SPH-based framework for modelling fluid-structure interaction

problems with finite deformation and fracturing

Md Rushdie Ibne Islama,∗
a
Department of Applied Mechanics, Indian Institute of Technology Delhi, New Delhi 110016, India

Abstract
arXiv:2310.03122v1 [cs.CE] 22 Sep 2023

Understanding crack propagation in structures subjected to fluid loads is crucial in various en-
gineering applications, ranging from underwater pipelines to aircraft components. This study
investigates the dynamic response of structures, including their damage and fracture behaviour un-
der hydrodynamic load, emphasizing the fluid-structure interaction (FSI) phenomena by applying
Smoothed Particle Hydrodynamics (SPH). The developed framework employs weakly compress-
ible SPH (WCSPH) to model the fluid flow and a pseudo-spring-based SPH solver for modelling
the structural response. For improved accuracy in FSI modelling, the δ-SPH technique is imple-
mented to enhance pressure calculations within the fluid phase. The pseudo-spring analogy is
employed for modelling material damage, where particle interactions are confined to their imme-
diate neighbours. These particles are linked by springs, which don’t contribute to system stiffness
but determine the interaction strength between connected pairs. It is assumed that a crack propa-
gates through a spring connecting a particle pair when the damage indicator of that spring exceeds
a predefined threshold. The developed framework is extensively validated through a dam break
case, oscillation of a deformable solid beam, dam break through a deformable elastic solid, and
breaking dam impact on a deformable solid obstacle. Numerical outcomes are subsequently com-
pared with the findings from existing literature. The ability of the framework to accurately depict
material damage and fracture is showcased through a simulation of water impact on a deformable
solid obstacle with an initial notch.
Keywords: Smoothed particle hydrodynamics, fluid-structure interaction, material damage and
fracture, pseudo-spring analogy.

1. Introduction
In recent years, there has been a growing focus on fluid-structure interaction (FSI), which
plays a prominent role in numerous engineering and industrial contexts. Examples of these ap-
plications include coastal engineering, the shipbuilding industry, and aviation. Understanding the
crack propagation in structures under fluid load is critical for enhancing safety, preventing envi-
ronmental disasters, reducing economic losses, and advancing engineering innovation in complex

