NUMERICAL SIMULATION 145

Numerical Model for Dynamic

Compaction of Sand Reclamations
Jia Mao, Xinyuan Cai, Hongqing The dynamic compaction method was used to analyze the consolidation of
Tian, Changchun, China the railway foundation. Based on the physical characteristics of sands, an
extended model was proposed by soil tests. This model is powerful for in-
vestigations of mechanical behaviour of sands under the dynamic compac-
tions. By comparing with the experimental results, it is indicated that the
dynamic compaction was an effective method for understanding the con-
solidation of railway foundations, moreover, the numerical model based
on the infinite element analysis was a powerful tool to find the practical
conditions without boundary reflection under the dynamic compactions.

Among highways and railways of coastal blowing sand reclamation project of coastal solution, so that u → 0 as r → ∞, and u → ∞
region constructions, many sections adopt regions, we built an infinite-finite element as r → 0. The interpolation provides terms of
blowing sand reclamation project in con- coupling model of foundation of blowing order 1/r, 1/r2, and, when the solution vari-
structing the foundation of roads. The key sand reclamation by introducing infinite able is a stress-like variable (such as the
problem of this kind of engineering is how element method as boundaries of three-di- pore liquid pressure in the analysis of flow
to deal with the capacity of blowing sand mensional numerical model. through a porous medium), as 1/r3. The far-
reclamation foundation economically and field behaviour of many common cases,
efficiently. There are many methods for Spatial Infinite such as a point load on a half-space, is
consolidating the foundation of blowing Element Method thereby included. This modeling is achieved
sand reclamation, such as vibro replace- by using standard cubic interpolation for
ment stone pile, dynamic compaction, wa- Infinite elements are used in boundary u(s) in -1 ≤ r ≤ 1, where s is a mapped coor-
ter soil whip pile, no filler vibration impact, value problems defined in unbounded do- dinate. We obtain a three-dimensional
among which the construction process of mains or problems in which the region of model of domains that reaches infinity by
dynamic compaction method is simple, and interest is small in size compared to the combining this interpolation in the s direc-
the effect of strengthening is good. For the surrounding medium and are usually used tion in a product form with standard linear
estimation of main construction parame- in conjunction with finite elements. or quadratic interpolation in orthogonal di-
ters (effective strengthening depth and ra- The static behaviour of the infinite ele- rections in the mapped space.
dius) in foundation treatment with dy- ments is based on modeling the basic solu- Three-dimensional infinite elements only
namic compaction method, Jin [1] built a tion variable, u (in stress analysis u is a dis- map infinite domain along one direction, as
three-dimensional finite element model placement component) with respect to spa- Figure 1 shows, elements along x direction
with LS-DYNA to get the numerical calcula- tial distance r measured from a “pole” of the and y direction are finite, while z direction
tion of single point pounder strike of kinet-
ics process (dynamic compaction method
estimation based on three-dimensional soil
dynamic numerical simulation). Li et al. [2]
obtained an estimation method and a for-
mula of dynamic compaction foundation
settlement in collapsible loess areas by
theory deduction of dynamic compaction
foundation settlement, error analysis,
measured data verification. Khaled [3] de-
veloped two-dimensional and three-dimen-
sional finite element models to study dy-
namic compaction in cohesive soils.
In order to analyze the effect of dynamic
compaction method, which is applied in Figure 1. Infinite element model, a) practical elements, b) parent elements

