Professional Documents
Culture Documents
Unified Method For Estimating The Ultimate Bearing
Unified Method For Estimating The Ultimate Bearing
Unified Method For Estimating The Ultimate Bearing
Abstract: This study presents a unified method for estimating the ultimate bearing capacity of shallow foundations resting on variably
saturated soils under a steady flow. The effective stress approach originally employed for predicting the bearing capacity of saturated soils
is extended by incorporating a suction stress-based representation. A closed-form equation is used to define the suction stress characteristic
curve, capturing changes in the effective stress because of varying soil saturation and matric suction. The proposed method uses the classic
effective shear strength parameters and two fitting parameters to represent the soil water characteristic curve. The method can be used for
different soil types, various degrees of saturation, and different surface flux boundary conditions. The results from the proposed formulation
were compared, and good agreement was observed against three sets of experimental results from model footing and plate load tests. A
parametric study was performed, and the ultimate bearing capacity profiles are presented for various surface flux boundary conditions and
depths of the water table. It is shown that different flow conditions can significantly affect the ultimate bearing capacity of clay. However,
different flow conditions have negligible effects on the ultimate bearing capacity of sand. DOI: 10.1061/(ASCE)GT.1943-5606.0001445.
© 2015 American Society of Civil Engineers.
Author keywords: Variably saturated soil; Bearing capacity; Effective stress; Suction stress.
Dw = 10 m
1m
0m
0m
1m
m
m
m
0m
m
1m
m
1
10
10
=1
0
10
=
=1
Dw = 1
=1
=1
w =
Dw =
=
Dw
w =
Dw =
=
w
Dw
D
D
m
w
m
m
D
D
=1
=1
=1
0.8
w
w
w
D
D
D
Clay Silt Sand
α = 0.01 (kPa )
-1 -1 -1
α = 0.005 (kPa ) α = 0.1 (kPa )
0.6
γ = 18 kN/m
3
γ = 18 kN/m
3
γ = 18 kN/m3
-8 -7 -5
ks = 5 x 10 m/s ks = 1 x 10 m/s ks = 3 x 10 m/s
0.4
-8
q = -3.14 x 10 m/s
0.2 q = 0 m/s
Downloaded from ascelibrary.org by University of California, San Diego on 12/28/15. Copyright ASCE. For personal use only; all rights reserved.
-8
q = 1.15 x 10 m/s
0.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Normalized Matric Suction, (ua-uw)/γ Dw
(a) (b) (c)
Fig. 2. Normalized suction profiles under various flow conditions for three hypothetical soils: (a) clay; (b) silt; (c) sand
water table increases (z > 0 m). Consequently, the hydrostatic case used for estimating qult for both fine- and coarse-grained soils
for nonzero flow conditions will underestimate the matric suction under various surface flux boundary conditions and with different
value as the depth of the water table increases beneath a foundation. degrees of saturation.
In Figs. 2(a and b), the influence of one-dimensional steady flow As demonstrated in Fig. 3, different cases are considered in this
conditions on the matric suction profiles for clay and silt is signifi- study to appropriately account for the depth of the water table and,
cant, with the greatest effect corresponding to the infiltration case consequently, various degrees of saturation. Changes in qult asso-
(q ¼ −3.14 × 10−8 m=s). Conversely, Fig. 2(c) shows that the ciated with variations in the depth of the water table have typically
steady-state flow rates only begin to influence the behavior of suc- been accounted for by modifying qo and γ (Bowles 1996). Terzaghi
tion in sands at a height of approximately z=Dw ¼ 0.3 above the and Peck (1948) first showed that the qult of shallow foundations
water table. could be reduced by approximately 50% when the depth of the
water table is approximately one footing width beneath the ground
surface. This necessitates modifications to qo and γ when Dw varies
Unified Bearing Capacity Formulation above and below the foundation, as exemplified in Fig. 3. As a re-
sult, Eq. (12) adjusts qo and γ as Dw varies above the foundation,
In this study, the suction stress-based effective stress approach is within the stress bulb, and below the stress bulb, as illustrated in
incorporated in Terzaghi’s (1943) bearing capacity equation Fig. 3 for Cases 1–4, respectively.
through the total cohesion concept. The total cohesion approach • Case 1: The water table is above the foundation, i.e., Dw;1 < Df .
has been extensively used in unsaturated slope stability methods For this case, (ua − uw ) and ðua − uw ÞAVR are 0 within the stress
(e.g., Fredlund and Rahardjo 1993; Springman et al. 2003; Gavin bulb. However, qo needs to be modified to account for the depth
and Xue 2010). A term referred to as total cohesion is introduced by of the water table above the foundation as
adding the apparent cohesion to the effective stress cohesion,
i.e., total cohesion ¼ c 0 þ capp . The total cohesion is then used qo ¼ H 1 γ þ H2 γ 0 ð13Þ
in Terzaghi’s bearing capacity equation to compute the ultimate
bearing capacity of variably saturated soils. In the unified effective where γ 0 = buoyant unit weight defined as the difference
stress theory, the apparent cohesion is defined as the shear strength between the saturated unit weight of soil (γ sat ) and γ w . Conse-
mobilized by σs through the internal friction angle capp ¼ quently, for this case, the proposed equation smoothly degener-
−σs tan ϕ 0 (Lu et al. 2009). Following this approach, the ultimate ates to Terzaghi’s classic bearing capacity formulation.