∗
Corresponding Author
Email address: [email protected] (Md Rushdie Ibne Islam)
Preprint submitted to Elsevier October 6, 2023
fluid-structure interaction scenarios. The intricacies of fluid-structure interaction (FSI) problems
often render them beyond the reach of analytical solutions. Furthermore, the high cost and logisti-
cal difficulties associated with experimental studies in FSI have spurred the adoption of numerical
modelling as an attractive alternative.
Over recent decades, various methodologies have been developed to tackle the complex chal-
lenges of fluid-structure interaction (FSI) problems. Although mesh-based methods such as finite
difference method (FDM), finite volume method (FVM), and finite element method (FEM) [1, 2, 3]
have achieved a degree of success for FSI simulations, they often require additional computation-
ally intensive numerical schemes (e.g., interface tracking or re-meshing etc.) when dealing with
free surfaces, moving boundaries, and deformable structures. The computation becomes even
more complex with propagating cracks and material separation as field variables exhibit disconti-
nuities. Traditional mesh-based methods such as FEM are unsuitable [4], and while discontinuous
enrichment helps in modelling the cracks [5], implementing these additional numerical strategies
is not only intricate but also computationally intensive, and they can frequently introduce instabil-
ity issues. As a promising alternative, Lagrangian particle-based methods have gained increasing
favour in FSI modelling. These methods are becoming more popular due to their meshless and
Lagrangian nature, which makes them well-suited for representing free-surface fluid flow and the
substantial deformation of solid structures. Moreover, Lagrangian particle-based meshless ap-
proaches offer a natural and efficient means of capturing moving interfaces and finite deformation
in structures encountered in FSI problems.
Smoothed Particle Hydrodynamics (SPH), initially designed to address challenges in astro-
physical contexts, has gained widespread acclaim as a leading meshless method [6, 7]. SPH
operates as a truly meshless and Lagrangian particle-based approach, where individual particles
represent the material points and carry the field variables[8, 9]. Each particle exclusively en-
gages with its neighbouring counterparts in this method through a kernel function. The extent of
this interaction is governed by the smoothing length, which defines the dimensions of the local
neighbourhood, also referred to as the influence domain. Notably, the kernel function exhibits a
characteristic bell-shaped profile designed to maximize interaction strength with immediate neigh-
bours and progressively diminish it as the distance between interacting particles increases. SPH
offers distinct advantages in handling scenarios involving free-surface flow, finite material defor-
mation, moving interfaces and boundaries. Its wide range of applications can be found in dynamic
fluid flow [10, 11, 12], geotechnical simulations [13, 14, 15, 16], explosive and impact events
[17, 18, 19, 20]. SPH also plays a prominent role in FSI simulations [21, 22, 23].
SPH methodologies can be of different types, such as weakly compressible SPH (WCSPH),
incompressible SPH (ISPH), total Lagrangian SPH (TLSPH) etc. WCSPH and ISPH are the most
prominent techniques for fluid flow, whereas standard SPH and TLSPH are used for modelling the
deformation of solids. In WCSPH, the time step size used for numerical integration is quite small,
whereas ISPH allows a larger time step size for integration. Another advantage of ISPH is that it
produces smooth pressure fields compared to WCSPH simulations [24, 25, 26, 27]. However, the
computational cost per step is relatively low in WCSPH. For large-scale simulations, implementing
parallelized techniques for ISPH [28, 29, 30] is more challenging than its WCSPH counterpart.
Nonetheless, the conventional WCSPH method is hindered by substantial pressure fluctuations.
While these fluctuations have a limited impact on flow kinematics, they pose significant challenges
2
in Fluid-Structure Interaction (FSI) modelling. This is because pressure fluctuations can lead to
inaccuracies in assessing the interacting forces at the fluid-structure interface. To address this
issue, several numerical schemes (e.g. δ-SPH method [31, 32], Rusanov flux [33] etc.) have been
introduced to get smooth pressure fields.
As for the solid phase, two methods are mainly used, i.e., the conventional SPH based on the
Eulerian kernel and TLSPH. In the conventional SPH approach, particle positions are updated at
each computational step, and kernel functions are computed based on these updated positions [8].
Consequently, the traditional SPH kernel function is often called the Eulerian because particles
can enter and exit its influence domain. This Eulerian kernel function is known to introduce a
well-recognized issue called tensile instability [34], leading to local particle clustering and the
formation of unphysical numerical fractures. To mitigate the tensile instability, a commonly used
correction method is the artificial pressure/stress technique [35, 36], which can effectively alle-
viate the issue. The tensile instability can be circumvented by calculating kernel functions using
reference particle positions [37]. The kernel function based on the reference configuration is the
Lagrangian kernel, and the corresponding SPH formulation is called TLSPH. TLSPH eliminates
tensile instability if computations are consistently based on the initial configuration [38, 39]. How-
ever, the original TLSPH method faces limitations in modelling scenarios involving significant
material distortion and separation due to negative Jacobians [20]. Moreover, the traditional SPH-
based frameworks provide better agreements than TLSPH when compared with the experimental
and other numerical results for finite deformation and failure of materials [40, 41, 20].
Most SPH-based FSI modelling uses different combinations of these methods [21, 22, 23]. De-
spite the success of SPH in modelling FSI problems, however, important issues on material damage
and fracture are yet to be addressed. Limited studies can be found in the literature on a stable, accu-
rate, efficient SPH framework for modelling FSI problems involving material damage and fracture
[42]. In this context, the research community increasingly embraces SPH and its extensions, pri-
marily due to their innate ability to model crack propagation [37, 43, 44]. Among these extensions,
the General Particle Dynamics (GPD) framework, built upon SPH, has gained widespread adop-
tion for simulating progressive failure in slopes and the fracturing of rocks [45, 46, 47]. Another
notable SPH extension is the pseudo-spring augmented SPH [48], which establishes connections
between each particle and its immediate neighbours through pseudo-springs. The advancement of
a crack front occurs when the pseudo-spring linking any two immediate neighbours is disrupted.
The crack paths can be modelled in this approach without refinement, enrichment or visibility cri-
teria. A slightly adapted version of this methodology has proven effective in simulating failures in
both brittle and ductile materials [18, 19].
This work presents a coupled WCSPH framework for fluid-structure interaction with deformable
structures undergoing damage and fracture. The WCSPH enhanced with numerical schemes to
improve accuracy and stability is used for simulating the fluid flow. A pseudo-spring analogy in
traditional SPH has been adopted for modelling crack initiation and propagation in structures. The
effectiveness of the proposed approach is demonstrated through several numerical illustrations.
The paper is organized as follows. Section 2 discusses the governing equations, their discretiza-
tion, boundary conditions and stability schemes for fluid simulation using WCSPH. Section 3
discusses the SPH method for solid deformation and the pseudo-spring analogy for modelling
material damage and fracture. Section 4 discusses the coupling strategy of WCSPH and pseudo-
3
spring-based SPH. Section 5 presents some numerical examples for verification and validation
purposes. The crack initiation and propagation in elastic obstacles with a notch due to water im-
pact is also demonstrated through an example. Finally, some conclusions are drawn in section
6.

