Sei sulla pagina 1di 6

Joseph, P. G. (2010). Geotechnique 60, No. 10, 807812 [doi: 10.1680/geot.9.P.

001]

TECHNICAL NOTE

A dynamical systems-based approach to soil shear


P. G . J O S E P H 

This note models monotonic soil shear as a dynamical Cette communication modelise le cisaillement monotone
system. It provides additional information to support the du sol comme un systeme dynamique. Elle fournit des
hypothesis that for monotonic soil shear, the rates of informations supplementaires a lappui de lhypothese
change of the shear stress, effective normal stress and dapres laquelle, en ce qui concerne le cisaillement mono-
void ratio are proportional to the applied shear and tone du sol, les taux de changement de la contrainte de
effective normal stresses with the proportionality values cisaillement, la contrainte normale efficace et lindice de
decaying with strain to become zero at the steady-state vide sont proportionnels aux contraintes de cisaillement
condition. This hypothesis provides close fits to stress et efficaces normales appliquees, les valeurs de propor-
strainvoid ratio curves from undrained shear tests of tionnalite sestompant au fur et a mesure de la deforma-
uncemented, resedimented clays and drained shear tests tion, et sont nulles a letat dequilibre. Cette hypothese
of uncemented sands and silts, using triaxial and true- fournit des resultats proches des courbes contraintes/
triaxial equipment, and various stress paths. For un- deformation /vide relatives aux resultats dessais de ci-
drained shear, model parameters vary smoothly with saillement non draine sur des argiles non cementees et
over-consolidation ratio; for drained shear, they vary in re-sedimentees et dessais de cisaillement draine sur des
an orderly way with relative density. The models value sables et des boues non cementes, en utilisant des appa-
lies in that a well-formed hypothesis, derived from the reils triaxiaux et triaxiaux vrais ainsi que differents
steady-state condition, provides a simple, alternative ap- chemins de contrainte. Pour le cisaillement non draine,
proach to current complex elastoplastic models based on les parametres du modele varient de facon reguliere avec
critical state theory. lOCR. Avec le cisaillement draine, ils varient de facon
ordonnee avec la densite relative. Linteret de ce modele
est quune hypothese bien concue, derivee de letat
dequilibre, fournit une methode simple qui se substitue
KEYWORDS: clays; constitutive relations; elasticity; plasticity, aux modeles elasto-plastiques complexes actuels, qui sont
sands; silts fondes sur la theorie de letat critique.

INTRODUCTION For the triaxial tests, q ( a  r )=2 and p


Joseph (2009) presents results of modelling drained and ( a r )=2 were used in fitting equations (1), while for the
undrained monotonic shear as a dynamical system. The true-triaxial tests, q 1=3[( 1  2 )2 ( 2  3 )2 ( 3 
current note presents additional supporting data, analysing 1 )2 ]1=2 and p ( 1 2 3 )=3 were used.
66 additional drained and undrained tests from various test
programmes on uncemented clays, silts and sand, and then
using a calibrated model to predict stressstrain curves for a APPLICATION OF THE MODEL TO UNDRAINED
new, independent data set. SHEAR
From Joseph (2009) The current paper applies equations 1 to 33 undrained
tests from Plant (1956) and Gens (1982). Plant conducted a
dq=d pAe  B  qJ e  D (1a) total of 18 undrained triaxial compression tests on resedi-
 B  D mented samples of London clay Eocene series high-
dp=d pLe  qKe (1b)
plasticity (CH) clay deposited in a marine environment
 B  D
dv=d pN e  qMe (1c) isotropically consolidated to different over-consolidation ra-
tios (OCRs). He typically reported 17 data points per test.
where A, L, N and J, K, M are the proportionality constants The coefficient of correlation of fit of equations (1) to the
and represent the initial measures of the resistance offered test data averaged 0.98 and 0.94 for q and p respectively.
by the soil structure to p, the maximum normal stress, and Gens conducted 15 undrained triaxial compression tests
q, the maximum shear stress respectively; D, B are the on resedimented samples of lower Cromer till a low-
exponential rates at which these proportionalities decay with plasticity (CL) sandy clay isotropically consolidated to
strain to zero. The change in sign for the void ratio in different OCRs. Gens typically reported less than 15 data
equation (1c) is so that a decrease in is positive. Joseph points per test. The coefficient of correlation of fit of
(2009) applied equations (1) to data from Sheahan (1991) equations (1) to the test data averaged 0.99 and 0.98 for q
and showed that they fit the observed data well. and p respectively.
Figure 1 shows typical fits of the model to q and p
respectively for sample tests from Plant and Gens, chosen so
Manuscript received 31 December 2008; revised manuscript
accepted 10 December 2009. Published online ahead of print 8 that their coefficient of correlation values were close to the
June 2010. average of their respective test sets. Fig. 2 plots against
Discussion on this paper closes on 1 March 2011, for further OCR the average parameter values from all of Plants tests,
details see p. ii. with the exception of the single test U7 at an OCR of 32,
 Engineering Solutions, Boxboro, Massachusetts, USA. and two outlier values for B and D from tests U10 and U14.

