Professional Documents
Culture Documents
Resistance in Sands
R. Salgado, M.ASCE1; and M. Prezzi, A.M.ASCE2
Downloaded from ascelibrary.org by UFRGS - Universidade Federal do Rio Grande do Sul on 01/09/20. Copyright ASCE. For personal use only; all rights reserved.
Abstract: A cavity expansion-based theory for calculation of cone penetration resistance qc in sand is presented. The theory includes a
completely new analysis to obtain cone resistance from cavity limit pressure. In order to more clearly link the proposed theory with the
classical cavity expansion theories, which were based on linear elastic, perfectly plastic soil response, linear equivalent values of Young’s
modulus, Poisson’s ratio and friction and dilatancy angles are given in charts as a function of relative density, stress state, and critical-state
friction angle. These linear-equivalent values may be used in the classical theories to obtain very good estimates of cavity pressure. A
much simpler way to estimate qc—based on direct reading from charts in terms of relative density, stress state, and critical-state friction
angle—is also proposed. Finally, a single equation obtained by regression of qc on relative density and stress state for a range of values
of critical-state friction angle is also proposed. Examples illustrate the different ways of calculating cone resistance and interpreting cone
penetration test results.
DOI: 10.1061/共ASCE兲1532-3641共2007兲7:4共251兲
CE Database subject headings: Computation; Cone penetration tests; Sand; In situ tests; Penetration resistance.
冉冊
Nij−1
rj Nij
Numerical Cavity Expansion Formulation ri = r j 共3兲
ri
Plastic Zone where r⫽effective radial stress; ⫽effective hoop stress around
Fig. 1 shows a horizontal cross section of an expanding cylindri- the expanding cylindrical cavity; Nij⫽flow number at the center
cal cavity in an infinite soil mass. Initially, before the cavity was of the element, given by
T4 13.6 16.7 2.68 0.575 0.924 1.50 0.54 0.36 34.8 647 2.27 0.43
T9 13.6 16.4 2.68 0.605 0.930 1.35 0.55 0.36 34.8 647 2.27 0.43
Y 13.1 16.1 2.65 0.611 0.985 1.27 0.23 — 35.1 900 2.17 0.40
Note: Data after Bolton 共1986兲, Lo Presti 共1987兲, and Lo Presti et al. 共1992兲. H⫽Hokksund; M0⫽Monterey No. 0; O⫽Ottawa Sand; Y⫽Toyoura Sand;
T⫽Ticino Sand; ␥d,min determined using ASTM D4254-83; and ␥d,max determined using ASTM D4253-83 for Monterey No. 0 sand and pluviation 共Miura
and Toki 1982兲 for the other sands.
IR,ij =
100
冋 冉 冊册
DR,ij
Q + ln
pA
100pij
− RQ 共5兲
in cavity expansion analysis. Starting from the plastic radius,
where stresses, strains and the radial displacement are known,
calculations proceed inward, toward the cavity, element by ele-
ment. At the elastic–plastic interface, the radial stress R and the
ij = c + 5IR,ij 共6兲 mean effective stress pR are given by 共Salgado et al. 1997b;
Salgado and Randolph 2001兲:
IR,ij
sin ij = 共7兲 2N p
6.7 + IR,ij R = p0 共8兲
Np + 1
where DR,ij⫽relative density ranging from 0 to 100% within the
thin ij shell; Q, RQ⫽fitting parameters that depend on the sand
characteristics; pij⫽mean effective stress at the center of the ij
shell; and PA⫽reference stress⫽100 kPa= 0.1 MPa⬇ 1 tsf. Al-
1
pR = 关1 + R兴 1 +
3
冉 冊
1
Np
R 共9兲
of a nonlinear elastic stress–strain relationship ensures this con- would produce equivalent linear material properties that would
sistency. It is possible, by integration of strains, to obtain an then be fed into the criterion to determine whether a limit condi-
equivalent linear value of G for the elastic zone. This value of G tion has been reached, what happens at the cavity wall happens in
can be used to calculate the strains for the first element of the response to material properties that are local and vastly different
plastic zone 共the element adjacent to the elastic–plastic interface兲, from equivalent linear parameters. It turns out that it is key to
which allows the results of the calculations for the elastic zone to capture this local response if we desire an accurate cavity expan-
feed into the calculations for the plastic zone. sion analysis.
