Professional Documents
Culture Documents
Lin Et Al 2020 Generalized Plastic Mechanics Based Constitutive Model For Estimation of Dynamic Stresses in Unsaturated
Lin Et Al 2020 Generalized Plastic Mechanics Based Constitutive Model For Estimation of Dynamic Stresses in Unsaturated
Abstract: This paper presents a generalized plastic mechanics–based constitutive model for the estimation of dynamic stresses in unsatu-
rated subgrade soils under repeated vehicle loading conditions. The stresses in subgrade soils are taken as the sum of two components: dy-
namic stress and total suction. The increment of dynamic stress consists of three subcomponents, including coaxial stress increment, stress
increment due to rotations of principal stresses under vehicle loading, and stress increment due to stress accumulation under repeated loading.
The total suction in unsaturated subgrade soils is considered to comprise two subcomponents: wet suction characterizing the capillary effect
between liquid and solid phases; and structure suction quantifying the adsorptive and bonding effects between soil particles. The proposed
dynamic model is then solved using the numerical nonlinear incremental calculation method. A case study is presented in the end to elaborate
on the application of the proposed model in practice. The calculated stresses in subgrade soils using both the proposed model and the con-
ventional equivalent static load model are compared against the measured values in the field. Through the case study, the advantages and the
practical value of the proposed model are demonstrated. DOI: 10.1061/(ASCE)GM.1943-5622.0001704. © 2020 American Society of Civil
Engineers.
Author keywords: Subgrade; Unsaturated soils; Constitutive model; Effective dynamic stress; Stress accumulation.
Car
Location B D
pr irec Direction of major
in tio
cip n principal stress
al of
Embankment str m
es ajo
s r σ1B
σ1A
Soil element σ 3A
Downloaded from ascelibrary.org by "National Institute of Technology, Calicut" on 01/17/24. Copyright ASCE. For personal use only; all rights reserved.
σ 3B σ 3B
Subgrade soils
σ1A
(a) (b) σ A
3
σ1B
Fig. 1. Stress changes in subgrade soils due to a moving car: (a) schematics of car driving down the road; and (b) rotation of major principal stress for
a soil element.
Similarly, the stress increment dσr can be defined, but both ma-
Decomposition of Dynamic Stress Increments
trices T1 and T1T would be a function of rotation angle θ. Differen-
As shown in Fig. 1, when a car moves from Location A to Location tiating Eq. (1) gives
B, the direction of principal stress rotates, and the direction of major
principal stress always points to the center of the distributed dy- dσ r = d(T1 ∧ T1T ) = dT1 ∧ T1T + T1 ∧ dT1T (4)
namic load. The dynamic stress increments can be decomposed
From Eq. (2), the derivative of T1 is written
into two components: coaxial part and rotational part (Liu et al.
1998). Hence, the complex problem associated with the rotation −sin θ −cos θ
of dynamic stress increments is simplified as the combination of dT1 = dθ (5)
cos θ −sin θ
a coaxial model and a pure rotation problem for principal stress in-
variants. In a two-dimensional (2D) plane, as shown in Fig. 3, the The stress increment dσr in the major principal stress space can
two principal stresses σ1 and σ2 have direction vectors of N1 and be derived by
T1T dσ r T1 = T1T (dT1 ∧ T1T + T1 ∧ dT1T )T1 = (T1T dT1 ) ∧ (T1T T1 ) + (T1T T1 ) ∧ (dT1 T1T )
0 −1 σ1 0 σ1 0 0 −1 0 dθ(σ 1 − σ 2 ) (6)
= dθ I +I dθ =
1 0 0 σ2 0 σ2 1 0 dθ(σ 1 − σ 2 ) 0
Hence, the total stress increment dσ in the 2D space is the sum- is written
mation of dσc and dσr as follows: ⎛ ⎞⎛ ⎞
σ1 0 0 N1
σ = (N1 N2 N3 )⎝ 0 σ2 0 ⎠⎝ N2 ⎠ = T ∧ T T (8)
0 0 σ3 N3
k1 0 0 k2
dσ = dσ c + dσ r = T1 T1T + T1 T1T (7)
0 k3 k2 0 The stress increment dσ in the major principal stress space has
the following form:
⎛ ⎞
where k1 = dσ1; k3 = dσ2 ; and k2 = dθ (σ1–σ2). M1 A1 C1
Similarly, for three-dimensional (3D) stress increments σ1, σ2, T T dσT = ⎝ A1 M2 B1 ⎠ (9)
and σ3 with direction vectors of N1, N2, and N3, the stress matrix C1 B1 M3
Increments
0 0 0 0 0 C1
⎜ ⎟ ⎜ ⎟ All plastic multipliers should be determined first to calculate the
+ T ⎝ 0 0 B1 ⎠T T + T ⎝ 0 0 0 ⎠T T (11)
plastic deformations within the soil induced by rotational stress in-
0 B1 0 C1 0 0 crements. For coaxial plastic strain increments, the plastic potential
functions Q1c = σ1, Q2c = σ2, and Q3c = σ3 should be substituted
The parameters in matrix form are defined by M1 = dσ1; M2 =
into Eq. (15), and the corresponding plastic multipliers are evalu-
dσ2; M3 = dσ3; A1 = dθ1 (σ1–σ2) = dτ12; B1 = dθ2 (σ2–σ3) = dτ23;
ated as follows:
and C1 = dθ3 (σ1–σ3) = dτ13.
