Discussions and Closures

to have failed in general shear mode and not in local or punching

Discussion of “Axial Compression of
shear mode. In these cases when size of excavation is larger than
Footings in Cohesionless Soils. II: the size of footing, the performance and bearing capacity of the
Bearing Capacity” by Sami O. Akbas footing is significantly influenced by vertical effective stress (q0 )
and Fred H. Kulhawy at a depth of B=2 directly below the foundation in the excavated
November 2009, Vol. 135, No. 11, pp. 1575–1582. trench.
DOI: 10.1061/(ASCE)GT.1943-5606.0000136 Using Eqs. (10) through (14) as given in the Akbas and
Downloaded from ascelibrary.org by Pontificia Universidad Catolica del Peru on 08/04/24. Copyright ASCE. For personal use only; all rights reserved.

Kulhawy paper on which this discussion is written, the discusser

Ramesh C. Gupta, M.ASCE1 performed a parametric study to determine in which of the
1
Structure and Bridge Division, Central Office, Virginia Dept. of Transpor-
1200
tation, 1401 East Broad St., Richmond, VA 23219. E-mail: Ramesh
B/L = 0.5
.gupta@vdot.virginia.gov B=1m D=0
800
Irc

Irr, Irc
The authors are to be commended for presenting an extensive Irr 0.5 m
database in their original paper of full-scale field load tests for 400
D=1m
examining bearing capacity of footings in cohesionless soils and 2m
4m
for providing modifications to the bearing capacity equations. 0
Using this database, the authors have identified whether soil 25 30 35 40 45
(a) φ, degrees
behaves as a rigid-plastic material and whether the soil fails in
1200
general shear mode. This database will greatly help researchers B/L = 0.75
to make further advances to the research related to the bearing B=1m
D=0
capacity. 800
Irc
Irr, Irc

The discusser noticed few typographical errors in Cases 6=13,

6=14, 6=15, 26=3, 35=1, 35=2, 36=1, 36=2, and 36=3 of Table 2 in 400
Irr 0.5 m

the original paper. The typographical error in the spreadsheet D=1m

2m
primarily occurs in the calculation of average effective vertical 4m

stress (q0 ) evaluated at a depth of B=2 below the foundation; 0

25 30 35 40 45
this error then continued to create errors in calculation of (b) φ, degrees

rigidity index (I r ), Δ, I rr , and modifiers, ζ qr and ζ γr . Values of unit 1200

weight (γ) are taken from Table 1 in Akbas and Kulhawy (2009) B/L = 1
B=1m D=0
and values of L, B, D, E s1%B , and ϕtc , for the calculations for this 800
discussion, are taken from the original paper. In Table 1, the
Irr, Irc

Irc
corrected values are given by the discusser. When these correc- Irr 0.5 m
400
tions are made, it is seen that all the 106 cases presented in the D=1m
2m
database in Table 2 of the original paper show that the reduced 4m
0
rigidity index (I rr ) is greater than the critical rigidity index (I rc ); 25 30 35 40 45
therefore, the soil in all cases appears to behave as a rigid-plastic (c) φ, degrees

material, and the soil has appeared to have failed in general shear
mode. Even the loose or medium dense sand with ϕtc values of 30° Fig. 1. For footing width of 1 m, I rr and I rc versus ϕ curves at various
depths of embedment (D) with width to length ratio (B=L) of (a) 0.5,
(Cases 3=2, 3=3, and 3=4) and of 34° (Cases 4=3, 4=4, and 4=6
(b) 0.75, and (c) 1
through 4=11 in Table 2 in the original paper), respectively, appears

Table 1. Correct Values after Correcting Typographical Errors in Table 2 in the Original Paper
pa ¼ 101:3 kPa q0 ¼ vertical effective stress at a depth of B=2 below the foundation
Case B (m) L (m) D (m) γ (kN=m3 ) Es1%B (kPa) ϕtc (degrees) Ir q0 (kPa) Δ I rr I rc ξ qr ξγr ξ qd ξ qs
6=13 1.00 1.00 0.00 16.8 20,530 41.5 1,062 8.4 7.256E-05 986 280 1 1 0.6 1.88
6=14 1.00 1.00 0.00 16.8 17,640 41.5 913 8.4 7.256E-05 856 280 1 1 0.6 1.88
6=15 1.00 1.00 0.00 16.8 17,820 41.5 922 8.4 7.256E-05 864 280 1 1 0.6 1.88
26=3 0.75 circ.a 0.00 21.5 47,100 42.5 2,452 8.1 4.974E-05 2,185 325 1 1 0.6 2.33
35=1 0.61 0.61 0.30 17.3 15,130 39.9 665 10 0.0001317 611.4 222 1 1 0.6 1.84
35=2 0.61 0.61 0.30 17.3 11,450 39.9 503 10 0.0001317 472 222 1 1 0.6 1.84
36=1 0.31 0.31 0.00 17.3 12,560 40.8 2,087 2.7 2.779E-05 1,973 253 1 1 0.6 1.86
36=2 0.76 0.76 0.00 17.3 26,840 41.0 1,806 6.6 6.49E-05 1,617 260 1 1 0.6 1.87
36=3 0.31 0.31 0.00 17.3 22,790 40.8 3,787 2.7 2.779E-05 3,426 253 1 1 0.6 1.86
a
circ. = circular footing.

