.
With continuing shear, individual elements of clay each proceed towards the CSL
after failure of the sample as a whole and eventually reach ultimate states on the
CSL.
The term ultimate can apply only to the state of a sample as a whole if the strains
remain essentially homogeneous throughout the sample.
The term ultimate state must be thought of as the final state with would be
reached by homogeneous shearing of an element or sample of clay in which the
arrangement of the particles remained essentially random.

Critical state line

-h exp

where: hpc is proportion to p

-
qc= -h exp
+hpc

Hvorslev surface is expressed as:

Hvorslev surface

For NC samples, by undrained and drained TC tests, the failures occur on the
CSL.
Roscoe surface, a state boundary surface, limits the states of NC sample in q :
p : Q space when they move from NCL to CSL.
What are the concepts that are necessary to be modified or to be extended for
OC samples?

(pe)3

A3

Find pe

B3

CSL

(pe)2

B3,C3

A1

B2,C2

NCL

Introduction

Q3

Q2

Q1

Shape of Roscoe surface

The Roscoe surface

Q2 < Q1

Weald clay

-
-h exp increases as Q decreases

So, if two specimens were at failure by
drained tests at the same p but different
Q1 and Q2, where Q1 > Q2, qf is higher for
Q2 than for Q1

Peak strength, qf, exhibits. Then, q


decrease with an increase in Ha.
q seems to decrease to a constant
value at the end of test; however, we
must check for the accuracy of the
measured stress and strain at large
shear strain values as sample is no
longer uniform.
The Hvol initially decreases then
increases substantially with Ha.
Also, Hvol seems to be constant at the
end of test; however, Hvol at large shear
strains is obtained after sample
becomes non-uniform.

Complete state boundary surface

Note:
Hvorslev and Roscoe state boundary surfaces intersect and the line of
intersection is CSL.

Working hypothesis:
Hvorslev surface is the state boundary surface for HOC samples in the same way
that the Roscoe surface is the state boundary surface for NC and LOC samples.

Hvorslev surface

Drained

Drained test on OC sample

The data from both undrained and drained tests presented in q/qe : p/pe
space show that the Roscoe surface is unique for all compression tests,
irrespective of the applied loading paths.

Shape of Roscoe surface

Summarily, curve surface in q : p : Q space traced out by families of drained


and undrained tests is the same.
This surface is called Roscoe surface.

The Roscoe surface


Undrained:
All test paths have the same shape in q
: p space, but of different sizes,
because the initial isotropic stress, pe.
If q and p are scaled down by dividing
by pe, all test paths would reduce to the
single curve.
CSL becomes a point in q/qe : p/pe
space

ultimate?

OC sample
Undrained TC

Up to F, strain in sample is
homogeneous. However, after F, strain
will be concentrated in the weaker
regions of the specimen (i.e., within
shear band).
Therefore, stresses and strains,
calculated from boundary
measurements and assumption that
specimen is uniform, are not accurate.

At the end of test, q is still decreasing


and Hvol is still dilating. Therefore, the
sample has not reached a steady-state
condition at which stresses are constant
and there is no change in volume.
Failure state = q reaches the maximum.
Ultimate state = state that large shear
strains can occur with no change in
stresses or in volume.
An ultimate state may or may not be
achieved at the end of test.

Test path moves above CSL to the


failure point in q : p space. Then, it
moves back along the same path
toward CSL.
Therefore, the state boundary surface
for OC samples must lies above CSL in
q : p space.
If a series of drained TC tests were
performed on OC samples, the failure
points in q : p space refer to different
constant Q sections of the state
boundary surface.
Yet, for each constant Q section, we
expect that the state boundary surface
has the same shape while is of different
sizes.
Therefore, we would better scale q and
p to account for changes in Q during
testing.
On the other hand, we could not see
that the failure point is a peak in the
case of undrained TC test on OC
sample.

If sample is tested undrained, the critical


state is the state at which the maximum
q can be sustained (n.b., no peak).
Therefore, undrained test path on HOC
sample would rise vertical then traverse
the state boundary surface until failure
occurred at the CSL, like the case of
LOC sample.
However, there is possibility that failure
of a sample occurs prematurely, soon
after sample reaches the Hvorslev
surface.