The initial soil stiffness is represented by the small-strain shear Chadwick 共1959兲 and latter Yu and Houlsby 共1991兲 took the
modulus G0. The small-strain shear modulus G0 of sand can be steady state as that for which the limit of the ratio of the initial to
computed from the following expression 共Hardin and Black the current cavity radius approaches zero. This does indeed rep-
1968兲: resent the notion of cavity creation and is well suited to analytical
冉 冊
solutions based on a simple constitutive relation, for which math-
G0 共eg − e兲2 m ng
ematical limits can be taken, but is difficult to use effectively in
= Cg 共13兲
pA 1+e pA numerical solutions. The numerical difficulty resides in ad-
equately capturing this limit process. If calculations are done by
where e⫽initial void ratio; m = p⫽initial mean effective stress;
expanding a cavity from some initial radius, the criterion tells us
and Cg, eg, ng⫽intrinsic variables of the soil.
that we would need to keep expanding this cavity forever in order
to reach a solution. So the numerical difficulty is in answering the
Relationship to Analytical Cavity Expansion Solutions question: at what cavity expansion radius short of infinity have
we obtained a sufficiently good approximation? Although ap-
Limit Cavity Pressure proximations can be found, it would be difficult to control the
In the discussion that follows, we focus on the closed-form ana- accuracy of a numerical solution based on this criterion.
lytical solutions of Carter et al. 共1986兲 and Yu and Houlsby In this paper, the steady-state criterion used is that the calcu-
共1991兲 for linear elastic, perfectly plastic soil. In a cavity creation lated displacement at the inner boundary of the cylindrical ele-
process, once the cavity is expanded from zero initial radius and ment must equal its inner radius. This expresses the physical
reaches a steady-state condition, ongoing expansion happens at notion of cavity creation rigorously and has the advantage of
constant limit cavity pressure. It is necessary to express math- being easily implemented in a numerical formulation. So compu-
ematically the requirements for a steady-state condition to be tations proceed from the plastic radius towards the cavity. The
reached and thus for the limit pressure to be calculated. plastic radius, the starting radius, can be any number, as cavity
Closed-form analytical solutions can be obtained for the limit expansion in the free field is a self-similar problem and what
cavity pressure if the assumptions of linear elasticity and perfect matters is the ratio of R to a, and not their absolute values. We
plasticity can be made. The limit pressure can be linked with the took R = 100 in our calculations.
radial stress at the elastic-plastic interface using Eq. 共3兲 with ri
= a, r j = R and a single value of N 共which, for a perfectly plastic Equivalent Linear Values of Plastic and Elastic Parameters
material, is constant throughout the plastic zone兲: Engineers wishing to use the analytical cavity expansion solutions
冉冊 共N−1兲/N need guidance on which values of shear modulus and friction and
R dilatancy angles to use. The present analysis can produce these
pL = R 共14兲
a L values. It is of particular interest to investigate how the equivalent
In Eq. 共14兲, N must be an equivalent linear flow number in linear values of , , E, and vary with initial soil state, i.e.,
order to produce the same limit pressure as a full nonlinear analy- initial relative density and stress state. For this purpose, calcula-
sis. The subscript L indicates that the R / a ratio is a limiting tions are done for the following values of the intrinsic values Cg,
共maximum兲 value. Eq. 共14兲 cannot be used directly to calculate eg, ng, emin, emax, Q, and RQ:
the limit pressure because the ratio of the plastic to cavity radius Cg = 650, eg = 2.2, ng = 0.45
is not known a priori; however, combined with a suitable steady-
state criterion, an equation can be derived, based on Eq. 共14兲, that emin = 0.60, emax = 0.95
can be used to calculate the limit pressure.
