polymers
Article
Research on Resilient Modulus Prediction Model and
Equivalence Analysis for Polymer Reinforced Subgrade Soil
under Dry–Wet Cycle
Yingcheng Luan 1,2 , Wei Lu 1, * and Kun Fu 2
Citation: Luan, Y.; Lu, W.; Fu, K.
Research on Resilient Modulus
Prediction Model and Equivalence
Analysis for Polymer Reinforced
Subgrade Soil under Dry–Wet Cycle.
Polymers 2023, 15, 4187. https://
doi.org/10.3390/polym15204187
Academic Editor: Denis Rodrigue
Received: 25 September 2023
Revised: 20 October 2023
Accepted: 21 October 2023
Published: 23 October 2023
Copyright: © 2023 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Polymers 2023, 15, 4187. https://doi.org/10.3390/polym15204187
1 Research Center of Geotechnical and Structural Engineering, Shandong University, Jinan 250061, China
2 School of Transportation, Southeast University, 2# Southeast University Road, Jiangning District,
Nanjing 210096, China
* Correspondence: lu.wei@sdu.edu.cn
Abstract: The subgrade soil of asphalt pavement is significantly susceptible to changes in moisture
content, and therefore many projects introduce polymer-based reinforcement to ensure soil perfor-
mance. This paper aims to incorporate a variable representing the dry–wet cycle into the prediction
model of resilient modulus of polymer reinforced soil. The polymer adopted is a self-developed
subgrade soil solidification material consisting of sodium dodecyl sulfate and polyvinyl oxide. The
current resilient modulus prediction model is improved, notably involving the effects of the dry–wet
cycle. Combined with finite element method (FEM) analysis, the actual stress state of pavement
and the coupling effect of dry–wet cycle and vehicle load on the resilient modulus are studied. The
deterioration in resilient modulus with the variation in seasonal climate and load response is also
investigated. Results show that the deviator stress is negatively correlated with the resilient modulus
while the bulk stress has a linearly positive relation. The decreasing rate at low deviator stress is
larger than that at the high level. Moreover, the dry–wet cycle can reduce the resilient modulus and
the reducing amplitude is the largest at the first dry–wet cycle. FEM analysis shows that the middle
position of the subgrade slope has the largest initial resilient modulus with decreasing amplitude
in the first year of dry–wet cycles, while the upper position shows a smaller change. The variation
in resilient modulus is closely related to the changes in cumulative volumetric water content. Con-
sidering that different positions of subgrade bear the external vehicle load, the equivalent resilient
modulus is more realistic for guiding the subgrade design.
Keywords: polymer reinforced subgrade soil; dry-wet cycle; resilient modulus; pavement structure;
finite element analysis; moisture content
1. Introduction
The subgrade is the foundation of the pavement structure and bears the loads from
the weight of vehicles and pavement structure. A change in bearing capacity will greatly
affect the stability and service life of a pavement structure [1,2], and the bearing capacity is
commonly represented by the resilient modulus, which is affected by the dry–wet cycles,
resulting in the reduction in strength of the subgrade soil [3,4]. For this reason, many
projects introduce polymer-based reinforcement to ensure soil performance.
The physical condition of optimal moisture content (OMC) is usually adopted to
achieve the maximum compaction density and a better engineering performance [5]. This
indicates that variation in seasonal moisture content and suction due to exposure to the
external environment have significant effects on the mechanical performance of subgrade
soil, especially for the resilient modulus [6,7]. Many models have been proposed to predict
the resilient modulus of the subgrade soil and other bases of unbonded aggregate materials,
which is generally in non-linear form and influenced by many factors, such as stress
state, hydraulic hysteresis, physical microstructure, etc. [8–10]. In order to study the
https://www.mdpi.com/journal/polymers
Polymers 2023, 15, 4187 2 of 18

effect of moisture content on the resilient modulus, in a prediction model employed to

study the direct effect of moisture, suction was considered to be significant, and adopted
as an independent term [3,11,12]; suction also has an influence on confining or shear
stress [13–15]. In addition, the soil specimens that suffered from drying and wetting
have a larger permanent deformation and lower resilient modulus compared to those
specimens that are not subjected to the fluctuation in moisture content [16]. Atmospheric
rainfall changes periodically, resulting in the annual occurrence of subgrade settlement and
deformation, which endangers the highway capacity and driving safety [17]. Therefore, the
effects of the dry–wet cycle on the resilient modulus should be included in a more realistic
prediction model.
Due to the fact the resilient modulus changes with the variation in seasonal rainfall
and the experimental conditions are limited, large differences in environmental factors and
stress conditions exist between the experimental and actual pavement conditions, indicating
that the experiment cannot represent the actual pavement service condition. This limitation
can be broken through by conducting a finite element method (FEM) analysis and the actual
state of the pavement structure can thus be better characterized. Hu et al. have carried
out a slope stability analysis using ABAQUS FEM software while considering the dry–wet
cycle [18]. Chen et al. have developed a numerical model to analyze the unsaturated flow
in porous media with repeated dry–wet cycles [19]. Yaqi et al. proposed an analytical
framework to predict the longitudinal shrinkage crack by incorporating different dry/wet
paths into numerical modeling [20]. Other research is mainly concentrated on the effects of
the dry–wet cycle on engineering performance. However, limited efforts have been made
to explore the coupling effect of traffic load and dry–wet cycles on the resilient modulus
of subgrade.
The resilient modulus of unsaturated soil is affected by the stress state, humidity
state, and dry–wet cycle. In this paper, the current resilient modulus prediction model is
improved, and the effects of dry–wet cycle are included. Based on this constitutive model of
resilient modulus, the finite element method (FEM) is adopted to simulate the actual stress
state of the pavement and the effects of the dry–wet cycle are studied by introducing the
UMAT user-defined subroutine. The deterioration of resilient modulus with the variation
in seasonal climate and load response is next investigated, and the resilient modulus of the
whole subgrade can be reasonably estimated. This will provide a basis for the design of the
soaked subgrade.

2. Resilient Modulus Prediction Model

2.1. Development of Resilient Modulus Model
The conventional prediction models of resilient modulus depend on bulk stress, or
deviatoric stress Refs. [21–23], which was proposed by the National Cooperative Highway
Research Program (NCHRP) and is used by many researchers (Equation (1)) to study the
mechanical response of the unbonded material [24]:
  q2   q3
C1 Doct
Ey = q1 A a +1 (1)
Aa Aa

where C1 is the term of bulk stress; Doct , the term of deviatoric stress; Aa , the atmospheric
pressure; and q1 , q2, and q3 , the regression coefficients. This model shows the resilient
modulus is dependent on the stress state. However, the unbonded aggregate is also a kind
of moisture-dependent material. The saturation degree and matric suction are regarded
to be the influencing factors for the resilient modulus and have thus been included in
the prediction models. AASHTO employed the saturation degree to characterize the
moisture sensitivity, which has been adopted in the MEPDG [25]. Heath et al. incorporated
the normalizing matric suction to predict the resilient modulus [26]. Therefore, both the
stress state and moisture variation could be considered in the resilient modulus model.
However, Sahin et al. have pointed out that the moisture-sensitivity not only depends on
Polymers 2023, 15, 4187 3 of 18

the saturation degree but also the matric suction, and both of them were therefore included
in a new model as shown in Equation (2) [13]:
 q2   q3
C1 − 3φsω

Doct
Ey = q1 A a (2)
Aa Aa

where φ is the volumetric water content; ω, the matric suction related to the water content;
and s, the saturation parameter. The matric suction can be obtained by the soil–water
characteristic curve (SWCC). The SWCC of a targeted soil could be obtained by the model
proposed as shown in Equation (3) [27]:

φs
φ=     β γ (3)
ω
ln e + αA a

where φs is the saturated volumetric water content; and α, β, and γ, the fitting coefficients.

2.2. Resilient Modulus with Dry–Wet Cycle

By contrast, there are limited studies that focus on the effect of drying–wetting cycles
on the resilient modulus. Combined with SWCC, the cumulative increment of volumetric
water content is used to characterize the effects of the drying–wetting cycle. The volumetric
water content is calculated according to Equation (4), while the increment form is shown in
Equation (5) and is represented by the absolute value. The cumulative increment of the
volumetric water content has an increasing trend with the increase in cycle number and
amplitude. Since a constant amplitude of the dry–wet cycle is adopted in this paper, the
expression can be expressed as Equation (6).