Complete state boundary surface (undrained)

Critical state line

OC sample
Drained TC

Drained test on OC sample

Undrained

Samples 2 4 were isotropically


consolidated to lower Q then swelled
back to the Q1 value. Therefore, samples
2 4 are overconsolidated.
Undrained tests were performed. Thus,
pe is the same for all samples.
The Roscoe surface coincides with the
test path for Rp = 1.0.
For samples 2 4, the staring points for
each test is below the Roscoe surface.
Lightly OC samples fail at the same
point as the NC sample, at the CSL.
The Roscoe surface forms a boundary
beyond which the test paths do not go.

Roscoe surface as state boundary surface

Shape of Roscoe surface

pce =exp -

Drained:
Test path moves through an infinity of
constant Q sections of the Roscoe
surface, each of different sizes.
Shape of each constant Q section is the
same for different values of Q.
Therefore, q and p from drained test at
a moment can be scaled down by pe on
NCL at that Q value.
pe is called the equivalent pressure.
At any value of Q, pe can be determined
from the equation for NCL as:

Contracting

CSL

HOC
sample

For drained test, Parry (1958) plotted


GHvol/GHs against pu/pf at failure.
If sample fails at critical state (pf = pu),
rate of volume change at failure is zero.
If sample fails with pf < pu (i.e., HOC
sample), volume is expanding at failure.
Thus, if sample fails at A, it is expanding
and moving towards the CSL at a higher
Q.
If sample fails at B with pf > pu, it is
contracting at failure and moving
towards the CSL at a lower Q.
Rate of movement towards the CSL at
failure increases as the distance
between the failure state of the sample
and the CSL increases.

1
qc=1c ; pc= 1c
3
? qc/pc= qc pce pc pce =3

Size of each constant Q section is


different for different Q values and is
proportional to pe.
Drained and undrained test results may
be compared directly in q/pe : p/pe
space.
At failure, data from both drained and
undrained tests lie on a single line.
The line is limited on the right by a point
representing CSL in connection with the
Roscoe surface.
If soil cannot withstand tensile effective
stress, the highest value of q/p is
corresponding to V3 = 0. Then, for TC
test:

As Q decrease, the size of the constant


Q section increases.
Point of intersection of the constant Q
sections and the drained plane moves
from NCL to Roscoe surface to CSL to
Hvorslev surface and away from CSL.

Complete state boundary surface (drained)

Expanding

Contracting

Expanding

Critical state line

Hvorslev surface

There is a similarity between the


Roscoe surface and the NCL.
For NCL, at a particular stress level,
samples cannot exist at a higher Q than
that on the NCL at that stress.
In fact, NCL is a part of the Roscoe
surface lying in the q = 0 plane.
The Roscoe surface separates states
which samples can achieve from states
which samples can never achieve; thus,
Roscoe surface is a state boundary
surface.

Roscoe surface as state boundary surface

Shape of Roscoe surface

A1

(pe)1

B2,C2

NCL

B1,C1

CSL

D2

A2

(pe)2

B3,C3

(pe)3

HOC
sample

Roscoe
surface

For undrained test, Parry (1958) plotted


Gu/pf)/GHs) against pu/pf at failure.
pf is used to normalise Gu of different
samples fail at different pf.
Rate of change in u at failure is largest
for samples which fail most furthest
away from the CSL and the sign of
change in u is such as to move the
sample toward the CSL.
If sample fails at A with pf < pu, it has a
negative u and is moving to the right
towards the CSL.
If sample fails at B with pf > pu, it has a
positive u and is moving to the left
towards the CSL.