The total stress increment in the 3D space, dσ, is then derived by dλkc = dεkp (k = 1, 2, 3) (20)
summing dσc and dσr
⎛ ⎞ Similarly, the plastic potential functions Q1r = σ1, Q2r = σ2,
dσ 1 dθ1 (σ 1 − σ 2 ) dθ3 (σ 1 − σ 3 ) Q3r = σ3, Q4r = τ12, Q5r = τ13, and Q6r = τ23 are substituted into
⎜ ⎟ Eq. (16) to derive the remaining six plastic multipliers for rotational
dσ = dσ c + dσ r = T ⎝ dθ1 (σ 1 − σ 2 ) dσ 2 dθ2 (σ 2 − σ 3 ) ⎠T T
plastic strain increments
dθ3 (σ 1 − σ 3 ) dθ2 (σ 2 − σ 3 ) dσ 3 ⎧ p p p
(12) ⎪
⎪ dλ1r = dε11r1 + dε11r2 + dε11r3
⎨ p p p
dλ2r = dε22r1 + dε22r2 + dε22r3
In the aforementioned expression, dθ1, dθ2, and dθ3 represent p p p (21)
⎪
⎪ dλ = dε33r1 + dε33r2 + dε33r3
the rotation increments induced by rotational stress increments ⎩ 3r p p p
dλ4r = dε12r1 , dλ5r = dε13r3 , dλ3r = dε23r2
dσr1, dσr2, and dσr3 with respect to the directions of minor,
major, and intermediate principal stress, respectively. Interested Zheng et al. (2002) derived the plastic strain increments induced
readers can find more details about the decomposition of stress in- by the corresponding stress increments of dσr1, dσr2 and dσr3 with
crements in 3D in Zheng (2000) and Zheng et al. (2002). respect to the directions of minor, major, and intermediate principal
stress as follows:
⎧ p p
Plastic Potential and Yield Surface ⎪
⎪ dε = Er1
1
|dθ1 |, dε22r1 = Er1 2
|dθ1 |
⎪ 11r1
⎪
⎪
⎪ dε p 3 p
= Er1 |dθ1 |, dε12r1 = Er1 |dθ1 |
4
Both the coaxial and rotational dynamic stress increments can ⎪
⎨ 133r1
cause plastic deformations within subgrade soils. Hence, the plastic Er1 = R1 σ 1 K f 1 , Er1
2
= σ 2 K f 1 , Er1
3
= 0.3σ 3 K f 1
⎪
strain increment dɛ p can be defined as ⎪
⎪ 1 )2 + (E 2 )2 + (E 3 )2 ]
⎪
⎪
4
Er1 = (9D2 + 2B2 )(σ 1 − σ 2 )2 − 2[(Er1
⎪
⎪ √
r1 r1
p
dε p = dεcp + dεrp = dεcp + dεr1 p
+ dεr2 p
+ dεr3 ⎩
(13) K f 1 = B 3|σ 1 − σ 2 |/(R1 σ 1 + σ 2 + 0.3σ 3 )
where dεcp = the plastic strain increment induced by the coaxial (22)
stress increment dσc; dεrp = the plastic strain increment induced ⎧ p p
by the rotational stress increment dσr; and dεr1 p p
, dεr2 p
, and dεr3 = ⎪
⎪ dε = Er2
1
|dθ2 |, dε22r2 = Er22
|dθ2 |
⎪ 11r2
⎪ p p
the plastic strain increments due to dσr1, dσr2, and dσr3. ⎪
⎪ dε = E 3
|dθ 2 |, dε = E r2 |dθ2 |
4
⎨ 133r2 r2 23r2
The elastoplastic strain increment (dɛ) is then written as the Er2 = 0.3σ 1 K f 2 , Er2
2
= R1 σ 2 K f 2 , Er2
3
= σ3 K f 2
summation of elastic (dɛe) and plastic strain increments (dɛp)
⎪
⎪
⎪ 4 2 1 2 2 2
⎪ Er2 = (9D2 + 2B2 )(σ 2 − σ 3 ) − 2[(Er2 ) + (Er2 ) + (Er2 ) ]
3 2
⎪
⎪ √
p
dε = dεe + dε p = dεe + dεcp + dεr1 p
+ dεr2 p
+ dεr3 (14) ⎩
K f 2 = B 3|σ 2 − σ 3 |/(0.3σ 1 + R1 σ 2 + σ 3 )
Following the work of Yang and Li (2002) and Lai et al. (2010), (23)
the generalized plastic potential functions for the coaxial and rota-
⎧ p p
tional parts are expressed as follows: ⎪ dε11r3 = Er31
|dθ3 |, dε22r3 = Er3
2
|dθ3 |
⎪
⎪
⎪
⎪ dε p
= E 3
|dθ |, dε p
= E 4
|dθ3 |
3
∂Qkc ⎪
⎨ 1 33r3 r3 3 13r3 r3
p
dεijc = dλkc (15) Er3 = R1 σ 1 K f 3 , Er3 2
= 0.3σ 2 K f 3 , Er33
= σ3 K f 3
∂σ ij
k=1 ⎪
⎪
⎪ 4 2 1 2 2 2
⎪ Er3 = (9D2 + 2B2 )(σ 1 − σ 3 ) − 2[(Er3 ) + (Er3 ) + (Er3 ) ]
3 2
6 ⎪
⎪ √
p ∂Qkr ⎩
dεijr = dλkr (16) K f 3 = B 3|σ 1 − σ 3 |/(R1 σ 1 + 0.3σ 2 + σ 3 )
k=1
∂σ ij
(24)
where dλ = plastic multiplier (in a total of nine multipliers); the
plastic potential functions Q in six directions should be nonassoci- Dynamic Constitutive Model
ated and are defined as follows:
The constitutive model can be developed by using an elastoplastic
Qkr = Qkc = σ k , k = 1, 2, 3 (17) flexibility matrix to characterize the stress–strain relationship,
