Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

Available online at www.sciencedirect.

com

ScienceDirect
Advances in Space Research xxx (xxxx) xxx
www.elsevier.com/locate/asr

DEM-FEM-MBD coupling analysis of landing process of lunar


lander considering landing mode and buffering mechanism
Shunying Ji ⇑, Shaomin Liang
State Key Laboratory of Structural Analysis for Industrial Equipment, Dalian University of Technology, Dalian 116024, China

Received 6 December 2020; received in revised form 2 March 2021; accepted 24 March 2021

Abstract

The safe lander dynamics is an important part of the lunar landing mission. In this paper, a discrete element method (DEM) for lunar
soil is set up, and the finite element method (FEM) for the lander is set up by shell elements and beam elements. The lander is regarded as
a multibody system composed of a cabin, legs and footpads, and its motion characteristics are solved by multibody dynamics (MBD). A
DEM-FEM-MBD coupling algorithm is developed to simulate the landing process of the lander considering landing mode and buffer
mechanism, whose correctness is verified by comparison with a full-scale experiment involving lunar lander on earth. The effects of
the mass, landing velocity and attitude of the lander on the safe landing are discussed. The buffering mechanism and influencing factors
of lunar soil are analyzed. The results show that the impact force peak and impact depth gradually increase with the increase in the land-
ing velocity and mass of the lander. Two kinds of inclined landing modes are defined and compared with vertical landing. It is found that
the force on the landing leg that first contacts the lunar soil is significantly greater than that on the landing leg, that contacts later. The
impact force peak on the lander under the two inclined landing modes is similar, but the impact depth of the 1-2-1 mode is significantly
greater than that of the 2-2 mode. In the process of landing, lunar soil has the function of buffering dissipation. The energy dissipation
rate is affected by the physical characteristics of lunar soil and the mechanical energy of the lander.
Ó 2021 Published by Elsevier B.V. on behalf of COSPAR.

Keywords: DEM-FEM-MBD coupling algorithm; Lunar lander; Safe soft landing; Lunar soil energy absorption ratio; Buffering dissipation mechanism

1. Introduction reduce the impact load on the lander and protect the
onboard equipment and astronauts, there are two main
The moon is the nearest celestial body to the earth and landing buffer technologies currently studied, one of which
the starting point of human space exploration (Wang et al, is the instantaneous reduction in the vertical descent veloc-
2019). In lunar exploration missions, the safe and stable ity (Kwon et al, 2016; Ma et al, 2018), which includes the
lander dynamics on the lunar surface not only affects the use of gliders, landing buffer rockets (Zhu et al, 2017)
safety of the equipment and personnel on board but is also and retractable brakes. Another approach is to dissipate
the key to lunar exploration projects and the premise for energy over a limited distance, including the use of materi-
the smooth implementation of future work (Li et al, als with elastic deformation or inelastic deformation or
2016; Zhu et al, 2015). The lander is subjected to a high structures to absorb energy (Moghadam et al, 2019), such
impact load during the process of landing, which is affected as cushion airbags and aluminum honeycomb (Wang
by many factors (Liu et al, 2008; Zheng et al, 2018). To et al, 2019). Aluminum honeycomb buffering has the char-
acteristics of a simple structure, high reliability and small
⇑ Corresponding author.
influence by temperature, which is widely used in the early
E-mail address: jisy@dlut.edu.cn (S. Ji). stages of lunar exploration projects.

https://doi.org/10.1016/j.asr.2021.03.034
0273-1177/Ó 2021 Published by Elsevier B.V. on behalf of COSPAR.

Please cite this article as: S. Ji and S. Liang, DEM-FEM-MBD coupling analysis of landing process of lunar lander considering landing mode and
buffering mechanism, Advances in Space Research, https://doi.org/10.1016/j.asr.2021.03.034
S. Ji, S. Liang Advances in Space Research xxx (xxxx) xxx

In addition, due to the small range of the lunar surface The DEM-FEM-MBD coupling method is effective in
observed before landing (Arslantas and Theil, 2018; Gong simulating the lander dynamics although the research on
et al, 2007) and the fact that the landing site on the lunar this method is scarce at present. However, the DEM-
surface cannot be accurately determined (Seo et al, 2016), FEM coupling algorithm has been widely used in many
the lunar lander may land on a complex lunar surface with engineering problems (Recuero et al, 2016; Recuero et al,
slopes or covered with rocks and craters (Wang et al, 2019; 2017; Tu et al, 2017; Zeng et al, 2020; Zheng et al, 2017).
Zheng et al, 2018). Thus, it is necessary to assess the lan- Since the 1990s, the DEM-FEM coupling algorithm has
der’s safety and stability regarding complex landing sur- been developed and improved gradually and has irreplace-
faces under a random initial position and attitude before able advantages in working with multiscale problems of the
landing (Zheng et al, 2018). coupling between granular materials and engineering struc-
Lunar soil is similar to the soil on the earth’s surface in tures (Wang et al, 2013; Wang and Ji, 2018; Yang et al,
that it exists in a granular state with discontinuities 2020). This method gives full play to the characteristics
(Katagiri et al, 2015; Metzger et al, 2018). Lunar soil plays of the DEM and FEM and has obvious advantages in solv-
an important buffer role during the landing process (Li ing multiscale interface coupling problems. The application
et al, 2009). At present, lunar soil is generally regarded as of a planetary wheel and soil (Knuth et al, 2012; Michael
a continuous medium to analyze the deformation and other et al, 2015; Nakashima et al, 2010; Nakashima et al,
characteristics of lunar landing sites overall, which is unfa- 2007; Smith and Peng, 2013) or ball mill and screw con-
vorable for studying the discrete characteristics of lunar veyor (Shi and Xie, 2016) can effectively verify the feasibil-
soil, such as buffer energy absorption properties (Mishra ity of the DEM-FEM coupling method (Okubo et al,
and Prasad, 2018). In the 1970s Cundall and Strack pro- 2020). The FEM-MBD coupling algorithm has made out-
posed the discrete element method (DEM) (Cundall and standing contributions in the field of structural fatigue
Strack, 1979), which developed into an important tool strength calculations (Olivier et al, 2020; Sanborn et al,
for simulating granular materials. This method has the 2014; Sutoh et al, 2018). This method has significant
characteristics of simple calculations and high efficiency. advantages in studying the stress distribution and dynamic
The DEM can set up a numerical model of a discontinuous characteristics of structures under high load (El-Ghandour
medium according to the discrete characteristics, which can et al, 2016; Sanborn et al, 2014). The application of fatigue
be used to simulate changes in the mechanical parameters analysis to locomotive bodies, strength analysis of the
such as the velocity field, displacement field and force field gasoline engine connecting rod and coupled vibration of
during the motion of a granular material or system the vehicle bridge can be effective, and the feasibility of this
(Nakashima et al, 2011). The DEM regards the granular method has been confirmed (El-Ghandour and Foster,
material as a collection of individual particles and thus 2019). Based on this, the above two methods are developed
effectively overcomes the macro continuity assumption of and coupled, and a DEM-FEM-MBD coupling algorithm
the traditional continuum mechanical model and can con- is proposed to analyze the interaction between the lander
duct numerical simulations of the mechanical characteris- and the lunar soil.
tics of granular material from a micro perspective The software Altair/EDEM is mature to deal with
(Kulchitsky et al, 2016). Therefore, the DEM can be used motion properties of granular materials. The software
to set up a lunar soil model (Gao et al, 2017; Jiang et al, MSC/ADAMS has also been widely used to simulate the
2013; Li et al, 2010; Liu et al, 2020; Otto et al, 2018). dynamic characteristics of structures. It can be realized
The structural model of the lander can be set up by the by builing a coupling channel to transfer the calculation
finite element method (FEM) (Ma, 2018; Zheng et al, information when working with the interaction between
2018). The FEM has been fully developed and widely used the discrete element and the structure (Wang et al, 2018).
in engineering fields, and has achieved excellent results. However, the method is not mature and is unsuitable for
This method has a strong foundation and development in working with the problem of a large number of particles,
modeling, static analysis and dynamic analysis. In addi- because the transmission of information greatly reduces
tion, to analyze the buffer characteristics of the lander dur- the calculation efficiency, or even prevents calculation.
ing landing, it is generally necessary to set up its own buffer For the problem of asynchronous time steps, the calcula-
device, which is realized by the theory of multibody tion results may be inaccurate (Lommen et al, 2018).
dynamics (MBD) (Chen and Nie, 2008; Wang et al, In this paper, a DEM model of lunar soil and an FEM-
2005). MBD can be used to study systems consisting of MBD coupling algorithm of the lander are developed with
multiple bodies, each of which has mass, inertia and CUDA C++, and a DEM-FEM-MBD coupling algorithm
degrees of freedom. These bodies are connected by joints, is built. The dynamic characteristics of the lander and the
cables, contacts, or other kinematic or force constraints. motion characteristics of lunar soil are analyzed in detail.
Overall, the MBD method has proved to be a useful tool Then, the main factors affecting the lander dynamics are
for the motion analysis of multibody systems (Lommen analyzed, including the landing velocity, mass of the lander
et al, 2018). Therefore, an FEM-MBD coupling model and landing attitude. Finally, the energy dissipation mech-
can be set up to analyze the dynamic characteristics of anism of lunar soil is analyzed.
the landing process of the lander.
2
S. Ji, S. Liang Advances in Space Research xxx (xxxx) xxx