The line at which failure points located


is AB. It is limited on the left by the line
OA having slope = 3, corresponding to
tensile failure.
Line AB = Hvorslev surface.
On Hvorslev surface, q is a function of
both p and Q of the specimen at failure,
unlike knowing only p for failure on
CSL.

a Q relation is the normal view of


drained plane itself.
There is a series of a Q relations for
different values of p0

Complete state boundary surface (drained)

Critical state line

Hvorslev surface

A3

CVE661 Advanced Soil Mechanics

Q3

Q2

Q1

The behaviour of overconsolidated samples:


The Hvorslev surface

Shape of Roscoe surface

Samples have been isotropically


consolidated to different maximum
pressures and then isotropically swelled
until they are at the same p0.
Therefore, along OA, samples will have
different OCR but drained test paths will
lie on the same drained plane.
Expectedly, OC sample will initially
deform so that Q decreases slightly as q
and p increase. Then, samples will
reach the state boundary surface (SBS),
and traverse across it towards the CSL.
CSL in not the point on the SBS which
has the highest value of q.
Drained test on OC sample will give
qmax at a stage just before the sample
moves down the Hvorslev surface
towards the CSL.
For real soil, the deformation after
failure becomes non-uniform and only
elements of clay in the shear band
move down the Hvorslev surface; the
remainder does not soften significantly.

The path of OC soil whose state lies


below the SBS must remain on a
particular elastic wall.
Therefore, the path followed by sample
of an OC soil will follow the line of
intersection of the elastic wall and the
appropriate drained or undrained
planes.

Elastic range

Therefore, plastic behaviour is specified by:


1. Yield surface;
2. Flow rule; and
3. Hardening law

Flow rule defines a relationship between


the gradient GHpa/GHpc of strain increment
vector and the stress vector, V.
Plastic potential is defined such that the
plastic strain increment vectors are
orthogonal to it.
Similarly, flow rule may be specified as a
relationship between a plastic potential
and a yield curve.
If the plastic potential and the yield curve
coincide, the material has associated
flow rule and normality condition applies
such that strain increment vectors are
normal to a yield curve.

Ha

Slope = E

Ha

Elastic range

Undrained TC

Va

Drained TC

Slope = E

Essential plasticity theory

Va

Consequently, for drained test on a soil in the elastic range in which p changes
during loading, the stress-strain behaviour is expectedly non-linear. On the other
hand, for undrained test on a soil in the elastic range in which Gp = 0, the stressstrain behaviour is theoretically linear.

Ec pc= 3 1-2c  constant

In many cases, the change in X during a loading path which causes only elastic
strains, and which therefore, remains on a particular elastic wall, is relatively small.
Therefore, we get:

Calculation of elastic strains

1
p'+0 q'
Kc
1
s =0 p'+
q'
3Gc

vol =

Plasticity for soils

Eu
3 1-2 u
Eu
2 1+ u

Ku =
Gu =

A sample whose state is anywhere on


the elastic wall can move everywhere on
the elastic wall, causing only elastic
strains.
Plastic strains occur only when the
sample state touches SBS.
Therefore, SBS serves also as a yield
surface.
A typical yield curve is given by the
intersection of an elastic wall with SBS.
If sample states lie on a particular elastic
wall, they are also lie on a particular
swelling line.

In terms of invariants of total stress, we obtain:


1
vol =
p+0 q
Ku
where:
1
s =0 p+
q
3G

3 = 1/E u > - u 1 - u 2 + 3 @

2 = 1/E u > - u 1 + 2 - u 3 @

1 = 1/E u > 1 - u 2 - u 3 @

From Hookes law for undrained loading in terms of total stress, we have:

u = - r a

E u = a a

For undrained loading, it is convenient to relate increments of strain to increments


of total stress, ignoring any pore pressure generated.
For undrained elastic behaviour in terms of total stress, we define an undrained
Youngs modulus Eu and an undrained Poissons ratio Qu. Therefore, for an
undrained TC test with 'V3 = 'Vr = 0, we get:

Calculation of elastic strains for undrained loading

Undrained test:
The path DG rises vertically from D to
G, which is on the SBS.
If loaded beyond G, plastic strain occurs
as state of sample traverses the SBS
along the intersection between the
undrained plane and the SBS towards
its ultimate state at F on CSL.
As GHvol = 0, Gp = 0 on the elastic wall.