Vw mw /ρw w · ms w · Gs
φ= = = = (4)
V 1+e (1 + e ) · ρ w 1+e

In the above equation, Vw is the volume of water; V, the volume of dry soil particles; mw ,
the moisture content; ρw , the density of water; e, the void ratio; Gs , the specific gravity of
soil particles; and w, the rate of water content.
Z N
dw · Gs
∑ dφ = 0 1+e
dN (5)

N · dw
∑ dφ = 1+e
· Gs (6)

In the above equations, N is the number of cycles; and dw, the variation in rate of
water content.
Two aspects are considered when constructing the prediction model: (1) the effects
from the drying–wetting cycle should be considered, that is, the resilient modulus shows
a converse relation with the times and amplitudes of the drying–wetting cycle; (2) the
cumulative volumetric water content and the matric suction in the SWCC should be in-
volved. The prediction models are presented in the form of power function and exponential
function as shown in Equations (7) and (8). The final form is then determined by the
experimental data fitting results.
 q2   q3  N
 q5 
C1 − 3φsω dw · Gs
 Z
Doct
Ey = q1 A a +1 1 + q4 dN (7)
Aa Aa 0 1+e
 q2   q3  N
C1 − 3φsω dw · Gs
  Z 
Doct
Ey = q1 A a +1 1 + q4 exp q5 dN (8)
Aa Aa 0 1+e
 rial consisting of sodium dodecyl sulfate and polyvinyl oxide. The tested results of the
polymer-based reinforcement agent are shown in Table 1.

Table 1. Tested results of polymer-based reinforcement agent.

Polymers 2023, 15, 4187 4 of 18
Testing Reference Tested
Tested Item
Standards Standards Results
Density (g/cm3) D ± 0.03 1.41
3. MaterialspHand Experiments A ± 1.0 7.6
3.1. Raw
Soluble solidsand Specimen Preparation
Materials S ± 2.0 96.5
3.1.1. Soil
Unconfined compres- Soil stabilizing
≥2.5 3.2
This
sion paper takes a typical
strength polymer-reinforced soft subgrade soil section under dry–wet
admixtures
cycle conditions
Unconfined as an example
compres- (Figure[28]
CJ/T 486-2015 1). During the road construction process, due to
thesive
dry–wet cycling ≥120 148
strength ratio effect, the mechanical properties of the soil cannot meet the design
requirements,
Seven-day water so polymer-based
sta- reinforcement agents are introduced as subgrade soil
≥105
reinforcement materials. The polymer adopted is a self-developed 115
soil solidification ma-
bility coefficient ratio
terial consisting of sodium dodecyl sulfate and polyvinyl oxide. The tested results of the
Note: D, A and S denote the density, pH value and solid content of product control value, respec-
polymer-based
tively.
reinforcement agent are shown in Table 1.

Figure1.1. The
Figure The reinforcement
reinforcementprocess
processof
ofsoil.
soil.

Table Due to the

1. Tested introduction
results of polymer-based
of polymer-based reinforcement
reinforcement agent. agents, the properties of re-
inforced soil are quite different from the original soil. The experimental soil belongs to the
clay and comesTested
fromItem Testing
a test section in Pukou District, Reference
Nanjing. The specific Tested
physical prop-
Standards Standards Results
erties are shown in Table 2.
Density (g/cm3 ) D ± 0.03 1.41
pH A ± 1.0 7.6
Soil stabilizing
Soluble solids S ± 2.0 96.5
admixtures
Unconfined compression strength ≥2.5 3.2
CJ/T 486-2015 [28]
Unconfined compressive strength ratio ≥120 148
Seven-day water stability coefficient ratio ≥105 115
Note: D, A and S denote the density, pH value and solid content of product control value, respectively.

Due to the introduction of polymer-based reinforcement agents, the properties of

reinforced soil are quite different from the original soil. The experimental soil belongs to
the clay and comes from a test section in Pukou District, Nanjing. The specific physical
properties are shown in Table 2.

Table 2. The physical indexes of soil.

Optimum
Specific Maximum Dry Plastic Liquid Plasticity
Moisture
Gravity (g/cm3 ) Density (g/cm3 ) Limit (%) Limit (%) Index (%)
Content (%)
2.701 1.78 17.0 23.63 35.29 11.66

3.1.2. Filter Paper Test

The filter paper test was applied to measure the matric suction in accordance with
ASTM D5298. A series of moisture contents were tested so as to establish the optimal
moisture content: these were 12%, 15%, 17%, 18%, 20%, 23%, and 25%, respectively. The
cylindrical specimens were formed with the dimension of Φ 3.91 cm × 8 cm. Three pieces
of filter paper for each experiment were prepared and placed between two sections of a soil
sample that had been sawn in half. Only the middle filter paper was used for measuring
 3.1.2. Filter Paper Test
The filter paper test was applied to measure the matric suction in accordance with
ASTM D5298. A series of moisture contents were tested so as to establish the optimal
moisture content: these were 12%, 15%, 17%, 18%, 20%, 23%, and 25%, respectively. The
Polymers 2023, 15, 4187 5 of 18
cylindrical specimens were formed with the dimension of Φ 3.91 cm × 8 cm. Three pieces
of filter paper for each experiment were prepared and placed between two sections of a
soil sample that had been sawn in half. Only the middle filter paper was used for measur-
the
ing variation in moisture
the variation weight,
in moisture while
weight, the other
while two two
the other werewere
to prevent the pollution
to prevent of the
the pollution of
middle filter paper. The entire sample was then sealed and stored at the temperature
the middle filter paper. The entire sample was then sealed and stored at the temperature of
60 ◦ C for 7 days to ensure a sufficient balance of moisture between the filter paper and the
of 60 ℃ for 7 days to ensure a sufficient balance of moisture between the filter paper and
soil sample.
the soil The The
sample. specimen preparation
specimen is shown
preparation in Figure
is shown 2. 2.
in Figure

(a) (b) (c)

Figure 2.
Figure 2. The
The specimen
specimen preparation:
preparation: (a)
(a) formation,
formation, (b)
(b) filter
filter paper,
paper, (c)
(c) seal.
seal.

3.2. Triaxial
3.2. Triaxial Test
Test ofofDynamic
DynamicResilient
ResilientModulus
Moduluswith
withDry–Wet
Dry–WetCycle
Cycle
3.2.1. Dry–Wet Cycle
3.2.1. Cycle
In order totostudy
studythe theeffect
effectof of drying–wetting
drying–wetting cycles
cycles on resilient
on the the resilient modulus
modulus of com- of
compacted
pacted soil,soil,
a testa test
withwith different
different numbers
numbers of drying–wetting
of drying–wetting cyclescycles
was was carried
carried out. out.
The
The resilient
resilient modulus
modulus was was
thenthen measured.
measured. The initial
The initial waterwater content
content of 12%of was
12%set
was asset
theas the
start-
starting
ing point point
andand
the the amplitude
amplitude valuevaluewaswas ±2%.
±2%. FourFour
timestimes of cycle
of cycle were
were determined,
determined, i.e.,i.e.,
1,
1,
2, 2, 3 and4.4.The
3 and Themoisture
moisturecontent
contentfirst
firstincreased
increasedfrom
fromthetheinitial
initial value
value of
of 12% to 14%, thenthen
the specimens
specimens werewere dried
dried in anan oven
oven down
down to to 10%,
10%, followed
followed by by continuing
continuing to to add
add water
water
back
back upup to 12% for a cycle: namely,
namely, 12% 12% to 14% to 10% to 12%. The specific implementation
implementation
was
was as as follows:
follows: (1)
(1) humidification:
humidification: the the required
required water
water content
content was
was calculated
calculated and
and injected
injected
into
into the
the sample
sample by by aa syringe
syringe toto reach
reach the the target
target peak.
peak. When
When adding
adding water,
water, the
the surface
surface of of
the
the sample
samplewas wascovered
coveredwithwitha filter
a filterpaper
paperto reduce
to reducethe the
erosion of the
erosion ofsample surface,
the sample then
surface,
sealed and moisturized
then sealed for 24 h;
and moisturized for(2)24dehumidification: the specimen
h; (2) dehumidification: was placed
the specimen wasinplaced
the oven in
for dehumidification after the humidification process was completed;
the oven for dehumidification after the humidification process was completed; then the then the specimen
was weighed
specimen wasevery
weighed 10 min to 10
every measure
min to the variation
measure in moisture
the variation content until
in moisture it reached
content until it
10%, and was then sealed for 24
reached 10%, and was then sealed for 24 h.h.