⎪
⎪ ⎪
⎪ ⎪
⎪ dσ 2 ⎪⎪
⎪
⎪
dσ 2 ⎪
⎪ ⎪
⎪ ⎪
⎪
where [Ce] = elastic flexibility matrix; and A = 2(1 + μ). ⎪
⎨ dσ ⎬ ⎪ ⎨ dσ ⎪
⎪ ⎬
The yield surface for the coaxial model is assumed to have the 3 3
× = [Crp ] (28)
⎪
⎪ ⎪ ⎪ ⎪
following form: ⎪ dτ12 ⎪
⎪ ⎪
⎪ ⎪ dτ12 ⎪
⎪
⎪ ⎪
⎪
⎪
⎪ dτ23 ⎪⎪ ⎪
⎪ dτ23 ⎪⎪
εip = fi (σ 1 , σ 2 , σ 3 ) (26) ⎪
⎪ ⎪
⎪ ⎪
⎪ ⎪
⎪
⎩ ⎭ ⎩ ⎭
dτ13 dτ13
Hence, the stress–strain relationship for coaxial strain incre-
ments can be written with the coaxial plastic flexibility matrix The elastoplastic flexibility matrix [Cep] is then calculated as
[Ccp] as follows: follows:
⎡ ⎤ [Cep ] = [Ce ] + [Ccp ] + [Crp ] (29)
∂f1 ∂f1 ∂f1
0 0 0
⎧ p ⎫ ⎢ ∂σ 1 ∂σ 2 ∂σ 3 ⎥⎧ ⎫ ⎧ ⎫ The direction vectors T1, T2, and T3 for major, intermediate, and
⎪ dε1 ⎪ ⎢ ⎥⎪ dσ 1 ⎪ ⎪ dσ 1 ⎪
⎪ ⎪ ⎢ ⎥ ⎪ ⎪ ⎪ ⎪
⎪ dε p ⎪
⎪
⎪ ⎪ ⎢ ∂f2 ∂f2 ∂f2 0 0 0 ⎥⎪ ⎪ dσ 2 ⎪
⎪ ⎪
⎪ ⎪
dσ 2 ⎪
minor principal stresses are defined as follows:
⎪ 2 ⎪⎪ ⎢ ⎥ ⎪
⎪ ⎪
⎪ ⎪
⎪ ⎪
⎪
⎨ p ⎬ ⎢ ∂σ 1 ∂σ 2 ∂σ 3 ⎥ ⎨ ⎬ ⎨ ⎬
dε3 ⎢ ⎥ dσ dσ T1 = [ L1 L2 L 3 ]T , T 2 = [ M1 M2 M3 ] T ,
p = ⎢ ∂f ∂f ∂f ⎥
3
= [C cp ] 3
(27)
⎪
⎪ dε12 ⎪
⎪ ⎢
3 3 3
0 0 0 ⎥⎪⎪ dτ12 ⎪
⎪ ⎪
⎪ dτ12 ⎪
⎪ T3 = [ N1 N 3 ]T
⎪
⎪ ⎪ ∂σ 1 ∂σ 2 ∂σ 3 ⎪ ⎪ ⎪ ⎪ N2 (30)
⎪ dε23 ⎪
⎪ ⎢ ⎥⎪⎪ dτ23 ⎪ ⎪ dτ23 ⎪
⎪
p
⎪ ⎪ ⎪
⎩ p ⎪ ⎭ ⎢ ⎢ 0 0 0 0 0 0 ⎥⎪
⎥ ⎩ ⎪
⎭ ⎪
⎩ ⎪
⎭
dε13 ⎣ 0 0 0 0 0 0 ⎦ dτ13 dτ13 Hence, the elastoplastic flexibility matrix in a general stress
space (Zheng et al. 2002) can be written as follows:
0 0 0 0 0 0
[C ′ep ] = [TA ][Cep ][TA ]T (31)
Similarly, the stress–strain relationship for rotational strain
increments can be derived with the rotational plastic flexibility where
⎡ ⎤
L21 M12 N12 2L1 M1 2M1 N1 2L1 N1
⎢ L2 M22 N22 2L2 M2 2M2 N2 2L2 N2 ⎥
⎢ 22 ⎥
⎢ L M32 N32 2L3 M3 2M3 N3 2L3 N3 ⎥
[TA ] = ⎢ 3
⎢ L1 L2
⎥ (32)
⎢ M1 M2 N1 N2 L 1 M2 + L 2 M 1 M1 N 2 + M2 N 1 L1 N2 + L2 N1 ⎥
⎥
⎣ L2 L3 M2 M3 N2 N3 L 2 M3 + L 3 M 2 M2 N 3 + M3 N 2 L2 N3 + L3 N2 ⎦
L1 L3 M1 M3 N1 N3 L 1 M3 + L 3 M 1 M1 N 3 + M3 N 1 L1 N3 + L3 N1
′
Matrix inversion is then conducted on [Cep ] to get the elasto- The initial boundary condition at time t = 0 is written
′
plastic stiffness matrix [Dep ] in an arbitrary stress space considering
the influence of rotation of principal stress. p(x, y, z, t)|t=0 = 0 (34)
Initial and Boundary Conditions due to Moving Traffic Calculation of Effective Stress within Unsaturated
Subgrade Soils
For subgrade soils subjected to moving vehicles, a schematic view
of a half-space is defined as depicted in Fig. 4. The dynamic load
induced by a moving car is simplified as a rectangular uniformly Effective Stress Formulation
loaded area. In the Cartesian coordinate system, the dynamic The formulation of effective stress for unsaturated soils can be in-
load p (x, y, z, t) is defined by corporated into the elastoplastic constitutive model to analyze the
behavior of subgrade soils under moving vehicles. Following the
p(x, y, z, t) study of Tang (2000), the terms of wet suction and structure suction
q, vt − c ≤ x ≤ vt + c, 0 ≤ y ≤ b, z = 0 are used rather than the term of matric suction to characterize the
= (33) influence of the degree of saturation on the strength of unsaturated
0 other
soils. As shown in Fig. 2, the wet suction is induced by the contrac-
where x, y, and z represent three coordinates; parameters b and c are tile skin (Fredlund and Rahardjo 1993) between liquid and solid
defined in Fig. 4 to show the dimensions of the dynamic load; t = phases, and the structure suction can characterize the adsorptive
time; and v = velocity of the moving car. and bonding effects between solid particles. The generalized
z
for use in the analysis.