Wyx
Wyz

Vy

9pc 1-2c
2 1+c

Va
Elastic range

Ha

Undrained TC

Plasticity for soils


Other elastic walls are each associated
with different swelling lines.
Therefore, yield curve for a sample is
obtained from the sample state in X: p
space.
If loading QS causes 'X, a new yield
curve in q : p space will applicable and
the sample has strain hardened as the
yield curve expands.

By working in terms of total stress, we avoid the need to calculate the pore
pressure. However, the total stress technique is valid only for the case when the
soil is saturated and undrained (no volume change). Otherwise, calculations must
proceed in terms of effective stress.

Because Gp = 0, Eu is a constant in the elastic range and undrained stress-strain


behaviour of elastic soil will be linear.

Eu =

From Qu = 0.5 we get:


Ec
Gc=
=G u = E u 3
2 1+c
Therefore:
3
Eu =
Ec
2 1+c

G u =Gc

Because Gq = Gq, we get:

For undrained loading, GHvol = 0 and therefore Ku is infinity Qu = 0.5

Slope = Eu

Drained test:
The path DG rises as a curve and
volume decreases as p increases
during standard drained TC test.
If loaded beyond G, plastic strains will
occur as state of sample traverses the
SBS along the intersection between
drained plane and the SBS towards its
ultimate state at F on CSL.

Hypothesis:
Plastic strains occur only when the
sample is traversing the SBS.
Therefore, for paths below SBS, the
strains are purely elastic.
On path DBCE, plastic strain occurs
during section BC which lies on the
Roscoe surface.

For soil:
Strains are recoverable (elastic) along
the swelling lines DB and EC.
Plastic strains occur over path BC which
is on NCL, a part of SBS.

Calculation of elastic strains for undrained loading

Calculation of elastic strains

Vx

Wxz

Calculation of elastic strains

Wxy

Wzx

Vz

elastic

flo

Wzy

elastic

c
sti

Gl

Elastic wall

NC sample A must fail without change


in Q at C on the CSL.
If OC sample B has the same initial Q as
the one of sample A and if the sample B
deforms uniformly, sample B also fails
at C on the CSL.
NC sample A fails with a large positive
u, while the OC sample B fails with a
negative u.

Pore pressure changes in undrained test

pla

dia.=d
Area = A

ic flo

For metal:
If V < Vy, the deformation is linear elastic
and strains caused by loading are fully
recovered by unloading.
If V > Vy, additional plastic strain occurs;
when unloading from G to B, some
(elastic) strain is recovered.
If reloading from B and V < Vg, the
deformation is linear elastic.
The Vy and Vg are called yield stresses.
Effects of plastic straining to increase
the yield stresses from Vy to Vg is called
strain hardening.

In Q : p space, locus of failure states of


the OC samples will lie to the left of the
CSL with the deviation from the CSL
increasing as Rp increases.
In q : p space, each OC sample fails
when it reaches the appropriate
constant Q section of the Hvorslev.
LOC samples fail on the CSL.

in

For ideal isotropic elastic soil:

plast
tain
con

Elastic wall

Same pmax but swell to


different p0 (i.e., swell
from pmax at point 5)

Complete state boundary surface (drained)

ta

Calculation of elastic strains

At the early stage of test, it is necessary to make a distinction between elastic


and plastic strains and to develop a criterion which determines whether a
particular loading path produces elastic or plastic strains.
By applying theories of elasticity and plasticity to the stress-strain behaviour of
soil, estimation of the shear and volumetric strains caused by loading can be
made.