2. DEM-FEM-MBD coupling algorithm of the lunar lander Under the impact load, lunar soil particles may be bro-
ken, which can be modeled by setting a parallel bonding
Considering the complexity of DEM-FEM-MBD cou- model between spherical elements and establishing certain
pling algorithm, the modeling process is divided into three crushing criteria (Potyondy and Cundall, 2004). In the par-
parts: set up the discrete element model of lunar soil, build- allel bonding model, two spherical particles are bonded by
ing the finite element model of lander, and determining the setting virtual parallel keys, as shown in Fig. 1. Compared
connection mode of multibody structure of lander. The with the traditional point bonding model, the parallel bond
three parts will be described in detail. Finally, the imple- model can transfer not only forces but also moments. Here,
mentation of the coupling algorithm based on parameter an elastic bonded disk is used to simulate the parallel bond
transfer is introduced. of the particle element. The radius of the disk is the mean
of the radii of the two bonded particles. The forces and
2.1. DEM model of the lunar soil moments between the bonded elements can be decomposed
into normal and tangential components as:
According to the discrete characteristics of lunar soil, a
F b ¼ F nb þ F sb ð1aÞ
simulation model of lunar soil can be set up by using dis-
crete elements. Although the construction technology of Mb ¼ M nb þ M sb ð1bÞ
arbitrary shape particles is developing rapidly, the content
where F b andM b are the force and moment on the bonded
of this method is not mature. The calculation process of the
area, respectively, and F nb ; F sb ; M nb and M sb are the normal
contact force between particles with arbitrary shape is very
complex and is not conducive to large-scale calculation (Ji and tangential components of the bond zone force and
et al, 2019; Zeng et al, 2015). To improve the simulation moment.
efficiency, randomly arranged spherical elements can be When the particle element is bonded, the force and tor-
used to simulate lunar soil (Li and Ji, 2018), as shown in que increment will be generated in the bonding area due to
Fig. 1. The physical properties of lunar soil are affected the change of element position and rotation angle in each
by many factors, such as particle size, porosity, cohesion, calculation step, which will affect the bonding state
internal friction angle, bulk density, particle specific gravity between particles. The increment of force and moment
and particle morphology (Wang et al, 2020). In this paper, can be calculated by the following formula:

the force between lunar soil particles can be calculated by DF nb ¼ ðk n A DU n Þnij ð2aÞ
the Hertz-Mindlin nonlinear contact model based on

spherical elements. The contact force between particles DF sb ¼ ðk s A DU s Þsij ð2bÞ
mainly includes the elastic force, viscous force and sliding 
friction force based on the Coulomb friction criterion DM nb ¼ ðk n J Dbn Þnij ð2cÞ
(Ramırez et al, 1999), which is consistent with the physical 
process of particle collision (Bui et al, 2009). The existence DM sb ¼ ðk s I Dbs Þsij ð2dÞ
of the electrostatic force between lunar soil particles has an
where DF nb , DF sb , DM nb and DM sb are the normal and
obvious effect on static lunar soil. Accordingly, the electro-
tangential force and torque increments of the bonding
static force is not considered when studying the dynamic
zone, respectively; DU n ¼ V ij  nij Dt and DU s ¼ V ij  sij Dt
response of lunar soil under an impact load (Pei et al,
2016; Yang et al, 2017). are the relative displacement in the normal and tangential 
directions of two particles, respectively, Dbn ¼ xj  xi 
 
nij Dt, and Dbs ¼ xj  xi  sij Dt are the normal and
tangential rotation angles of the two particles respectively;
V ij is the relative velocity between the two particles.
i j
k n ¼ kknni þk
kn
n
j is the normal stiffness coefficient of spherical

elements. The normal stiffness coefficient of parallel bond-



RR
ing model is k n ¼ pE Ri þR
i j
j
. k s ¼ k n =2 is the tangential stiff-
ness coefficient of spherical elements. The tangential
 
stiffness coefficient of parallel bonding model is k s ¼ k1n ;
1 is the coefficient associated with m ; m is the Poisson’s ratio
of granular materials, A is the area of the bonded disk, J is
the polar moment of inertia of the bonded disk, and I is the
moment of inertia of the bonded disk, which can be calcu-
lated by the following equation:
1 1
Fig. 1. Discrete element model of lunar soil and parallel bonding model A ¼ pR2 J ¼ pR4 I ¼ pR4 ð3Þ
between spherical elements. 2 4
3
S. Ji, S. Liang Advances in Space Research xxx (xxxx) xxx