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / SEPTEMBER 2011 / 855

J. Geotech. Geoenviron. Eng., 2011, 137(9): 855-856

 cases the soil is likely to fail either in the general shear mode or in in the original paper presented by the authors, wherein the all
local or punching shear mode. In this study the value of embedment depths are less than 1 m.
modulus of elasticity of sand for a particular value of ϕ is selected Fig. 2 also shows that the zone below the I rc line where I rr is less
from a table given in the AASHTO LRFD Bridge Design than I rc , i.e., the zone in which the soil is likely to fail in local or
Specifications (AASHTO 2007). Fig. 1 shows I rr and I rc versus punching shear mode increases with the increase in the width of the
ϕ charts at various values of embedment depth (D) for B=L ratios footing. The soil with ϕ even greater than 40° may fail in local or
of 0.5, 0.75, and 1 when the width of footing is 1 m. Fig. 2 shows I rr punching shear mode when embedment depths are greater than 1 m
(see Fig. 2).
and I rc versus ϕ charts at various values of embedment depth (D)
for B=L ratio of 0.75 when the widths of footing are 0.5 m, 1 m,
and 1.5 m. References
In Figs. 1 and 2, the area above the I rc line represents a soil layer
Downloaded from ascelibrary.org by Pontificia Universidad Catolica del Peru on 08/04/24. Copyright ASCE. For personal use only; all rights reserved.

with I rr values greater than I rc . The soil layer in this area shall be- Akbas, S. O., and Kulhawy, F. H. (2009). “Axial compression of footings
have as a rigid-plastic material and is likely to fail in general shear in cohesionless soils. I: Load-settlement behavior.” J. Geotech.
Geoenviron. Eng., 135(11), 1562–1574.
mode. However, the area below the I rc line represents a soil layer
AASHTO. (2007). AASHTO LRFD bridge design specifications,
with I rr values less than I rc . The soil layer in this lower area, be- 4th Ed., Washington, DC.
cause of lower stiffness, is likely to fail in local or punching shear
mode. According to these concepts, even dense soils can fail in
punching or local shear mode with an increase in the depth of
the embedment. A crude example of this is the punching shear Closure to “Axial Compression of Footings in
mode that occurs during cone penetration even in very dense soils. Cohesionless Soils. II: Bearing Capacity” by
Fig. 1 also shows that the zone of punching shear mode increases Sami O. Akbas and Fred H. Kulhawy
with a decrease in the B=L ratios, i.e., this zone is greater for B=L November 2009, Vol. 135, No. 11, pp. 1575–1582.
ratio of 0.5 than that for B=L ratio of 1. Fig. 1 also shows that the DOI: 10.1061/(ASCE)GT.1943-5606.0000136
soils for the footings with embedment depths of less than 0.5 m
(even less than 1 m in most of the cases), are likely to behave Sami O. Akbas, M.ASCE1; and Fred H. Kulhawy,
as a rigid-plastic material and fail in general shear mode even Dist.M.ASCE2
though the soil may be very loose (with ϕ values even as low 1
Associate Professor, Dept. of Civil Engineering, Gazi Univ., Celal Bayar
as 25 or 30°) or medium dense. This is confirmed by the database Bulvari, Maltepe, Ankara, Turkey 06570 (corresponding author).
E-mail: soakbas@gazi.edu.tr
2
2500
Professor Emeritus, School of Civil and Environmental Engineering,
B/L = 0.75 Hollister Hall, Cornell Univ., Ithaca, NY 14853-3501. E-mail: fhk1@
B = 0.5 m
2000 cornell.edu
D=0
1500
Irr, Irc