bearing capacity of variably saturated soils can be estimated as
follows: B
0 0 0
qult ¼ fc þ að1 − Se;AVR Þ tan ϕ þ ½ðua − uw ÞSe AVR tan ϕ gN c ξ c z1
Dw,1 Case 1 H1
~ γ ξγ
þ qo N q ξ q þ 0.5γBN ð12Þ Df
Dw,2 H2
z2
where a represents the air-entry pressure, i.e., inverse of α, in kPa; Dw,3
Case 2
and Se;AVR represents the average effective degree of saturation, 1B
Dw,4
which is a function of the average matric suction ðua − uw ÞAVR , z3
1.5B
within the proximity of the stress bulb zone, i.e., 0 to 1.5B, Case 3
below a shallow foundation. In Eq. (12), the second term,
að1 − Se;AVR Þ tan ϕ 0 , embodies the contribution of matric suction z4
Case 4
up to the air-entry pressure. Additionally, the third term,
½ðua − uw ÞSe AVR , i.e., σsAVR , delineates the contribution of the
Fig. 3. Different cases for estimating the ultimate bearing capacity on
average suction stress beyond the air-entry pressure. By selecting
basis of depth of water table
the appropriate input parameters, the proposed formulation can be
Δx Δx
z b b
x x d x d
Δy
1.5B Dwu (xc, yc) 1.5B Δy 1.5B
z
(xc, yc) c
y = f(x); Dwu a
x = f(y)
y = f(x); Dwl
z x = f(y)
y y y
(a) (b) (c)
Downloaded from ascelibrary.org by University of California, San Diego on 12/28/15. Copyright ASCE. For personal use only; all rights reserved.
Fig. 4. Integration of average matric suction within stress bulb: (a) Case 1; (b) Cases 2 and 3; (c) Case 4
Z Z Z
• Case 2: The water table is between the bottom of the foundation 1 1 d 1 Duw
ȳ ¼ yc dA ¼ yfðyÞdy ¼ yfðyÞdy ð15bÞ
and a depth equal to the width of the foundation below that, A A A c A Dlw
i.e., 0 < ðDw;2 − Df Þ < B. For Case 2, ðua − uw ÞAVR and the
related terms are calculated by considering the matric suction where x̄ = horizontal coordinate of the centroid of the matric suc-
profile from the base of the foundation, i.e., Df , down to tion profile and characterizes the average matric suction within the
(Dw;2 − Df ). Moreover, qo and γ should be modified as stress bulb; ȳ defines the vertical coordinate of the centroid of the
matric suction profile within the stress bulb, A defines the area of
~ f
qo ¼ γD ð14aÞ the stress bulb bounded by the horizontal and vertical limits a − d,
on x̄ and ȳ, respectively; and xc and yc = horizontal and vertical
‘moment arms’ (or coordinates), respectively, of the differential
γ~ ¼ γ 0 þ fðγ − γ 0 Þ ð14bÞ elements, as illustrated in Fig. 4.
In Eq. (15a) and Fig. 4, ðua − uw Þu , i.e., horizontal limit b, char-
where γ~ represents the modified average unit weight; and f acterizes the upper limit of integration, which is the matric suction
represents the ratio of (Dw;2 − Df ) and B, varying from 0, at the depth of the foundation for Cases 1–4. In Eq. (15a), the lower
i.e., fully saturated, to 1, i.e., water table at the depth of limit of the integral, i.e., horizontal limit a, is defined as ðua − uw Þl .
Dw;2 − Df ¼ B. In Eq. (15b) and Fig. 4, Duw , i.e., vertical limit d, is the upper limit of
• Case 3: This case represents a condition where B < ðDw;3 − integration and represents the distance from the water table to the
Df Þ < 1.5B. As a result, Eq. (12) analyzes ðua − uw ÞAVR depth of the foundation. Similarly, Dlw , i.e., vertical limit c, repre-
and the related terms by considering the matric suction profile sents the lower limit of integration. As previously demonstrated in
from the base of the foundation, i.e., Df , down to (Dw;3 − Df ). Cases 1–4, the limits of the integrals in Eqs. (15a) and (15b) are
No modification is needed for qo and γ in this case. defined as follows:
• Case 4: The water table is very deep in this case, • For Case 1 [Fig. 4(a)], the upper and lower limits in both
i.e, ðDw;4 − Df Þ ≥ 1.5B. In such a case, the water table will have equations coincide, representing a saturated case; therefore,
no effect on the depth of the stress bulb zone, and Eq. (12) ana- the average suction is 0.
lyzes ðua − uw ÞAVR and the pertinent terms by considering the • For Cases 2 and 3 [Fig. 4(b)] ðua − uw Þl is equal to the matric
profile from Df down to 1.5B. suction calculated at (Dw − Df ), Duw ¼ Dw − Df , and Dlw ¼ 0.
The modifications proposed for qo and γ are consistent with • For Case 4 [Fig. 4(c)], ðua − uw Þl is the matric suction calcu-
those suggested in classic bearing capacity studies, which account lated at the base of the stress bulb, i.e., 1.5B, Duw ¼ Dw − Df ,
for variations in the depth of the water table (e.g., Terzaghi and and Dlw is nonzero and is equal to the distance between the depth
Peck 1948). of the water table and the base of the stress bulb.