where R is the radius of the bonded disk. Based on the computational parameters are listed in Table 1. The elastic
beam theoretical model, the failure criterion of the disk is modulus of lunar soil collected from the moon is
established, in which the maximum tensile stress rmax and 48.863 MPa, and that of artificial lunar soil is generally
maximum shear stress smax act on the disk and can be 11.1 ~ 80.0 MPa (Alshibli and Hasan, 2009). Therefore,
expressed as: the elastic simulation selected in this paper is 48.8 MPa.
F nb jM sb j
rmax ¼ þ R ð4Þ
A I 2.2. FEM model of the lunar lander
F sb jM nb j
smax ¼ þ R ð5Þ
A J Referring to the Chang’e-4 lander (Fig. 2(a)), a struc-
tural model of the lander is set up, which consists of a cabin
When the maximum tensile stress rmax and maximum shear
and 4 sets of landing gear. Each set of landing gear includes
stress smax of the bonded disk between particles reach the
a primary landing leg, two secondary landing legs, and a
failure strength, the bonding between particles fails, which
footpad, as shown in Fig. 2(b). The landing leg is filled with
leads to particle breakage on the macro scale.
aluminum honeycomb. The footpad and cabin are com-
Generally, the landing time of a lander is short, and the
posed of shell elements of a certain thickness. The primary
impact area of the shock wave generated by it is limited.
struts are composed of beam elements.
Accordingly, to improve the calculation efficiency, a finite
The Crank-Nicolson method is used to solve the
simulation domain of lunar soil particles can be estab-
dynamic equation of the structure:
lished, and the non-reflection boundary conditions around
and at the bottom are used to simulate the infinite space
M€ut þ C u_ t þ Kut ¼ F t ð7Þ
domain. The dynamic compaction method provides a sim-
ulation method for the propagation distance of the shock
where M; C and K are the mass, damping and stiffness
wave in granular media (Closs and Cook, 1987):
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi matrices of the structure, respectively, and F t is the exter-
D ¼ a ðMH =1000Þ ð6Þ _ €
nal load on time t. ut ; ut and ut are the displacement,
where D is the propagation distance of the shock wave in velocity and acceleration matrices of the structure at the
_ €
the lunar base during the impact of the lander, whose unit instant of time t, respectively. ut and ut can be obtained
is m; ais the factor determined by the physical properties of by the central difference method:
the lunar soil, and its value range is generally 0.42–0.8; M is
(
the mass of the lander structure, whose unit is kg; and H is u_ t ¼ gDt
1
ðut  utDt Þ  1g u_ tDt
g
the equivalent free-falling height of the lander, whose unit ð8Þ
is m, which can be calculated by the landing velocity. To €ut ¼ gDt
1
ðu_ t  u_ tDt Þ  1g
g
€utDt
satisfy the simulation domain and ensure the calculation
efficiency, the particle size of the lunar soil discrete element where Dt is the time step and utDt , u_ tDt and €utDt are the
model is larger than that of the actual lunar soil particle. In displacement, velocity and acceleration of the structure,
the study, the numerical model of lunar soil is used to respectively, at time t  Dt. g is a computational parameter
replace the real lunar weathering layer, and the interaction that ensures the stability of the calculation results
model between lunar soil model and structure is realized by (g ¼ 0:5).
certain algorithm. Therefore, the difference of particle size To obtain the stiffness matrix of the structure, the flat
can be tolerated, which can ensure the correctness of the shell element of the lander structure can be considered a
calculation results by adjusting the calculation parameters. combination of planar membrane elements and plate bend-
At present, experimental studies of lunar soil usually ing elements, as shown in Fig. 3. For isotropic shell ele-
adopt volcanic rock as the simulated lunar soil. In this ments, the deformation of the planar membrane elements
paper, the parameters of the simulated lunar soil are used and the plate bending element is relatively independent,
for the numerical simulation (Hou et al, 2018), and the so the solution of the original element can converge
through the fine division of the elements. The global stiff-
ness matrix is singular as the elements are coplanar or
Table 1
Computational parameters of lunar soil in the DEM simulation. nearly coplanar. A generalized coordinated triangular flat
shell element composed of the triangular membrane ele-
Parameter Symbol Value
ment GT9 and triangular thin plate element TMT with
Particle size R 45 cm
an additional rigid body rotation angle can be used to elim-
Number of particles n 190,200
Maximum tensile stress of the disk rmax 1.2 MPa inate the singularity (Li, 2015). The TMT element is univer-
Maximum shear stress of the disk smax 0.3 MPa sal for triangular thin and thick plates and is suitable for
Elastic modulus of lunar soil E 48.8 MPa small deformation problems. The element stiffness matrix
Friction coefficient between lunar soil l 0.44 can be composed of the stiffness matrix of the two elements
Resilience coefficient of lunar soil e 0.35
as:
Density of lunar soil q 1370 kg=m3

4
S. Ji, S. Liang Advances in Space Research xxx (xxxx) xxx

(a) Chang'e-4 lander (b) Finite element model of the lander

Fig. 2. Chang’e-4 lander and its finite element model for the simulation.

(a) Triangle planar membrane element (b) Triangle plate bending element

Fig. 3. Triangular shell element for constructing the lander structure.

2 3
k m;11 k m;12 0 0 0 k m;13 comb materials can be divided into two types: out-of-plane
6k k m;22 0 k m;23 7 compression and in-plane compression. Because the aver-
6 m;21 0 0 7
6 7 age stress of out-of-plane compression is large, it is often
6 0 0 k b;11 k b;12 k b;13 0 7
Ke ¼ 6
6
7
7 ð9Þ used in the design of the buffer. The secondary buffer device
6 0 0 k b;21 k b;22 k b;23 0 7 uses two kinds of honeycomb material to cushion the
6 7
4 0 0 k b;31 k b;32 k b;33 0 5 impact force in series. Here, a prismatic joint is used to
k m;31 k m;32 0 0 0 k m;33 connect the outside cylinder and the inside cylinder of the
landing leg to realize the simulation of its buffer device.
where K e is the stiffness matrix of the shell element, k m;ij is Fig. 5(b) and 5(c) show the simple diagrams of the relation-
the stiffness coefficient of the planar membrane element, ship between the load and the energy absorption and the
and k b;ij is the stiffness coefficient of the plate bending compression of aluminum honeycomb material.
element. When characterizing the relative motion between the
parts of the lander and the motion state of the whole struc-
ture, it is necessary to set up the global coordinate system
2.3. MBD model of the lunar lander for the overall motion and the local coordinate system
for the motion of each part and to establish the transfor-
To realize the buffer device of the landing leg and ensure mation relationships between each local coordinate system
structural coordination during the landing, a multibody and the global coordinate system. Taking any joint and its
structure system of the lander is built. The lander is divided two connecting bodies as an example, the coordinate sys-
into a cabin, landing leg and footpad. The cabin and land- tems of the multibody structure are plotted in Fig. 6.
ing leg are connected by a revolute joint, and the landing The local coordinates of the inside connecting body,
leg and footpad are connected by a spherical joint. The outside connecting body, inside joint and outside joint
topology and connection mode between the three parts are established. The body-fixed bases of the inside connect-
are indicated in Fig. 4, where B0 represents the cabin, B1 ,  
ing body and outside connecting body are e11 ; e12 ; e13 and
B4 , B7 and B10 represent the cylinder, that upper strut fixed  2 2 2
e ; e ; e , respectively. The inside joint coordinate is
to the lander main body, B2 , B5 , B8 and B11 represent the  11 21 31   
cylinder, that lower moveable part of the lander. B3 , B6 , h1 ; h2 ; h3 , and the outside joint coordinate is h21 ; h22 ; h23 .
B9 and B12 represent the footpads. The conversion between the outside joint and outside
The landing leg is filled with a honeycomb structure to connecting body, the inside connecting body and inside
realize the buffering function, and a secondary buffer device joint, and the inside joint and outside joint can be
is designed, as plotted in Fig. 5. The deformation of honey- expressed as:
5
S. Ji, S. Liang Advances in Space Research xxx (xxxx) xxx

(a) Topology (b) Connection modes

Fig. 4. Components and connection modes of the lander multibody system.

(a) The secondary buffer device of the landing leg

(b) Simplified collapse diagram of secondary aluminum (c) Energy diagram of crushing process
honeycomb material

Fig. 5. Schematic diagram of the secondary buffer device of the landing leg.

2 3 2 3 2 3 2 3 2 3 2 3 2 3 2 13
e21 h21 h11 e11 h21 h11 e21 e1
6 27 T 6 7 6 7 6 7 6 7 6 7 6 27 6 17
4 e2 5 ¼ H 22 4 h22 5; 4 h12 5 ¼ H 11 4 e12 5; 4 h22 5 ¼ H 21 4 h12 5 4 e2 5 ¼ H 22 H 21 H 11 4 e2 5
T
ð11Þ
e23 h23 h13 e13 h23 h13 e23 e13
ð10Þ
A2;1 ¼ H T22 H 21 H 11 , so,
where H T22 ; H 11 and H 21 are transformation matrices, 2 23 2 13
e1 e1
which can be obtained by the Euler angle or quaternions. 6 27 6 17
4 e2 5 ¼ A2;1 4 e2 5 ð12Þ
The conversion relationship between the inside connecting
body and the outside connecting body can be expressed as: e23 e13
6
S. Ji, S. Liang Advances in Space Research xxx (xxxx) xxx

Fig. 7. Translational transfer of the mass centers of contiguous bodies.