2. Weakly compressible smoothed particle hydrodynamics (WCSPH) for fluid flow

The foundational equations governing the dynamic fluid flow encompass the principles of mass
conservation and momentum preservation principles. These equations encompass:

dρ ∂vβ
= −ρ β , (1)
dt ∂x
dvα 1 ∂p 1 ∂ταβ
=− + + gα , (2)
dt ρ ∂xα ρ ∂xα
where ρ is the material density. In the moving Lagrangian frame, we represent the time derivative
d
as . At an individual material point, we describe the spatial coordinates using the notation xα
dt
for the component indexed as α, while the velocity at this point is denoted as vα . ταβ denotes the α
and β elements of the viscous stress:

∂vα ∂vβ
!
αβ
τ = µf + , (3)
∂xβ ∂xα
where µ f is the dynamic viscosity of the fluid. gα is the α component of the body force. p is
the pressure, and we derive the value of p by employing a weakly compressible equation of state
model [10], formulated as follows in this study:
" !γ
ρ
#
p = p0 −1 , (4)
ρ0
c2 ρ
where γ = 7, ρ0 represents the reference material density and p0 = 0γ 0 with c0 representing the
speed of sound.
SPH, classified as a collocation method, divides the domain into a collection of material points,
commonly known as particles, whether they are distributed regularly or irregularly. The partial
differential equations associated with the conservation equations are converted into a set of equiv-
alent ordinary differential equations. These transformed equations are then solved using one of
the numerous numerical integration techniques. Field variables pertaining to a particle situated at
the material point xiα are determined by considering neighboring particles positioned at xαj . This
computation relies on the utilization of a kernel function denoted as W(q, h). Notably, this kernel
function serves as an approximation of the Dirac-delta function. The parameter q is defined as the
normalized distance between xiα and xαj , with this normalization being relative to the smoothing
length h i.e., q = (||xiα − xαj ||)/h. In this work, we use the Wendland C2 kernel function [49]:

4
 
(q + 0.5)(2 − q)4 ,
 if q ≤ 2
W(q, h) = αd 

(5)
0,
 otherwise
where αd = 7/(32πh2 ) in 2D. To maintain simplicity in our discussion, from this point onward, we
will denote the kernel function W(q, h) for a particle pair i and j as Wi j , and its derivative as Wi j,β .
The discretized forms of the conservation equations are as follows:
β
dρi X X mj xi j
= m j vβi j Wi j,β + δhc0 2 (ρi − ρ j ) β Wi j,β , (6)
dt j j
ρj ||xi − xβj ||2 + 0.01h2

dvαi X  ταβ ταβ

   
 X  pi p j 
αβ
m j  2 + 2  Wi j,β + gα ,
j
= m j  2 + 2 − πi j δ  Wi j,β −
i  (7)
dt j
ρi ρj j
ρi ρ j

where vβi j = vβi − vβj . There are two distinct terms on the right-hand side of the equation 6. The first
term signifies the SPH discretization of equation 1, while the second term introduces an additional
numerical diffusion component referred to as δ-SPH [31]. We use δ = 0.1 in this paper. The
δ-SPH term ensures a smooth pressure field in WCSPH simulations. πi j is the artificial correction
term and maintains numerical stability in the presence of shock. The following form is used in this
work [50]:

−β1 c̄i j µi j + β2 µ2i j


, if vαi j xiαj ≤ 0



πi j =  ρ̄

(8)

 ij

0, otherwise



where,
hvαi j xiαj
µi j = , (9)
||xiα − xαj ||2 + 0.01h2
where xiαj = xiα − xαj , c̄i j represents the average sound speed calculated across particles i and j and
ρ̄i j = 0.5(ρi + ρ j ).