Calculating the Average Matric Suction within the Parametric Analysis and Examples
Stress Bulb
A set of parametric studies was performed to investigate the varia-
By taking into account variations in the depth of the water table and tion in qult for hypothetical fine- and coarse-grained soils under
various surface flux boundary conditions, the proposed method various surface flux boundary conditions. Because of a space limi-
uses numerical integration of differential elements to calculate tation, the results are presented for only two hypothetical soil types,
the average matric suction, i.e., abscissa of the centroid, within namely, clay and sand, with variations in the depth of the water
the stress bulb zone below a foundation. As demonstrated in Fig. 2, table, i.e., 0 ≤ Dw ≤ 10 m, beneath the depth of the foundation.
as the depth of the water table increases beneath a foundation, the Input parameters for the hypothetical soils and different flow rates
matric suction profile becomes nonlinear for fine- and coarse- were selected on the basis of values reported by Lu and Likos
grained soils under various surface flux boundary conditions. As (2004, 2006).
a result, the geometric center, i.e., centroid, of the matric suction
profile within the stress bulb can be calculated as Effect of Soil Types and Flow Rates on the Ultimate
Z Z Z Bearing Capacity
1 1 b 1 ðua −uw Þu
x̄ ¼ xc dA ¼ xfðxÞdx ¼ xfðxÞdx ð15aÞ The results shown in Figs. 5–8 demonstrate the effect of Dw , q,
A A A a A ðua −uw Þl
and ϕ 0 on ðua − uw ÞAVR , Se;AVR , and qult in clay and sand.
95 3000 o
25
c' = 10 kPa φ'=
80 o
α = 0.005 (kPa )
-1
25
φ' =
90 n = 1.8 2000
3
γ = 18 kN/m o
0
-8
ks = 5 x 10 m/s φ' = 2 o
40 -8 φ' = 20
85 q = -3.14 x 10 m/s 1000 o
q = 0 m/s φ' = 25
Downloaded from ascelibrary.org by University of California, San Diego on 12/28/15. Copyright ASCE. For personal use only; all rights reserved.
-8
q = 1.15 x 10 m/s o
φ' = 20
0 80 0
0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12
Depth of Water Table, Dw (m)
(a) (b) (c)
Fig. 5. Results for surface foundation (Df ¼ 0) on hypothetical clay under various flow rates: (a) ðua − uw ÞAVR ; (b) Se;AVR ; (c) qult
Average Effective Degree of Saturation, Se,AVR (%)
o
25
95 3000
o
80 c' = 10 kPa 20
-1 φ' =
α = 0.005 (kPa )
90 n = 1.8 2000 o
γ = 18 kN/m
3 φ' = 25
-8 o
ks = 5 x 10 m/s φ' = 2
0
40 -8
85 q = -3.14 x 10 m/s
1000 o
q = 0 m/s φ' = 20
-8
q = 1.15 x 10 m/s
0 80 0
0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12
Depth of Water Table, Dw (m)
(a) (b) (c)
Fig. 6. Results for embedded foundation (Df ¼ 1.5 m) on hypothetical clay under various flow rates: (a) ðua − uw ÞAVR ; (b) Se;AVR ; (c) qult
Average Effective Degree of Saturation, Se,AVR (%)
c' = 0 kPa
Ultimate Bearing Capacity, qult (kPa)
α = 0.1 (kPa )
-1
100 100
n=4
γ = 18 kN/m
3
3000
-5
ks = 3 x 10 m/s
80 80 -8
q = -3.14 x 10 m/s
q = 0 m/s
-8
60 60 q = 1.15 x 10 m/s 2000 o
φ' = 35 (x 1.1)
40 40
1000 o
φ' = 30 (x 1.1)
20 20
0 0 0
0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12
Depth of Water Table, Dw (m)
(a) (b) (c)
Fig. 7. Results for surface foundation (Df ¼ 0) on hypothetical sand under various flow rates: (a) ðua − uw ÞAVR ; (b) Se;AVR ; (c) qult
n=4 3000
3
γ = 18 kN/m
80 80 -5
ks = 3 x 10 m/s
-8
q = -3.14 x 10 m/s
q = 0 m/s o
60 60 -8 2000 φ' = 30
q = 1.15 x 10 m/s
40 40
1000
20 20
Downloaded from ascelibrary.org by University of California, San Diego on 12/28/15. Copyright ASCE. For personal use only; all rights reserved.
0 0 0
0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12
Depth of Water Table, Dw (m)
(a) (b) (c)
Fig. 8. Results for embedded foundation (Df ¼ 1.5 m) on hypothetical sand under various flow rates: (a) ðua − uw ÞAVR ; (b) Se;AVR ; (c) qult
In Figs. 5 and 7, the results are representative of the surface 1% for up to Dw ¼ 10 m for the evaporation, hydrostatic, and
foundations, i.e., Df ¼ 0, in clay and sand, respectively, with a infiltration conditions, respectively. A lower value for Se;AVR com-
width, B, of 1 m. Figs. 6 and 8 show the counterpart results in clay pliments higher values for ðua − uw ÞAVR and σsAVR , leading to a
and sand, respectively, for an embedded foundation with Df ¼ larger qult. Specifically, the 15% drop in Se;AVR under the evapo-
1.5 m and B ¼ 1 m. The following parameters were used for ration condition contributes to more than a 50% increase in qult
the hypothetical clay: γ ¼ 18 kN=m3 , n ¼ 1.8, α ¼ 0.005 kPa−1 , [Fig. 5(c)] under the same condition.