Fig. 4. Diagram of a half-space subgrade soil subjected to the dynamic Sa = aw2 + bw + c (37)
load induced by a moving car.
where a, b, and c are three fitting coefficients based on the results of
suction tests (Zhang et al. 2012b).
effective stress for unsaturated soils (Tang 2000, 2001) can then be The structure suction Sc is mainly induced by cementation
written bonds between solid phases, which can be evaluated by microstruc-
tural analysis of clayey subgrade soils (Tang 2000, 2001; Tang
σ ′ = (σ − ua ) + S = (σ − ua ) + (Sa + Sc ) (35)
et al. 2003).
where ua = pore-air pressure; and Sa and Sc = the wet suction and Once the generalized effective stress matrix {σ′ } is derived, the
the structure suction, respectively. shear strength τ can be estimated by
It should be emphasized that the generalized effective stress of
Tang (2000) has a similar formulation compared to the equation of τ = c′ + (σ − ua ) tan φ′ + (Sa + Sc ) tan φ′ (38)
Bishop’s effective stress. However, the effective stress parameter χ
in Bishop’s formula is controversial (Sheng 2011). The proposed where c′ and φ′ = effective cohesion and effective friction angle, re-
effective stress formula can distinguish the effects of capillary spectively. Different equations have been introduced in the litera-
and adsorptive components directly. The term of wet suction is ture for predicting the shear strength of unsaturated soils
caused by the contractile skin at the liquid and solid interface, (Fredlund and Morgenstern 1977; Vanapalli et al. 1996; Öberg
which corresponds to a spherical stress, but cannot transfer shears. and Sällfors 1997; Xu and Cao 2015; Ahmadi Naghadeh and
The magnitude of wet suction primarily depends on the water sur- Toker 2019). Zhang et al. (2014) presented a detailed literature re-
face tension and its contact angle with soil particle. On the contrary, view on different constitutive models that can be used to character-
the structure suction is induced by cementation bonds between ize the stress–strain behavior for unsaturated soils. Again, it is
solid phases, which results in a spherical stress that can transfer emphasized that this study follows the framework of Tang (2000).
shears. The generalized effective stress is then written in matrix For saturated subgrade soils below the groundwater table, the
form as follows: generalized effective stress matrix becomes
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
{σ ′ }T = {σ}T − {ua }T + {Sa }T + {Sc }T σ x −uw τxy τxz σ x τxy τxz uw 0 0
⎡ ⎤ ⎢
σ x − ua + Sa + Sc τxy + Sc τxz + Sc {σ ′ }= ⎣ τyx σ y − uw τyz ⎥ ⎢
⎦ = ⎣ τyx σ y τyz ⎥ ⎢
⎦ − ⎣ 0 uw 0 ⎦
⎥
⎢ τyx + Sc σ y − ua + Sa + Sc τyz + Sc ⎥ τzx τzy σ z −uw τzx τzy σ z
=⎣ ⎦ 0 0 uw
τzx + Sc τzy + Sc σ z − ua + Sa + Sc (39)
(36)
where {σ}T = σ x σ y σ z τxy τyz τxz ; {ua }T = ua ua ua 0 0 0 ; Cumulative Effect of Dynamic Stress
{Sa }T = Sa Sa Sa 0 0 0 ; and {Sc }T = Sc Sc Sc Sc Sc Sc . Based on the results of field tests, Lin (2012) indicated that as the
number of load cycles (N ) induced by moving vehicles increases,
Strength of Unsaturated Subgrade Soils the cumulative dynamic stress of subgrade soils increases as a lin-
ear function of N on a logarithmic scale
All matrices in Eq. (36) can be solved separately to construct the
generalized effective stress formula. The procedures for calculating [σ N ] = [m] ln N + [n] (40)
the shear strength of unsaturated subgrade soils are summarized as
follows: where [m] and [n] are the fitting parameters.