2.1. No-slip boundary condition

To maintain the no-slip solid boundary condition along solid walls, we introduce boundary
particles that contain extrapolated information about velocity and pressure [11]. We represent the
solid wall boundaries using distinct boundary particles spanning 2h (Fig. 1). The particles at the
walls are assigned the same initial properties (inter-particle distance, particle mass and density)
as the fluid particles. The field variables of the solid boundary wall particles are extrapolated
from the adjacent fluid particles and remain fixed in their initial positions. These solid boundary
wall particles participate like regular particles for field variable computation of fluid particles.
The pressure values at the solid boundary wall particles are obtained using the information from
neighbouring fluid particles in the following form:

5
 2h

Support
domain

Fluid particle
Boundary wall particle

Figure 1: Diagram of the boundary treatment at the solid wall

p f W(xw f ) + (gβ − aβw ) f ρ f xwβ f W(xw f )

P P
f
pw = P , (10)
f W(xw f )

where the solid boundary wall particles are denoted by subscript w and f represents the fluid
particles. aβw represents the β component of the specified acceleration of the solid boundary wall
particles. The following equation is used to calculate the density of the solid boundary wall parti-
cles:
! γ1
pw
ρw = ρ0 +1 . (11)
p0

3. Pseudo-spring based SPH for solid deformation

The conservation equations for material deformation due to external loading are

dρ ∂vβ
= −ρ β , (12)
dt ∂x
dvα 1 ∂σαβ
= , (13)
dt ρ ∂xβ
where the Cauchy stress tensor’s component corresponding to indices α and β is symbolized as
σαβ . The above conservation equations are discretized as:
dρi X
= m j vβi j Wi j,β , (14)
dt j

6
 dvαi X  σαβ σαβ
 

m j  i2 + 2 − πi j δαβ − Paij δαβ  Wi j,β ,
j
= (15)
dt j
ρi ρj
where Paij is the artificial pressure correction term, and its purpose is to prevent the occurrence of
tensile instability, which occurs when particles tend to cluster together, leading to the development
of unrealistic cracks. This adjustment introduces a short-range repulsive force using the given
expression [35]:

 |Pri | |Pr j |  W(di j ) n̄

 " #
Pi j = γ  2 + 2 
a
, (16)
ρi ρj W(∆p)
where γ symbolizes the adjustment parameter, and n̄ is defined as W(0)/W(∆p), where ∆p signifies
the average particle spacing in the initial configuration.

3.1. Constitutive model for elastic structure

The Cauchy stress tensor, denoted as σαβ , consists of two main components: the hydrostatic
pressure p and the deviatoric stress S αβ (σαβ = S αβ − pδαβ ). We have employed a linear equation
of state to calculate the hydrostatic pressure in deformable solids [51] as p = K ρρ0 − 1 , K being
the bulk modulus. The rate of change of the deviatoric stress S αβ is calculated is determined by
the following equation:
!
1 αβ γγ
Ṡ = 2µ ϵ̇αβ − δ ϵ̇ + S αγ ωβγ + S γβ ωαγ ,
αβ
(17)
3
Here, µ is the shear modulus. The above Jaumann stress rate is used to ensure frame independence.
The strain rate tensor and spin tensor are represented as ϵ̇ αβ and ωαβ , respectively. These tensors
can be calculated as follows:
1  αβ βα  1  αβ βα 
ϵ̇ αβ = l +l , ωαβ = l −l , (18)
2 2
On the other hand, the velocity gradient tensor lαβ is determined by:
mj
lαβ
X
= − (vαi − vαj )Wi j,β . (19)
ES PH
j
ρj