Fig. 6. Local coordinates of the inside connecting body, outside connect-


ing body, inside joint and outside joint and the fixed global coordinate respectively, and Q and P are the joint points of the two
system and transformations between each part. bodies.
Hence, the displacement of the centroid of the external
body can be expressed as:
Hence, when the multibody system has multiple struc-
tures, the transformation relationship between the body- ri ¼ rLðiÞ þ riq þ riqp  rip ð16Þ
fixed basis vector of the first structure and the global coor-
where riqp is the relative displacement between the two joint
dinate base vector ðg 1 ; g2 ; g3 Þ can be expressed as:
2 i3 2 3 points. The rotation between contiguous bodies can be
e1 g1 obtained by:
6 i7 6 7
4 e2 5 ¼¼ Ai;i1 Ai1;i2    A1;0 4 g2 5 ð13Þ xi ¼ xri þ xLðiÞ ð17Þ
i
e3 g3
According to the superposition principle, the rotation of
The quaternion is effective for the transformation rela- body i caused by the rotation of the first inside connecting
tionship between coordinates. The transformation matrix body can be calculated from:
X
H can be obtained by quaternions Qðq0 ; q1 ; q2 ; q3 Þ and xi ¼ xrk þ x0 ð18Þ
expressed as: k2i
2 2 3
q0 þ q21  q22  q23 2ð q1 q2 þ q0 q 3 Þ 2ð q1 q3  q0 q2 Þ where xrk is the angular velocity of body k relative to its
6 7inside connecting body LðkÞ, x . is the angular velocity
H ¼ 4 2ð q1 q2  q0 q3 Þ q0  q0 þ q0  q0
2 2 2 2
2ð q2 q3 þ q0 q1 Þ 5 0
2 of body 0.
2ð q1 q3 þ q0 q2 Þ 2ð q2 q3  q0 q 1 Þ q20  q21  q22 þ q3
ð14Þ
2.4. DEM-FEM-MBD coupling algorithm of lunar lander
The updated quaternions can be expressed as:
At each time step of the DEM-FEM-MBD coupling
2 3nþ1 2 3n 2 3n 2 3n
q0 q0 q0 q1 q2 q3 0 algorithm, the contact detection and node equivalent force
6 7
q1 7 6 7 6 7 6x 7 calculation are carried out by the DEM and FEM-MBD
6 6 q1 7 Dt 6 q1 q0 q3 q2 7 6 17
6 7 ¼6 7 þ 6 7 6 7 coupling sections, which are transferred to FEM-MBD
4 q2 5 4 q2 5 2 4 q2 q3 q0 q1 5 4 x2 5
for the stress computation and the dynamic calculation.
q3 q3 q3 q2 q1 q0 x3 Due to the relationship between the multibody structures,
ð15Þ the movement is restricted in some directions, which
reflects the motion characteristics of the multibody struc-
whereðx1 ; x2 ; x3 Þ denote the components of relative rota-
  ture. The displacement and strain update of the structure
tional velocity in the three directions of ei1 ; ei2 ; ei3 . More- are transformed into the displacement of the coupling
over, the introduction of quaternion update is relatively interface as the displacement boundary condition of the
simply; a more detailed description of the quaternion discrete element. The program flowchart of the DEM-
update can be found in the work of Fritzer (Fritzer, 2001). FEM-MBD coupling algorithm is shown in Fig. 8. The
Since the motion of the contiguous body drives the information transfer on the coupling section is the key
motion of the body itself, the motion of the contiguous problem of the coupling algorithm.
body is divided into translation and rotation. The transmis- When the lunar soil interacts with the lander structure,
sion diagram of the centroid motion of contiguous bodies the contact position between the lunar soil and lander
is plotted in Fig. 7, where CLðiÞ and Ci are the centroids structure is random. Therefore, it is necessary to determine
of the inside connecting body and outside connecting body, the equivalent nodal load of the lander under load. The
7
S. Ji, S. Liang Advances in Space Research xxx (xxxx) xxx

Fig. 8. Program flowchart of the DEM-FEM-MBD coupling algorithm.

lunar soil is composed of spherical elements, and the lander


structure is composed of beam elements and flat shell ele-
ments. The node equivalent force of the spherical discrete
elements and flat shell elements can be obtained according
to the principle of virtual work (Wellmann et al, 2008):
f i ¼ N T fDEM ði ¼ 1; 2; 3Þ ð19Þ
where f i is the equivalent node force of the element, fDEM is
the force of the DEM, and N is the shape function of the
element. Here, the shape function of the element is con-
structed by using the area coordinates of the triangle:
N i ¼ Li ¼Ai =A ð20Þ Fig. 9. Equivalent element force computation model of a triangular shell
element.
where Li is the area coordinate of the triangle, Ai is the area
enclosed by the action point and the other two element fF e gT ¼ ½NT  ff i0 ; f j0 ; f k0 gT ð21Þ
nodes, and A is the area of the triangular element, as plot-
ted in Fig. 9. where fi0 is the contact force of the beam element in the
The contact force and contact point coordinates of the 0

spherical element and beam element are obtained in the local coordinate system in the direction of i , which is com-
local coordinate system, as plotted in Fig. 10. According posed of normal and tangential parts; ½N is the transfor-
to Hertz-Mindlin theory, assuming that both ends of the mation matrix of the equivalent node force; and ½N A ; N B 
beam element are fixed, the equivalent node forces of the is composed of
two elements in the local coordinate system can be
obtained by static equilibrium: ð3a þ bÞb2
NA
ij ¼   dij ði; j ¼ 1; 2; 3Þ ð22Þ
L3
8
S. Ji, S. Liang Advances in Space Research xxx (xxxx) xxx

ger than that of the experiment, which is mainly due to the


rigid structure in Altair/EDEM and the self-buffering effect
of the lander is not realized. In addition, the immersion
depth calculated by Altair/EDEM is significantly larger
than that of the experiment. The above results show that
the DEM-FEM-MBD coupling algorithm is reliable.