Irc
1000
The writers thank the discusser for his interest in our paper. Regard-
Irr ing the point that was raised for nine of the 106 tests analyzed, there
D = 0.5 m
500 are no typographical errors in Table 2. But perhaps the description
D=1m
4m
2m was not complete enough.
0
25 30 35 40 45 Seven of the nine cases noted by the discusser have zero embed-
(a) φ, degrees ment depth, while Cases 35=1 and 35=2 are embedded to a depth
1200 of 0.3 m. However, eight of the nine cases (except 26=3) were
B/L = 0.75
B=1m
placed in narrow excavations during the load tests, with only a
D=0
800
small clear distance between the excavation walls and the edge
Irc of the footings. Therefore, initially, it was not clear whether the
Irr, Irc

Irr D = 0.5 m effective unit weight of the soil should be multiplied by (B=2)
400
D=1m or (B=2 þ the excavation depth) when estimating the average ver-
2m
3m
tical stress (q0 ) at a depth of B=2 below the foundation. Therefore,
4m
0
25 30 35 40 45
the documentation about the load test conduct and the resulting
(b) φ, degrees load-settlement behavior were evaluated carefully for each case
800 to decide on the appropriate methodology to calculate q0 . For those
B/L = 0.75
B = 1.5 m
cases where the distance between the excavation walls and the foot-
600 ing edge was less than or equal to 1:5B (Cases 6=13, 6=14, and
6=15), it was decided to calculate the q0 values using an effective
Irc
D=0
Irr, Irc

400 depth of (B=2 þ excavation depth). As a second criterion, and for

Irr D = 0.5 m
D=1m the load tests that were not presented with sufficient detail to de-
200 2m
termine this distance, the load-settlement curves also were exam-
4m
ined. Using the available information on the effective stress friction
0
25 30 35 40 45 angle, each case was evaluated to determine whether the observed
(c) φ, degrees
load-settlement behavior would be likely with a surface footing
Fig. 2. For various depths of embedments (D), I rr and I rc versus ϕ
assumption or not, based on comparison with comparable and
curves for footing widths of (a) 0.5, (b) 1, and (c) 1.5 m with width
better-documented case histories. According to this qualitative
to length ratio (B=L) of 0.75 assessment, Cases 35=1, 35=2, 36=1, 36=2, and 36=3 were also
analyzed using (B=2 þ excavation depth).

856 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / SEPTEMBER 2011

J. Geotech. Geoenviron. Eng., 2011, 137(9): 855-856

 The previously mentioned question did not exist for Case 26=3. tests. As shown in Fig. 1, this spatial variation can be decomposed
Similar to Cases 26=1 and 26=2, which were not mentioned by the conveniently into a smoothly varying trend function ½tðzÞ
and a
discusser, q0 values were calculated directly using the effective fluctuating component ½wðzÞ
as follows (Phoon et al. 1995):
(buoyant) unit weight for these surface footings.
As with all studies using extensive databases derived from many ξðzÞ ¼ tðzÞ þ wðzÞ ð1Þ
sources, some judgments have to be made in a very small number
of case histories. As long as the numbers are small, there will be no where ξ = in situ soil property; and z = depth. The inherent soil
significant influence on the overall results and conclusions. variability can be represented by the fluctuating component, if
wðzÞ is modeled as a homogenous random field, as suggested
by Vanmarcke (1983). For this reason, the discussers would like
to emphasize the combined effect of deterministic and stochastic
Discussion of “Probabilistic Analysis of
Downloaded from ascelibrary.org by Pontificia Universidad Catolica del Peru on 08/04/24. Copyright ASCE. For personal use only; all rights reserved.

variability that is called inherent variability on the consolidation

Coupled Soil Consolidation” by Jinsong behavior of natural alluvial deposits.
Huang, D. V. Griffiths, and Gordon A. Fenton For this aim, a simple uncoupled (Terzaghi) approach was
invoked to show the effect. The involved parameters, k and mv ,
March 2010, Vol. 136, No. 3, pp. 417–430 were considered equally important and embodied into a single
DOI: 10.1061/(ASCE)GT.1943-5606.0000238 coefficient of consolidation just for simplicity. Therefore, mv was

Pouya Pishgah Gilani1 and Reza Jamshidi Chenari2

1
M.Sc. student, Civil Engineering Dept., Univ. of Guilan, Iran (correspond-
ing author). E-mail: pouya.pishgah@gmail.com
2
Assistant Professor, Civil Engineering Dept., Univ. of Guilan, Iran.
E-mail: jamshidi_reza@guilan.ac.ir