3.2. Definition of immediate neighbour particles for approximation

The kernel functions employed in SPH exhibit their maximum values near the centre particle
of their compact support. As one moves away from this centre particle, the magnitude of these
functions rapidly diminishes. Consequently, particles located close to a reference particle denoted
as i have a significantly greater impact on the approximation than those positioned near the outer
boundary of the kernel support. Considering this behaviour, the field variables at particle i are
estimated by summing the contributions solely from its immediate neighbouring particles in our
work. Therefore, only those particles that can be directly connected to particle i through straight

7
lines, without intersecting any other particles within the domain, are considered for the approx-
imation [48, 18]. When employing a rectangular particle distribution, this criterion implies that
any interior particle is influenced by its eight nearest neighbours, a boundary particle is affected by
five nearest neighbours, and three nearest neighbours influence a corner particle. Here, a gradient
correction method [52] is employed to mitigate the truncation errors emerging due to the incom-
plete or partial support domain. In SPH, the inclusion of a gradient correction method serves the
purpose of attaining both zeroth-order consistency (C 0 consistency near boundary particles) and
first-order consistency (C 1 consistency in interior particles). In this study, we substitute Wi j,β with
Ŵi j, β in order to accomplish this, with Ŵi j, β being computed as follows::
X mj
Ŵi j,β = Bβα W i j,α with Bi = A −1
and Aβα
= − xiβj Wi j,α . (20)
i i i
j
ρ j

We checked the error caused by this reduction in the number of interacting particles by approxi-
mating the function sin πx
2
, 0 < x < 1 and its derivatives. This exercise revealed that the use of only
neighbouring particles introduced negligible error.

3.3. Pesudo-spring analogy in SPH

In our framework, it is presumed that the closest neighbouring particles are linked to the ith
particle through what we refer to as pseudo springs. These pseudo springs are introduced solely
for the purpose of modelling interactions between connecting particles and do not impart any
additional stiffness to the system. This technique is more detailed in [48] and [18]. These pseudo
springs are responsible for defining the level of interaction denoted as fi j between the connected
particles. Specifically, when the material connecting particles i and j is undamaged, the value of fi j
is set to 1. However, it is assumed that these pseudo springs will fail when certain predetermined
criteria are met, such as reaching a critical axial stress or strain along the i j line or some other
relevant parameter. When such failure occurs, we set the value of fi j to 0, and this change from 1
to 0 is considered permanent. Consequently, the presence of these permanently damaged or failed
pseudo springs allows for tracking the crack path within the domain. To accommodate the evolving
interactions between particles as a result of these pseudo springs, kernel functions utilized in SPH,
denoted as Ŵi j , along with their respective derivatives Ŵi j,β used in approximating field variables,
are replaced by modified versions incorporating the interaction level fi j . These modified functions
are expressed as fi j Ŵi j and fi j Ŵi j,β reflecting the changing influence of particle connections due
to the presence of damaged or failed pseudo springs. To visualize the path of a crack, we employ
fringe plots of a damage variable denoted as D. This variable is defined for a given particle, say
particle i, as the ratio of the count of pseudo springs for which fi j = 0 (indicating failure) to the
total count of initial pseudo springs connected to that particle. When D = 1, it signifies that all
the pseudo springs linked to particle i have experienced permanent failure or damage, essentially
representing the complete failure of particle i. On the other hand, values of D within the range
0 < D < 1 imply that the material associated with particle i has suffered partial damage. It is
important to note that a crack can propagate even when Di < 1, underscoring the idea that damage
can extend beyond individual particles, affecting their connections and interactions.