All dynamic stress components {σ}T under pavements can be The cumulative effect of dynamic stress (σNx, σNy, and σNz) is
calculated using numerical integration with the elastoplastic consti- then included in the formulation of generalized effective stress as
tutive model and boundary conditions. follows:
⎡ ⎤
σ x + σ Nx − ua + Sa + Sc τxy + Sc τxz + Sc
{σ ′ }N = ⎣ τyx + Sc σ y + σ Ny − ua + Sa + Sc τyz + Sc ⎦ (41)
τzx + Sc τzy + Sc σ z + σ Nz − ua + Sa + Sc
O y = 0.5 m, 1.0 m, 2.0 m Four types of vehicles were selected to cover a wide range of
y load levels, and details are given in Table 2. Vehicles were con-
Inclinometer
trolled to drive at speeds of 10, 20, and 30 km/h. In the field
Settlement plate
tests, the maximum number of load cycles was chosen as 50. Stress
Downloaded from ascelibrary.org by "National Institute of Technology, Calicut" on 01/17/24. Copyright ASCE. For personal use only; all rights reserved.
(a) (b)
Fig. 6. Earth pressure transducers for measuring dynamic stresses in three orthogonal directions: (a) view of the earth pressure sensors; and (b) view
of cables.
50
80 80
Water content (%)
40
60 60
30
40 40 20
20 20 10
0
0 0 0 20 40 60 80 100
0 1 2 3 4 5 Degree of saturation (%)
Depth from the ground surface (m)
Fig. 8. Variations of structure suction with a degree of saturation.
Fig. 7. Variations of water content and degree of saturation with depth
from the ground surface.
method was used to solve the problem of subgrade soils under re-
The volumetric yield surface is assumed as an ellipse (Zheng peated traffic loads. The technical details of the numerical nonlinear
et al. 2002) as follows: incremental method can be found in Yin (2007). In this study,
based on the measurements of water content in the field within
p2 q2 the first 5 m from the ground surface and the corresponding degree
+ =1 (45)
a22 b22 of saturation as shown in Fig. 7, the term of wet suction can be cal-
culated using Eq. (42). Fig. 8 shows the correlation between struc-
where a2 and b2 are two fitting parameters as a function of volumet-
ric plastic strain εvp . Based on the triaxial testing data of subgrade ture suction and degree of saturation from which the contribution of
soils (Lin 2012), a2 and b2 are defined as structure suction can be estimated. Using parameters reported in
Table 3, the matrix [σN] expressed in Eq. (40) can be computed.
a22 = 6.47 × 103 εvp + 2.30 (46) As such, all parameters in the generalized effective stress formula-
tion expressed in Eq. (41) can be determined.
b22 = 1.71 × 103 εvp − 0.18 (47) The comparison of measured and calculated dynamic stresses in
subgrade soils in the x-direction is presented in Fig. 10. Similarly,
the measured dynamic stresses in subgrade soils are compared
against calculations in the y and z-directions as illustrated in Figs.
Dynamic Stress Calculation
11 and 12, respectively. Moreover, the conventional static method
The constitutive model proposed in this analysis was implemented is used to compute the stresses in subgrade soils. The results are
in a computer program, and the numerical nonlinear incremental also plotted in Figs. 10–12 for comparison. Note that a factor of
0 0 0
Downloaded from ascelibrary.org by "National Institute of Technology, Calicut" on 01/17/24. Copyright ASCE. For personal use only; all rights reserved.
0 1 2 0 1 2 0 1 2
10 10 10 10 10 10 10 10 10
(a) Number of load cycles, N (b) Number of load cycles, N (c) Number of load cycles, N
Fig. 9. Relations between the cumulative dynamic stress in subgrade soils induced by a fully loaded dump truck and the number of load cycles: (a) in
the x-direction with y = 0.5 m, z = 1.0 m, and v = 20 km/h; (b) in the y-direction with x = 0.0 m, z = 1.0 m, and v = 20 km/h; and (c) in the z-direction
with x = 0.0 m, y = 0.5 m, and v = 20 km/h.
Stress (kPa)
Direction m n Coefficient of determination, R 2 Calculated, total stress, dynamic
6 Calculated, effective stress, static
z = 0.5 m 0.4754 0.8424 0.9704 4
Calculated, total stress, static
z = 1.0 m 0.5226 0.3394 0.9953
z = 2.0 m 0.4148 –0.0645 0.9546 2
z = 5.0 m 0.0835 0.0130 0.9029 N=1
0
y = 0.5 m 0.6912 –0.2706 0.8971 (a) 0 1 2 3 4 5
y = 1.0 m 0.5037 –0.6046 0.9594
10
y = 2.0 m 0.1808 –0.2505 0.8007
x = 0.0 m 0.4762 –0.1864 0.9221 8
Stress (kPa)
1.2 is applied to the traffic load on the road surface, i.e., 44.5 × 1.2 = (b) Length of the roadway, x (m)
53.4 kPa. For both the conventional static method and the proposed
generalized plastic mechanics–based constitutive model, two types Fig. 10. Comparison of measured and calculated dynamic stresses in
of calculations were conducted: (a) total stress analysis; and (b) ef- subgrade soils in the x-direction: (a) after 1 load cycle; and (b) after
50 load cycles.
fective stress analysis.
