Professional Documents
Culture Documents
Abstract: In this paper, two very simple and accurate equations are introduced that enable geotechnical engineers to determine the undrained
ultimate bearing capacity of shallow strip foundations resting on two-layered clays. Results of the analyses show that in the case of weak-over-
Downloaded from ascelibrary.org by Colorado State Univ Lbrs on 01/07/16. Copyright ASCE. For personal use only; all rights reserved.
strong clay, usually general shear failure occurs, and in the case of strong-over-weak clay, the soil will most likely experience punching failure
with a large amount of plastic settlement before it reaches its ultimate bearing capacity. This study applied the finite-difference method using a
computer program to evaluate the undrained bearing capacity for all cases analyzed. The Mohr-Coulomb failure criterion was used for model-
ing the plastic behavior of soil. Simple equations are presented for cases of both strong-over-weak clay and weak-over-strong clay representing
the bearing capacity of shallow foundations resting on two-layered clays. Results of this study are verified by acceptable statistical methods, and
advantages of the presented equations in this paper are discussed. DOI: 10.1061/(ASCE)GM.1943-5622.0000615. © 2016 American Society of
Civil Engineers.
Author keywords: Bearing capacity; Failure mechanism; Strip footings; Numerical modeling; Two-layered clays; Design.
Fig. 1. Boundary conditions of the symmetrical model used for numerical analyses performed in this study
Fig. 2. Variations of Nc for the case of weak-over-strong clays for various H/B ratios
Fig. 3. Comparison between Eq. (3) and the numerical results for the
parameter slope Fig. 4. Comparison between Eq. (4) and the numerical results for the
parameter Nc,max
8
< cut
5:14 þ slope 1 As mentioned, Fig. 2 shows that the ultimate bearing capacity
Nc ¼ min cub (2)
: for the case of weak-over-strong clays did not constantly increase as
Nc;max
cut /cub decreased. Therefore, another equation with the purpose of
determining Nc,max as a function of H/B is needed. Considering the
The parameter slope is a function of H/B. Considering the slopes Nc,max for each H/B case in Fig. 2, Eq. (4) represents a best-fit curve
of each H/B case in Fig. 2, Eq. (3) represents a best-fit curve for cal- for determining Nc,max, as follows:
culating slope in Eq. (2), as follows:
" 6 #
! H
0:25 Nc;max ¼ 5:14 1 þ 1:25 1 (4)
slope ¼ 5:14 (3) B
H = 0:75
B
R2 for Eq. (4) is 0.996. Fig. 4 shows a comparison between Eq. (4)
2
The coefficient of determination (R ) for Eq. (3) is 0.984. Fig. 3 and the numerical results of Nc,max obtained in this study.
compares the suggested values for the parameter slope with the nu- Finally, substitution of Eqs. (3) and (4) in Eq. (2) results in the
merical results presented in Fig. 2. following practical equation, which can be used to determine the
Fig. 5. Variations of Nc for the case of strong-over-weak clays for various cut/cub and H/B ratios
ultimate bearing capacity of strip footings for the case of weak- footing in the top layer was less restricted in moving laterally, and
over-strong clays: hence did not act as a rigid column. Nevertheless, a partial punching
8 shear failure mechanism in the top layer and a large general shear
>
> 0:25 cut
>
> 1 þ 0:75 1 failure mechanism in the bottom layer were still observed (e.g.,
>
< H cub Item B in Fig. 5).
Nc ¼ 5:14 min B
6 (5) Fig. 5 displays the variations of Nc with respect to H/B. Each
>
>
>
> H curve represents variations of this parameter for a specific cut /cub
>
: 1 þ 1:25 1 B ratio. The suggested Eq. (6) has a format similar to Eq. (2). Therefore,
the equation presented for this case should be in the format
According to the comparison made between the prediction of 8
< H
ultimate bearing capacity by Eq. (5) and the numerical results, the cub 5:14 þ slope
Nc ¼ min B (6)
coefficient of determination equals 0.994, which indicates a very cut : N
c;max
good agreement between Eq. (5) and the numerical results for the
case of weak-over-strong clays.
