You are on page 1of 7

CHAPTER

ELEVEN
THE BEHAVIOUR
OF OVERCONSOLIDATED
.
.
SAMPLES: THE HVORSLEV SURFACE

'
111 INTRODUCTION

We found in Cbapter 10 that, for normall), consolidated specimens, there


eltists a critical state line at which specimens fail, and a Itate boundary
surface, the Roscoe surface, which Jimits tbe state of specimens in q' : p' : "
space"IS they move from the normal consolidation line to the critical state
line. We must now ask ourselves whether the same concepts, perhaps with
modification, appl)' for overcoDsolidated specimens.

111 DRAINED TESTS


We will consider the behaviour of specimens which have been isotropicaJl)'
consolidated 10 lome mean normal effective stress p:Uu and then allowed
to swell isotropically to some lower mean normal slress P~. as indicated in
Fig. J \-1. The overco~solidation ratio Rp is defined as p;..... fp~. and so

Normal consolida tion line

"'f--.,......
5...."111,,,
line

P,

/'_.

Fh:\lre 11-1 Compression and swcllinslines

"

BEllAVIQUk OF OVEItCONSO UOATED SAMru:s 211

IOOr------------------,
0') D

....

34.S kN m- I
rltioR, . 2'-

~rconsolidlt)on

7.

~,o/~.

I
o

j'
10

'0

-.,------------------,
-,
!~ o'=,..c.--------~----__\
');-0---,!;;:o---,f.;-o--,A.I.T.
. f.

UBRARY

(per ~nt)

F'lwe 112 Tesl dall rrom draine4 lest on In overconsolidlled umple of Weald elay
(Ifter Bishop and Henkel, 1962, p. 121)

normally consolidated samples have Rp - I while Rp is large fOf heavily


oycrconsolidatcd samples.
A typlyal drained test on a heavily ovcrconsolld8t~d ,sample (Rp _ 24) of
Weald clay was discussed in Sec. 94 and q': l" and c,,: " curves ,nfC
reproduced in Fig: 11-2. A significant feature of the q' : Sa curve is thai the
sample exhibits a peak strength, denoted by q;. afler which the value of q'
falls as ea incruses. The value of q' seems to be decreasing towards a steady
value at the end orlhe test, but, as discussed in Sec. 9-4, we must be suspicious
about the acc::uracy of the data of both stress and slrain at large shear strains.
We nole also that the sample initially decreases in volume, before it expands
substantially as the test proceeds, There is a hinl that the rate of volume
expansion is decreasing towards the end of the lesl, but again we cancol
rely on the data at this stage of the lest.
We can now plot ~e stress path for the test in q' : p' space as shown in
Fig. 11 -3. TIle Ie.~t path moves nbove the projeclio n of Ihe criticnl lillie line
to the failure poinl before moving back along the Slime puth lowRnh the
projection of the critical state line. The state boundary surface for over
consolidated specimens must therefore have a projection in q': p' space thlll
lies above the projection of the critical state line.
'

I
",

,,

U2 THE MECHANICS

or SOtu

,.r---------------------~----~
FaillJ~

I,

ClilicaISlat~ .line,
y

,.
E

2.

".

".

".

".

".

20

".

".

".

".
End of
luI

//

".

p~

",0

60

p'(kN m- 1)

FI,ute 113 Tell path follo_d in 1M drained lest of Fl._ 11-2

We could consider n family of drained tests to obtain morc information


about the shape of the stalc boundary surfact. However, the dilficulty with ,
such a family of tests, and indeed with drained tests in general, is that the
specific volume of the specimens is changing throughout the testl. The
projections of the test paths, or the failure point, into q' : p' space then refer

to different constant" stions of the state boundary surface. By aDalolY


with the Roscoe surface, we expect that each constant" section of the state
boundary surface is of different site, though we hope of the same sha.pe. The
most convenient way of investigating the shape of the state boundary
surface is, thert:fore, to scuJe the stresses to allow for changes in specific
volume that occur during the tests.

11-3 THE HVORSLEV SURFACE


In order to scale the stresses, we argue as before that the size of each
constant tI section of the state bOu-ndary surface will' be difi'ere~or each
v~lue of v, and will be in Pf9~ruOI\ ~ ~ Cquivalent";tms'p!:~ue

~~:~~~~:~~~~t~~:~l\!~~f~~~JbT:~~~~~
by' plotting stress' pattis in
:
space; drained and uodhtined tests
q'lp~ p'lp~

may tben be compared. directly. .


;.:
. . ....
This method of scaling was adopted by Hvonlev, who was concerned
with the correlation of failure strengths of specimens tested in the shear box.

If.HAVIOUIl OF OVf.R.CONSOLIDATf.D SAMPUS 213 ,

We can consider the failure states of specimens in the Iria xi:lI apparatus and
plot the dat. on normalized slrcsll axes, as shown in Fig. 11-4. The data were
obtained frpm a series of compression tests on overconsolidated Weald
clay. and IltO takCJ? from Parry (1960). It is clear that the data of both drained
and' undralned ,tests lie on a single, line in q'/p~: p '/p~ space. The line is
limited on tISrigpt-hand end by the point representing the critical state line
at the top Fdgc of the Roscoe state boundary surface. By the following
. argument, the line of failure points is also limited on its left hand end. The
maximum Vlllue of q'/p' would be when ai was large and as was small. If the
soil could 110t wilhstand tensile effective stresses, the highest value of q' Jp'
that .could be observed would corrcspond to
O. Then, for a triaxial
compression test,
(II-I)

a; _

(11-2)

nnd
q'/~~ - 3.