3.2.2. Dynamic Resilient Modulus

3.2.2. Dynamic Resilient Modulus
The GDS unsaturated soil triaxial apparatus can simulate the mechanical response
The GDS unsaturated soil triaxial apparatus can simulate the mechanical response of
of the soil under dynamic loading. The Haversine wave loading mode was used as it can
the soil under dynamic loading. The Haversine wave loading mode was used as it can best
best reflect the typical stress state inside the road. The period was set as 1.0 s, in which
reflect the typical stress state inside the road. The period was set as 1.0 s, in which the load
the load pulse was 0.1 s and the rest period was 0.9 s. The pre-loading was performed to
eliminate the occurrence of large plastic deformation at the initial stage and to simulate the
stress history of subgrade soil. The loading sequence is listed in Table 3, while the dynamic
triaxial test process is shown in Figure 3 and the testing process is shown in Figure 4.
The resilient modulus was carried out by AASHTO T-307. For each loading se-
quence, the data of the last five loading cycles were taken to calculate the dynamic resilient
modulus when the strain response became stable. The calculation of Mr is shown in
Equations (9)–(11).
P
σd = i (9)
A
Polymers 2023, 15, 4187 6 of 18

In the above equation, σd is the axial stress; Pi , the average amplitude of the last five axial
loadings
Polymers 2023, 15, x FOR PEER REVIEW (N); and A, the average cross-sectional area of the upper and lower surfaces 6ofofthe 19
specimen (mm2 ).
∆
ε0 = i (10)
l0
pulse was 0.1 s and the rest period was 0.9 s. The pre-loading was performed to eliminate
In
thethis equation,ofε0large
occurrence is theplastic
recoverable at the initial∆istage
strain (mm/mm);
deformation , the average axial deformation
and to simulate the stress
amplitude of the last five loadings (mm); and
history of subgrade soil. The loading sequence l , the measurement distance of displacement
0 is listed in Table 3, while the dynamic tri-
sensor (mm).
axial test process is shown in Figure 3 and the σ testing process is shown in Figure 4.
Mr = d (11)
ε0
Table 3. Loading sequence of dynamic triaxial test.
Here, Mr is the dynamic resilient modulus.
Sequence Num- Confining Stress Deviator Stress Vertical Stress
Load Times/s
Table 3. ber σ3of(kPa)
Loading sequence dynamic triaxialσtest.
d (kPa) 𝝈𝟏 (kPa)
Pre-loading 30 55 85 1000
Sequence Confining Stress Deviator Stress Vertical Stress
1 60 30 90 Load100
Times/s
Number σ3 (kPa) σd (kPa) σ1 (kPa)
2 45 30 75 100
Pre-loading 30 55 85 1000
1
3 60
30 30
30
60
90
100
100
24 4515 30
30 45
75 100
100
35 3060 55
30 115
60 100
100
46 1545 30
55 45
100 100
100
5 60 55 115 100
7 30 55 85 100
6 45 55 100 100
7 8 3015 55
55 70
85 100
100
89 1560 75
55 135
70 100
100
9 10 6045 75
75 135
120 100
100
10 45 75 120 100
11 30 75 105 100
11 30 75 105 100
1212 1515 75
75 90
90 100
100
1313 6060 105
105 165
165 100
100
1414 4545 105
105 150
150 100
100
1515 3030 105
105 135
135 100
100
16 15 105 120 100
16 15 105 120 100

Figure 3. Diagram of cycle

Figure 3. cycle load.
load.
Polymers 2023, 15,
Polymers 2023, 15, 4187
x FOR PEER REVIEW 7 7of
of 19
18

(a) (b) (c)

Figure
Figure 4.
4. Dynamic
Dynamic triaxial
triaxial test
test process:
process: (a)
(a) specimens,
specimens, (b)
(b) sample
sample placement,
placement, (c)
(c) triaxial test.
triaxial test.

The resilient
4. Analysis modulus
of Resilient was carried
Modulus Modelout by AASHTO T-307. For each loading se-
quence, the data of
4.1. The Curve of SWCCthe last five loading cycles were taken to calculate the dynamic resili-
ent modulus when the strain response became stable. The calculation of Mr is shown in
The SWCC was obtained for the tested soil material according to the testing results
Equations (9)–(11).
Polymers 2023, 15, x FOR PEER REVIEW
of the filter paper test, as shown in Figure 4. The fitting accuracy is 0.93. Where the 8water
of 19

content is given, the matric suction will have been determined from Figure 5.
Pi
σd = (9)
A
In the above equation, σd is the axial stress; Pi, the average amplitude of the last five axial
loadings (N); and A, the average cross-sectional area of the upper and lower surfaces of
the specimen (mm2).

Δi
ε0 = (10)
l0
In this equation, ε0 is the recoverable strain (mm/mm); Δi, the average axial deformation
amplitude of the last five loadings (mm); and l0, the measurement distance of displace-
ment sensor (mm).

σd
Mr = (11)
ε0
Figure 5. Soil water characteristic curves for soil material.
Here, Mr is the dynamic resilient modulus.
4.2. Results of Triaxial
Triaxial Test
Test
4. Analysis of Resilient Modulus Model
4.2.1.
4.2.1. Effect of Deviator Stress
Effect of Deviator Stress
4.1. The Curve of SWCC
Figure
Figure 6 presents the trends
6 presents the trends between
between thethe resilient
resilient modulus
modulus and and the
the deviator
deviator stress
stress
for Thespecimens
SWCC was obtained for the tested soil material according tothe
thelegend
testing results
for the specimens without dry–set cycle, in which reference to C in the legend is
the without dry–set cycle, in which reference to C in is to
to the
the
of the filter paper
abbreviation test, as shown
for confining
confining stress.in Figure 4. The modulus
Theresilient
resilient fitting accuracy is 0.93. Where the to
water
abbreviation for stress. The modulus hashas a negative
a negative relation
relation to the the
de-
content
deviator isstress.
given,This
the may
matric suction will have been determined from
inaaFigure 5.
viator stress. This may bebe becausehigher
because higher deviator
deviator stressresults
stress results in largervertical
larger vertical strain,
strain,
and the increment in vertical strain is greater than the increased value
and the increment in vertical strain is greater than the increased value of deviator stress,of deviator stress,
which leads to
which leads to aa reduction
reductionin inresilient
resilientmodulus
modulus[29].[29].Taking
Taking thethe group
group of of C-60
C-60 kPakPa as
as an
an example, the decreasing rates are 14.1%, 19.0%, and 25.4% when
example, the decreasing rates are 14.1%, 19.0%, and 25.4% when the deviator stress in- the deviator stress
increases from 30 kPa to 55 kPa, 75 kPa, and 105 kPa, respectively. This is consistent with
creases from 30 kPa to 55 kPa, 75 kPa, and 105 kPa, respectively. This is consistent with
other findings [30,31]. In addition, the decreasing rate has a larger value at low deviator
other findings [30,31]. In addition, the decreasing rate has a larger value at low deviator
stress while a smaller decreasing amplitude can be found at high deviator stress, indicating
stress while a smaller decreasing amplitude can be found at high deviator stress, indicat-
that the resilient modulus has a dependence on the deviator stress. This may explain
ing that the resilient modulus has a dependence on the deviator stress. This may explain
why the vertical compressive strain should be controlled for the subgrade design. The
why the vertical compressive strain should be controlled for the subgrade design. The
stress state can be calculated once the pavement structure is determined, and the vertical
stress state can be calculated once the pavement structure is determined, and the vertical
compressive strain is limited to a specified range to meet the requirements of minimum
value for resilient modulus.