ks ¼ 5 × 10−8 m=s, c 0 ¼ 10 kPa, and two values of ϕ 0 (20 and As it is shown in Fig. 5(c), qult shows a similar bilinear trend for
25°); and for the hypothetical sand, the parameters were: γ ¼ fine-grained soils, with the intersection point equal to the depth of
18 kN=m3 , n ¼ 4.0, α ¼ 0.1 kPa−1 , ks ¼ 3 × 10−5 m=s, c 0 ¼ 0, the stress bulb, i.e., 1.5B. By incorporating the effect of q in Eq. (12),
and two values of ϕ 0 (30° and 35°). A total of three flow rates qult at each ϕ 0 value varies by approximately 70, 65, and 45% for up
of q, 3.14 × 10−8 (infiltration), 0 (hydrostatic), and 1.14 × to Dw ¼ 10 m for the evaporation, hydrostatic, and infiltration con-
10−8 m=s (evaporation), were used for both soil types in the ditions, respectively. More specifically, qult at ϕ 0 ¼ 25° decreases
parametric study. from approximately 2,700 kPa for the evaporation condition q ¼
The effect of dilation for the surface foundations, i.e., Df ¼ 0, 1.15 × 10−8 m=s to approximately 950 kPa for the infiltration con-
in sand (e.g., Fig. 7) was considered by increasing ϕ 0 by 10%, dition q ¼ −3.14 × 10−8 m=s. Likewise, qult at ϕ 0 ¼ 20° decreases
i.e., 1.1ϕ 0 , as recommended by Steensen-Batch et al. (1987) and from approximately 1,450 to 550 kPa for the evaporation and infil-
Oh and Vanapalli (2011). However, as the depth of a foundation tration conditions, respectively. This shows that the infiltration rate
increases below the ground surface, i.e., Df > 0, the overburden q ¼ −3.14 × 10−8 m=s results in the largest effect on qult in variably
pressure acts as a confining pressure, therefore decreasing the in- saturated clayey soils. However, the effect of q on qult is not as
fluence of dilation. For that reason, the dilatancy effect was not pronounced for lower values of Dw , which corresponds to minimal
considered when predicting qult in sand at Df > 0. It has been values for ðua − uw ÞAVR, Se;AVR , and σsAVR .
noted in the literature (e.g., Oh and Vanapalli 2013) that the effect The results in Figs. 6(a–c) characterize ðua − uw ÞAVR , Se;AVR ,
of dilation is not as pronounced for foundations in fine-grained and qult , with the variables Dw , q, and ϕ 0 for an embedded foun-
soils. Thus, the original ϕ 0 value was employed in this study for dation with Df ¼ 1.5 m and B ¼ 1 m. The shift away from the
calculating qult , as shown in Figs. 5 and 6, for foundations resting ordinate, i.e., Dw ¼ 1.5B, represents the increase in Dw that is re-
on fine-grained soils. quired to achieve the same equilibrium conditions as presented in
The values shown in Fig. 5(a) represent ðua − uw ÞAVR at the Figs. 5(a–c). For an increase in Df , it is shown that the intersection
limit state versus Dw . The trend in Fig. 5(a) is almost identical to point in each curve shifts to Dw ¼ 1.5B þ Df . As demonstrated,
the matric suction profile presented in Fig. 2(a). Such behavior is qult at Df ¼ 1.5 m, in comparison to qult at the surface, is increased
expected when implementing numerical integration to calculate by approximately 60% under each surface flux boundary condition.
ðua − uw ÞAVR . As it is shown, ðua − uw ÞAVR exhibits a bilinear trend More specifically, qult at ϕ 0 ¼ 25° for the embedded foundation
for fine-grained soils, with the intersection point equal to the depth of descends from approximately 4,450 kPa for the evaporation
the stress bulb, i.e., 1.5B, as represented by the gray vertical refer- condition q ¼ 1.15 × 10−8 m=s to approximately 2,000 kPa for
ence axis. Beyond this particular point, ðua − uw ÞAVR for each flow the infiltration condition q ¼ −3.14 × 10−8 m=s. Likewise, qult
condition increases roughly 90% for up to Dw ¼ 10 m. It is apparent at ϕ 0 ¼ 20°, relative to qult at ϕ 0 ¼ 25°, decreases by approximately
that the most significant effect of q on ðua − uw ÞAVR in clay takes 50% from 2,400 to 1,100 kPa for the evaporation and infiltration
place under infiltration conditions. However, the effect of q on the conditions, respectively. As seen in Fig. 6(c), qult at each flow con-
rate of change in ðua − uw ÞAVR is more pronounced, i.e., sharper dition begins to depart from one another as Dw increases past the
slope, under the evaporation condition as Dw increases from 0 to stress bulb zone, i.e., Dw > 1.5B þ Df . With an increase in Dw ,
10 m below the depth of the foundation. the largest effect of q on qult is observed under the evaporation
Fig. 5(b) illustrates the relationship between Se;AVR and Dw . As condition, i.e., steepest slope.
desaturation commences (Se;AVR < 100%) near the intersection The results in Fig. 7 demonstrate the effect of Dw , q, and ϕ 0
point in Fig. 5(b), Se;AVR varies by approximately 15, 10, and on ðua − uw ÞAVR , Se;AVR , and qult for a surface foundation in sand.