8
4. Coupling of WCSPH and Pseudo-spring based SPH
This section discusses the coupling strategy of the WCSPH and pseudo-spring SPH. In this
coupling methodology, we tackle the fluid flow problem by applying WCSPH, bolstered by in-
corporating δ-SPH techniques, as comprehensively discussed in Section 2. Concurrently, our ap-
proach to solving structural deformation and failure leverages the pseudo-spring SPH, elaborated
in Section 3. We employ particles with identical initial spacing for the discretisation to maintain a
seamless and harmonious treatment of fluid and solid phases.
An explicit contact force algorithm is essential for accurately simulating the complex multi-
body interactions between the fluid and deformable structure. In this study, we employ a soft re-
pulsive particle contact model [53]. This model incorporates a distance-dependent repulsive force,
characterized by a finite magnitude, acting upon particles (both fluid and deformable structure) as
they approach each other. This force is mathematically expressed as follows:
xiαj
Fiαj = 0.01c ζ f (η)
2
(21)
ri2j
ri j
η= (22)
0.75hi j
ri j
ζ =1− , 0 < ri j < ∆p (23)
∆d

if 0 < η ≤ 2/3,




2/3,
if 2/3 < η ≤ 1,

(2η − 1.5η2 ),


f (η) = 

(24)




0.5(2 − η2 ), if 1 < η ≤ 2,

0,
 otherwise.
The distance between two particles at the fluid-structure boundary, i.e., one fluid particle and an-
other particle from the deformable structure, is denoted by r. This softer repulsive force effectively
mitigates non-physical particle penetration while simultaneously reducing the occurrence of pres-
sure disturbances [53, 54]. While calculating the contact force and modelling the fluid-structure in-
teractions, the SPH particles from the deformable structural domain interact with the fluid particles
only through equations 21 - 24 and vice-versa. Finally, we add the interaction forces to the discrete
fluid and structure momentum conservation equations 7 and 15. A predictor-corrector integration
method is utilized to solve the discretized equations governing the fluid-structure interaction (FSI)
problem, with the time step determined through the Courant–Friedrichs–Lewy condition.

5. Numerical examples
Within this section, we present several numerical examples. We utilize the proposed cou-
pled WCSPH-pseudo-spring SPH method to simulate scenarios involving free-surface flow, elastic
solids undergoing significant deformation, and fluid-structure interactions with deformable struc-
tures. To assess the accuracy of our simulations, we compare the numerical results with analytical

9
solutions and experimental and numerical data available in the existing literature. In the last exam-
ple, we simulate the material damage and fracture in the structure due to water wave interaction.
For all the simulations, we utilize the WCSPH technique in conjunction with δ-SPH correction
to handle the fluid phase, while we adopt the pseudo-spring SPH approach to represent the de-
formable structure.

5.1. Dam break: collapse of a water column

The phenomenon involving the collapse of a water column was initially explored in [55] and
has since been extensively examined through numerical simulations. Additionally, an analytical
investigation of this problem was conducted in [56]. These studies have become the benchmark
tests routinely utilized to validate various computational frameworks simulating the free surface
flow of water. The test setup is shown in Fig. 2 with W = H = 0.057 m and L = 4H. The
water density is considered to be 1000 Kg/m3 . The dynamic viscosity coefficient (µ f ) is 0.05 Pa
s. Three sizes of inter-particle spacing are used in the simulations with ∆p = 0.00057 m, 0.0014
m and 0.0029 m. The rigid wall and water interaction is modelled through the boundary condition
described in section 2.1.

gravity

H
Water

Figure 2: Setup for the dam break test

The position of the water-front toe measured from the left wall is shown in Fig. 3a at different
time steps with the inter-particle resolutions and compared with the experimental results [55]. The
non-dimensional time coefficient is calculated as τ = √1 with g being the gravity force and the
H/g
x
non-dimensional distance is with x being the current position of the water-front toe measured
H
from the left wall. It can be observed that the present simulations agree well with the experimental
result. The positional time history of the water-front toe is compared with other results available
in the open literature in Fig. 3b with ∆p = 0.00057 m.
In Fig. 8, we present the contours illustrating the velocity and pressure distribution of the
water at different time steps. The simulation effectively captures the behaviour of free-surface flow
influenced by the gravitational force of the dam. The simulation notably depicts the dam breaking,
leading to water flow along the dry bed, culminating in an impact against a vertical rigid wall.

10
 (a) Effect of ∆p (b) Comparison with other results with ∆p = 0.00057 m

Figure 3: Time history of the water-front toe (measured from the left wall) and the effect of inter-particle distance on
the time history of the water-front toe in the dam break test (τ = √ t ).
H/g

Subsequently, the water rises, falls, and overturns backwards onto the underlying water. These
flow patterns and pressure distributions closely mirror findings from previous research [57, 22, 58].