Polymers 2023, 15, 4187
example, the decreasing rates are 14.1%, 19.0%, and 25.4% when the deviator stress in-
creases from 30 kPa to 55 kPa, 75 kPa, and 105 kPa, respectively. This is consistent with
other findings [30,31]. In addition, the decreasing rate has a larger value at low deviator
stress while a smaller decreasing amplitude can be found at high deviator stress, indicat-
ing that the resilient modulus has a dependence on the deviator stress. This may explain 8 of 18
why the vertical compressive strain should be controlled for the subgrade design. The
stress state can be calculated once the pavement structure is determined, and the vertical
compressive
compressivestrain
strainisislimited
limitedtotoaaspecified
specifiedrange
rangetotomeet
meetthe
therequirements
requirementsof
ofminimum
minimum
value for resilient modulus.
value for resilient modulus.

Polymers 2023, 15, x FOR PEER REVIEW 9 of 19

Thetrend
Figure6.6.The
Figure trendbetween
betweenresilient
resilientmodulus
modulusand
anddeviator
deviatorstress.
stress.
4.2.2. Effect of Confining Stress
4.2.2. Effect of Confining Stress
Figure 7 shows the variation in resilient modulus with the confining stress: the spec-
Figure 7dry–set
imen without shows the variation
cycle in resilient
was used. modulus of
The abbreviation with thethe
D in confining stress: the spec-
legend represents the
imen without dry–set cycle was used. The abbreviation of D in the
deviator stress. The confining stress has a lateral restraint effect on the specimens legend represents
and
the deviator
promotes stress.
a better The capacity.
bearing confiningThus,
stressthe
has a lateral
resilient restraint
modulus effect onpositively
is linearly the specimens
cor-
and promotes a better bearing capacity. Thus, the resilient modulus is linearly
related with the confining pressure. This is in agreement with previous studies [32,33]. positively
correlated with the confining pressure. This is in agreement with previous studies [32,33].

Figure
Figure 7. 7.
TheThe variation
variation inin resilient
resilient modulus
modulus with
with confining
confining stress.
stress.

4.2.3.Effect
4.2.3. Effect
ofof StressState
Stress State
Figure8 8shows
Figure showsthethechanges
changesininresilient
resilientmodulus
moduluswith withthe
thevariation
variationinindeviator
deviatorstress
stress
and bulk stress. The figures in the legend represent the number of dry–wet
and bulk stress. The figures in the legend represent the number of dry–wet cycles. For the cycles. For the
same curve in Figure 8a, the deviator stress is fixed, the increasing bulk stress
same curve in Figure 8a, the deviator stress is fixed, the increasing bulk stress indicates indicates that
thethe
that confining
confiningstress is increased,
stress is increased,while
whilethe
thecurves
curvesininFigure
Figure8b8brepresent
representthe
thevariation
variationin
deviator stress. It can be seen that the resilient modulus increases with the
in deviator stress. It can be seen that the resilient modulus increases with the confining confining stress,
and has a negative relationship to the deviator stress. Previous studies have found that the
stress, and has a negative relationship to the deviator stress. Previous studies have found
deviator stress has a more significant effect than the confining stress for fine-graded soils,
that the deviator stress has a more significant effect than the confining stress for fine-
graded soils, especially for clay soil [34]. The deviator stress reflects the shear effect which
is prone to resulting in a softening effect for the subgrade soil [32].
 Figure 8 shows the changes in resilient modulus with the variation in deviator stress
and bulk stress. The figures in the legend represent the number of dry–wet cycles. For the
same curve in Figure 8a, the deviator stress is fixed, the increasing bulk stress indicates
that the confining stress is increased, while the curves in Figure 8b represent the variation
Polymers 2023, 15, 4187 in deviator stress. It can be seen that the resilient modulus increases with the confining 9 of 18
stress, and has a negative relationship to the deviator stress. Previous studies have found
that the deviator stress has a more significant effect than the confining stress for fine-
graded soils,for
especially especially
clay soilfor clay
[34]. soildeviator
The [34]. Thestress
deviator stress
reflects thereflects the shear
shear effect effect
which whichto
is prone
isresulting
prone to in
resulting in a softening effect for the subgrade
a softening effect for the subgrade soil [32]. soil [32].

(a) (b)
Polymers 2023, 15, x FOR PEER REVIEW 10 of 19
Figure 8. 8.
Figure Resilient modulus
Resilient with
modulus the
with change
the inin
change stress
stressstate:
state:(a)(a)
deviator stress,
deviator (b)
stress, confining
(b) confiningstress.
stress.

4.2.4. Effect
4.2.4. EffectofofDry–Wet
Dry–WetCycle CycleNumber
Number
The confining
The confiningstressstressininthe
thesubgrade
subgrade is is generally
generally distributed
distributed fromfrom
25 kPa25 kPa tokPa,
to 40 40 kPa,
from which 30 kPa was selected. Figure 9 shows the relations between the variation in in
from which 30 kPa was selected. Figure 9 shows the relations between the variation
resilient modulus
resilient modulusand andthe thenumber
numberofofthe thedry–wet
dry–wet cycle.
cycle. TheThe effects
effects of the
of the dry–wet
dry–wet cyclecycle
can result in a reduction in resilient modulus. The reducing amplitude caused by the first the
can result in a reduction in resilient modulus. The reducing amplitude caused by
first cycle
cycle is theislargest,
the largest,
at 37%. atAfter
37%.that,
After that,
the the reducing
reducing rate decreases,
rate decreases, and themod-
and the resilient resilient
modulus
ulus almost almost
reachesreaches a stable
a stable statefour
state after aftercycles.
four cycles.
Another Another
study hasstudyalsohas also that
found found thethat
the resilient
resilient modulus
modulus significantly
significantly decreases
decreases withinwithin the first
the first eighteight cycles,
cycles, followed
followed by aby a
slightly decreasing
slightly decreasingamplitude
amplitude[35]. [35].Ceratti
Ceratti etet
al.al. revealed
revealed that
that thethe resilient
resilient modulus
modulus at the
at the
optimalmoisture
optimal moisturecontent
contentdecreases
decreases during
during upup to the
to the firstfirst
fourfour dry–wet
dry–wet cycles cycles [36].isThis
[36]. This
is consistent
consistent withwith
thethe findings
findings in this
in this paper.
paper. Cyclic
Cyclic dry–wet
dry–wet conditions
conditions willwill result
result in the
in the
breakdown and andbuildup
buildupofofsoil soilparticles
particles[37,38].
[37,38]. This is because of the irreversible
This is because of the irreversible swelling swelling
and the increase in in saturation
saturation degree
degree caused
causedby bythe
thedry–wet
dry–wetcycle cyclethat
thatlead
leadtotoinstability
instability of
interlocking
of interlocking structure
structure andanda softening
a softening behavior
behavior of of
thethe
soil [39].
soil After
[39]. Afterthat,
that, thethe
soil is more
soil is
susceptible
more to yield.
susceptible In addition,
to yield. In addition,the smaller
the smaller the deviator
the deviator stress, the the
stress, larger stable
larger value
stable
of resilient
value modulus.
of resilient modulus. TheTheresilient modulus
resilient modulus is is
about
about6666MPaMPawhen deviatorstress
when the deviator stress is
3030
is kPa, while
kPa, whileit it
is is
5252 MPa,
MPa,5050MPa,MPa,andand3030MPaMPaatatthe thedeviator
deviatorstress
stressofof5555kPa,kPa,7575kPa
kPaand
and kPa,
105 105 kPa, respectively.
respectively.

Figure
Figure 9.
9. The
Thevariation
variationininresilient
resilientmodulus
moduluswith
withnumber of of
number dry–wet cycles.
dry–wet cycles.

4.3. Determination of Resilient Modulus Prediction Model

The testing data was used to calibrate the model coefficients involved in Equations
(7) and (8), and the fitting parameters are presented in Table 4. The coefficient of determi-
nation values (R2) is larger than 0.9 for Equation (7), while the prediction model given by
 Polymers 2023, 15, 4187 10 of 18

4.3. Determination of Resilient Modulus Prediction Model

The testing data was used to calibrate the model coefficients involved in Equations (7) and (8),
and the fitting parameters are presented in Table 4. The coefficient of determination values
(R2 ) is larger than 0.9 for Equation (7), while the prediction model given by Equation (8)
has a lower correlation. Thus, the final form of the resilient modulus prediction model in
this paper is Equation (7).