56 (2014) 2 © Carl Hanser Verlag, München Materials Testing

146 NUMERICAL SIMULATION

is infinite. After using coordinate transfor- N3 = 0.125(1 + ξ)(1 + η)(ζ2 - ζ) The D-P model considers that when ma-
mation, we can map the practical element of N4 = 0.125(1 - ξ)(1 + η)(ζ2 - ζ) terial is in elastic phase (F < 0) or unload-
xyz coordinate in geometric to a spatial cube N5 = 0.25(1 - ξ)(1 - η)(1 - ζ2) ing phase (F = 0, and δF < 0), stress-strain
element with a length of each side of 2. N6 = 0.25(1 + ξ)(1 - η)(1 - ζ2) relation is shown in Equation (7):
N7 = 0.125(1 + ξ)(1 - η)(1 - ζ2)
The conversion relationship between the N8 = 0.25(1 - ξ)(1 + η)(1 - ζ2) (4) σij = KσεKKδij + 2Gδeij (7)
whole coordinates x – y – z and local coor-
dinate is: Constitutive Model If F = 0, and loading (δF < 0), stress-strain
n n n
of Sand-Soil relation is shown in Equation (8):
x = ∑ Mi x i y = ∑ Mi y i z = ∑ Mizi (1) Gsij
i=1 i=1 i=1 For nonmetal particle material such as soil δσ ij = Kσε kkδ ij + 2GδEij − dλ(−3Kα)δ ij +
and rock, we can adopt D-P model which J2
Gsij
with: n: node number, Mi: mapping func- can simulate nonmetal material, it extends
δσ ij = Kσε kkδ ij + 2GδEij − dλ(−3Kα)δ ij + (8)
tion, ui, vi, wi: nodal coordinates, its function based on ideal elastic-plastic J2
model.
where Drucker-Prager ideal elastic-plastic with:
model is one of the earliest constitutive G
M1 = (1 - ξ)(1 - η)(-ζ)/[2(1 - ζ)] models for elastic-plastic which adapt to −3Kaδε kk + Smnemn
J2
M2 = (1 + ξ)(1 - η)(-ζ)/[2(1 - ζ)] geotechnical material, its characteristics dλ =
M3 = (1 + ξ)(1 + η)(-ζ)/[2(1 - ζ)] are that model parameters are few and cal- 9Ka2 + G
M4 = (1 - ξ)(1 + η)(-ζ)/[2(1 - ζ)] culation is simple. It is widely used in engi-
M5 = (1 - ξ)(1 - η)(1 + ζ)/[4(1 - ζ)] neering and is proved to anastomose prac- dij: Kronecker symbol, when i = j, dij = 1,
M6 = (1 + ξ)(1 - η)(1 + ζ)/[4(1 - ζ)] tical well in most cases. Its yield criteria when i ≠ j, dij = 0, sij: stress tensors, eij:
M7 = (1 + ξ)(1 + η)(1 + ζ)/[4(1 - ζ)] expression is shown in Equation (5): strain tensors, K: volume elastic modulus
M8 = (1 - ξ)(1 + η)(1 + ζ)/[4(1 - ζ)] (2) and G: shear elastic modulus.
F(σ ij ) = J2 −aI1 − K = 0 (5)
When practical elements are mapped to be Linear Drucker-Prager Model
parent elements, we can analyze character- with: J2: second invariant of stress devia-
istics of the parent elements, suppose the tion tensors, I1: first invariant of stress de- The linear model is written in terms of all
parent elements use the same shape func- viation tensors, α, k: material constants for three stress invariants. It provides for a pos-
tion as 8-nodes spatial elements, they can Drucker and Prager derived relation be- sibly noncircular yield surface in the devia-
couple with the 8-nodes spatial finite ele- tween α, k and C, j: material constants of toric plane to match different yield values in
ments. Selection of displacement model is Mohr-Coulomb criteria as shown in Equa- triaxial tension and compression, associated
as follows: tion (6): inelastic flow in the deviatoric plane, and
n n n
separate dilation and friction angles.
3 cosϕ sinϕ
u = ∑ N i ui v = ∑ Ni v i w = ∑ Ni w i (3) K= a= (6) Yield Criterion. The linear Drucker-
i=1 i=1 i=1
2
3 + sin ϕ 3 3 + sin2 ϕ Prager criterion is written as:

with: n: node number, Mi: mapping func- Drucker and Prager described in 1952 the F = t - ptanβ - d = 0 (9)
tion, xi, yi, zi: nodal coordinates, yield condition that the yield surface is a
cone in stress space, as can be seen in Figure with:
where 2, on π-plane, its yield curve is a circle which
is inscribed as Mohr-Coulomb yield curve; in ⎡ 3⎤
1 1 ⎛ 1⎞ ⎛ r ⎞
N1 = 0.125(1 - ξ)(1 - η)(ζ2
- ζ) stress space, its yield surface is a cone, t = q ⎢1+ − ⎜ 1− ⎟ ⎜ ⎟ ⎥ (10)
2 ⎢ K ⎝ k⎠ ⎝ q⎠ ⎥
N2 = 0.125(1 + ξ)(1 - η)(ζ2 - ζ) center axis and isocline are coincident. ⎣ ⎦