807
Delivered by ICEVirtualLibrary.com to:
IP: 130.102.42.98
On: Sun, 10 Oct 2010 16:49:35
808 JOSEPH
20 Gens IC-1-11 q/vc measured 45 A 05428 OCR05257
R2 093
Gens IC-1-11 p/vc measured
Plant U11 q/vc measured B 01053 OCR01589
18 R2 085
Plant U11 p/vc measured 40 J 01823 OCR0310
p/vc,q/vc calculated R2 083
16
D 00988 OCR01631
R2 098 R2 098

14 35 K 1212 OCR02461
R2 084

L 02557 OCR06769
12 R2 093

R2 096 30
p/vc,q/vc

10
R2 099
25

A, B, J, D, K, L
08

06
20

04 R2 098

15
02

0
0 10 20 30 10
Strain: %

Fig. 1. Average measured plotted against calculated stress


strain curves from Plant (1956) and Gens (1982) 05

Fig. 3 plots against OCR the average parameter values for


the tests by Gens.
0
Figures 2 and 3 show that the parameters of equations (1)
0 2 4 6 8 10 12 14 16 18
for the Plant and Gens data did not correlate as well with
OCR
OCR when compared to the Sheahan data analysed by
Joseph (2009). The correlations of the parameter values with Fig. 2. Model fit for Plant (1956) data
OCR decreased with the number of measurements reported
per test an average of 17 points per test for Plant and less
than 15 for Gens, as opposed to the average of 38 reported 35.78 as opposed to 32.88 for tests with thick lubricated
by Sheahan. ends. Shapiro observed little visual difference in volumetric
behaviour for the two sets of tests. For the purposes of this
analysis, it was considered useful to analyse how different
APPLICATION OF THE MODEL TO DRAINED SHEAR end conditions during a test would influence the values
The current paper also applies equations 1 to 33 drained determined for the model parameters. All tests on the sands,
tests from Shapiro (2000). Shapiro tested Nevada II sand both triaxial and true-triaxial, used thick lubricated ends.
and a mixture of Nevada II sand with 20% by weight of For the triaxial tests the average coefficients of correlation
ATC silt. The Nevada II sand was uniform fine quartz. The were 0.98 for q, 0.98 for p and 0.995 for for the pure
ATC silt was non-plastic quartz with a specific gravity of sand, and 0.99 for q, 0.97 for p and 0.99 for for the sand
2.74. Shapiro conducted four drained triaxial tests on the silt mixture. For the true-triaxial tests, they were 0.99 for q,
sand and eight drained triaxial tests on the sandsilt mixture 0.99 for p and 0.89 for for the pure sand, and 0.997 for q,
at nominal initial confining stresses of 50, 100 and 200 kPa 0.997 for p and 0.99 for for the sandsilt mixture.
and at a single nominal density. Nine of these triaxial tests Figures 46 show typical fits of the model to , and
were compression tests, while three were extension tests. respectively for drained compression and extension tests on
Shapiro also conducted 21 drained true-triaxial tests at dif- sand and silt materials for sample tests. Fig. 7 likewise plots
ferent b values, 16 on the silty sand and five on the sand. average parameter values from Shapiros triaxial compression