It can be seen that both calculations can capture the salient na-
ture of subgrade soils under moving vehicles well compared to field 20
Measured
measurements (i.e., the trends of all curves are similar). In general, Calculated, effective stress, dynamic
Stress (kPa)
all calculated stresses are greater than measured values, but calcu- 15 Calculated, total stress, dynamic
lations of effective stress analysis are always higher than those Calculated, effective stress, static
10 Calculated, total stress, static
obtained from total stress analysis. Stresses calculated using the tra-
ditional static method remain the same for both N = 1 and N = 50 5
loading cycles, as it cannot consider the cumulative effect of N=1
0
stresses with load cycles. Although the static method seems to be (a) 0.0 0.5 1.0 1.5 2.0
conservative at the beginning, it could underestimate the stresses
20
in subgrade soils in the long term when the loading cycle N be-
comes larger. The advantage of the proposed approach over the
Stress (kPa)
15
conventional method is apparent. In the z-direction, the difference
between measurements and calculations using the proposed ap- 10
load cycle and by 15.4%–21.4% after 50 load cycles. In the (b) Width of the roadway, y (m)
y-direction, the error level falls within 29.1%–36.4% after 1 load
cycle and 18.7%–26.7% after 50 load cycles. Design of pavement Fig. 11. Comparison of measured and calculated dynamic stresses in
subgrade soils in the y-direction: (a) after 1 load cycle; and (b) after
foundation with overestimated dynamic stress obtained by the pro-
50 load cycles.
posed method could result in conservative analysis.
Stress (kPa)
40
30 References
20
10
N = 50 Ahmadi Naghadeh, R., and N. K. Toker. 2019. “Exponential equation for
0 predicting shear strength envelope of unsaturated soils.”
0 1 2 3 4 5 Int. J. Geomech. 19 (7): 04019061. https://doi.org/10.1061/(ASCE)
Depth from the ground surface, z (m) GM.1943-5622.0001435.
(b)
Brown, S. 1996. “Soil mechanics in pavement engineering.” Géotechnique
Fig. 12. Comparison of measured and calculated dynamic stresses in 46 (3): 383–426. https://doi.org/10.1680/geot.1996.46.3.383.
Brown, S. F., C. Bell, and B. Brodrick. 1977. Permanent deformation of
subgrade soils in the z-direction: (a) after 1 load cycle; and (b) after
flexible pavements. Rep. No. DAERO-75-G-023. Nottingham,
50 load cycles. England: Univ. of Nottingham.
Bufler, H. 1971. “Theory of elasticity of a multilayered medium.” J. Elast.
1 (2): 125–143. https://doi.org/10.1007/BF00046464.
The difference between calculations and measurements can be Chai, J.-C., and N. Miura. 2002. “Traffic-load-induced permanent
attributed to two reasons: (a) some model parameters are obtained deformation of road on soft subsoil.” J. Geotech. Geoenviron. Eng.
through curve fitting of measurements, and some are determined 128 (11): 907–916. https://doi.org/10.1061/(ASCE)1090-0241(2002)
empirically, where error could be introduced inevitably; and (b) 128:11(907).
the proposed approach considers the subgrade soil as a homoge- Christopher, B. R., C. W. Schwartz, and R. Boudreau. 2006. Geotechnical
neous half-space, and the layered nature of the soil cannot be aspects of pavements. Woodbury, MN: Ryan R. Berg & Associates.
simulated. Collins, I., A. Wang, and L. Saunders. 1993. “Shakedown in layered pave-
Last, it is recalled that the empirical values for model parameters, ments under moving surface loads.” Int. J. Numer. Anal. Methods
Geomech. 17 (3): 165–174. https://doi.org/10.1002/nag.1610170303.
e.g., coefficients in Eqs. (42)–(47) and Table 3, are project-specific.
Eason, G. 1965. “The stresses produced in a semi-infinite solid by a moving
Cautions must be practiced when the same empirical values are used surface force.” Int. J. Eng. Sci. 2 (6): 581–609. https://doi.org/10.1016
for analyses of other projects for subgrade soils under traffic loads. /0020-7225(65)90038-8.
Otherwise, much higher errors could be resulted. Feng, S.-J., X.-L. Zhang, Q.-T. Zheng, and L. Wang. 2017. “Simulation
and mitigation analysis of ground vibrations induced by high-speed
train with three dimensional FEM.” Soil Dyn. Earthquake Eng. 94:
204–214. https://doi.org/10.1016/j.soildyn.2017.01.022.
Conclusions Fredlund, D. G., and N. R. Morgenstern. 1977. “Stress state variables for
unsaturated soils.” J. Geotech. Eng. Div. 103 (5): 447–466.
The analysis of dynamic stress in subgrade soils induced by re- Fredlund, D. G., and H. Rahardjo. 1993. Soil mechanics for unsaturated
peated moving vehicles is normally conducted using empirical soils. New York: Wiley.
solutions. Alternatively, numerical simulation can be performed Gallipoli, D., and A. Bruno. 2017. “A bounding surface compression model
to approximate the behavior of subgrade soils. However, the cumu- with a unified virgin line for saturated and unsaturated soils.”
lative effect of dynamic stress in subgrade soils due to repeated Géotechnique 67 (8): 703–712. https://doi.org/10.1680/jgeot.16.P.145.
dynamic loads from surface traffic is often neglected in current Gens, A. 2010. “Soil–environment interactions in geotechnical engineer-
practice, as well as the rotation of principal stress with moving ing.” Géotechnique 60 (1): 3–74. https://doi.org/10.1680/geot.9.P.109.
vehicles. It is also imperative to analyze the behavior of subgrade Hardy, M., and D. Cebon. 1993. “Response of continuous pavements to
moving dynamic loads.” J. Eng. Mech. 119 (9): 1762–1780. https://
soils with a constitutive model in the framework of unsaturated
doi.org/10.1061/(ASCE)0733-9399(1993)119:9(1762).