4. Dynamic response analysis of the lander during landing

To account for the short duration of the landing pro-


cess, the simulation time is set to 1.0 s. The calculation
parameters of the lander are listed in Table 3. Fig. 11 shows
the dynamic characteristics of the lunar soil and lander
structure during the landing, from which we can see the
Fig. 10. Equivalent element force computation model of a beam element. movement of lunar soil particles under impact. When
t = 40 ms, the lunar soil at the contact point obtains a cer-
tain velocity. When t = 80 ms, the lunar soil around the
ð3b þ aÞa2
N Bij ¼   dij ði; j ¼ 1; 2; 3Þ ð23Þ contact point also achieves a velocity, which indicates that
L3 momentum transfer also occurs between lunar soil parti-
where L is the length of the beam element, aı; and;b are the cles. When t = 120 ms, the lunar soil at the landing site
distances from the contact point to the nodes at both ends, obviously splashes; and when t = 1000 ms, the lunar soil
and dij is a Kronecker symbol. particles basically reach a stable state.
The dynamic time history curve of the lander during
landing is obtained by using a landing velocity of 4 m=s,
3. Validation as presented in Fig. 12. The load on the lander fluctuates
at the initial stage of contact between the lander and the
To verify the correctness of the above methods, the lunar soil due to the buffer device. In addition, the figure
landing process of the lander is studied and compared with shows that the landing time of the lander is approximately
the vertical landing experiment on the earth (Hou et al, 0.6 s, and the impact depth is approximately 45 cm. Here,
2018). In the experiment, a full-scale manned lunar lander the impact depth is the height of the center of mass falling.
with a buffer device is used. The experiment is realized on Fig. 12(d) presents the change in the mechanical energy of
the earth in vertical landing. The gravity acceleration is the lunar soil. At the moment of contact with the lander,
g¼  9:81m=s2 , the mass of the lander is 3000 kg, and the lunar soil obtains a higher mechanical energy that
the landing speed is 6.2 m/s. decreases slowly. When the lunar soil contacts the lander,
The results of experiment and simulation are listed in momentum transfer occurs at the contact point. The
Table 2. The simulation result of single leg impact force mechanical energy of the lander decreases rapidly, and
is slightly larger than the experiment value. In the experi- the particles gain more momentum. Through the interac-
ment, the triggering loads of the two stage buffer device tion of particles as a result of collision and friction,
of the lander are 50 kN and 90 kN respectively, and the momentum is transferred to the surrounding particles so
maximum compression capacity of the buffer device is that more particles can obtain mechanical energy. The col-
200 mm. The maximum load on the single leg is less than lision between particles causes plastic deformation and
the critical load that allows the second stage buffer to work, even the breakage of particles, which makes the energy
so the second stage buffer does not work. In the simulation, change irreversibly, and the friction between particles
the maximum load of the single leg is 94 kN, which is transforms mechanical energy into heat energy. In this
greater than the trigger load of the two-stage buffer device, way, the mechanical energy of the whole collision system
so the two-stage buffer devices play a role. Therefore, the is transformed and absorbed so that the lunar soil acts as
immersion depth of the lander into the soil is slightly shal- a buffer for the lander.
lower than the experimental results. The single leg impact During landing, the footpad first contacts the lunar soil
force calculated by Altair/EDEM (Hou, 2018) is much lar- and receives the longitudinal load of the lunar soil. Fig. 13
plots the von Mises stress distribution of the footpad at dif-
ferent times. Since the stress of each footpad is basically the
Table 2
Comparison the result of the experiment and simulation.
same at the same time, one of them is selected for analysis.
The stress is mainly concentrated at the bottom of the foot-
Contents Simulation Experiment Altair/EDEM
pad, that is, the contact with the lunar soil and the stress
Maximum force of leg 94 kN 80 kN 132 kN value is relatively large at the initial contact time. Then,
Immersion depth 88 mm 90 mm 145 mm
the stress gradually decreases.

9
S. Ji, S. Liang Advances in Space Research xxx (xxxx) xxx

Table 3
Computational parameters of the lander in the FEM-MBD simulation.
Parameter Symbol Value
Lander mass M 1000 kg
Elastic modulus of the lander Es 207 GPa
Friction coefficient between the lander and lunar soil ls 0.3
Resilience coefficient of the lander es 0.1
Poisson’s ratio c 0.3
Density of the lander qs 7860 kg/m3
Acceleration due to gravity g 1.63 m/s2
Number of elements / 4084

t = 0 ms t = 40 ms t = 80 ms

t = 120 ms t = 570 ms t = 1000 ms

(a) Impact force vs. time (b) Velocity of the lander vs. time

(c) Impact depth vs. time (d) Variation in the mechanical energy of the lunar soil
with time in the landing system

Fig. 12. Dynamic characteristics of the lander and mechanical energy change of lunar soil during landing.

5. Analysis of the influencing factors on the landing process cussed in detail. Finally, the slope landing and lunar soil
energy absorption are analyzed.
The impact depth, impact force peak value and acceler-
ation of the lander are mainly affected by the impact veloc- 5.1. Influence of the landing velocity
ity, soil compaction and mass of the lander. The impact
velocity and the mass of the lander directly affect the initial One of the factors affecting the landing is the impact
mechanical energy of the landing system, which will be dis- velocity. The variation in the peak impact force on the lan-
10
S. Ji, S. Liang Advances in Space Research xxx (xxxx) xxx

t = 70 ms t = 140 ms t = 240 ms t = 360 ms

Fig. 13. von Mises stress nephogram of one footpad at different times during landing.

der during landing is presented in Fig. 14, which shows that


the impact force peak rises with an increase in the impact
velocity. The increase rate of the peak impact force is low
when the velocity is low, and the impact force peak
increases with increasing linear velocity when the velocity
is high. Fig. 15 presents the relationship between the
impact depth and impact velocity. Here, the impact depth
refers to the descent distance of the mass center of the lan-
der after the lander contacts the lunar soil. With an
increase in the impact velocity, the kinetic energy of the
lander increases, and it will reach a greater depth; that is,
the vertical displacement will increase.
The variation in the energy absorbed by the lunar soil
with the impact velocity is presented in Fig. 16. At the
Fig. 15. Impact depth of the lander at different landing velocities.
moment when the mechanical energy of the particle system
is a maximum, considering the mechanical energy change
of the lander, the energy absorbed by the buffer device
and the strain energy of the footpad, the energy absorbed cal energy of the system and the greater the energy
by the particle system is calculated by using the law of absorbed by the particles.
the conservation of energy. The energy absorbed by the
particles increases linearly with an increase in the impact 5.2. Influence of the lander mass
velocity. This is mainly because when the landing velocity
is higher, the impact depth of the lander is greater, which An impact velocity of 4 m/s is chosen to study the influ-
makes the structure have a larger contact area with the ence of different masses of the lander on the landing pro-
lunar soil to transfer momentum. Then, more granular cess. Fig. 17 presents the variation in the peak impact
media can obtain kinetic energy, thus consuming the force with the mass of the lander. The impact force peak
mechanical energy of the landing system. In addition, the increases with an increase in the mass of the lander; when
higher the impact velocity, the greater the initial mechani- the mass reaches a certain value, the peak value tends to
be stable. Accordingly, when the mass of the lander reaches
a certain value, the mass of the lander is no longer the influ-
encing factor of the impact load. The impact depth
increases linearly with an increase in mass, as presented
in Fig. 18. Because the impact force does not increase when
the mass increases, the acceleration of the lander changes
slowly, and it takes a long time for the lander to reduce
its velocity to 0. Thus, the impact depth is greater.
Fig. 19 presents the energy absorbed by the lunar soil
with the variation in the lander mass. It can be seen that
the energy absorbed by the granular media increases lin-
early with an increase in the mass of the lander. When
the mass of the lander is large, the impact depth also
increases. The lander transfers more mechanical energy to
the granular media, which converts the mechanical energy
into other forms of energy through the collision and sliding
Fig. 14. Peak impact force on the lander at different landing velocities. friction between the particles to dissipate the energy of the
11
S. Ji, S. Liang Advances in Space Research xxx (xxxx) xxx

Fig. 16. Energy absorbed and dissipated by lunar soil at different landing Fig. 19. Energy absorbed and dissipated by lunar soil for different lander
velocities. masses.