Polymers 2023, 15, x FOR PEER REVIEW

Table 4. Fitting results of prediction model. 11 of 1

Equation q1 q2 q3 q4 q5 R2
(7) 2340.93 0.40 −2.43 −0.61 0.28 0.93
5. FEM
(8) analysis
1123.89 0.41 −2.49 0.99 −6.09 0.78
5.1. Establishment of FEM Model
5.5.1.1.
FEMFEM Model
Analysis
5.1. Establishment
The standardof FEM
of a Model
two-way four-lane highway and the design speed of 100 km/h wer
5.1.1. FEM Model
adopted. The section of the submerged subgrade is shown in Figure 10. The width of th
top The standard
surface of am.
is 24.5 two-way four-lane
The overall highway structure
pavement and the design
abovespeed
theofsubgrade
100 km/his were
4 cm AC1
adopted. The section of the submerged subgrade is shown in Figure
(asphalt concrete with a nominal maximum particle size of 13 mm), 6 cm AC20 10. The width of the (asphal
top surface is 24.5 m. The overall pavement structure above the subgrade is 4 cm AC13
concrete with a nominal maximum particle size of 20 mm), 12 cm ATB25 (asphalt-treated
(asphalt concrete with a nominal maximum particle size of 13 mm), 6 cm AC20 (asphalt
base with
concrete a nominal
with a nominalmaximum
maximum particle sizeofof2025mm),
particle size mm),1216cmcm continuously
ATB25 graded crush
(asphalt-treated
aggregate
base layer, and
with a nominal 32 cm cement
maximum particlestabilized gravel.
size of 25 mm), 16The slope of subgrade
cm continuously gradedwas
crushset to 1:1.
aggregate layer, and 32 cm cement stabilized gravel. The slope of subgrade was set tostructur
[40], and the height to 8 m. The foundation size was 60 m × 10 m. The pavement
was [40],
1:1.5 simplified
and theas a plane
height strain.
to 8 m. The The typicalsize
foundation plane
wasstrain
60 m × CPE4
10 m.element with complet
The pavement
structure
integral was
was simplified
used. As as fora the
plane strain. The
seasonal typical
soaking plane strain
subgrade, the CPE4 element
subgrade andwithfoundation
complete integral was used. As for the seasonal soaking subgrade, the
should consider the coupling effect of the seepage field with the change in water level subgrade and
foundation should consider
Thus, the freedom degree the coupling
of pore effect should
pressure of the seepage field with
be considered. the change
Finally, in
the completely
water level. Thus, the freedom degree of pore pressure should be considered. Finally, the
integrated plane strain pore pressure element CPE4P was used. The two sides of subgrad
completely integrated plane strain pore pressure element CPE4P was used. The two sides
and
of foundation
subgrade were set as
and foundation weretheset
water
as thehead
waterboundary.
head boundary.

Figure 10.Finite
Figure10. element
Finite model
element of immersed
model subgrade.
of immersed subgrade.

5.1.2. Material Parameter

5.1.2. Material Parameter
The linear elastic model was used in the numerical calculation of the pavement
TheThe
structure. linear elastic model
mechanical was used
parameters in the
that are numerical
needed calculation
in the modeling of themodulus
are elastic pavement struc
ture. The mechanical parameters that are needed in the modeling are elastic
E, Poisson’s ratio µ and material density ρ. The values of each parameter are determined modulus E
Poisson’s
based on theratio µ and material
specifications density [41],
of JTG D50-2017 ρ. The values
which of each
are shown parameter
in Table 5. are determined
based on the specifications of JTG D50-2017 [41], which are shown in Table 5.

Table 5. FEM input parameters of pavement material.

Layer Material Type E (MPa) µ ρ (g/cm3)

AC13 11,000 0.25 2.42
Surface AC20 12,000 0.25 2.42
ATB25 9000 0.25 2.42
Graded crush ag-
 Polymers 2023, 15, 4187 11 of 18

Table 5. FEM input parameters of pavement material.

Layer Material Type E (MPa) µ ρ (g/cm3 )

AC13 11,000 0.25 2.42
Surface AC20 12,000 0.25 2.42
ATB25 9000 0.25 2.42
Base Graded crush aggregate 500 0.35 2.36
Subbase Cement stabilized gravel 9000 0.25 2.35
Polymers 2023, 15, x FOR PEER REVIEW Subgrade Foundation Soil 60 0.4 12 of 19
1.83

In addition, since the subgrade and foundation are affected by the changes in water
coefficient can refer
level, related to the parameters
hydraulics empirical values
such as ofthe
representative rock and soil
saturated permeability listed inand
coefficient the
specification
correspondingof JTG/T D33-2012
reduction [42]. and
coefficient, Thematrix
soil used in this
suction paper is low
are necessary liquid limit
for solving clay,
the fluid–
andsolid
the saturated permeability
coupling problem in FEM. coefficient is setofasthe
The selection 5 ×saturated
10−5 mm/s. Considering
permeability that thecan
coefficient var-
refer
iation into the empirical
water values
level studied of representative
in this rock and
paper is in months, thesoil listed in thecoefficient
permeability specification of
is con-
JTG/T
verted D33-2012
to the value [42]. Them/mon.
of 0.129 soil usedThein this papercalculation
default is low liquid limit clay,
formula andreduction
of the the saturated
coef-
permeability
ficient in ABAQUS coefficient
is shownis set
inas 5 × 10−5(12),
Equation mm/s.
which Considering
is adopted that
inthe
thisvariation
paper. Thein water
volu-
metric water content could be converted to the saturation degree. Next, the values to
level studied in this paper is in months, the permeability coefficient is converted of the
ma-
tricvalue
suctionof 0.129 m/mon. The
and volumetric default
water calculation
content formula ofaccording
were calculated the reduction
to thecoefficient
SWCC. Itinis
ABAQUS is shown in Equation (12), which is adopted in this paper. The volumetric water
worth noting that the negative pore pressure is generally used instead of the matrix suc-
content could be converted to the saturation degree. Next, the values of matric suction and
tion in the FEM.
volumetric water content were calculated according to the SWCC. It is worth noting that
ks = ( Sr )
3 of the matrix suction in the FEM.
the negative pore pressure is generally used instead
(12)
k = ( Sr ) 3 (12)
Here, ks is the permeability reductions coefficient of unsaturated soil; and Sr, the satu-
ration degree.
Here, ks is the permeability reduction coefficient of unsaturated soil; and Sr , the
saturation degree.
5.1.3. Periodic Change in Water Level
5.1.3.
ThePeriodic Change
local rainfall is in Waterconcentrated
mainly Level from May to October, showing the charac-
teristics of a long cycle and large rainfall. The rainy May
The local rainfall is mainly concentrated from to October,
season and watershowing the character-
depth are simplified
for simulation calculations. Assuming that the initial groundwater level is located at −1for
istics of a long cycle and large rainfall. The rainy season and water depth are simplified m,
thesimulation calculations.
rainfall starts from April Assuming
to May that
and the initial
then groundwater
continues level is located
until October, −1 m, the
and theatgroundwa-
ter rainfall starts from April to May and then continues until October, and the groundwater
level rises to the maximum value, which is located at 2 m above the surface of the earth.
level rises to the maximum value, which is located at 2 m above the surface of the earth. As
As the rainfall weakens, the groundwater level begins to fall back to the initial groundwa-
the rainfall weakens, the groundwater level begins to fall back to the initial groundwater
ter level and remains unchanged. The variation in the annual water level of the subgrade
level and remains unchanged. The variation in the annual water level of the subgrade is
is shown
shown in in Figure 11.
Figure 11.

2.5
2
1.5
Water Level /m

1
0.5
0
−0.5 1
-0.5 2 3 4 5 6 7 8 9 10 11 12
-1
−1
−1.5
-1.5 Month
Figure 11. Annual water level change diagram.

Figure 11. Annual water level change diagram.