Figure 2. Generalized von Mises yield surface Figure 3. Typical yield/flow surfaces

56 (2014) 2
 NUMERICAL SIMULATION 147

β(θ, ƒi): slope of the linear yield surface in stiffness matrix will give an acceptable CK17+300) has a blowing sand reclamation
the p-t stress plane and is commonly re- rate of convergence and the non-symmetric project of littoral area, set the total length of dy-
ferred to as the friction angle of the material, matrix scheme may not be needed. namic compaction area as 150 m, section from
d: cohesion of the material, K(θ, ƒi): ratio of DK15+315.64 to DK15+515.64 was regarded
the yield stress in triaxial tension to the yield Loading Form of as the test section. According to the depth
stress in triaxial compression and, thus, con- Dynamic Compaction range that standard penetration test reaches,
trols the dependence of the yield surface yield geotechnical were mainly divided into 3
on the value of the intermediate prin- According to predecessors’ bygone theory layers, from ground to each layer are as follows:
­cipal stress (Figure 3), θ: temperature and work and test demonstration, in the pro- 1) filling sand 2) fine sand, and 3) mud.
ƒi(i = 1, 2, ...): other predefined field variables. cess of impact of pounder and collision to
The cohesion d of the material is related foundation, contact stress has only one sig- Numerical Analysis
to the input data as nificant peak value on time-history curve,
and its duration is very short, it is around Analysis Model. Using ABAQUS finite ele-
⎛ ⎛ 11 ⎞⎞ ⎛ ⎛11 11 ⎞⎞ 33 ⎛ ⎛ 11⎞ ⎞
d d==⎜ 1−
⎜ 1− tanβ
tanβ σ σ = =
⎟⎟ c c ⎜⎜ + + tanβ
tanβ σσ =
⎟⎟ t t = τ ⎜ ⎜ 0.1
τ1+1+ ⎟ ⎟ s. Impact load is simplified to be the ment software, aiming at foundation of
⎝ ⎝ 33 ⎠⎠ ⎝ ⎝KK 33 ⎠⎠ 22 ⎝ ⎝ Kload
K⎠ ⎠ of triangle form, see Figure 5. t , t , blowing sand reclamation which uses dy-
n r
Pmax in the figure, measured on test fields, namic compaction method, we designed
1 ⎞ 3 ⎛ 1 ⎞
+ tanβ ⎟ σ t = τ ⎜ 1+ ⎟ (11) can also be estimated by the following for- two three-dimensional entity models ac-
3 ⎠ 2 ⎝ K⎠ mulas: cording to finite element model and infinite
element-finite element coupling model,
with: σc: uniaxial compression yield stress, V ms
V ms ⎛⎛ 11 11⎞⎞ aiming to realize simulation of dynamic
PPmax
max
== 00 22 ttnn == ππ m
m // ss ttrr == ⎜⎜  ⎟⎟ ttnn
σt: uniaxial tension yield stress and πr
πr ⎝⎝ 44 22⎠⎠ compaction process in test sections.
t: shear stress. Factors that influence foundation of
V ms ⎛ 1 1⎞
Plastic Flow. G isPthe flow
= 0 potential,
t n = cho-
π m / s tr = ⎜  ⎟ tn (13) blowing sand reclamation which uses dy-
max
sen in this model as πr 2
⎝ 4 2⎠ namic compaction method are many, and
also characteristics of soil body of deforma-
G = t-ptanψ1 (12) with: V0: landing speed of pounder, m: tion are very complicated, so the following
pounder quality, r: pounder radius and assumption was introduced:
with: ψ1(q, ƒi): dilation angle in the p-t s: elastic constant. 1. Soil body in the model can be regarded
plane. as homogeneous, isotropic, elastic-plas-
A geometric interpretation of ψ is shown S = 2rE(1 - μ2) (14) tic infinite spatial
in the p-t diagram of Figure 4. In the case of 2. No considering effect of groundwater
hardening defined in uniaxial compres- For the deformation modulus of soil is ad- 3. Simplify the pounder to be a force,
sion, this flow rule definition precludes di- justing continuously, contact-stress and which is applied on soil body.
lation angles tanψ > 3. This restriction is contact-time are in continuous changes, we 4. Force and tangential force of interface of
not seen as a limitation since it is unlikely, consider this point fully in the simulation, blowing sand reclamation can be ig-
this will be the case for real materials. adjusting contact stress and contact time nored.
Non-Associated Flow. Non-associated in each analysis step based on deformation Finite element model (model 1): top surface
flow implies that the material stiffness ma- modulus, every pounder strike is defined width of roadbed finite element model is
trix is not symmetric; therefore, the non- as one analysis step, each analysis step 20 m, width of bottom surface is 40 m,
symmetric matrix storage and solution lasts 0.3 s, total analysis time is 2.1 s. height is 10 m, length is 10 m, and grid
scheme should be used. If the difference size used is 1 m. Tamper weight is 150 kN,
between β and ψ is not large and the region Project Application tamping energy is 2600 kN × m.
of the model in which inelastic deformation Infinite-finite element coupling model
is occurring is confined, it is possible that a One section of a railway branch from Da- (model 2): based on finite element model,
symmetric approximation to the material LanPing to BaoShuiGang (CK12+000 ~ boundary elements of four weeks and bot-