faces with rocks, craters and slopes (Arslantas, 2018; Gong,


2007; Seo, 2016). Therefore, it is necessary to evaluate the
landing characteristics of the lunar lander. Owing to the
differences in the flatness of the landing site and the inclina-
tion direction of the lander, the landing modes are various.
According to the landing sequence of the supporting legs,
two typical inclined landing modes are defined as indicated
in Fig. 20. In the 2-2 landing mode, legs 1 and 2 contact the
lunar surface first, and then legs 3 and 4 contact the lunar
surface. In the 1-2-1 landing mode, leg 1 contacts the lunar
surface first, followed by leg 2 and leg 3 and finally leg 4.
Since leg 1 penetrates the lunar soil to a certain depth,
Fig. 17. Impact force peak on the lander for different lander masses.
leg 4 does not contact the lunar soil at the same time.
A landing velocity of 4 m/s is selected to compare the
landing process of 1-2-1, 2-2 and 4-0 landing modes. The
vertical landing mode is considered 4-0 landing mode. Dur-
ing the inclined landing, the slope of the lunar surface is set
to 5°. The motion status at different times is shown in
Fig. 21. The impact depth of the lander is shallower in
the 1-2-1 mode, and the splash of lunar soil is more severe
in the 2-2 mode.
The force of each landing leg under different landing
modes is presented in Fig. 22. It can be found that the loads
along the four legs of the lander in the 4-0 mode are basi-
cally the same. In the 1-2-1 mode, it is obvious that leg 1
first contacts the lunar soil and is subject to the largest
force, and then the three legs are under the same force. It
can be clearly seen that leg 1 and leg 2 bear the impact force
Fig. 18. Impact depth of the lander for different lander masses. before t = 0.03 s, after which leg 3 and leg 4 mainly bear the
impact force. A comparison of the sum of the loads of the 4
leg contributions among the three landing modes is pre-
whole system. In addition, when the mass of the lander is sented in Fig. 22(d). It can be found that the interaction
large, the initial mechanical energy of the system is high, time between the lander and lunar soil in the 4-0 mode is
so the energy absorbed by the particles is high. the shortest and that in the 2-2 mode is the longest. The
peak force in the 4-0 mode is significantly greater than that
5.3. Influence of the landing attitude and inclination of the in the tilted landing mode.
lunar surface When the slope angle of the lunar surface is different, the
impact force peak of the lander under the two landing
With the development of deep space exploration tech- modes varies with the slope angle, as shown in Fig. 23. It
nology, new lunar landers must land on complex lunar sur- is found that the impact load of the lander under the 2-2
12
S. Ji, S. Liang Advances in Space Research xxx (xxxx) xxx

(a) 2-2 landing mode (b) 1-2-1 landing mode

Fig. 20. Landing modes on a sloped lunar surface.

t = 0 ms t = 40 ms t = 120 ms t = 570 ms
(a) 4-0 landing mode

t = 0 ms t = 70 ms t = 90 ms t = 570 ms
(b) 1-2-1 landing mode

t = 0 ms t = 70 ms t = 100 ms t = 570 ms
(c) 2-2 landing mode

Fig. 21. Motion characteristics of the lander and lunar soil under three landing modes during landing.

mode is slightly greater than that under the 1-2-1 mode for role. Therefore, the proportion of energy absorbed by the
angle > 7.5°. The impact depth of the lander gradually lunar soil is affected by the initial mechanical energy and
increases with an increase in the slope angle, but the change friction coefficient of the lander (Ciantia et al, 2016).
is slow, and the impact depth in the 1-2-1 mode is signifi- Since the friction coefficient of the lunar soil directly
cantly greater than that in the 2-2 mode, as shown in determines the ability of the system to convert mechanical
Fig. 24. energy into heat energy, the friction coefficient has a direct
The buffering effect of particles is mainly due to the impact on the energy absorption effect of lunar soil. Fig. 25
strong extrusion and friction between the particles under presents the influence of the lunar soil friction coefficient on
the impact load, and the structure of the internal complex the lunar soil energy absorption for three landing modes.
force chain is broken and reorganized, consuming a signif- The mechanical energy of the lander before landing is E0.
icant amount of energy (Mwangi and Kanny, 2012). Due When the strain energy of the structure reaches the maxi-
to the plastic deformation and viscous effect between parti- mum, the strain energy of the structure is E1. At this time,
cles, the energy of the system changes irreversibly and is the mechanical energy of the lander is E2. The energy
absorbed. In the particle system, the contact force is trans- absorbed by the self-generating buffer device of the lander
mitted through the force chain, which makes the local is E3. According to the energy conservation, the energy
impact load continuously expand in space and then reduces absorbed by lunar soil is E4 = E0- E1- E2- E3 and the energy
the impact strength. In addition, the force chain has a sig- absorption ratio of lunar soil is E4 / E0. It can be seen that
nificant time effect on the force propagation, which delays the proportion of lunar soil absorbed energy decreases with
the instantaneous impact load in time, thus playing a buffer an increase in the lunar soil friction coefficient. A compar-
13
S. Ji, S. Liang Advances in Space Research xxx (xxxx) xxx

(a) 4-0 landing mode (b) 1-2-1 landing mode

(c) 2-2 landing mode (d) Comparison of the sum of the loads of the 4 leg
contributions among the landing modes

Fig. 22. Force comparison among the landing legs under different landing modes.

Fig. 23. Impact force peak on the lander at different slope angles of the Fig. 24. Impact depth of the lander at different slope angles of the lunar
lunar surface in two landing modes. surface in two landing modes.

ison of the three landing modes shows that the energy an increase in the friction coefficient, the running distance
absorption ratio of lunar soil under the 2-2 mode is slightly of the lander in lunar soil particles is relatively short, and
higher than that under the other two modes, which is due the energy absorbed by the lunar soil is relatively low.
to the long contact time between the lunar soil and lander The above results show that smooth particles have a better
structure under the 2-2 landing mode. There is sufficient buffering performance.
momentum transfer between the lunar soil and lander. The initial mechanical energy of the lander also has
The main role of friction is to prevent relative sliding some influence on the energy absorption effect of the lunar
and rolling between the lunar soil particles and the lander soil. The initial mechanical energy of the lander is charac-
and achieve the consumption of energy through relative terized by the landing velocity, and the effect on lunar soil
sliding at the contact surface. When the particle friction energy absorption is studied, as shown in Fig. 26. It can be
coefficient is small, the stability of the particle system is seen that when the initial mechanical energy of the lander is
weak, so it is easy to slide or even locally flow. The motion low, the absorption ratio of the lunar soil decreases with an
distance of the lander in the granular material is long. With increase in the initial mechanical energy and then increases
14
S. Ji, S. Liang Advances in Space Research xxx (xxxx) xxx

increase in landing speed and mass. In addition, the influ-


ence of different inclined landing modes is studied in detail.
The impact force peaks of the two inclined landing modes
are similar. However, the impact depth in the 1-2-1 mode is
significantly greater than that in the 2-2 mode. By compar-
ing the three landing modes, the order of the lunar soil
energy absorption ratio is 4-0 < 1-2-1 < 2-2. Finally, the
main factors influencing lunar soil energy absorption are
studied in detail. It is found that the proportion of energy
absorbed by lunar soil in total energy decreases with the
increase in friction coefficient. When the initial mechanical
energy of the lander is low, the energy absorption ratio of
the lunar soil decreases with an increase in the initial
mechanical energy and then increases gradually. These
Fig. 25. Effect of the friction coefficient of lunar soil on energy dissipation
by lunar soil under three landing modes.
results provide a theoretical basis for the analysis of the
mechanism of energy absorption and dissipation of lunar
soil, and provide a useful reference for the realization of
safely soft lunar landings.
In addition, the time step of the coupling algorithm in
this paper is the DEM time step, which ensures the accu-
racy of the calculation, but sacrifices the computational
efficiency. The main force of future work will be to solve
the asynchronous problem of the time step. This paper also
uses spherical particles to simulate lunar soil, which is
approximate to a certain extent, and cannot fully realize
the simulation of lunar soil morphology. Due to the limita-
tion of current theory and technology, it is impossible to
realize large-scale calculation of non-spherical particles,
which will be a main goal in the future.