5.1.4. Load and Mesh

The standard axle load of 0.7 MPa and loading radius of 0.213 m were applied in the
general mechanical analysis. Since the plane two-dimension FEM model was used in this
paper, the load had to be converted based on the equivalence principle and was deter-
Polymers 2023, 15, 4187 12 of 18

5.1.4. Load and Mesh

The standard axle load of 0.7 MPa and loading radius of 0.213 m were applied in the
general mechanical analysis. Since the plane two-dimension FEM model was used in this
paper, the load had to be converted based on the equivalence principle and was determined
as 117.37 kPa/s. The haversine wave with 0.1 s load pulse and 0.9 s rest period was adopted
which is consistent with that used in the dynamic triaxial test.

5.1.5. Calculation Logic

Polymers 2023, 15, x FOR PEER REVIEW 13 of 19
The prediction model of resilient modulus requires hydraulic parameters such as the
pore pressure and volumetric water content, which need to be obtained and stored as state
variables before calling the UMAT subroutine. The USDFLD subroutine is used to redefine
subroutine
the to extract
field variables the
of the output result
integration data
points. stored
It is on the
generally material
used integration
with the points in
built-in GETVRM
the database. The calculation process for resilient modulus in ABAQUS is shown
subroutine to extract the output result data stored on the material integration points in Figure
in the
12.
database. The calculation process for resilient modulus in ABAQUS is shown in Figure 12.

Figure12.
Figure 12.Algorithm
Algorithmflow
flowchart
chartof
ofFEM.
FEM.

5.2.
5.2.Result
ResultAnalysis
Analysis
5.2.1.
5.2.1. Analysis of
Analysis of Water
WaterLevel
LevelChange
Change
The
Thecloud
cloudmap
mapof ofthe
thedistribution
distributionof ofresilient
resilientmodulus
moduluswas wasextracted
extractedfor forApril
Aprilwhen
when
the water level started to rise, while October when the water level started
the water level started to rise, while October when the water level started to decline to decline after a
after
period of high-water level was used for comparison. Because of the symmetry
a period of high-water level was used for comparison. Because of the symmetry of the of the model,
half of the
model, cloud
half maps
of the cloudofmaps
each characteristic month are
of each characteristic shown.
month The resulting
are shown. cloud map
The resulting cloudis
shown in Figure 13.
map is shown in Figure 13.
The
Theresilient
resilient modulus
modulus shows
shows anan increasing
increasing trend
trend from
from the
the top
top surface
surface toto the
the inside.
inside.
The
Themodulus
modulusininApril
Aprilisishigher
higherthan that
than in in
that October.
October.This is because
This is becausethe the
water levellevel
water and and
the
humidity are higher in October. The spatial distribution of bulk stress and shear
the humidity are higher in October. The spatial distribution of bulk stress and shear stress stress is
consistent with that of resilient modulus. This is because the upper part of the subgrade is
is consistent with that of resilient modulus. This is because the upper part of the subgrade
less affected by the vehicle load compared to the gravity of subgrade. Thus, the minimum
is less affected by the vehicle load compared to the gravity of subgrade. Thus, the mini-
values of bulk stress and shear stress appear in the upper part of the subgrade. It can be
mum values of bulk stress and shear stress appear in the upper part of the subgrade. It
seen from the constitutive model that the resilient modulus is positively correlated with the
can be seen from the constitutive model that the resilient modulus is positively correlated
bulk stress and negatively correlated with the shear stress, that is, the two have opposite
with the bulk stress and negatively correlated with the shear stress, that is, the two have
effects. Although the shear stress at the depth of the foundation is at the maximum value,
opposite effects. Although the shear stress at the depth of the foundation is at the maxi-
the resilient modulus here is also the maximum value, which also shows the prominent
mum value, the resilient modulus here is also the maximum value, which also shows the
role of volume stress in the calculation of resilient modulus.
prominent role of volume stress in the calculation of resilient modulus.
Polymers 2023, 15, x FOR PEER REVIEW 14 of 19
Polymers 2023, 15, 4187 13 of 18

(a)

(b)

(c)
Figure
Figure13.
13.The
Themechanical
mechanicalresponse
responsetotosoaking
soakingembankment:
embankment:(a)(a)resilient
resilientmodulus,
modulus,(b)
(b)bulk
bulkstress,
stress,
(c) shear stress.
(c) shear stress.

5.2.2.
5.2.2.Analysis
AnalysisofofWater
WaterLevelLevelFluctuation
Fluctuation
The
The three elements A1, A2 and
three elements A1, A2 and A3A3 were
were selected
selectedtotorepresent
representthe thethree
three positions
positions of of
the
the upper,
upper, middle,
middle, and and lower
lower parts
parts of of
thethe slope,
slope, respectively,
respectively, as as shown
shown in Figure
in Figure 14,14,
andandthe
the variation
variation in the
in the bearing
bearing capacity
capacity of subgrade
of subgrade with under
with time time under the dry–wet
the dry–wet cycle wascycle was
studied.
studied. The resilient
The resilient modulus modulus and cumulative
and cumulative volumetric
volumetric waterwater content
content at threeat three positions
positions were
were
extracted, and the time–history curves were drawn as shown in Figures 15 and 16. It can Itbe
extracted, and the time–history curves were drawn as shown in Figures 15 and 16.
can be found
found that thethat the resilient
resilient modulusmodulus in theand
in the middle middle
bottom and
of bottom
the slopeofdecline
the slope decline
significantly,
significantly,
and the initial and the initial
resilient resilient
modulus modulus
at the positionatof
the
A2position of A2 is
is the largest, at the largest,
about 60 MPa.at about
In the
60first
MPa.yearIn of
thedry–wet
first yearcycle,
of dry–wet cycle, the
the decrease decrease
in the in the
resilient resilient
modulus modulus
is the is the most
most severe, with
a decrease
severe, with of up to 18 MPa.
a decrease For18the
of up to topFor
MPa. surface of the
the top subgrade,
surface of the the effect ofthe
subgrade, water
effectlevel
of
changes
water levelischanges
relatively small, with
is relatively a decreasing
small, amplitude
with a decreasing of only 5 of
amplitude MPa during
only 5 MPathe three-
during
year
the dry–wet dry–wet
three-year cycle. This is closely
cycle. This iscorrelated with the changes
closely correlated with theinchanges
cumulative volumetric
in cumulative
volumetric water content in Figure 15. The cumulative volumetric water content atofthe
water content in Figure 15. The cumulative volumetric water content at the positions A2
and A3 increases with time, thus the resilient modulus decreases accordingly.
 Polymers
Polymers 2023,
2023, 15,15, x FOR
x FOR PEER
PEER REVIEW
REVIEW 1515ofof1919

Polymers 2023, 15, 4187 positionsofofA2

positions A2and
andA3
A3increases
increaseswith
withtime,
time,thus
thusthe
theresilient
resilientmodulus
modulusdecreases
decreasesaccord-
accord-
14 of 18
ingly.
ingly.

Figure
Figure
Figure 14.
14.
14. Selection
ofof
Selection
Selection of characteristic
characteristic
characteristic element.
element.
element.

Figure 15.
Figure15.
Figure The
The
15.The relationship between
relationshipbetween
relationship resilient
betweenresilient modulus
resilientmodulus and
modulusand time.
andtime.
time.

Figure
Figure
Figure 16.
16.
16. The
The
The relationship
relationship
relationship between
between
between cumulative
cumulative
cumulative volumetric
volumetric
volumetric water
water
water content
content
content and
and
and time.
time.
time.