Q = 10, RQ = 1
Steady-State Criteria
Carter et al. 共1986兲 proposed a steady-state criterion based on Eq. These values are representative of typical silica sands. The
共14兲. For the pressure pL to indeed be a limit pressure, R / a needs relative density values considered are DR = 10, 20, 30, 40, 50, 60,
to be equal to the limit value 共R / a兲L. Stated in another way, in 70, 80, 90, and 100%. The coefficient of active earth pressure at
steady-state cavity expansion, even as the cavity radius a and the rest is in the 0.4–0.5 range in most cases in practice, with dense
plastic radius R increase, their ratio remains the same. That hap- sands tending to have lower, and loose sands having higher val-
pens if and only if: ues. For simplicity, a single value of 0.45 is taken for all values of
再 冋 冉冉 冊 冊册冎
K = 0.5 1 + tanh ln
0.000102
␥
0.492
共17兲
再 冋 冉冉 冊 冊册冎
n = 0.272 1 − tanh ln
0.000556
␥
0.4
共18兲
The equivalent linear shear modulus for steady-state cavity
expansion, normalized with respect to the small-strain shear
modulus, G / G0, is plotted in Fig. 3 for c = 29° and c = 36°. It
where ␥⫽engineering shear strain 共as a number, not a can be seen that G / G0 does not vary much with relative density or
percentage兲. c, but does vary significantly with lateral effective stress. To
The radial and hoop stresses are computed at the center of thin some extent, this is due to the nature of the Ishibashi and Zhang
cylindrical shell elements in the nonlinear elastic zone using the 共1993兲 model, but the calculations show that the cavity expansion
equations for an elastic hollow cylinder of infinite radius 共in the process does not introduce a strong dependence on variables other
case of cavity expansion in the field, as opposed to in a calibration than the stress. For h = 22.5 kPa, G / G0 lies in the 0.52–0.6 range
chamber, when the equations for a hollow cylinder with a defined for c = 29° and in the 0.5–0.58 range for c = 36°. For h
outer boundary are used兲 with internal pressure equal to R. These = 450 kPa, G / G0 lies in the 0.86–0.9 range for c = 29° and in the
stresses then are used to compute the corresponding strains. Here 0.8–0.88 range for c = 36°.
iterations are necessary, as we initially assume a value of G then The linear-equivalent flow number N in the plastic zone at
iterate with Eq. 共16兲 to guarantee that the calculated strain and the steady state can be determined from the rigorous analysis by con-
modulus are compatible. Fig. 2 shows the modulus degradation sidering directly Eq. 共14兲. For a given plastic radius R, the analy-
共G / G0 versus ␥兲 curves for various mean effective stress values sis produces values of R, pL and a at steady state. It follows that
produced by the Ishibashi and Zhang 共1993兲 model. N can be calculated as
冉冊 冉 冊
L
N= 共19兲
R pL
ln − ln
a L R
The linear-equivalent friction angle is then obtained from N
through the general equation relating flow number to friction
angle:
Downloaded from ascelibrary.org by UFRGS - Universidade Federal do Rio Grande do Sul on 01/09/20. Copyright ASCE. For personal use only; all rights reserved.
N−1
sin = 共20兲
N+1
Once is known, the linear-equivalent dilatancy angle can be
obtained from the values of and c using Eqs. 共5兲–共7兲. The
calculated using Eq. 共20兲 is plotted in Fig. 4. Fig. 4 shows a plot
of – c versus initial lateral stress for three values of critical-
state friction angle c: 29, 33, and 36°. This covers the full range
of c values for silica sands. It can be seen from Fig. 4 that the
plots are, in essence, independent of c for DR = 0 – 60% but start
separating for larger relative densities. Even so, for DR = 100%,
the difference between – c values for c = 29 and 36° is only of
the order of 1°. For very loose sand, the equivalent was allowed
to dip below 0° 共which it does by no more than a few degrees兲. In
this, we followed the argument put forth by Bolton 共1986兲, who
linked the occurrence of values below c using his equation to
the fact that, for very loose sands, the peak and critical-state fric-
tion angles 共which would be very close兲 would occur at such large
shear strains that it would be acceptable to use ⬍ c.