The paper written by Huang, Griffiths, and Fenton is very inte-

resting. However, the discussers would like to point out some
considerations.
The writers of the paper have only taken a lognormally distrib-
uted stochastic component of k and mv into account and probably
deliberately overlooked the deterministic component, which, it is
believed, cannot be neglected (Pishgah and Jamshidi 2011). The
discussers think it is better for engineering purposes if a simplifi-
cation is introduced: a model in which spatial variability (k, mv ) is
separated into two parts: (1) a known deterministic trend; and
(2) residual variability about that trend.
With regard to natural processes, all soil properties in situ will
vary vertically and horizontally; trend analysis is concerned with
Fig. 1. Inherent soil variability
distinguishing deterministic trends for k and mv based on in situ

Fig. 2. Average degree of consolidation versus time for inherent soil variability

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / SEPTEMBER 2011 / 857

J. Geotech. Geoenviron. Eng., 2011, 137(9): 855-856

 held constant while k was assumed to bear a linearly depth-decreasing The writers wish to thank the discussers for their interest in our
trend along with a sinusoidal fluctuating component that is believed to paper. They have raised an important issue relating to the potential
represent stochastic behavior. variation of mean soil properties in the vertical direction. The prob-
A simple example shows the effect of trend analysis on the re- lem with the discussers’ demonstration, however, is that they used
sult, described in the following. If one detrends the simple general an uncoupled approach, when it is well known that both the coef-
variation of k, as illustrated in Fig. 1, and assumes a depth-constant ficient of volume compressibility and the soil permeability play im-
initial pore pressure profile (ui ), the uncoupled equation for portant roles in the consolidation of heterogeneous soils. These two
isochrones will be properties must be treated independently, and cannot be embodied
X∞
 into a single coefficient of consolidation. Comparisons between
2u0 Mz
uz ¼ sin expðM 2 T V Þ ð2Þ coupled and uncoupled responses of consolidating layers were
M H
m¼0 shown in our paper (Figs. 3 and 4), and also in a related paper
Downloaded from ascelibrary.org by Pontificia Universidad Catolica del Peru on 08/04/24. Copyright ASCE. For personal use only; all rights reserved.

by the same writers (Huang and Griffiths 2010).

Introducing the linear and sinusoidal components into the T V for- The methodology presented in our paper could easily be ex-
mulation, separately and combined, the average degree of consoli-
tended to model the mean trend of soil properties mentioned by
dation for different analyses; namely, linearly varying, sinusoidal
the discussers. The one-dimensional model (Fig. 1 in the paper)
fluctuation, and the general property will be calculated and pro-
is reanalyzed using 11 elements with a total length of 1.1, and four
vided as depicted in Fig. 2. The superposition of the results of
analysis are also plotted for individual components. cases are considered with the unit weight of water set to 1.0 for
For the example presented previously, if different components simplicity in all cases.
of permeability are considered individually while leaving the other In Case 1, the soil permeability (k) and coefficient of volume
components for subsequent analyses, this may lead to erroneous compressibility (mv ) are deterministic and set to 1.0 for all depths,
results in view of average degree of consolidation. as shown in Fig. 1. The coefficient of consolidation (cv ) is thus also
Simple calculations showed that the problem is nonlinear with equal to 1.0 at all depths. The uncoupled approach gives the same
respect to k and mv . Therefore, it is strongly suggested that both answers as the coupled approach in this case. In all figures, values
deterministic and stochastic variability should be embodied in in the middle of elements are displayed.
an integrated analysis to draw conclusions about the effects of In Case 2, the soil permeability (k) and coefficient of volume
different parameters in parametric studies. compressibility (mv ) are also deterministic, but decrease linearly
The final remark is that for consolidation and even bearing with depth, as shown in Fig. 2. The coefficient of consolidation
capacity or slope stability problems, it is crucially important to (cv ) however, remains equal to 1.0 at all depths as in Case 1. In
simultaneously take different components of inherent variability this case, the uncoupled (Terzaghi) approach gives different solu-
into consideration. The combined effect of deterministic and sto- tions to the coupled approach.
chastic analyses will pave the way toward a better understanding
of the effects of the different parameters involved.

References
Phoon, K. K., Kulhawy, F. H., and Grigoriu, M. D. (1995). “Reliability
based design of foundations for transmission line structures.”
Rep. TR-105000, Electric Power Research Institute, Palo Alto, CA.
Pishgah, G. P., and Jamshidi, C. R. (2011). “Stochastic vs. deterministic
analysis of consolidation problem in natural alluvial deposits.” Proc.,
GeoRisk 2011: Geotechnical Risk Assessment and Management Conf.,
ASCE, Reston, VA.
Vanmarcke, E. H. (1983). Random fields: Analysis and synthesis, MIT
Press, Cambridge, MA.