5.2.3. Analysis of Monthly Equivalent Resilient Modulus

The weighted average equivalent method was adopted based on the transfer mecha-
nism of vehicle load. The load is larger near the load position, and the contribution to the
 according to Equation (13):

Eeq =
 M ⋅u i i
(13)
u i

Polymers 2023, 15, 4187 where Mi is the resilient modulus of each element; and ui, the vertical displacement of each 15 of 18
element.
The monthly equivalent resilient modulus of the subgrade was calculated and com-
pared with the average
equivalent resilient
resilient modulus
modulus and
at the the resilient
upper positionmodulus
should be oflarger
top surface, which
than that at the lower
are shown in Figure 18. The comparison shows that the fluctuation of three
position. The index of displacement distribution along the vertical direction of pavement equivalent
modulusstructure
values is was
the same, andtovaries
selected use asinaaccordance
weight function.with the Thechange
verticalin deformation
water level. The
of subgrade
subgradeininOctober
Octoberisisshown
affected in by the rise
Figure 17. Itincan
thebewater
seenlevel
that thein the previous
vertical months and
deformation of subgrade
the continuous
spreads high-water
downward level,
from resulting
the loadinposition.
larger humidity
With the and lower resilient
increase in depth,modulus
the deformation
compared with other
gradually months This
decreases. in the same year.law
distribution When the waterwith
is consistent levelthe
then begins to
modulus de- of each
weight
cline, thepart
resilient modulus increases. In addition, the three kinds of resilient
of the subgrade, that is, the position with large deformation should also display modulus all large
show periodic changes during the two years and weaken over time until eventually
modulus to resist the external load. Thus, the spatial distribution of vertical deformation stable.
This is consistent
inside the with the simulation
subgrade meets theresults of water level
requirements changes
of weight of representative
function. In specificele-
terms, the
ments. The average
resilient modulus
modulus and is maximum
maximum,vertical
followed by the equivalent
deformation resilientwithin
of all elements modulus,
the range of
while theload
resilient modulus
influence wereofexported,
the top surface
and theisvalue
minimum. The pavement
of equivalent modulus is assumed
was then tocalculated
bear the same force to
according at Equation
all positions when calculating the average modulus of the subgrade,
(13):
while the weakening effect of load along the Edepth ∑isM i · uconsidered,
not i which causes the
eq = (13)
modulus to be larger. The method for calculating equivalent ∑ i resilient modulus is more in
u
accordance withMthe
where actual stress condition of the subgrade. Therefore, it is more reason-
i is the resilient modulus of each element; and ui , the vertical displacement of each
able to use the
element.equivalent resilient modulus to guide the design of subgrade.

Figure 17. Vertical deformation of subgrade in October.

The monthly equivalent resilient modulus of the subgrade was calculated and com-
pared with the average resilient modulus and the resilient modulus of top surface, which
are shown in Figure 18. The comparison shows that the fluctuation of three equivalent
modulus values is the same, and varies in accordance with the change in water level. The
subgrade in October is affected by the rise in the water level in the previous months and
the continuous high-water level, resulting in larger humidity and lower resilient modulus
compared with other months in the same year. When the water level then begins to decline,
the resilient modulus increases. In addition, the three kinds of resilient modulus all show
periodic changes during the two years and weaken over time until eventually stable. This
is consistent with the simulation results of water level changes of representative elements.
The average modulus is maximum, followed by the equivalent resilient modulus, while
the resilient modulus of the top surface is minimum. The pavement is assumed to bear the
same force at all positions when calculating the average modulus of the subgrade, while
the weakening effect of load along the depth is not considered, which causes the modulus
to be larger. The method for calculating equivalent resilient modulus is more in accordance
with the actual stress condition of the subgrade. Therefore, it is more reasonable to use the
equivalent resilient modulus to guide the design of subgrade.
Polymers 2023, 15, x FOR PEER REVIEW 17 of 19

Polymers 2023, 15, 4187 16 of 18

Figure 17. Vertical deformation of subgrade in October.

Figure 18.
18. Comparison diagram of each
each equivalent
equivalent resilient
resilient modulus.
modulus.

6. Conclusions
This
This paper
paperaimed
aimedtotodevelop
develop a new
a new prediction
predictionmodelmodelof resilient modulus
of resilient for analyzing
modulus for ana-
the effect of the dry–wet cycle on the resilient modulus of polymer
lyzing the effect of the dry–wet cycle on the resilient modulus of polymer reinforced reinforced soil. This
soil.
included a FEM
This included analysis
a FEM which
analysis was performed
which was performed to analyze
to analyzethe actual stress
the actual statestate
stress of the
of
pavement,
the pavement, andandstudying thethe
studying effects
effects of of
dry–wet
dry–wet cycle.
cycle.TheThefollowing
followingconclusions
conclusions could
could
be
be drawn:
drawn:
(1)
(1) The
The increase
increase in in deviator
deviator stress
stress could
could result
result in
in aa reduction
reduction in in resilient
resilient modulus;
modulus; thethe
decreasing
decreasing rate at low deviator stress is larger than that at a high level; and the
rate at low deviator stress is larger than that at a high level; and the resilient
resilient
modulus
modulus is is linearly
linearly positively
positively correlated
correlated withwith the
the confining
confining pressure.
pressure.
(2)
(2) The effect of a dry–wet cycle could reduce the
The effect of a dry–wet cycle could reduce the resilient
resilient modulus,
modulus, whereby
whereby thethe
reducing
reducing amplitude
amplitudecaused causedbybythe
thefirst
firstdry–wet
dry–wetcyclecycleis is
thethe
largest,
largest,followed
followed by bya gradually
a gradu-
stabilising trend.
ally stabilising trend.
(3)
(3) The initial
The initial resilient
resilient modulus
modulus at at the
the middle
middle position
position of of slope
slope is is the
the largest
largest and
and it
it
also
also has the largest decreasing amplitude in the first year of dry–wet cycle; moreover, the
has the largest decreasing amplitude in the first year of dry–wet cycle; moreover, the
resilient
resilient modulus
modulus at at the
the upper
upper position
position hashas the
the most
most stable
stable trend.
trend.
(4) The FEM results show that the method for calculating
(4) The FEM results show that the method for calculating equivalent equivalentresilient
resilientmodu-
mod-
ulus is more in accordance with the actual stress condition of the subgrade, which is
lus is more in accordance with the actual stress condition of the subgrade, which is rec-
recommended to guide the design of the soaked subgrade.
ommended to guide the design of the soaked subgrade.
Author Contributions: Writing—original draft, methodology, and investigation, Y.L.; data curation,
Author Contributions: Writing—original draft, methodology, and investigation, Y.L.; data curation,
conceptualization, and investigation, W.L.; visualization, resources, and project administration, K.F.
conceptualization, and investigation, W.L.; visualization, resources, and project administration, K.F.
All authors have read and agreed to the published version of the manuscript.
All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by the China Postdoctoral Science Foundation (Grant/Award
Funding: This work was supported by the China Postdoctoral Science Foundation (Grant/Award
Number: 2022M721943), the Natural Science Foundation of Shandong Province (Grant/Award
Number: 2022M721943), the Natural Science Foundation of Shandong Province (Grant/Award
Number: ZR2022QE251), and the National Natural Science Foundation of China (Grant/Award
Number: ZR2022QE251), and the National Natural Science Foundation of China (Grant/Award
Number: 52309132 and 52209140).
Number: 52309132 and 52209140).
Data
Data Availability Statement: The
Availability Statement: The data
data presented
presented inin this
this study
study are
are available
available on
on request
request from
from the
the
corresponding author.
corresponding author.
Conflicts
Conflicts of Interest: The
of Interest: The authors
authors declare
declare no
no conflict
conflict of
of interest.
interest.

References
1. Rahman, M.M.; Islam, K.M.; Gassman, S.L. Correlations of permanent strain and damping coefficients with resilient modulus for
coarse-grained subgrade soils. Int. J. Geotech. Eng. 2021, 15, 714–723. [CrossRef]
2. Liu, X.; Zhang, X.; Wang, H.; Jiang, B. Laboratory testing and analysis of dynamic and static resilient modulus of subgrade soil
under various influencing factors. Constr. Build. Mater. 2019, 195, 178–186. [CrossRef]
Polymers 2023, 15, 4187 17 of 18