In both test series, Shapiro used lubricated ends to reduce tests. Table 1 lists the parameter values for the extension
end restraint. These consisted of a greased latex sheet be- tests, which were all conducted at a confining pressure of
tween specimen and end platens. For four of the eight 50 kPa and at the same nominal density. Fig. 8 plots against
triaxial tests on the sandsilt mixture and for eight of the 16 the stress ratio b, average parameter values from Shapiros
true-triaxial tests on the sandsilt mixture, Shapiro used true-triaxial tests on silty sand and Fig. 9 plots these values
thin lubricated ends to investigate end effects. For all the for the sand. The various sign changes are consistent with
remaining tests, he used thick lubricated ends. Thin the direction of shear.
lubricated ends were 0.015 cm thick, while thick lubricated In general, for the soils tested, the models parameters
ends were 0.030 cm thick. Little difference between the two varied in an orderly manner, with similar trends for sand,
ends were found for the extension tests, and for the com- silty sand and silt. Usually, absolute parameter values for the
pression tests, the main difference was that tests with thin sand were higher than for the silt. Absolute parameter values
lubricated ends had a slightly higher peak friction angle of tended to increase with relative density and end platen

Delivered by ICEVirtualLibrary.com to:


IP: 130.102.42.98
On: Sun, 10 Oct 2010 16:49:35
A DYNAMICAL SYSTEMS-BASED APPROACH TO SOIL SHEAR 809

17 A 178 OCR04753 R2 087 500


B 069 OCR03223 R2 047
16
J 455 OCR02097 R2 057
D 093 OCR04581 R2 056 R2 099
15 450
K 611 OCR05006 R2 044
14 L 039 OCR07628 R2 056

13 400

12
350 R2 090
11

10
A, B, J, D, K, L

300

Shear stress: kPa


9

8
250
7

6 200

5
150 R2 099
4

3
Ttx-NII-50-200-ATC270-20-CD-8
100 (true-triaxial test on silty sand)
2
Tx-NII-50-200-ATC270-20-CD-8
1 (triaxial test on silty sand)
R2 099 50 Tx-NII-50-200-CD-4
0 (triaxial test on sand)
Ttx-NII-50-200-CD-10
(triaxial test on silty sand)
0 5 10 15 20 0 Calculated values
OCR

Fig. 3. Model fit for Gens (1982) data 30 20 10 0 10 20 30 40 50


Strain: %

300 Fig. 5. Measured plotted against calculated normal stressstrain


curves for four tests from Shapiro (2000)
080
R2 099
250
R2 099
R2 099

200

075
R2 099
Shear stress: kPa

150
Void ratio

100 R2 088
R2 099

R2 099
070
50

Ttx-NII-50-200-ATC270-20-CD-8 Ttx-NII-50-200-ATC270-20-CD-8
(true-triaxial test on silty sand) (true-triaxial test on silty sand)
Tx-NII-50-200-ATC270-20-CD-8 Tx-NII-50-200-ATC270-20-CD-8
(triaxial test on silty sand) (triaxial test on silty sand)
0 Tx-NII-50-200-CD-4 Tx-NII-50-200-CD-4
R2 099 (triaxial test on sand) (triaxial test on sand)
Ttx-NII-50-200-CD-10 Ttx-NII-50-200-CD-10
(triaxial test on silty sand) (triaxial test on silty sand)
Calculated values
Calculated values
50 065
30 20 10 0 10 20 30 40 50 30 20 10 0 10 20 30 40 50
Strain: % Strain: %