in qult up to Dw ¼ 1.5B [Fig. 7(c)]. As discussed, a lower value Mohamed (2013), and Mohamed (2014) provide further discussion
for Se;AVR compliments a larger qult. For example, as desaturation on the specifics, e.g., dimensions and material, of the UOBCE.
begins near Dw ¼ 0.5 m in Fig. 7(b), a greater contribution to shear
strength is achieved, which results in a higher qult. More specifi-
cally, Se;AVR in Fig. 7(b) decreases approximately 30% up to Soil Properties
the maximum value of qult at Dw ¼ 2B [Fig. 7(c)]. The soil used for the MFTs and PLTs is characterized as poorly
As shown in Fig. 7(c), the effect of q on qult in sand is incom- graded Unimin sand. The average water content, moist unit weight,
parable in relation to the effect of q on qult in clay. In other words, and void ratio of the Unimin sand were approximately 20%,
the effect of q in sand shows little to no effect on qult . This behavior 19 kN=m3 , and 0.63, respectively. The classical effective shear
is directly related to the amount of suction stress that can be de- strength parameters were obtained through a series of laboratory
veloped in sand. In Fig. 7(c), the relationship between qult and direct shear tests conducted following ASTM D3080 (Mohamed
Dw is presented for different values of ϕ 0 and various surface flux 2014). The effective cohesion and internal friction angle of the
boundary conditions for foundations with Df ¼ 0 and B ¼ 1 m.
Unimin sand were found to be 0.6 kPa and 35.3°, respectively.
The behavior of qult in sand appears to be serpentine in nature
Mohamed and Vanapalli (2006) tested the foundations under
[Fig. 7(c)], with a linear increase up to the base of the stress bulb
different suction values by lowering the depth of the water table
at Dw ¼ 1.5B. This loci corresponds to a maximum σsAVR of
every 48 h or until equilibrium conditions were achieved for the
approximately −6.0 kPa, i.e., ðua − uw ÞAVR and Se;AVR equal
given condition. For each depth of the water table, Mohamed
9.8 kPa and 60%, respectively. Such behavior is anticipated in var-
and Vanapalli (2006) used a hydrostatic approximation to calculate
iably saturated sands because the effect of suction stress is limited,
the average matric suction within the stress bulb. In the current
i.e., approaches a maximum, to only a small portion above the
study, the reported depth of the water table (Mohamed 2014)
water table (Lu and Likos 2004). Specific to this case, the contri-
was used along with Eq. (15a) to determine the ðua − uw ÞAVR .
bution of the interparticle capillary stress begins to diminish once
A surface flux of approximately q ¼ 4.8 × 10−9 m=s and ks ¼
Dw supersedes 1.5B. Beyond the linear portion of the curve
4.3 × 10−2 m=s were used to match the measured matric suction
(Dw > 1.5B) in Fig. 7(c), qult reaches a global maximum of approx-
imately 1,800 kPa at ϕ 0 ¼ 35° and Dw ¼ 2B. Equally for ϕ 0 ¼ 30° values for each test, as reported by Mohamed and Vanapalli
and Dw ¼ 2B, qult reaches a maximum of almost 700 kPa. Vana- (2006) and Vanapalli and Mohamed (2013). Furthermore, the val-
palli and Mohamed (2013) observed similar behavior in their pre- ues representing q and ks were selected based off of the ranges
dicted and measured values for qult from model PLTs in sands of provided for the tested soil after Lu and Likos (2004, 2006).
analogous c 0 and ϕ 0 . Beyond Dw ¼ 2B in Fig. 7(c), qult at each ϕ 0 Fig. 9(a) presents the predicted and measured SWCC of the
value decreases approximately 6% and approaches a residual value Unimin sand. The data points for establishing the measured SWCC
at approximately Dw ¼ 5B. were obtained by positioning a Tempe cell apparatus at various
The results in Figs. 8(a–c) illustrate ðua − uw ÞAVR , Se;AVR , and depths within the UOBCE (Mohamed and Vanapalli 2006). In
qult , with variations in Dw , q, and ϕ 0 , for an embedded foundation the current study, the data points for establishing the predicted,
with Df ¼ 1.5 m and B ¼ 1 m in the hypothetical sand. As it was i.e., best fit, SWCC and corresponding fitting parameters, α and
previously demonstrated for clay, the shift away from the ordinate, n, were obtained using van Genuchten–Mualem’s model. More-
i.e., Dw ¼ 1.5B, represents the increase in Dw that is required over, Figs. 9(b and c) were obtained using Eqs. (5) and (9), respec-
to achieve similar conditions, as presented in Figs. 7(a–c). In tively. As illustrated in Fig. 9, a decrease and increase in Se and
Fig. 8(a), it is shown that ðua − uw ÞAVR varies under different q val- (ua − uw ), respectively, leads to an momentary increase in σs . More
ues as Dw ranges between 6 and 10 m. However, it is apparent that specifically, the rate of change in σs with Se significantly increases
qult in sand [Fig. 8(c)] is somewhat similar under each flow condi- for a higher Se. For example, a 10% change in Se for soil with an
tion. As illustrated in Fig. 8(c), qult at Df ¼ 1.5 m, in comparison to initial Se of 20% will make a relatively insignificant change in the
qult at Df ¼ 0, increases by more than 100%. Although the mag- value of σs . However, for the same amount of change (10%) in soil
nitudes of ðua − uw ÞAVR , Se;AVR , and qult vary for the surface with an initial Se of 80%, σs will decrease significantly. In Fig. 9(c),
(Fig. 7) and the embedded (Fig. 8) foundations in sand, the overall σs is approximately 0 for a Se close to 100% and increases with a
effect of q on ðua − uw ÞAVR , Se;AVR , and qult is relatively similar. decrease in Se until it reaches a maximum value of approximately
−4 kPa at Se ¼ 80%. Beyond this point, σs decreases rapidly and
again approaches 0 at Se ¼ 0%. This behavior is because the capil-
Validation against Experimental Tests lary interparticle tractions, which occur at the air-water interfaces
where surface tension exist, are the only suction stress component
To investigate the validity of the proposed method for estimating in sand (Lu and Likos 2006). Too dry or too wet conditions
the qult of shallow foundations in variably saturated soils, the fol- will diminish air-water interface areas or will result in very minor
lowing section associates the predicted versus the measured values capillary stresses.