soil mechanics. Huang, H., and S. Chrismer. 2013. “Discrete element modeling of ballast
In this investigation, a generalized plastic mechanics–based settlement under trains moving at ‘critical speeds’.” Constr. Build.
constitutive model for unsaturated subgrade soils was proposed, Mater. 38: 994–1000. https://doi.org/10.1016/j.conbuildmat.2012.09
which could consider the cumulative effect of dynamic stress and .007.
the rotation of principal stress. The degree of saturation was sepa- Huang, H., Y. Gao, and S. Stoffels. 2014. “Fully coupled three-dimensional
rated into a wet suction component to characterize the capillary ef- train-track-soil model for high-speed rail.” Transp. Res. Rec: J. Transp.
fect between liquid and solid phases and a structure suction part to Res. Board 2448 (1): 87–93. https://doi.org/10.3141/2448-11.
model the adsorptive effect between solid particles. An explicit Kaynia, A. M., C. Madshus, and P. Zackrisson. 2000. “Ground vibration
scheme in numerical analysis was developed to solve the systems from high-speed trains: Prediction and countermeasure.” J. Geotech.
Geoenviron. Eng. 126 (6): 531–537. https://doi.org/10.1061/(ASCE)
of nonlinear equations for obtaining approximations. A field test
1090-0241(2000)126:6(531).
was conducted on a roadway, above which four types of vehicles Kim, D., and J. R. Kim. 2007. “Resilient behavior of compacted subgrade
were used to simulate the influence of repeated traffic load. The ef- soils under the repeated triaxial test.” Constr. Build. Mater. 21 (7):
fectiveness of the proposed generalized plastic mechanics–based 1470–1479. https://doi.org/10.1016/j.conbuildmat.2006.07.006.
constitutive model for unsaturated subgrade soils under moving Lai, Y., Y. Yang, X. Chang, and S. Li. 2010. “Strength criterion and elasto-
traffic was demonstrated by comparing against experimental plastic constitutive model of frozen silt in generalized plastic
Liu, J., and J. Xiao. 2010. “Experimental study on the stability of railroad Tang, L.-S. 2000. “Structure suction and principle of general effective
silt subgrade with increasing train speed.” J. Geotech. Geoenviron. Eng. stress in unsaturated soils.” Acta Sci. Nat. Univ. Sunyatseni 39 (6):
136 (6): 833–841. https://doi.org/10.1061/(ASCE)GT.1943-5606 95–100.
.0000282. Tang, L.-S. 2001. “New suggestion on shear strength in unsaturated soil based
Liu, K., Q. Su, P. Ni, C. Zhou, W. Zhao, and F. Yue. 2018. “Evaluation on on suction between grains.” Chin. J. Geotech. Eng. 23 (4): 412–417.
the dynamic performance of bridge approach backfilled with fibre rein- Tang, L. S., H. Chen, H. Sang, S. Zhang, and J. Zhang. 2015.
forced lightweight concrete under high-speed train loading.” Comput. “Determination of traffic-load-influenced depths in clayey subsoil
Geotech. 104: 42–53. https://doi.org/10.1016/j.compgeo.2018.08.003. based on the shakedown concept.” Soil Dyn. Earthquake Eng. 77:
Liu, Y., Y. Zheng, and Z. Cheng. 1998. “The general stress strain relation 182–191. https://doi.org/10.1016/j.soildyn.2015.05.009.
of soils involving the rotation of principal stress axes.” Appl. Math. Tang, L. S., H. T. Sang, X. B. Deng, Z. G. Luo, and H. K. Chen. 2014.
Mech. 19 (5): 407–413. “Experimental study on cementation strength of unsaturated granite re-
Lu, Z., H. Yao, J. Liu, and Z. Hu. 2014. “Dynamic response of a sidual soil.” In 4th Int. Conf. on Green Building, Materials and Civil
pavement-subgrade-soft ground system subjected to moving traffic Engineering (GBMCE 2014), edited by J. C. M. Kao, W.-P. Sung,
load.” J. Vibroeng. 16 (1): 195–209. and R. Chen. 789–794. London: Taylor & Francis.
Mašín, D., and N. Khalili. 2008. “A hypoplastic model for mechanical re- Tang, L.-S., and S.-J. Wang. 2000. “Absorbed suction and principle of
sponse of unsaturated soils.” Int. J. Numer. Anal. Methods Geomech. effective stress in unsaturated soils.” Chin. J. Geotech. Eng. 22 (1): 83–88.
32 (15): 1903–1926. https://doi.org/10.1002/nag.714. Tang, L.-S., Y. Wang, P.-C. Zhang, and H.-R. Liao. 2003. “Experimental
Miller, G., S. Teh, D. Li, and M. Zaman. 2000. “Cyclic shear strength of study on suctions between grains in unsaturated cohesive soil.”
soft railroad subgrade.” J. Geotech. Geoenviron. Eng. 126 (2): 139– Chin. J. Geotech. Eng. 25 (3): 304–307.
147. https://doi.org/10.1061/(ASCE)1090-0241(2000)126:2(139). Tarefder, R. A., N. Saha, J. W. Hall, and P. T. Ng. 2008. “Evaluating weak
Monismith, C. L., N. Ogawa, and C. R. Freeme. 1975. “Permanent defor- subgrade for pavement design and performance prediction: A case
mation characteristics of subgrade soils due to repeated loading.” study of US 550.” J. GeoEng. 3 (1): 13–24.
Transportation Research Records, No. 537. Washington, DC: Vanapalli, S. K., D. G. Fredlund, D. E. Pufahl, and A. W. Clifton.