Fig. 1. Case 1: distribution of coefficient of consolidation

Closure to “Probabilistic Analysis of

Coupled Soil Consolidation” by Jinsong
Huang, D. V. Griffiths, and Gordon A. Fenton
March 2010, Vol. 136, No. 3, pp. 417–430.
DOI: 10.1061/(ASCE)GT.1943-5606.0000238

Jinsong Huang, M.ASCE1; D. V. Griffiths, F.ASCE2; and

Gordon A. Fenton, M.ASCE3
1
Research Associate Professor, Division of Engineering, Colorado School
of Mines, Golden, CO 80401 (corresponding author). E-mail: jinsong
.huang@newcastle.edu.au
2
Professor, Division of Engineering, Colorado School of Mines, Golden,
CO 80401. E-mail: d.v.griffiths@mines.edu
3
Professor, Dept. of Engineering Mathematics, Dalhousie Univ., P.O. Box Fig. 2. Case 2: distribution of coefficient of consolidation (cv ), perme-
1000, Halifax, Nova Scotia, Canada B3J 2X4. E-mail: Gordon
ability (k), and mean coefficient of volume compressibility (mv )
.Fenton@dal.ca

858 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / SEPTEMBER 2011

J. Geotech. Geoenviron. Eng., 2011, 137(9): 855-856

Downloaded from ascelibrary.org by Pontificia Universidad Catolica del Peru on 08/04/24. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Case 3: distribution of mean permeability (μk ) and coefficient Fig. 6. Case 4: distribution of mean coefficient of volume compressi-
of variation of permeability (ν k ) (five typical simulations are shown) bility (μmv ) and coefficient of variation of coefficient of volume com-
pressibility (ν mv ) (five typical simulations are shown)

Fig. 4. Case 3: distribution of mean coefficient of volume compressi-

bility (μmv ) and coefficient of variation of coefficient of volume com-
pressibility (ν mv ) (five typical simulations are shown) Fig. 7. Average degree of consolidation defined by pore pressure
(U avp ) and settlement (U avs ), mean average degree of consolidation de-
fined by pore pressure (μU avp ) and settlement (μU avs )

In Case 3, the soil permeability (k) and coefficient of volume

compressibility (mv ) are lognormally distributed random variables,
as shown in Figs. 3 and 4. The mean permeability (μk ) and the
mean coefficient of volume compressibility (μmv ) are both set to
1.0 for all depths. The coefficient of variation of permeability
(ν k ) is set to 0.5 and remains constant with depth. The coefficient
of variation of the coefficient of volume compressibility (ν mv ) is set
to 0.3, and also remains constant with depth. Although there is evi-
dence that the cross-correlation between k and mv is positive, it is
set to zero in this case for simplicity.
In Case 4, the mean permeability (μk ) and mean coefficient of
volume compressibility (μmv ) both decrease linearly with depth, as
shown in Figs. 5 and 6. They have the same coefficients of variation
and cross-correlation as in Case 3.
Five thousand Monte Carlo simulations were conducted for
Fig. 5. Case 4: distribution of mean permeability (μk ) and coefficient Cases 3 and 4. The effects of the mean trend of soil permeability
of variation of permeability (ν k ) (five typical simulations are shown)
and compressibility on the consolidation behavior are clearly

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / SEPTEMBER 2011 / 859

J. Geotech. Geoenviron. Eng., 2011, 137(9): 855-856

 shown in Fig. 7. In Case 3, the mean average degree of consolida- are input separately) for proper modeling of the consolidation
tion defined by settlement and pore pressure was indistinguishable. behavior of heterogeneous soils.
The difference becomes larger for larger coefficients of variation of
k and mv , as demonstrated in the paper.
References
In summary, the authors have shown that the random field ap-
proach described in the paper can easily be adapted to model a Huang, J., and Griffiths, D. V. (2010). “One-dimensional consolidation the-
mean trend with depth for any soil property. The reply has also ories for layered soil and coupled and uncoupled solutions by the finite
reiterated the importance of using a coupled approach (k and mv element method.” Geotechnique, 60(9), 709–713.
Downloaded from ascelibrary.org by Pontificia Universidad Catolica del Peru on 08/04/24. Copyright ASCE. For personal use only; all rights reserved.

860 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / SEPTEMBER 2011

J. Geotech. Geoenviron. Eng., 2011, 137(9): 855-856