80 80
40 40 -2
best - fit -1
20 20
Measured
Downloaded from ascelibrary.org by University of California, San Diego on 12/28/15. Copyright ASCE. For personal use only; all rights reserved.
-1
α = 0.176 kPa , n = 7.7
0 0 0
0 2 4 6 8 10 0 2 4 6 8 10 0 2 4 6 8 10
Matric Suction, (ua - uw) (kPa)
(a) (b) (c)
Fig. 9. (a) Measured (Mohamed and Vanapalli 2006) and predicted (van Genuchten 1980; Mualem 1976) SWCC for Unimin sand and corresponding;
(b) effective degree of saturation (Se ); (c) suction stress (σs )
Set 1: Surface MFTs value of qult under saturated conditions. As shown in Fig. 10(a)
for Set 1, the proposed method could predict the measured qult , with
Mohamed and Vanapalli (2006) conducted several surface MFTs, a NRMSE of 4%.
i.e., Df ¼ 0 m, to observe the variation of qult in sand for various
values of ðua − uw ÞAVR . The tests were conducted under two differ-
ent conditions, i.e., saturated and unsaturated, with four different Set 2: Surface PLTs
depths of the water table below the model footing of size 0.1 × Vanapalli and Mohamed (2013) conducted several surface PLTs,
0.1 m for B and length (L), respectively. i.e., Df ¼ 0 m, to assess the variation of qult beneath the surface
The relationship between the predicted qult and ðua − uw ÞAVR plate with B ¼ 0.15 m and L ¼ 0.15 m. The PLTs were conducted
for Set 1 is demonstrated in Fig. 10(a). For comparison purposes, at two different conditions, i.e., saturated and unsaturated, with four
the results are also presented for the qult estimated using Terzaghi’s different depths of the water table below the surface plate.
classic bearing capacity formulation. As previously noted, by con- The variation in the proposed qult and Terzaghi’s classical qult ,
sidering the effect of dilatancy, the corrected value for ϕ 0, with respect to ðua − uw ÞAVR is illustrated for Set 2 in Fig. 10(b).
i.e., 1.1ϕ 0 , was considered for computing the qult for surface MFTs Similar to Set 1, 1.1ϕ 0 was used in the calculation to account for
and PLTs. The values for ðua − uw ÞAVR were obtained through the effect of dilatancy in sand. Fig. 10(b) shows that the predicted
Eq. (15a), with variations in the depth of the water table below values for ðua − uw ÞAVR varying between 2 to 6 kPa increase qult
the depth of the foundation. The upper and lower limits were ob- by more than 6 times the value of qult at ðua − uw ÞAVR equal to 0.
tained by Eq. (5) following the discussion pertaining to Cases 1–4. Further, qult expresses a linear demeanor with ðua − uw ÞAVR for up
The predicted values for ðua − uw ÞAVR, ranging between 2 and to 6 kPa. The NRMSE for this set is 4%, and the predicted values
6 kPa, contribute to an increase in qult of more than 5 times the from the proposed method are almost identical to the measured
1400
Ultimate Bearing Capacity, qult (kPa)
1200
1000
800
600 Measured
Proposed Method
Terzaghi
400
NRMSE (Proposed) = 4% NRMSE (Proposed) = 4% NRMSE (Proposed) = 10%
NRMSE (Terzaghi) = 18% NRMSE (Terzaghi) = 18% NRMSE (Terzaghi) = 17%
200
0
0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12
Average Matric Suction (ua - uw)AVR (kPa)
(a) (b) (c)
Fig. 10. Variation in predicted and measured ultimate bearing capacity (qult ) versus average matric suction ðua − uw ÞAVR for (a) Set 1, B ¼ 0.1 m,
Df ¼ 0 m; (b) Set 2, B ¼ 0.15 m, Df ¼ 0 m; (c) Set 3, B ¼ 0.15 m, Df ¼ 0.15 m
qult at Df ¼ 0.15 m. between the proposed predicted and measured values for the em-
Fig. 10(c) embodies the relationship between the predicted, bedded PLTs indicates that the proposed method may slightly
i.e., proposed and Terzaghi’s methods, and measured values for underpredict qult for Df > 0 m under unsaturated conditions.