Transportation Research Board. 1996. “Model for the prediction of shear strength with respect to soil suc-
Mylonakis, G. 2001. “Winkler modulus for axially loaded piles.” tion.” Can. Geotech. J. 33 (3): 379–392. https://doi.org/10.1139/t96-060.
Géotechnique 51 (5): 455–461. https://doi.org/10.1680/geot.2001.51 Xenaki, V., and G. Athanasopoulos. 2008. “Dynamic properties and lique-
.5.455. faction resistance of two soil materials in an earthfill dam—Laboratory
Ni, P., G. Mei, and Y. Zhao. 2018a. “Influence of raised groundwater level test results.” Soil Dyn. Earthquake Eng. 28 (8): 605–620. https://doi.org
on the stability of unsaturated soil slopes.” Int. J. Geomech. 18 (12): /10.1016/j.soildyn.2007.10.001.
04018168. https://doi.org/10.1061/(ASCE)GM.1943-5622.0001316. Xu, Y. F., and L. Cao. 2015. “Fractal representation of effective stress of
Ni, P., I. D. Moore, and W. A. Take. 2018b. “Distributed fibre optic sensing unsaturated soils.” Int. J. Geomech. 15 (6): 04014098. https://doi.org
of strains on buried full-scale PVC pipelines crossing a normal fault.” /10.1061/(ASCE)GM.1943-5622.0000446. .
Géotechnique 68 (1): 1–17. https://doi.org/10.1680/jgeot.16.P.161. Yang, G.-H., and G.-X. Li. 2002. “Mathematical foundation of constitutive
Nuth, M., and L. Laloui. 2008. “Effective stress concept in unsaturated models of geotechnical material and generalized potential theory.” Rock
soils: Clarification and validation of a unified framework.” Soil Mech. 23 (5): 531–535.
Int. J. Numer. Anal. Methods Geomech. 32 (7): 771–801. https://doi Yang, L. A., W. Powrie, and J. A. Priest. 2009. “Dynamic stress analysis of
.org/10.1002/nag.645. a ballasted railway track bed during train passage.” J. Geotech.
Öberg, A., and G. Sällfors. 1997. “Determination of shear strength param- Geoenviron. Eng. 135 (5): 680–689. https://doi.org/10.1061/(ASCE)
eters of unsaturated silts and sands based on the water retention curve.” GT.1943-5606.0000032.
Geotech. Test. J. 20 (1): 40–48. https://doi.org/10.1520/GTJ11419J. Yang, Y.-B., and H. Hung. 2008. “Soil vibrations caused by underground
Powrie, W., L. A. Yang, and C. R. I. Clayton. 2007. “Stress changes in the moving trains.” J. Geotech. Geoenviron. Eng. 134 (11): 1633–1644.
ground below ballasted railway track during train passage.” Proc. Inst. https://doi.org/10.1061/(ASCE)1090-0241(2008)134:11(1633).
Mech. Eng. Part F: J. Rail Rapid Transit. 221 (2): 247–262. https://doi Yıldırım, H., and H. Erş an. 2007. “Settlements under consecutive series of
.org/10.1243/0954409JRRT95. cyclic loading.” Soil Dyn. Earthquake Eng. 27 (6): 577–585. https://doi
Puppala, A. J., S. Saride, and S. Chomtid. 2009. “Experimental and mod- .org/10.1016/j.soildyn.2006.10.007.
eling studies of permanent strains of subgrade soils.” J. Geotech. Yin, Y. 2007. Foundation of nonlinear finite element analysis.” Beijing:
Geoenviron. Eng. 135 (10): 1379–1389. https://doi.org/10.1061 Peking University Press.
/(ASCE)GT.1943-5606.0000163. Zafir, Z., R. Siddharthan, and P. E. Sebaaly. 1994. “Dynamic
Ravichandran, N., and K. K. Muraleetharan. 2009. “Dynamics of unsaturated pavement-strain histories from moving traffic load.” J. Transp. Eng.
soils using various finite element formulations.” Int. J. Numer. Anal. 120 (5): 821–842. https://doi.org/10.1061/(ASCE)0733-947X(1994)
Methods Geomech. 33 (5): 611–631. https://doi.org/10.1002/nag.737. 120:5(821).
Reddy, C. N. V. S., and N. V. R. Moorthy. 2005. “Significance of bearing Zhang, H. W., R. Santagiuliana, and B. A. Schrefler. 2012a. “Return mapping
capacity of clayey subgrade in flexible pavement design.” algorithm for an enhanced generalized plasticity constitutive model of par-
Int. J. Pavement Eng. 6 (3): 183–189. https://doi.org/10.1080 tially saturated soils.” Comput. Geotech. 45: 53–61. https://doi.org/10
/10298430500137194. .1016/j.compgeo.2012.05.003.
Russell, A., and N. Khalili. 2006. “A unified bounding surface plasticity Zhang, L. L., D. G. Fredlund, M. D. Fredlund, and G. W. Wilson. 2014.
model for unsaturated soils.” Int. J. Numer. Anal. Methods Geomech. “Modeling the unsaturated soil zone in slope stability analysis.”
30 (3): 181–212. https://doi.org/10.1002/nag.475. Can. Geotech. J. 51 (12): 1384–1398.
plastic mechanics: Generalized plastic mechanics.” Beijing: China Methods Geomech. 39 (11): 1141–1164. https://doi.org/10.1002/nag
Architecture and Building Press. .2355.