It is obvious that in this case, maximum ultimate bearing
capacity occurred when the failure fully occurred in the top layer,
Strong-over-Weak Clays and the weak bottom layer had no effect on it (Nc = 5.14).
Fig. 5 shows the numerical results of bearing capacity factor ðNc Þ Therefore, Nc;max can be determined using Eq. (7), as follows:
for a series of numerical analyses at different values of H/B and cut
cut /cub ratios. In this case, the parameter Nc is defined as the ratio of Nc;max ¼ 5:14 (7)
cub
the ultimate bearing capacity (qu) to the undrained shear strength of
the weak bottom layer (cub). Variations of Nc with respect to H/B Fig. 5 shows that the parameter slope for this case was a function
are shown in Fig. 5, where each curve represents variations of Nc
of cut /cub and increased as cut /cub increased. To this end, Eq. (8) is
for a specific cut /cub ratio. For example, for the case of cut /cub = 5,
presented as follows:
the detrimental effect of the weak bottom layer on the ultimate bear-
ing capacity persisted for H/B ratios up to 2, whereas for bigger H/B 0:75
cut
ratios, the effect of the weak bottom layer on the bearing capacity slope ¼ 5:14 0:75 1 (8)
cub
disappeared because the failure mechanism was fully contained
within the top layer (Nc = 5.14 or Nc ¼ 5 5:14 ¼ 25:70). Also, it
The coefficient of determination for Eq. (8) is 0.994. A compari-
should be noted that all of the curves in Fig. 5 pass through the com-
son between Eq. (8) and the numerical results carried out in this
mon point of Nc ¼ 5:14 for H/B = 0, which conforms to theoretical
study is presented in Fig. 6.
solutions available for the bearing capacity of single-layered homo-
By substituting Eqs. (7) and (8) in Eq. (6) and converting Nc to
geneous clays (Prandtl 1920).
Nc , the ultimate bearing capacity for the case of strong-over-weak
The results of the analyses indicate that, depending on the value
clays can be determined using Eq. (9), as follows:
of the H/B and cut /cub ratios, full or partial punching shear failure
8"
may occur in the strong top layer. Similar to what is shown in Item > 0:75 #
< cut H cub
A of Fig. 5 for relatively strong, thin crusts (H/B 0.5 and cut /cub 1 þ 0:75 1
Nc ¼ 5:14 min c B cut
3), full punching shear failure occurred, which was characterized >
:
ub
Fig. 6. Comparison of Eq. (8) and the numerical results for the param-
eter slope, for strong-over-weak clays
Discussion ∂Nc
¼0 (13)
∂r
In this section, a comparison is made between the results obtained
in this study and those from several other researchers: lower- and
upper-bound solutions of Merifield et al. (1999), upper-bound solu- ∂Nc
¼0 (14)
tion of Michalowski (2002), semi-empirical solution of Meyerhof ∂u
and Hanna (1975), upper-bound solution of Chen (1975), and FEM
results of Zhu (2004) and Bandini and Pham (2011). This solution gives Nc = 5.53 for the case of single-layered ho-
It should be noted that Merifield et al. (1999), Michalowski mogeneous clay, which is 8% higher than Prandtl's (1920) solution.
(2002), Zhu (2004), and Bandini and Pham (2011) presented Meyerhof and Hanna (1975) provided semi-empirical solutions
their results in the form of charts and figures for various values for determining the ultimate bearing capacity of two-layered clays
of the H/B and cut /cub ratios. Before continuing with the discus- based on their small-scale model footing tests. Eq. (15) was pro-
sion, other approaches employed in solving this problem are posed as follows for the case of strong-over-weak clay:
reviewed.