qult with respect to ðua − uw ÞAVR . The value of NRMSE ¼ 10%
for Set 3 in Fig. 10(c) indicates that the goodness of fit in the pro-
posed qult , in comparison to Sets 1 [Fig. 10(a)] and 2 [Fig. 10(b)], is Summary and Conclusions
not as close to the qult measured. However, the accuracy is still
sensible considering inherent limitations and simplifying assump- A unified method for estimating the ultimate bearing capacity of
tions associated with an analytical closed-form solution. Moreover, shallow foundations in variably saturated soils under numerous sur-
the proposed predicted values of qult for ðua − uw ÞAVR equal to 2 face flux boundary conditions was presented. Terzaghi’s equation
and 6 kPa are merely 1.1 times smaller than the measured values. was extended by implementing a suction stress-based effective
Conversely, the predicted qult under the saturated condition is stress. A total of three independent sets of experimental results
approximately 2.1 times greater than the measured value, which from model footing and plate load tests were used to validate
indicates that Eq. (12), in addition to Terzaghi’s qult , has a tendency the proposed method. Soil suction and its variation with different
to overpredict qult under saturated conditions when Df > 0. As flow conditions, the effective degree of saturation, and the suction
is shown in Fig. 10, qult becomes nonlinear for a greater range stress were demonstrated to have a significant effect on the ultimate
of matric suction. bearing capacity of shallow foundations.
This comparison illustrates the significant effect of matric suc- Values of the normalized root-mean-square error were provided
tion in estimating the ultimate bearing capacity of shallow founda- to evaluate the variation between the predicted and measured
tions resting on variably saturated soils. Specifically, the results ultimate bearing capacities for Sets 1–3. The overall trends and cor-
from Sets 1–3 indicate that qult can be significantly underestimated responding values of the normalized root-mean-square error sug-
by not considering the effect of matric suction. However, great cau- gest that the proposed method forecasts the ultimate bearing
tion must be taken in counting on suction for design purposes. It is capacity rather well for the surface model footings, i.e., Sets 1
because of uncertainties associated with and a huge variation of and 2, of various sizes. Specifically, the proposed method appeared
matric suction during the life span of a foundation (e.g., Vahedifard to predict the ultimate bearing capacity slightly better for Set 2,
et al. 2015a). Moreover, the range of possible matric suction values whereas the predicted ultimate bearing capacity values for the em-
can be quite large depending on the SWCC parameters used in the bedded case, i.e., Set 3, were not as accurately predicted. Further-
analysis. The proposed method presents a useful tool for explaining more, the results from Sets 1–3 indicate that the ultimate bearing
capacity can be significantly underestimated by not considering the
effect of matric suction.
1200 The effect of flow rates, q, on the ultimate bearing capacity
for hypothetical clay and sand was studied. For each soil type,
Measured Ultimate Bearing Capacity, qult (kPa)
the values for the ultimate bearing capacity were compared under
1000
infiltration (q < 0), hydrostatic (q ¼ 0), and evaporation (q > 0)
conditions. As expected, the ultimate bearing capacity under var-
800 iably saturated and various flux boundary conditions significantly
dominated the ultimate bearing capacity under saturated conditions
as much as 5–7 times. Studying the effect of q for each soil type
600 showed that the use of different q values can significantly affect the
ultimate bearing capacity of shallow foundations in clay. However,
the effect of q on the ultimate bearing capacity diminished sand.
400 1:1 line
Set 1, B=0.1 m, Df = 0 m In summary, the study showed that the ultimate bearing capacity
NRMSE = 4% for shallow foundations in variably saturated soils under different
Set 2, B=0.15 m, Df = 0 m
200
flow conditions could be estimated rather well when taking into
NRMSE = 4%
Set 3, B=0.15 m, Df = 0.15 m account the role of matric suction. The new analytical approach
NRMSE = 10% only requires two controlling parameters, i.e., α and n, and can
0 be implemented to provide an accurate and practical framework
0 200 400 600 800 1000 1200 to analyze the ultimate bearing capacity for shallow foundations
Predicted Ultimate Bearing Capacity, qult (kPa) resting on various soil types at different degrees of saturation.
The proposed method presents a useful tool for explaining the true
Fig. 11. Predicted versus measured ultimate bearing capacity (qult ) for
behavior of foundations and the back-analysis of current founda-
Sets 1–3
tions. However, great caution must be taken in counting on suction
deposit.” Geotech. Test. J., 2(2), 219–226. Steensen-Bach, J. O., Foged, N., and Steenfelt, J. S. (1987). “Capillary
Fredlund, D. G. (1996). “The scope of unsaturated soil mechanics: An over- induced stresses—Fact or fiction?” Proc., 9th European Conf. on
view.” Proc. 1st Int. Conf. on Unsaturated Soils, E. E. Alonso and Soil Mechanics and Foundation Engineering, Balkema, Rotterdam,
P. Delage, eds., Vol. 3, Balkema, Rotterdam, Netherlands, 1155–1177. Netherlands, 83–89.
Fredlund, D. G., and Morgenstern, N. R. (1977). “Stress state variables for Terzaghi, K. (1943). Theoretical soil mechanics, Wiley, New York.
unsaturated soils.” J. Geotech. Engrg. Div., 103(5), 447–466. Terzaghi, K. and Peck, R. B. (1948). Soil mechanics in engineering prac-
Fredlund, D. G. and Rahardjo, H. (1993). Soil mechanics for unsaturated tice, Wiley, New York.
soils, Wiley, New York, 517. Vahedifard, F., Leshchinsky, B., Mortezaei, K., and Lu, N. (2015a). “Active
Gardner, W. R. (1958). “Some steady state solutions of the unsaturated earth pressures for unsaturated retaining structures.” J. Geotech. Geo-
moisture flow equation with application to evaporation from a water environ. Eng., 10.1061/(ASCE)GT.1943-5606.0001356, 04015048.
table.” Soil Sci., 85(4), 228–232. Vahedifard, F., Leshchinsky, D., Mortezaei, K., and Lu, N. (2015b).