Figure 4. Yield surface and flow direction Figure 5. Strong pounder dynamic effect model

56 (2014) 2
148 NUMERICAL SIMULATION

cess of former 7 strikes, surface settle-

ment data of dynamic compaction were
summarized in Table 2, and dynamic
compaction settlements were plotted as
shown in Figure 8.
In the numerical analysis of consolidat-
ing the capacity of foundation of blowing
sand reclamation, settlement amount of
model 1 which adopts finite element
method is smaller than of model 2 which
adopts infinite element-finite element
method and put infinite element as bound-
ary condition. Values of model 2 are close
to measured values of the test field, this
shows that it can simulate the practical
Figure 7. Coupling method of infinite and boundary conditions well by introducing
Figure 6. Finite element method finite elements infinite elements to the analysis model.
In the process of dynamic compaction,
when the times of rammer strike is near 7,
settlement tends to be stable, measured
value is 3 cm and reaches the settlement
requirement of dynamic compaction.

Figure 8. The relation- Conclusions

ship between roughness
coefficient JRC Based on the analysis on the capacity of
and ascent angle β foundation of blowing sand reclamation by
of structural planes
dynamic compaction method, and the com-
parison with the measured values, we can
draw the following conclusions:
1. It can reflect the practical condition well
if infinite elements as boundaries are
introduced.
2. Boundary conditions have a big influ-
Sand layer E/MPa μ β(°) K ψ(°) εp σ/kPa ence on the calculation results of finite
element method.
0~1 m 8 0.3 58.5 0.778 0 0 100
3. Compared with the measured values,
1~2 m 14.5 0.3 58.5 0.778 0 0 140 values of model 2 which put infinite ele-
2~5 m 19 0.3 58.5 0.778 0 0 180 ment as boundary condition are close to
5~6 m 8 0.3 58.5 0.778 0 0 100 the measured values.
4. If the times of rammer strike is near 7,
6~8 m 14.5 0.3 58.5 0.778 0 0 140
the settlement tends to be stable and
8-12 m 18 0.3 58.5 0.778 0 0 180 reaches the settlement requirement of
Table 1. Physical and mechanical parameters of filling sand layer dynamic compaction.