It is interesting to explore how the ratio 共R / a兲L of the plastic
radius to the cavity radius at steady state 共when the limit pressure
has been reached兲 varies with initial soil state. This is a quantity
of interest not only in that it is an integral part of the calculations
of limit pressure but also in that the question of how far the
effects of cavity creation extend is often asked in practical prob-
lems. For example, if experiments are being conducted on pen-
etrometers, the question of how far a boundary can be placed
without excessively distorting the mechanism the experiments
aim to study is an important one, and the extent of the plastic
zone can guide us on that account.
Fig. 5 shows 共R / a兲L versus lateral effective stress for c = 29
and 36° and DR = 10– 100%. Note first that the values of 共R / a兲L
are practically independent of c. Note also that the plastic radius
to cavity radius ratio is greater than 100 for dense sands and as
high as 70 for medium dense sands at low confining stresses
共0 – 50 kPa兲, which correspond to depths often of interest in geo-
technical engineering problems. This means that the use of cali-
bration chambers or centrifuges to study cone penetration must be
guided by rigorous theory so that due account can be taken of
boundary effects. Insufficient distance between boundaries and
the cone allows boundaries to interfere with the formation of the Fig. 4. Equivalent linear friction angle as function of initial soil state:
plastic zone around the cone, and this interference is substantial at 共a兲 c = 29°; 共b兲 c = 33; and 共c兲 c = 36°
low confinements.
The plots of 共R / a兲L can be used to obtain suitable values of
共R / a兲L for use in Eq. 共14兲. Since the radial stress at the elastic– Penetration Resistance Analysis
plastic interface follows from Eq. 共8兲, limit cavity pressures can
then be calculated using Eq. 共14兲. A more direct way to obtain
Basic Analysis for Cone Penetrometers
limit pressure is to use the charts in Fig. 6, in which limit pressure
versus initial lateral effective stress is plotted for DR = 10– 100% The penetration resistance of a penetrometer with flat or conical
and c ranging from 29 to 36°. There is a gradual shift to the right tip may be calculated from the cylindrical limit pressure pL deter-
in the pL versus stress curves as c increases from 29 to 36°. The mined in the previous section. Spherical cavity pressures have
maximum limit pressure, calculated for DR = 100% and 1 MPa also been proposed in the past for the same purpose, although in
vertical stress 共450 kPa lateral stress兲, is approximately 15 MPa. papers typically focusing on clays. For sands, Mitchell and Tseng
冦
sin共⌬ − c兲
exp共⌬ tan T兲 if T ⱖ c
sin c
C = 共23兲
sin共⌬ − c兲
if T ⬍ c
sin c
duced by cone penetration outside the immediate area below the where
Fig. 6. Cylindrical cavity limit pressure as a function of initial soil state for a normally consolidated sand with critical-state friction angle equal
to: 共a兲 29; 共b兲 30; 共c兲 31; 共d兲 32; 共e兲 33; 共f兲 34; 共g兲 35; and 共h兲 36°
=1−
NT − 1
NT
共27兲 冉
Q1 = pL exp共2⌬ tan T兲 1 + C − C
r0
rc
冊 −1
共28兲
Fig. 6. 共Continued兲.
共1 + C兲+1 − C共 + 1兲 − 1
qc = 2f v pL exp共2⌬ tan T兲 共30兲
C2 共 + 1兲
with
fv =
1
2
冋冉 冊 冉 冊
1+
1
Nc
+ 1−
1
Nc
共cos ␦c cot c − sin ␦c兲 册 共31兲
The starting point for finding the average mean effective stress
in the transition zone, required for the calculation of T, is to
realize that qc⫽average vertical stress acting on the face of the
cone. Working back from that, we find that the average radial
stress acting on the conical face of the penetrometer is given by
共1 + C兲+1 − C共 + 1兲 − 1
¯r = 2pL 共32兲
C2 共 + 1兲
and the associated mean effective stress can be written as Fig. 7. Assumed slip mechanism for cone penetration in sand
Consider the slipline starting at the point on the conical sur- where ⫽angle ranging from 0 to ⌬ between an arbitrary log-
face where v = qc 共which we may choose to call the “average” spiral radius s and s0 = BC.
slip line兲. The mean effective stress varies along this slip line Taking the average of p̄ over the slip line, we obtain the aver-
following: age mean stress p̄T in the transition zone:
Downloaded from ascelibrary.org by UFRGS - Universidade Federal do Rio Grande do Sul on 01/09/20. Copyright ASCE. For personal use only; all rights reserved.