Fig. 4. Measured plotted against calculated shear stressstrain Fig. 6. Measured plotted against calculated void ratiostrain
curves for four tests from Shapiro (2000) curves for four tests from Shapiro (2000)

Delivered by ICEVirtualLibrary.com to:


IP: 130.102.42.98
On: Sun, 10 Oct 2010 16:49:35
810 JOSEPH
05

04

A
03

02

004
003
002
B

001
0
004
003
002
D

001

0
08
07
06
J

05
04
03
08
07
06
K

05
04
05

04
L

03

02

00012
M

00008

00004

0
00006

00004
N

00002

0
50 100 150 200
Normal stress: kPa

Silty sand thin ends (C) Silty sand thick ends (C) Sand thick ends (C)

Fig. 7. Model fit for Shapiro (2000) standard triaxial compression tests

lubrication thickness, and decrease with increase in initial


Table 1. Parameter values for Shapiro (2000) triaxial extension
tests (initial confining stress 50 kPa) confining stress. For the triaxial tests, parameter values and
the effects of initial density, end platen lubrication thickness
Test Silty sand Silty sand Sand and sand as opposed to silt decreased as the initial confining
(thin ends) (thick ends) (thin ends) pressure increased; for true-triaxial tests, they decreased
similarly as the stress ratio moved from zero (compression)
A 0.427 0.747 0.469 to one (extension).
B 0.021 0.012 0.596
J 0.902 0.132 1.209
D 0.013 0.044 0.787 APPLICATION OF CALIBRATED BOSTON BLUE CLAY
K 0.900 0.068 1.209
MODEL
L 0.426 0.359 0.469
M 0.003 0.0003 0.002 Fayad (1986) conducted undrained shear tests at four
N 0.002 0.001 0.0007 OCRs on resedimented, Ko consolidated samples of Boston
Blue clay. Fayad used manual stress increments to maintain

Delivered by ICEVirtualLibrary.com to:


IP: 130.102.42.98
On: Sun, 10 Oct 2010 16:49:35
A DYNAMICAL SYSTEMS-BASED APPROACH TO SOIL SHEAR 811
08

06

A
04

02

0
032

024
B

016

06

04
D

02

0
24
16
J

08
0

16

08
K

06

04
L

02

0
0
M

00008

00016
00004

00002
N

00002
0 05 10
Stress ratio, b

Silty sand thin ends (50 kPa)


Silty sand thick ends (50 kPa)
Silty sand thin ends (200 kPa)

Fig. 8. Model fit for Shapiro (2000) true-triaxial tests on silty sand

Ko conditions as opposed to Sheahan who used computer- batches different to Sheahans and, most importantly, that
controlled stress increments. Also, Fayads test samples were Fayad used a manual procedure for Ko consolidation. The
reconstituted from previous test programmes and from maximum discrepancies in the fit are for q at OCR 8 where
batches different from Sheahans. the calculated maximum of 1.48 compares against the meas-
Figure 10 shows the application of the model calibrated in ured value of 1.74, a discrepancy of 15%, and for p at OCR
Joseph (2009) using data from Sheahan (1991) to predict 4 where the calculated maximum of 2.2 compares against
Fayads results. Table 2 shows the parameter values from the measured value of 1.82, a discrepancy of 21%.
Joseph (2009) obtained from fitting the model to Sheahans
data, at OCRs corresponding to Fayads tests. These para-
meter values were used to fit Fayads data, starting from an CONCLUSION
initial strain of 0.5%, chosen to reduce the effect of errors The high correlation coefficients and orderly variation of
in initial conditions resulting from the manual stress incre- the model parameters obtained on fitting these equations to
ments used for the Ko consolidation. drained and undrained tests on sands, silts and clays at
The fits are reasonable given that Fayads test samples various densities, OCRs and with various stress paths sup-
were reconstituted from previous test programmes and from port the validity of the model proposed. The value of the