3. Yan, R.Z.; Ke, G.J.; Zhao, S.L.; Xia, J.L. Experimental study on dynamic resilient modulus of subgrade with circulating fluidized
bed combustion ash as filler. Constr. Build. Mater. 2023, 397, 132427. [CrossRef]
4. Liu, F.R.; Zhou, Z.W.; Ma, W.; Zhang, S.J.; Sun, Z.Z. Dynamic Parameters and Hysteresis Loop Characteristics of Frozen Silt Clay
under Different Cyclic Stress Paths. Adv. Mater. Sci. Eng. 2021, 2021, 3763181. [CrossRef]
5. Coronado, O.; Caicedo, B.; Taibi, S.; Correia, A.G.; Fleureau, J. A macro geomechanical approach to rank non-standard unbound
granular materials for pavements. Eng. Geol. 2021, 119, 64–73. [CrossRef]
6. Han, Z.; Vanapalli, S.K. Relationship between resilient modulus and suction for compacted subgrade soils. Eng. Geol. 2016, 211,
85–97. [CrossRef]
7. Hoyos, L.R.; Suescun-Florez, E.A.; Puppala, A.J. Stiffness of intermediate unsaturated soil from simultaneous suction-controlled
resonant column and bender element testing. Eng. Geol. 2015, 188, 10–28. [CrossRef]
8. Ekblad, J.; Isacsson, U. Influence of water on resilient properties of coarse granular materials. Road Mater. Pavement Des. 2006, 7,
369–404. [CrossRef]
9. Cary, C.E.; Zapata, C.E. Pore Water Pressure Response of Soil Subjected to Dynamic Loading under Saturated and Unsaturated
Conditions. Int. J. Geomech. 2016, 16, D4016004. [CrossRef]
10. Salour, F.; Erlingsson, S. Resilient modulus modelling of unsaturated subgrade soils: Laboratory investigation of silty sand
subgrade. Road Mater. Pavement Des. 2015, 16, 553–568. [CrossRef]
11. Caicedo, B.; Coronado, O.; Fleureau, J.M.; Correia, A.G. Resilient Behaviour of non Standard Unbound Granular Materials. Road
Mater. Pavement Des. 2009, 10, 287–312. [CrossRef]
12. Ng, C.W.W.; Zhou, C.; Yuan, Q.; Xu, J. Resilient modulus of unsaturated subgrade soil: Experimental and theoretical investigations.
Can. Geotech. J. 2013, 50, 223–232. [CrossRef]
13. Gu, F.; Sahin, H.; Luo, X.; Luo, R.; Lytton, R.L. Estimation of Resilient Modulus of Unbound Aggregates Using Performance-
Related Base Course Properties. J. Mater. Civ. Eng. 2015, 27, 04014188. [CrossRef]
14. Nokkaew, K.; Tinjum, J.M.; Likos, W.J.; Edil, T.B. Effect of Matric Suction on Resilient Modulus for Compacted Recycled Base
Course in Postcompaction State. Transp. Res. Record 2014, 2433, 68–78. [CrossRef]
15. Yang, S.R.; Huang, W.H.; Tai, Y.T. Variation of resilient modulus with soil suction for compacted subgrade soils. In Proceedings of
the 84th Annual Meeting of the Transportation-Research-Board, Washington, DC, USA, 9–13 January 2005; pp. 99–106.
16. Rasul, J.M.; Ghataora, G.S.; Burrow, M. The effect of wetting and drying on the performance of stabilized subgrade soils. Transp.
Geotech. 2018, 14, 1–7. [CrossRef]
17. Lu, Z.; Fang, R.; Chen, L.H.; She, J.B.; Xian, S.H. Long-Term Deformation of Highway Subgrade under Coupling Effect of Traffic
Load and Drying-Wetting Cycles. Int. J. Geomech. 2020, 20, 04019168. [CrossRef]
18. Hu, C.M.; Yuan, Y.L.; Mei, Y.; Wang, X.Y.; Liu, Z. Comprehensive strength deterioration model of compacted loess exposed to
drying-wetting cycles. Bull. Eng. Geol. Environ. 2020, 79, 383–398. [CrossRef]
19. Chen, P.; Wei, C.F. Numerical Procedure for Simulating the Two-Phase Flow in Unsaturated Soils with Hydraulic Hysteresis. Int.
J. Geomech. 2016, 16, 04015030. [CrossRef]
20. Yaqi, W.Y.; Abdallah, I.; Nazarian, S.; Puppala, A.J. Moisture Content-Based Longitudinal Cracking Prediction and Evaluation
Model for Low-Volume Roads over Expansive Soils. J. Mater. Civ. Eng. 2015, 27, 04014263. [CrossRef]
21. El-Ashwah, A.S.; Mousa, E.; El-Badawy, S.M.; Abo-Hashema, M.A. Advanced characterization of unbound granular materials for
pavement structural design in Egypt. Int. J. Pavement Eng. 2022, 23, 476–488. [CrossRef]
22. Andrei, D.; Witczak, M.; Schwartz, C.; Uzan, J. Harmonized Resilient Modulus Test Method for Unbound Pavement Materials.
Transp. Res. Rec. J. Transp. Res. Board 2004, 1874, 29–37. [CrossRef]
23. Uzan, J. Characterization of granular material. Transp. Res. Record 1985, 1022, 52–59.
24. NCHRP. Harmonized Test Methods for Laboratory Determination of Resilient Modulus for Flexible Pavement Design; National Cooperative
Highway Research: Washington, DC, USA, 2003.
25. AASHTO. Mechanistic-Empirical Pavement Design Guide: A Manual of Practice; AASHTO: Washington, DC, USA, 2008.
26. Heath, A.C.; Pestana, J.M.; Harvey, J.T.; Bejerano, M.O. Normalizing behaviour of unsaturated granular pavement materials. J.
Geotech. Geoenviron. Eng. 2004, 130, 896–904. [CrossRef]
27. Fredlund, D.G.; Xing, A.Q. Equations for the soil-water characteristic curve. Can. Geotech. J. 1994, 31, 1026. [CrossRef]
28. CJ/T 486-2015; Soil Solidification Admixture Urban Construction Industry Standard (CJ). China Standard Press: Beijing,
China, 2015.
29. Xin, G.; Zhang, A.; Wang, Z.; Shen, Q.; Mu, M. Influence of Humidity State on Dynamic Resilient Modulus of Subgrade Soils:
Considering Repeated Wetting-Drying Cycles. Adv. Mater. Sci. Eng. 2021. [CrossRef]
30. Yao, Y.; Luo, S.; Qian, J.; Li, J.; Xiao, H. Soil-Water Characteristics of the Low Liquid Limit Silt considering Compaction and
Freeze-Thaw Action. Adv. Civ. Eng. 2020, 13. [CrossRef]
31. Yao, Y.; Ni, J.; Li, J. Stress-dependent water retention of granite residual soil and its implications for ground settlement. Comput.
Geotech. 2021, 129, 11. [CrossRef]
32. Zhang, J.; Peng, J.; Zheng, J.; Yao, Y. Characterisation of stress and moisture-dependent resilient behaviour for compacted clays in
South China. Road Mater. Pavement Des. 2020, 21, 262–275. [CrossRef]
33. Han, Z.; Vanapalli, S.K.; Ren, J.; Zou, W. Characterizing cyclic and static moduli and strength of compacted pavement subgrade
soils considering moisture variation. Soils Found. 2018, 58, 1187–1199. [CrossRef]
Polymers 2023, 15, 4187 18 of 18

34. Li, D.Q.; Selig, E.T. Resilient modulus for fine-grained subgrade soils. J. Geotech. Eng.-ASCE 1994, 120, 939–957. [CrossRef]
35. Zaman, M.M.; Zhu, J.; Laguros, J.G. Durability effects on resilient moduli of stabilized aggregate base. Transp. Res. Record 1999,
1687, 29–38. [CrossRef]
36. Ceratti, J.A.; Gehling, W.; Nunez, W.P. Seasonal variations of a subgrade soil resilient modulus in southern brazil. In Proceedings
of the 83rd Annual Meeting of the Transportation-Research-Board, Washington, DC, USA, 11–15 January 2004; pp. 165–173.
37. Allam, M.M.; Sridharan, A. Effect of Wetting and Drying on Shear-Strength. J. Geotech. Eng. Div. 1981, 107, 421–438. [CrossRef]
38. Guney, Y.; Sari, D.; Cetin, M.; Tuncan, M. Impact of cyclic wetting-drying on swelling behavior of lime-stabilized soil. Build.
Environ. 2007, 42, 681–688. [CrossRef]
39. Chu, X. A Review on the Resilient Response of Unsaturated Subgrade Soils. Adv. Civ. Eng. 2020, 2020, 7367484. [CrossRef]
40. JTG D30-2015; Specifications for Design of Highway Subgrades. M.O.T. China: Beijing, China, 2015.
41. JTG D50-2017; Specifications for Design of Highway Asphalt Pavement. M.O.T. China: Beijing, China, 2017.
42. JTG/T D33-2012; Specifications for Drainage Design of Highway. M.O.T. China: Beijing, China, 2012.

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.