Declaration of Competing Interest


Fig. 26. Effect of the landing velocity on energy dissipation by lunar soil
under three landing modes.
The authors declare that they have no known competing
gradually. Under the three landing modes, the order of the financial interests or personal relationships that could have
absorption ratio of lunar soil is 4-0 < 1-2-1 < 2-2; that is, appeared to influence the work reported in this paper.
the absorption ratio of lunar soil in the 2-2 mode is the
highest. Acknowledgements
In addition, other physical properties of lunar soil and
This study is financially supported by the National Nat-
the friction coefficient between lunar soil and lander also
ural Science Foundation of China (Grant No. 11872136)
play an important role in the energy dissipation of the
whole system, which will not be discussed in detail here. and the Fundamental Research Funds for the Central
Universities (Grant No. DUT19GJ206).

6. Conclusions Compliance with ethical standards

A DEM-FEM-MBD coupling algorithm is developed to On behalf of all authors, the corresponding author
study the motion characteristics of the lander and the states that there is no conflict of interest.
dynamic characteristics of the lunar soil during the landing
process at different attitudes. The lunar soil is simulated by References
DEM. The FEM of the multibody structure of a lander is
developed. The second buffer device of the landing leg is Alshibli, K.A., Hasan, A., 2009. Strength Properties of JSC-1A Lunar
designed and realized by prism joint. Subsequently, the cor- Regolith Simulant. J. Geotech. Geoenviron. Eng. 135 (5), 673–679.
Arslantas, Y.E., Theil, S., 2018. Attainable Landing Area Computation of
rectness of the DEM-FEM-MBD coupling algorithm is a Lunar Lander with Uncertainty by Reachability Analysis. Adv.
verified by comparing it with a full-scale experiment of a Aerospace Guid., Navig. Control. 41 (1), 497–513.
lunar lander on earth. Moreover, the main factors affecting Bui, H.H., Kobayashi, T., Fukagawa, R., Wells, J.C., 2009. Numerical
the landing safety of the lander are studied. It is found that and experimental studies of gravity effect on the mechanism of lunar
the impact force peak and the depth increase with the excavations. J. Terramech. 46 (3), 115–124.

15
S. Ji, S. Liang Advances in Space Research xxx (xxxx) xxx

Chen, J., Nie, H., 2008. Overloading of landing based on the deformation Lommen, S., Lodewijks, G., Schott, D.L., 2018. Co-simulation framework
of the lunar lander. Chin. J. Aeronaut. 21, 43–47. of discrete element method and multibody dynamics models. Eng.
Ciantia, M.O., Arroyo, M., Butlanska, J., Gens, A., 2016. DEM Comput. 35 (3), 1481–1499.
modelling of cone penetration tests in a double-porosity crushable Ma, L., Wang, K., Xu, Z., Shao, Z., Song, Z., Biegler, L.T., 2018.
granular material. Comput. Geotech. 73, 109–127. Trajectory optimization for lunar rover performing vertical takeoff
Closs, D.J., Cook, R.L., 1987. Multi-Stage Transportation Consolidation vertical landing maneuvers in the presence of terrain. Acta Astronaut.
Analysis Using Dynamic Simulation. Int. J. Phys. Distrib. Mater. 146, 289–299.
Manage. 17 (3), 28–45. Metzger, P.T., Anderson, S., Colaprete, A., 2018. Experiments Indicate
Cundall, P.A., Strack, O.D.L., 1979. A discrete numerical model for Regolith is Looser in the Lundar Polar Regions than at the Lunar
granular assemblies. Géotechnique. 29 (1), 47–65. Landing Sites. Earth Space 2018: Eng. Extreme Environ. 4, 1–9.
El-Ghandour, A., Hamper, M.B., Foster, C.H., 2016. Coupled finite Michael, M., Vogel, F., Peters, B., 2015. DEM–FEM coupling simulations
element and multibody system dynamics modeling of a three-dimen- of the interactions between a tire tread and granular terrain. Comput.
sional railroad system. J. Rail Rapid Transit. 7, 1–12. Methods Appl. Mech. Eng. 289, 227–248.
El-Ghandour, A.I., Foster, C.D., 2019. Coupled finite element and Mishra, S.K., Prasad, K.D., 2018. Numerical evaluation of surface
multibody systems dynamics modelling for the investigation of the modifications at landing site due to spacecraft (soft) landing on the
bridge approach problem. J. Rail Rapid Transit. 1, 1–15. moon. Planet. Space Sci. 156, 57–61.
Fritzer, H.P., 2001. Molecular symmetry with quaternions. Spectrochim. Moghadam, M.G.E., Shahmardan, M.M., Norouzi, M., 2019. Magneto-
Acta Part A Mol. Biomol. Spectrosc. 57 (10), 1919–1930. rheological damper modeling by using dissipative particle dynamics
Gao, X., Tang, D., Yue, H., Jiang, S., Deng, Z., 2017. Influence of friction method. Comput. Particle Mech. 7 (3), 567–592.
on sampling disturbance of lunar surface in direct push sampling Mwangi, F.M., Kanny, K., 2012. Development of granular-medium-based
method based on DEM. Adv. Space Res. 59 (12), 3036–3044. impact energy management system. Int. J. Crashworthiness 17 (4),
Gong, S., Li, J., Baoyin, H., Gao, Y., 2007. Lunar landing trajectory 401–414.
design based on invariant manifold. Appl. Math. Mech. 28 (2), 201– Nakashima, H., Fujii, H., Oida, A., Momozu, M., Kanamori, H., Aoki,
207. S., Yokoyama, T., Shimizu, H., Miyasaka, J., Ohdoi, K., 2010.
Hou, X., Xue, P., Wang, Y., Cao, P., Tang, T., 2018. Theoretical and Discrete element method analysis of single wheel performance for a
discrete element simulation studies of aircraft landing impact. J. Braz. small lunar rover on sloped terrain. J. Terramech. 47 (5), 307–321.
Soc. Mech. Sci. Eng. 40 (3), 114–129. Nakashima, H., Fujii, H., Oida, A., Momozu, M., Kawase, Y., Kanamori,
Ji, S., Wang, S., Peng, Z., 2019. Influence of external pressure on granular H., Aoki, S., Yokoyama, T., 2007. Parametric analysis of lugged wheel
flow in a cylindrical silo based on discrete element method. Powder performance for a lunar microrover by means of DEM. J. Terramech.
Technol. 356, 702–714. 44 (2), 153–162.
Jiang, M., Shen, Z., Thornton, C., 2013. Microscopic contact model of Nakashima, H., Shioji, Y., Kobayashi, T., Aoki, S., Shimizu, H.,
lunar regolith for high efficiency discrete element analyses. Comput. Miyasaka, J., Ohdoi, K., 2011. Determining the angle of repose of
Geotech. 54, 104–116. sand under low-gravity conditions using discrete element method. J.
Katagiri, J., Matsushima, T., Yamada, Y., Tsuchiyama, A., Nakano, T., Terramech. 48 (1), 17–26.
Uesugi, K., Ohtake, M., Saiki, K., 2015. Investigation of 3D Grain Okubo, K., Rougier, E., Lei, Z., Bhat, H.S., 2020. Modeling earthquakes
Shape Characteristics of Lunar Soil Retrieved in Apollo 16 Using with off-fault damage using the combined finite-discrete element
Image-Based Discrete-Element Modeling. J. Aerosp. Eng. 28 (4), method. Comput. Particle Mech. 7, 567–592.
04014092. Olivier, B., Verlinden, O., Kouroussis, G., 2020. A vehicle track soil model
Knuth, M.A., Johnson, J.B., Hopkins, M.A., Sullivan, R.J., Moore, J.M., using co-simulation between multibody dynamics and finite element
2012. Discrete element modeling of a Mars Exploration Rover wheel in analysis. Int. J. Rail Transport. 8 (2), 135–158.
granular material. J. Terramech. 49 (1), 27–36. Otto, H., Kerst, K., Roloff, C., Janiga, G., Katterfeld, A., 2018. CFD–
Kulchitsky, A.V., Johnson, J.B., Reeves, D.M., 2016. Resistance forces DEM simulation and experimental investigation of the flow behavior
during boulder extraction from an asteroid. Acta Astronaut. 127, 424– of lunar regolith JSC-1A. Particuology. 40, 34–43.
437. Pei, C., Wu, C., Adams, M., 2016. DEM-CFD analysis of contact
Kwon, J.W., Lee, D.H., Bang, H., 2016. Virtual Trajectory Augmented electrification and electrostatic interactions during fluidization. Powder
Landing Control Based on Dual Quaternion for Lunar Lander. J. Technol. 304, 208–217.
Guid., Control, Dynam. 39 (9), 2044–2057. Potyondy, D.O., Cundall, P.A., 2004. A bonded-particle model for rock.
Li, C., Umbanhowar, P.B., Komsuoglu, H., Koditschek, D.E., Goldman, Int. J. Rock Mech. Min. Sci. 41 (8), 1329–1364.
D.I., 2009. Sensitive dependence of the motion of a legged robot on Ramırez, R., Poschel, T., Brilliantov, N.V., Schwager, T., 1999. Coeffi-
granular media. Proc. Natl. Acad. Sci. 106 (24), 9932-9932. cient of restitution of colliding viscoelastic spheres. Phys. Rev. E 60 (4),
Li, F., Ye, M., Yan, J., Hao, W., Barriot, J.-P., 2016. A simulation of the 4465–4472.
Four-way lunar Lander-Orbiter tracking mode for the Chang’E-5 Recuero, A., Shabana, M., A., A., Mohil, P., Ulysses, C., 2016. ANCF
mission. Adv. Space Res. 57 (11), 2376–2384. Continuum-Based Soil Plasticity for Wheeled Vehicle Off-Road
Li, W., Huang, Y., Cui, Y., Dong, S., Wang, J., 2010. Trafficability Mobility. J. Comput. Nonlinear Dynam., 11, 4, 044504.
analysis of lunar mare terrain by means of the discrete element method Recuero, A., Serban, R., Peterson, B., Sugiyama, H., Jayakumar, P.,
for wheeled rover locomotion. J. Terramech. 47 (3), 161–172. Negrut, D., 2017. A high-fidelity approach for vehicle mobility
Li, Y., 2015. A new family of nonconforming finite elements on simulation: Nonlinear finite element tires operating on granular
quadrilaterals. Comput. Math. Appl. 70 (4), 637–647. material. J. Terramech. 72, 39–54.
Li, Y., Ji, S., 2018. A geometric algorithm based on the advancing front Sanborn, G., Choi, J., Yoon, J.S., Rhim, S., Choi, J.H., 2014. Systematic
approach for sequential sphere packing. Granular Matter 20 (4), 58– Integration of Finite Element Methods Into Multibody Dynamics
69. Considering Hyperelasticity and Plasticity. J. Comput. Nonlinear Dyn.
Liu, T., Bai, Z., Zhao, Y., 2020. Research on the Retention Characteristics 9, 041012-1-041012-11.
of the Stratification Information of Lunar Soil Drilling Sampling. Adv. Seo, M.G., Hong, S.M., Tahk, M.J., 2016. Lunar Lander Landing Site
Space Res. 66, 2428–2445. Decision in Low-Fuel Case. Matec Web Conf. 54, 09003.
Liu, X.L., Duan, G.R., Teo, K.L., 2008. Optimal soft landing control for Shi, F., Xie, W., 2016. A specific energy-based ball mill model: From batch
moon lander. Automatica. 44 (4), 1097–1103. grinding to continuous operation. Miner. Eng. 86, 66–74.