(a) Time = 0.12 s (b) Time = 0.12 s

(c) Time = 0.19 s (d) Time = 0.19 s

(e) Time = 0.25 s (f) Time = 0.25 s

Figure 4: Contours of velocity magnitude and pressure distribution at 0.12, 0.19 and 0.25 s in dam break test

11
5.2. Oscillation of beam
In this example, we show through a transverse oscillation of a beam that the present formu-
lation is able to capture the deformation of deformable solids accurately. We consider an elastic
cantilever beam (Fig. 5) of length L = 10 m and thickness d = 1 m.

P d

Figure 5: Schematic sketch of the beam under transverse oscillation

2 4
The frequency of the oscillation is computed as ω2 = 12ρ(1−ν
Ed k
2 ) [36]; where, ρ = 7850kg/m is
3

the material density, E = 211GPa is the elastic modulus and ν = 0.3 is the Poisson’s ratio. Wave
number k is computed from the condition cos(kL)sin(kL) = −1; for the first mode, kL = 1.875.
Initially, the beam is set in motion with the following velocity function.
vy M {cos(kx) − cosh(kx)} − N {sin(kx) − sinh(kx)}
= Vf . (25)
c0 Q
where, c0 is the sound speed in the medium, V f is the transverse velocity set as V f = 0.05, M =
sin(kL) + sinh(kL), N = cos(kL) + cosh(kL) and Q = 2(cos(kL) sinh(kL) − sin(kL) cosh(kL)).
Simulations are performed with inter-particle spacing ∆p = 0.05 m and ∆p h
= 1.5. γ = 0.3 has
been used to suppress the tensile instability in the present problem. The numerically computed
time periods differ from the theoretical time period (0.114 s) by 7.2% (in the case of SPH with
γ = 0.3, the time period found to be 0.122 s). It can be concluded that the present SPH formulation
with γ = 0.3 yields results close to the analytical solutions.

5.3. Dam break - large deformation of an elastic gate

The investigation into the deformation of an elastic gate due to water pressure was undertaken
in [21], employing both experimental and numerical methods. This problem is further modelled
using different numerical techniques in [22, 58, 23]. The set-up of the dam break flow through
an elastic rubber gate is shown in Fig. 6. Initially, the water column is at rest, measuring 0.14 m
in height (H) and 0.1 m in width (W). On the other hand, the elastic rubber gate has dimensions
of 0.079 m in length (L) and 0.005 m in thickness (D). In this example, the water density is
1000 Kg/m3 , and the dynamic viscosity coefficient (µ f ) is taken as 0.05 Pa s. The density of the
elastic rubber gate is 1100 Kg/m3 . The elastic modulus is 12 MPa, and the Poisson’s ratio is 0.45.
The initial computational domain is discretized with ∆p = 0.0008 m.
The water column applies force to the deformable rubber gate securely clamped to a rigid
wall from above (see Fig. 6). After releasing the elastic rubber gate, the water initiates contact
with the gate and exits the tank through the gap between the gate and the unyielding bottom wall.
The contact between the elastic rubber gate and the water is modelled through the soft repulsive
particle contact model discussed in section 4, whereas the interaction between water and rigid wall
is modelled using the boundary condition described in section 2.1. Fig. 7 illustrates the temporal
12
 W

gravity
Rigid wall

Water
H
Elastic gate L

Figure 6: Setup of the dam break flow through an elastic gate

evolution of horizontal and vertical displacements observed at the free end of the gate. Our results
have been compared with the experimental data [21] and other numerical results [21, 22]. The
process unfolds in the following manner: initially, the water’s pressure pushes the elastic gate
aside, allowing the water to flow out. During the early stages, the horizontal displacement of
the elastic gate increases rapidly. As the water depth in the enclosure decreases, the pressure
force acting on the elastic gate diminishes, causing the gate to return towards its initial position
slowly. Overall, our simulation closely aligns with the experimental and numerical data, indicating
a significant level of agreement. We achieve a more favourable agreement in the early stage of the
gate opening, but our simulation tends to underestimate the displacements slightly as the gate
progresses toward closure.