Delivered by ICEVirtualLibrary.com to:


IP: 130.102.42.98
On: Sun, 10 Oct 2010 16:49:35
812 JOSEPH
Table 2. Parameter values used to predict Fayads test data
All tests conducted at the same
25
initial confining stress of 50 kPa
and with thick platens OCR 1 2 4 8

A 11.987 8.649 6.442 4.642


20
B 1.001 0.899 0.815 0.732
J 25.807 16.075 10.483 6.517
D 0.945 0.837 0.750 0.663
Parameter values

15 K 28.087 11.584 5.206 2.139


L 11.472 6.561 3.962 2.261

10

theory. At present, the physical basis of equations (1) is yet


to be determined and, in this sense, the model qualifies as
05 curve fitting.

0 ACKNOWLEDGEMENT
The author thanks the British Library, London, UK for
making available the theses of both Plant (1956) and Gens
0 02 04 06 08 10
(1982).
Stress ratio, b
A B D J K L M N

Fig. 9. Model fit for Shapiro (2000) true-triaxial tests on sand NOTATION
A, J, K, L, M, N initial proportionality constants of the model
30 (dimensionless)measures of the initial
resistance offered by the soil structure to the
25 initial stresses
B, D decay constants (dimensionless) measures of
20 the change with strain of the resistance offered
by the soil structure
b stress ratio defined as ( 2  3 )( 1  3 )
p/vc

15
e base of the natural logarithm
OCR over-consolidation ratio
10
p effective normal/confining stress, defined as
( 1 3 )=2 for a triaxial test and as
05 ( 1 2 3 )=3 for a true-triaxial test
q shear stress, defined as ( 1  3 )=2 for a triaxial
0 test and as 1=3[( 1  2 )2 ( 2  3 )2
0 5 10 15 20 ( 3  1 )2 ]1=2 for a true-triaxial test
r2 coefficient of correlation
shear strain
30 void ratio
1 major effective principal stress
25 2 intermediate effective principal stress
minor effective principal stress
20
OCR 1 (meas.) vc vertical effective consolidation stress
OCR 1 (calc.)
q/vc

15 OCR 2 (meas.)
OCR 2 (calc.) REFERENCES
10 OCR 4 (meas.) Fayad, H. F. (1986). Aspects of the volumetric and undrained
OCR 4 (calc.) behavior of Boston Blue clay. MS thesis, Massachusetts Institute
05 OCR 8 (meas.) of Technology, Cambridge, Massachusetts, USA.
Gens, A. (1982). Stressstrain and strength of a low plasticity clay.
OCR 8 (calc.)
0 PhD thesis, Imperial College, London, UK.
0 5 10 15 20 Joseph, P. G. (2009). A constitutive model of soil based on a
Strain: % dynamical systems approach. J. Geotech. Environ. Engng 135,
No. 8, 11551158.
Fig. 10. Predicted plotted against measured stressstrain curves Plant, J. R. (1956). Shear strength properties of London clay. MSc
for Fayad (1987) data thesis, Imperial College, London, UK.
Shapiro, S. (2000). The effects of nonplastic fines on the three-
dimensional behavior of sand. MS thesis, Clarkson University,
New York, USA.
model lies in that it provides a simple, alternative approach Sheahan, T. C. (1991). An experimental study of the time-dependent
to current complex elastoplastic models based on critical undrained shear behavior of resedimented clay using automated
state theory. In addition, the model connects soil shear to a stress path triaxial equipment. ScD Thesis, Massachusetts Insti-
powerful mathematical framework the dynamical systems tute of Technology, Cambridge, Massachusetts, USA.

Delivered by ICEVirtualLibrary.com to:


IP: 130.102.42.98
On: Sun, 10 Oct 2010 16:49:35

Potrebbero piacerti anche