16
S. Ji, S. Liang Advances in Space Research xxx (xxxx) xxx

Smith, W., Peng, H., 2013. Modeling of wheel–soil interaction over rough Wellmann, C., Lillie, C., Wriggers, P., 2008. Homogenization of granular
terrain using the discrete element method. J. Terramech. 50 (5), 277– material modeled by a three-dimensional discrete element method.
287. Comput. Geotech. 35 (3), 394–405.
Sutoh, M., Wakabayashi, S., Hoshino, T., 2018. Landing Behavior Yang, P., Zang, M., Zeng, H., Guo, X., 2020. The interactions between an
Analysis of Lunar Probe Based on Drop Tests and RFT in a Vacuum. off-road tire and granular terrain: GPU-based DEM-FEM simulation
IEEE Rob. Autom. Lett. 3 (1), 360–366. and experimental validation. Int. J. Mech. Sci. 179 105634.
Tu, F., Ling, D., Hu, C., Zhang, R., 2017. DEM–FEM analysis of soil Yang, Y., Zi, C., Huang, Z., Wang, J., Lungu, M., Liao, Z., Yang, Y., Su,
failure process via the separate edge coupling method. Int. J. Numer. H., 2017. CFD-DEM investigation of particle elutriation with
Anal. Meth. Geomech. 41, 1157–1181. electrostatic effects in gas-solid fluidized beds. Powder Technol. 308,
Wang, C., Nie, H., Chen, J., Lee, H.P., 2019. The design and dynamic 422–433.
analysis of a lunar lander with semi-active control. Acta Astronaut. Zeng, H., Xu, W., Zang, M., Yang, P., 2020. Calibration of DEM-FEM
157, 145–156. model parameters for traction performance analysis of an off-road tire
Wang, L., Li, S., Zhang, G., Ma, Z., Zhang, L., 2013. A GPU-Based on gravel terrain. Powder Technol. 362, 350–361.
Parallel Procedure for Nonlinear Analysis of Complex Structures Zeng, Y.W., Jin, L., Du, X., Gao, R., 2015. Refined modeling and
Using a Coupled FEM/DEM Approach. Mathem. Probl. Eng. 12, 1– movement characteristics analyses of irregularly shaped particles. Int.
16. J. Numer. Anal. Meth. Geomech. 39 (4), 388–408.
Wang, S., Deng, Z., Hu, M., Gao, H., 2005. Dynami cmodel building and Zheng, G., Nie, H., Chen, J., Chen, C., Lee, H.P., 2018. Dynamic analysis
simulation for mechanical main body of lunar lander. J. Central South of lunar lander during soft landing using explicit finite element
Univ. Technol. 12 (3), 329–334. method. Acta Astronaut. 148, 69–81.
Wang, S., Marmysh, D., Ji, S., 2020. Construction of irregular particles Zheng, Z., Zang, M., Chen, S., Zhao, C., 2017. An improved 3D DEM-
with superquadric equation in DEM. Theor. Appl. Mech. Lett. 10 (2), FEM contact detection algorithm for the interaction simulations
68–73. between particles and structures. Powder Technol. 305, 308–322.
Wang, S.L., Ji, S., 2018. Coupled DEM–FEM Analysis of Ice-Induced Zhu, H., Tian, H., Cai, G., 2017. Hybrid uncertainty-based design
Vibrations of a Conical Jacket Platform Based on the Domain optimization and its application to hybrid rocket motors for manned
Decomposition Method. Int. J. Offshore Polar Eng. 28 (2), 190–199. lunar landing. Chin. J. Aeronaut. 30 (2), 719–725.
Wang, X., Yang, S., Li, W., Wang, Y., 2018. Vibratory finishing co- Zhu, H., Tian, H., Cai, G., Bao, W., 2015. Uncertainty analysis and
simulation based on ADAMS-EDEM with experimental validation. probabilistic design optimization of hybrid rocket motors for manned
Int. J. Adv. Manuf. Technol. 96, 1175–1185. lunar landing. Sci. China Technolog. Sci. 58 (7), 1234–1241.

17

You might also like