冦
p̄i exp共2⌬ tan T兲 − exp共⌬ tan T兲
if T ⱖ c
2 tan T cot T exp共⌬ tan T兲 − 1
p̄T = 共35兲
exp共2⌬ tan T兲 − 1
p̄i if T ⬍ c
2⌬ tan T
Computation of qc from pL is done iteratively as follows: The predictive power of cavity-expansion-based cone resis-
1. Assume value for T 共and thus for T and NT兲; tance theory is further illustrated by Fig. 10 for Ticino sand. Fig.
2. Compute , ⌬, and C using Eqs. 共27兲, 共24兲, and 共23兲, 10 shows qc curves obtained for Ticino sand using CONPOINT
respectively; plotted together with data points from calibration chamber tests
3. Compute T using the Bolton relationship 关Eqs. 共5兲–共7兲 with on Ticino sand. The CONPOINT curves approximate the cone
pij = p̄T兴; resistance measured in calibration chambers reasonably well for
4. Iterate until convergence of assumed and calculated values of
comparable relative densities.
T is reached; and
5. Compute qc using Eqs. 共30兲 and 共31兲.
There is no single value of friction angle in the immediate
neighborhood of the cone, as the stress and strain fields are com- Correlations for Penetration Resistance
plex and have sharp gradients. Along the conical surface and the
penetrometer shaft, where shear strains are quite large, is equal
to c both for contractive and dilative sands. Accordingly, the Cone Resistance Charts
interface friction angle is also a critical-state value ␦c that is taken
as a fraction of c. However, near the cone surface but slightly The preceding analysis allows us to develop charts and correla-
away from the steel–soil interface, stresses are very high and the tions for qc in terms of key intrinsic and state variables of the
sand is likely to undergo some crushing, which is difficult to sand. Fig. 11 shows qc in terms of relative density and stress state
model. for eight different values of c. These charts can be used directly
Bolton 共1986兲 suggests it may be necessary to allow the to estimate qc if the sand properties are similar to those used to
calculated using Eqs. 共5兲–共7兲 to be less than c under some con-
prepare the charts 共recall that these values are Cg = 650, eg = 2.2,
ditions, as very contractive sands would require such large strains
ng = 0.45, emin = 0.60, emax = 0.95, Q = 10, and RQ = 1兲. For different
to develop c that the operative, mobilized friction angle would
be less than c under many conditions of practical interest. For values of these parameters, use of CONPOINT is indicated. Al-
dilative sands, either sands with high relative densitites or low to ternatively, with knowledge of K0 共perhaps from knowledge of
moderate initial confining stresses, values tend to be greater the geological history of the soil deposit兲 and qc, the relative
than c even within the transition zone. Fig. 8 shows values of density may be estimated.
T – c as a function of relative density and lateral stress for c Observing Fig. 11, we note that qc values range from zero at
= 29, 33, and 36°. The curves are similar, with T – c tending to the ground surface due to lack of confinement to as much as
increase with increasing relative density, decreasing stress and 80 MPa at a vertical effective stress of 1 MPa in a normally-
decreasing c. Note that this approach indirectly accounts for consolidated deposit with DR = 100% 共admittedly, an eminently
sand “compressibility” 共a term found in the literature to describe theoretical case兲. For cases of more relevance in on-shore prac-
what, in more appropriate usage, is contractiveness兲.