(11-3)

The locus of failure .points can then be idealized a,s line AD in FiS. II'.
The locus is limited on its len-hand side by the line OA which has slope 3,
cormpondinc to tensile failure, and on its riaht~hand side by the critical

0.6

o Dr.llMid
11 Undrlined

"0.'

l/..:... . . .

o;u

.Db.
5'

Rotcoe Rlrf.ce
.....
,
,

,,

"

\
\

\
\

0.2

Notmal tonsolid. llon


line

0.2

0.'

0.6

0.8

1.0

p'lp~

FI...... 11--4 Failure siaies of drained .nd undr.lned tests on ovcrC'onsolidaled samples
01 Weald cta~ (data from Parry, 1960)

114 Til! "'lCHANICS OF SOIU

.....- . ,

;'-

.,

A
)

O~~

____________________

~C~~

p' lp:
F1lllrt 11' The complete .tlte boUndlry 'Urflce in ,'J,; : p'I'~ .,.ce

state line (point B) and the Roscoe surface (BC). Of course, if the soil could
sustain tensile effective stre~, the line corresponding to (ensile failure
would lie to the left of OA, and might be curved. This latter possibility is
relevant for many cohesive powders , whose handling is important in the
chemical engineering industry.
We shall call the locus AB of failure points in Fig, 11-5 tbe Hvor.dev
sur/oee. The sianificant feature of the surface with which Hvorslev wu
particularlycOiicernecfii-thiii the
iire~ithsPeQmenatfiilure is
a [unction both of the mean normal stress
and of the s~ific- vo iiiiMV
of the specimen at failure. Tlte specific 'v~i~me 'appean' i~"Fig: 11-5 ihrougb
its influence on the equivalent stress p~ , which depends directly on s~fic
~?Iume. The point can be illustrated if
idealite the Hvorslev surface-;; a
siraight line whose equation is

sh;;'

i,

ori.

we

q' /p~ - g + h(p' /p'J,

(11-4)

where g and h are soil constants as shown in Fig. 11 -6. Equation 11 -4 may
he rewriuen as

q';s, gp~+ hp'

(11-5)

p; - "p((N-')/~J

(11-6)

q' - gexp [(N - v)/~)+hp' .

(11-7)

and, using Eq. (10-12),

We may substitute 10 get

The Hvorslev surface intersects the critical slate line given by Eqs (10-1) and
(10-2) al q~. pj and v, where
.

q; "" Mp j: v f

r-).Inpj

(11-8)

au.v~a OF OVU.OONSOUoATIO SANI"US 215

.,

~
"

,~

OIt1uJ ,tate line

.i.

"

"
,,

Slope. II

aDd hence, from Eq. (11-7)


(M-h)p; _ gexp [( N~r)

g _ (M-h)exp

+ In p;]

(r-N) ,
-~-

( II ')
(1110)

, '(r;" )+ lip' ,

( 1111)

Equation (II-II) slates explicitly ~bat the deviator st~$S at failure of an


overconsolidated s~men is made up of two components. The first
com~~~.!,.!j!!2.1.is .p.~~~rtional to ~ean normal effective .tres,:-i.nfio
im\Y"be thouaht of as belna frictional by nature, while the s~ond compone-nt
M':":: lIlexp I(r-I!)/~D depe~ds only on the current specific volume, and the
nlue of certain soil constants, The form of the exponential term is such that
the ' second component' of strenath Increases as the specific volume of the
specimen decreases. Thus if two specimens were brought to failure in drained
tests at the same value of p' but at different specific volumes VI and vt , where
Vl >"~' tbe failure of the specimens would occur at different values ofq', a5
shown in 'Fig, 11-7. The lines AIBI and A,B. are sections of the Hvorslev
surface corresponding to the two specific voli.tmes "I and Vt.
We misbt expect shat samples at failure have progressed outwards as
far as they can in q': p' : "'pace, Therefore. we will now adopt as n working
hypothesi. the idea that the Hvonlev .urface is the .tale boundary surface

f9~ . h~~ili:...!lX~r~:~)!:!!~I~~~~~fiimplesIRthe ~"m.!:

waY"itiltth~~ it9~oe

su!~~~ ,~~ .!~!I!. ~!!~!..?.!.urf~ ror normally and lightly C!verconsolidated


~mples,

Thus the equation of the Hvonlev surface is _


q' - (M- II)exp

,,

y!'r ~C;* ~hat the Hvorsley and Roscoe ~tate boundary surfaces
antersect and that the line or intersection is the critical slate line.

216 TilE MECHAN ICS Of SOILS

B,

B,

"Tesl I
p'
FI ... re I I' Failure

~Iales

or dr ained specimens &1 differe nl specillc volumes

Example 11-1 Ca lculation of Ihe values of Ihe deviator stress al fail ure
for samples which fail on the HVOr1lev surface
Three samples A. B. a nd C are brought to failure on thu Hvonlcv surface
at the foll owing combinations of 11 and p' : Sample A. 11 - 1.90,
p' '"" 200 kN m- 1 ; Sample 8. 11 _ 1.90. p' _ SOO kN m- I ; Sample C,
/)-2.05, p' _ 200kNm- t The clay has N - 3.2S, .\ - 0.2, r - J.16,
M _ 0 .94. and II _ 0.615. Calculate the values of q' at failure of eack
sample.

For states on the Hvprslev surface. the deviator stress is given by


Eq. (II-II )'
q' - (M - h)"p I(r-.)/ ~ I +hp'.

Calculated values arc given in Table Ell-I.


Table EU-l
s.mplo

A'

p ' (kN
q' (kN

m- 'l
m-' )

1.90
200
, 279

1:90

"'"

'"

c
2.o,
200
2<)]