Gavin, K. and Xue, J. (2010). “Design charts for the stability analysis of “Effective stress-based limit equilibrium analysis for homogenous
unsaturated soil slopes.” Geotech. Geolog. Eng., 28(1), 79–90. unsaturated slopes.” Int. J. Geomech., 10.1061/(ASCE)GM.1943-
Griffiths, D. V., and Lu, N. (2005). “Unsaturated slope stability analysis 5622.0000554, in press.
with steady infiltration or evaporation using elasto-plastic finite Vanapalli, S. K., Fredlund, D. G., Pufahl, D. E., and Clifton, A. W. (1996).
elements.” Int. J. Numer. Anal. Meth. Geomech., 29(3), 249–267. “Model for the prediction of shear strength with respect to soil suction.”
Kumbhokjar, A. S. (1993). “Numerical evaluation of Terzaghi’s.” J. Geo- Can. Geotech. J., 33(3), 379–392.
tech. Eng., 10.1061/(ASCE)0733-9410(1993)119:3(598), 598–607. Vanapalli, S. K., and Mohamed, F. M. O. (2007). “Bearing capacity of
Lu, N. (2008). “Is matric suction a stress variable?” J. Geotech. Geoen- model footings in unsaturated soils.” Experimental unsaturated soil
viron. Eng., 10.1061/(ASCE)1090-0241(2008)134:7(899), 899–905. mechanics, T. Schanz, ed., Springer, Berlin, 483–493.
Lu, N., Godt, J. W., and Wu, D. T. (2010). “A closed-form equation for Vanapalli, S. K., and Mohamed, F. M. O. (2013). “Bearing capacity and
effective stress in unsaturated soil.” Water Resour. Res., 46(5), settlement of footings in unsaturated sands.” Can. Geotech. J., 5(1),
W05515. 595–604.
Lu, N., Kim, T.-H., Sture, S., and Likos, W. J. (2009). “Tensile strength of van Genuchten, M. Th. (1980). “A closed-form equation predicting the hy-
unsaturated sand.” J. Geotech. Geoenviron. Eng., 135(12), 1410–1419. draulic conductivity of unsaturated soils.” Soil Sci. Soc. Am., 44(5),
Lu, N., and Likos, W. J. (2004). Unsaturated soil mechanics, Wiley, 892–898.
New York. Vesic, A. B. (1973). “Analysis of ultimate loads of shallow foundations.”
Lu, N., and Likos, W. J. (2006). “Suction stress characteristic curve for J. Soil Mech. and Found. Div., 99(1), 45–73.
unsaturated soil.” J. Geotech. Geoenviron. Eng., 10.1061/(ASCE) Wuttke, F., Kafle, B., Lins, Y., and Schanz, T. (2013). “Macroelement
1090-0241(2006)132:2(131), 131–142. for statically loaded shallow strip foundation resting on unsaturated
Mohamed, F. M. O. (2014). “Bearing capacity and settlement behaviour of soil.” Int. J. Geomech., 10.1061/(ASCE)GM.1943-5622.0000254,
footings subjects to static and seismic loading conditions in unsaturated 557–564.
sandy soils.” Ph.D. dissertation, Univ. of Ottawa, Ottawa. Yeh, T.-C. (1989). “One-dimensional steady state infiltration in hetero-
Mohamed, F. M. O. and Vanapalli, S. K. (2006). “Laboratory investigations geneous soils.” Water Resour. Res., 25(10), 2149–2158.
for the measurement of the bearing capacity of an unsaturated coarse- Zhou, A., and Sheng, D. (2015). “An advanced hydro-mechanical constit-
grained soil.” Proc., 59th Canadian Geotechnical Conf., Canadian utive model for unsaturated soils with different initial densities.”
Geotechnical Society, Vancouver, BC, Canada, 219–226. Comput. Geotech., 63, 46–66.
Mualem, Y. (1976). “A new model for predicting hydraulic conductivity of Zhou, A.-N., Sheng, D., Sloan, S. W., and Gens, A. (2012a). “Interpretation
unsaturated porous media.” Water Resour. Res., 12(3), 513–522. of unsaturated soil behaviour in the stress-saturation space. I: Volume
Oh, W. T., and Vanapalli, S. K. (2011). “Modelling the applied vertical change and water retention behaviours.” Comput. Geotech., 43,
stress and settlement relationship of shallow foundations in saturated 178–187.
and unsaturated sands.” Can. Geotech. J., 48(3), 425–438. Zhou, A.-N., Sheng, D., Sloan, S. W., and Gens, A. (2012b). “Interpretation
Oh, W. T., and Vanapalli, S. K. (2013). “Interpretation of the bearing capac- of unsaturated soil behaviour in the stress-saturation space. II: Constit-
ity of unsaturated fine-grained soil using the modified effective and the utive relationships and validations.” Comput. Geotech., 43, 111–123.