tice, a depth of 30 m with a deep water table would give us a
As the analysis is applicable to both field and calibration
more credible “large” stress. That would correspond to a vertical
chamber conditions, the quality of the predictions made using the
analysis can be checked by comparison with calibration chamber effective stress on the order of 600 kPa and a lateral effective
penetration tests. As an illustration of the excellent match be- stress of the order of 270 kPa. For a very high critical-state fric-
tween predicted and measured values, Fig. 9 shows how well qc tion angle of 36° and a relative density of 90%, qc would be about
values predicted using CONPOINT compare with values mea- 70 MPa, which would constitute a point beyond refusal for a
sured in chamber tests done on West Kowloon sand. These studies typical CPT system. For a sand with 29° critical-state friction
were done by Lee et al. 共1999兲 in the course of the engineering of angle, the same relative density 共DR = 90% 兲 and stresses would
large, hydraulically placed marine sand fills in Hong Kong, for lead to qc = 35 MPa, which might be reachable by sufficiently
which the ability to effectively interpret CPT results was very powerful systems even without friction reducers.
important.
Fig. 8. Transition zone friction angle T as function of initial soil state: 共a兲 c = 29°; 共b兲 c = 33°; and 共c兲 c = 36°
A ratio of interest is the ratio of qc to cylindrical cavity limit Most expressions in the form of Eq. 共36兲 have been obtained
pressure pL, which can be extracted from Figs. 6 and 11. The ratio by regressions of qc values obtained from penetration tests done
ranges from 3.2 for loose sands with high c and large h values in large calibration chambers 共Bellotti et al. 1982; Villet and
to 7.3 for dense sands with low c and low h values. Mitchell 1981; Ghionna and Jamiolkowski 1991兲. Proposal of C1,
C2, and C3 values for Eq. 共36兲 based on all available calibration
Simple Correlations chamber test results have been made by, for example, Baldi et al.
A simpler approach to penetration resistance estimation is to use 共1983, 1985, 1986兲. However, as different sands have different
an equation of the form: intrinsic variables and qc is sensitive to these variables, particu-
larly to c, the use of a single expression of the form of Eq. 共36兲
qc
pA
= C1 冉 冊
h
pA
C2
exp共C3DR兲 共36兲
to calculate qc for all silica sands cannot be expected to have high
accuracy. Additionally, it is impossible to find constants C1, C2,
and C3 that can truly fit qc values for all possible relative densities
where DR⫽relative density, ranging from 0 to 100%; and lateral effective stresses. In fact, they cannot be constants at
pA⫽reference stress 共=100 kPa= 0.1 MPa兲 in the same units as
all if they are to provide a satisfactory fit. They must themselves
h; and C1, C2, and C3⫽regression coefficients. This type of ex-
be functions of the relative density and the critical-state friction
pression has been proposed by Baldi et al. 共1983, 1985, 1986兲,
Schmertmann 共1976兲, and others. angle.
Fig. 9. Comparison of predicted with measured values of cone resis- Fig. 10. Cone resistance calculated both using CONPOINT with
tance for calibration chamber tests on West Kowloon sand 共adapted Ticino Sand properties and cone resistance measured in calibration
from Lee et al. 1999兲 chamber tests on normally, 1D consolidated samples of Ticino Sand
with DR within the 10–20%, 50–60%, and 90–100% ranges 共data
adapted from Salgado 1993 and Salgado et al. 1997a,b兲
Fig. 11. Cone penetration resistance using CONPOINT as a function of initial soil state for a normally consolidated sand with c equal to: 共a兲
29; 共b兲 30; 共c兲 31; 共d兲 32; 共e兲 33; 共f兲 34; 共g兲 35; and 共h兲 36°
qc
= 1.64 exp关0.1041c + 共0.0264 − 0.0002c兲DR兴
pA
⫻ 冉 冊
h
pA
0.841−0.0047DR
共37兲
冉 冊
DR = ⱕ 100% h = 88 kPa. From Fig. 12, qc = 16 MPa. Taking all of this into Eq.
h 共38兲, we get DR = 62%. Using Fig. 11, DR = 61%. This value of DR
0.0264 − 0.0002c − 0.0047 ln
pA can now be used, for example, to estimate friction angles or as-
共38兲 sess the susceptibility of the soil to liquefaction.