qu cub ca H
Chen (1975) presented Eq. (10) by using the theories of limit Nc ¼ ¼ 5:14 þ2 (15)
analysis while assuming a circular failure mechanism for calculat- cut cut cut B
ing an upper bound for the ultimate bearing capacity of two-layered
clays, as follows: The second term in this equation accounts for the effect of the
2 punching shear failure of the stronger top layer, where ca varies from
r u þ nu i 1 (homogeneous soil) to 0.8 (cut /cub = 5). For the case of weak-over-
Nc ðr; u Þ ¼ 2 (10)
B ðr =B Þsin u 1 =2 strong clays, they have proposed the following equation:
qu cub H 2
where Eqs. (11) and (12) give Nc ¼ ¼ 5:14 þ 5:14 1 1 5:14 (16)
cut cut Hf
H
u i ¼ cos1 cos u þ (11)
r where H is the top layer's depth; Hf is the depth of the failure surface
beneath the footing in the thick bed of upper soil; and the depth ratio
of Hf/B may be set to 1 for clays (Meyerhof 1955).
cub The results of the aforementioned approaches in solving this prob-
n¼ 1 (12)
cut lem are compared graphically in Figs. 7–15. To save space in these
figures, lower-bound and upper-bound solutions are abbreviated as
Finally, by solving Eqs. (13) and (14) simultaneously and substi- LB and UB, respectively, in each figure's legend. Details of expla-
tuting the values for u and r into Eq. (10) the least upper bound for nations for the comparison of the results obtained in this study and
Nc, is as follows: those obtained by other researchers follow in subsequent sections.
Fig. 10. Variations of Nc for the case of H/B = 0.125 (1 cut =cub 5)
Weak-over-Strong Clays
Comparison is made between the suggested Eq. (5) in this study
and the relationships proposed by other researchers, namely, the
lower- and upper-bound solutions of Merifield et al. (1999), the
upper-bound solutions of Michalowski (2002) and Chen (1975),
the semi-empirical solution of Meyerhof and Hanna (1975), and the
FEM results of Zhu (2004) and Bandini and Pham (2011). The fol-
lowing conclusions are made.
As can be seen in Fig. 2, the ultimate bearing capacity for
this case either solely depended on H/B (Nc,max) or was a func-
Fig. 9. Variations of Nc for the case of H/B = 0.5 (cut /cub 1) tion of both cut /cub and H/B. Figs. 7–9 indicate that for cases
where Nc only depended on H/B, the numerical results obtained
Fig. 11. Variations of Nc for the case of H/B = 0.25 (1 Fig. 12. Variations of Nc for the case of H/B = 0.5 (1 cut =cub 5)
cut =cub 5)
in this study were close to the average of the upper- and However, because no maximum limit for Nc [such as Nc,max in Eq.
lower-bound solutions of Merifield et al. (1999). However, (4)] was predicted in their equation for the case of weak-over-strong
because Nc became a function of cut /cub as well (sloped por- clays, Nc constantly increased as the cut /cub ratio decreased, and a
tion of each curve in Fig. 2), the results obtained in this study high percentage of error was especially produced for cut /cub ratios
were closer to the upper-bound solutions of Merifield et al. less than 0.6.
(1999) and Michalowski (2002), especially for smaller H/B val- In comparison with the FEM results, as can be seen in Figs. 7–9,
ues (Figs. 7 and 8). The graphical comparison in Figs. 7–9 Zhu (2004) estimated relatively higher values for the ultimate
also indicates that the method of Chen (1975) results in the bearing capacity, which occasionally exceeded the upper-bound
least accurate upper-bound solution. solutions of Merifield et al. (1999). However, the results of
As for Meyerhof and Hanna's (1975) semi-empirical solution, it Bandini and Pham (2011) and this study are very close. However,
can be concluded that their solution gives a good approximation of Bandini and Pham (2011) did not analyze this problem for cut /cub
the ultimate bearing capacity for the sloped portion of Fig. 2. ratios smaller than 0.5, and this could lead to significant errors in
Fig. 13. Variations of Nc for the case of H/B = 1 (1 cut =cub 5) Fig. 14. Variations of Nc for the case of H/B = 1.5 (1 cut =cub 5)
qu ¼ cut Nc
Conclusion