Introduction

Higher Rp

Intersection of
drained plane
with the floor

Complete state boundary surface (drained)

n
co

Plasticity for soils

Constant Vc

Metal

Soil

we obtain:

2 1+c
qc
9pc 1-2c

Swelling lines are associated with elastic


walls. Therefore, the above equation for
GHpvol also gives the increment of plastic
volumetric strain from a state anywhere
on the elastic wall above DB to a state
anywhere on the elastic wall above EC.

pvol = - D = D -E D

=E -D

Magnitude of plastic strain is obtained by


comparing X at the same value of p.
For example, plastic volumetric strain
increment from D to E is obtained from:

Constant Va

Vc =0

(also valid for undrained loading path)

s =

Gc 3 1-2c
=
we obtain:
Kc 2 1+c
pc 3 1-2c
Gc=
2 1+c
Therefore:

From

Kc= pc

Therefore, bulk modulus K is:

vol = pc pc

From vol = -

=- pc pc

=k -lnpc

As DG is on a swelling line on Q : p space:

Calculation for Hs and Hvol for ideal isotropic


elastic soil in drained TC along DG:

General:
In fact, there are other paths by which
the sample could move from D to E; all
of them require that the sample moves
across the SBS.
In contrast, there are paths that can be
followed by sample at D without causing
plastic strain.

Constant p path:
If shearing from D to E at constant p, in
order to cause plastic strain, q
increases so that test path rises up and
touches the Roscoe surface at G, above
D. Then, test paths traverse the SBS to
K, above E.
q value at G fixes the value of q which
must be applied to the sample at D in
order to cause plastic strain.

wet of critical = sample that has a


moisture content higher than that of a
sample on the CSL at the same p.
dry of critical = the opposite to wet of
critical.
This categorization groups together
samples with similar pore pressure and
volume change behaviour.

The Q at F is substantially lower than


that at D; therefore, NC sample D must
compress during shear.
The Q at G is substantially higher than
that at E; therefore, OC sample E must
expand during shear.
OC sample E fails at H on the Hvorslev
surface before it reaches the CSL at G.
However, Q at H cannot be determined
from geometry alone.

Three essential features of plasticity theory:


1) Yielding; 2) Hardening; and 3) Flow
For a hypothetical metal-like material, we may assume:

Essential plasticity theory

Calculation of elastic strains

Elastic wall

Volume changes in drained test

Kc= pc = Ec 3 1-2c

All paths that remain on the curved


vertical plane above the swelling line
BD, but below the SBS, will cause only
the elastic strain.
This curved surface, BJIH is called the
elastic wall.
There are a number of elastic walls,
each associated with a particular
swelling line. That is, it depends on
loading history.
If state of sample is below the SBS, its
behaviour is assumed to be elastic and
stresses and strains may be related by
the theory of elasticity.
If state of sample lies on the SBS, both
elastic and plastic strains may occur
and plastic strains may be calculated by
the theory of plasticity.

s =

2 1+c
qc
9pc 1-2c

vol = pc pc

Plasticity for soils

Hardening

Hardening

Essential plasticity theory

Therefore:

Flow rule relates the gradient (GHps/GHpvol)


of the plastic strain increment vector QR
with the stress vector OQ.
The direction of the plastic strain
increment vector QR is independent of
the direction of the applied stress
increment but depends only on the
vector of applied stresses.
For soils, it is likely that different flow
rules apply for different parts of the yield
curve. Therefore, it is likely that one flow
rule applies to the portion of the yield
curve corresponding to the Hvorslev
surface and another flow rule, possibly
the normality rule, applies for the portion
corresponding to the Roscoe surface.

All states of stress at yield and at failure


lie respectively on a yield curve and a
failure envelop
If the stress state lies inside the yield
curve, the material behaviour is only
elastic.
The expansion of yield curve YaYc to
GaGc due to plastic strain along YG is
strain hardening or work hardening.
Hardening law relates the change of
yield stress with the plastic strain during
yielding.
If the state of sample traverses the yield
surface, the sample suffers plastic
strains, calculated by hardening law.
If the state of sample lies on failure
envelop, strain hardening ceases and no
elastic strain.

valid only for increments of loading sufficiently


small so that the value of E may be assumed
to be constant.

E will not be constant and the soil behaviour, even assumed that it is isotropic elastic,
will not be linear.

c= - r a

If we performed drained TC test in which 'V3 = 'Vr = 0, we get:


Ec= ca a

We get:
Ec= 3pc 1-2c

From:

Calculation of elastic strains

Elastic wall

Behaviour of soils before failure

CVE661 Advanced Soil Mechanics

Anda mungkin juga menyukai