References

1 J. Li, W. F. Teng: Estimation of construction

parameters of dynamic compaction method
tom can be replaced by spatial infinite ele-
Times Test Model 1 Model 2 based on 3D soil dynamic numerical simula-
ments, if infinite element belongs to bound- tion. Yangtze River 42 (2011), No. 15,
1 35.7 36.0 40.0
ary element, there is no need to set bound- pp. 50-52
2 19.0 21.0 25.0 ary conditions, as shown in Figures 6 and 7. 2. W. Li, Q. Gu, L. Su, B. Yang: Finite element
3 16.7 12.0 14.0 Material Model Parameters. According analysis of dynamic compaction in soft foun-
to the test, physical and mechanical param- dation. Procedia Engineering 12 (2011),
4 12.3 10.0 12.0
eters of filling sand layer from top to down pp. 224-228
5 8.7 5.0 7.0 3 K. Mostafa: Numerical modeling of dynamic
are illustrated in Table 1.
6 6.0 3.0 5.5 compaction in cohesive soils. A Dissertation
The simulation and analysis of the dy-
Presented to The Graduate Faculty of The Uni-
7 3.0 1.0 2.0 namic compaction process of the test sec- versity of Akron (2010), pp. 2-3
total 101.4 88.0 105.5
tions by two models, and comparison of 4 J. P. Borg, J. R. Cogar, A. Lloyd, A. Ward,
analysis results with measured values of D. Chapman, K. Tsembelis: Computational
Table 2. Settlement of dynamic compaction [cm] field dynamic penetration test, in the pro- simulations of the dynamic compaction of

56 (2014) 2
 NUMERICAL SIMULATION 149

porous media. International Journal of Abstract

Impact Engineering 33 (2006), pp. 109-118
5 L. Chen: Application and Research of Ground
Treatment of Encircling the Sea to Land Ein numerisches Modell für die dynamische Kompaktierung von Sand-
Project. Special Structure 27 (2010), No. 2, aufarbeitung. Das dynamische Kompaktierungsverfahren wurde ange-
pp. 1-5
6 D. C. Drucker, W. Prager: Soil mechanics and
wendet, um die Verdichtung der Eisenbahnfundamente zu analysieren.
plastic analysis for limit design. Quarterly of Ein erweitertes Modell basierend auf den physikalischen Sandeigenschaf-
Applied Mathematics 10 (1952), No. 2, ten sowie auf Bodentests wird in diesem Beitrag aufgestellt. Gerade für
pp. 157-165
Untersuchungen des mechanischen Verhaltens von Sand bei dynamischer
Bibliography Kompaktierung ist dieses Modell sehr aussagekräftig. Der Vergleich mit
DOI 10.3139/120.110537
den experimentellen Ergebnissen zeigt, dass dynamische Kompaktierung
Materials Testing eine effektive Methode darstellt, um den Verdichtungsprozess von Eisen-
56 (2014) 2, page 145-149
bahnfundamenten zu verstehen. Außerdem eignete sich das auf Infinite
© Carl Hanser Verlag GmbH & Co. KG
ISSN 0025-5300 Elemente Berechnungen basierende Modell hervorragend zur Auffindung
der geeigneten Realbedingungen ohne Randreflexion bei dynamischen
The Authors of This Contribution
Kompaktierungen.
Jia Mao, born in 1973, PhD, associate professor at
College of Transportation, Jilin University, Chang-
chun, China. The research interest of Jia Mao is
computer simulation of traffic and transportation
engineering.
Xinyuan Cai, born in 1989, graduate student
at College of Transportation, Jilin University,
Changchun, China. The research interest of
Xinyuan Cai is also computer simulation of traffic
and transportation engineering.
Hongqing Tian, born in 1971, master degree,
senior engineer in Navy Logistic Equipment and
Technology Institute, Beijing, China. The research
You will find the article and additional material by entering the document number MP110537
interest of Hongqing Tian is mechanical and elec-
on our website at www.materialstesting.de
tronic behaviour of traffic and transportation
engineering.

56 (2014) 2