Figure 7: Comparison of time histories of the horizontal and vertical displacements of the free end of the gate

The simulation frames with the present framework at specific time points are presented in Fig.
8 and compared with the experimental snapshots. It can be noted that the FSI coupling process
with nonlinear characteristics is effectively replicated. The pressure distribution is also shown, and
a consistent pressure distribution is observed. The simulation as a whole maintains stability, with
no occurrences of instability or simulation failure observed. The maximum stress is observed on
13
the inner side of the anchored gate’s end, where the maximum bending moment is concentrated.
(see Fig. 9).

5.4. Dam break flow impacting on flexible obstacle

In this segment, we undertake another numerical simulation to explore the dynamic interaction
of a deformable structure in a fluid-structure interaction (FSI) scenario. Specifically, we simulate
the scenario where a vertical water column collapses and strikes a flexible elastic wall, investigat-
ing the resulting complex dynamics. The system’s configuration is shown in Fig. 10.
The water column in this setup exhibits specific geometric parameters: its width, denoted as W,
measures 0.146 m, while its height, represented as H, measures 0.292 m. The gap between the two
vertical walls, which serves as the spatial constraint for the column, amounts to 4 times the width,
or 0.584 m. A deformable elastic plate occupies the central position within this confined space,
fixed at its lower end. The plate is positioned at a horizontal distance of L, equivalent to W, from
the approaching water column. The elastic plate has distinct dimensions, with a thickness denoted
as a measuring 0.012 m and a height denoted as b measuring 0.08 m. The computational domain
is discretized with ∆p = 0.0025 m. The initial state of the experiment involves the water column
being abruptly released, setting in motion its trajectory towards a collision with the stationary
elastic plate. The density values for the water and the deformable elastic obstacle are initially set
to be 1000 Kg/m3 and 2500 Kg/m3 . Furthermore, the deformable elastic obstacle is characterized
by an elastic modulus E of 106 N/m2 and a Poisson ratio ν of 0.
Fig. 11 comprehensively depicts various aspects of the scenario, showcasing the evolution
of the water pressure, free-surface profile, and the deformation of the elastic obstacle at distinct
time intervals. At the outset, the water flows freely, resembling a typical dam break scenario, with
the flow front exhibiting low pressure. However, as the water collides with the elastic obstacle, a
substantial surge in pressure becomes apparent, generating significant impact forces. This leads
to a considerable deformation in the elastic obstacle, instigating notable changes in the dynamics
of the flowing water. As the water progresses over the wall, the pressure gradually subsides,
eventually reaching a state of hydrostatic equilibrium. Concurrently, the elastic wall rebounds
from its deformed state. Notably, upon the initial impact of water on the tank’s rigid wall, a
localized high-pressure zone re-emerges. This method effectively captures the intricate interplay
between the fluid and the deformable structure, resulting in a qualitative agreement between the
numerical present simulations and the other observations from literature [22, 59, 42]. The stress
distribution in the deformable obstacle is also shown in Fig. 11. Following the impact around
0.2 s, the upstream face of the wall experiences tension, while the opposite face is subjected to
compression. The maximum stress can be found near the fixed support, consistent with other
literature findings. Also, there is no instability observed in the obstacle.
To further confirm the accuracy of the SPH–PD method, we analyzed the deflection process
occurring at the upper-left corner of the elastic plate over time. In Fig. 12, we present the variations
in horizontal displacement observed at the upper-left corner of the elastic wall. Furthermore,
we include numerical results from previous literature [22, 58, 60, 61], to enable a comparative
assessment. It is evident from the results that the current approach effectively anticipates and
replicates the overall response of the obstacle when subjected to hydrodynamic forces induced by
the collapsing water column.
14
 (a) Time = 0.04 s (b) Time = 0.04 s

(c) Time = 0.08 s (d) Time = 0.08 s

(e) Time = 0.12 s (f) Time = 0.12 s

(g) Time = 0.16 s (h) Time = 0.16 s

Figure 8: Qualitive comparison between experimental results [21] and present work at different time steps

15
 (a) Time = 0.16 s (b) Time = 0.16 s

Figure 9: Pressure and stress distribution in the water and elastic gate

gravity

Elastic obstacle
Water
H
L b

a
4W

Figure 10: Setup for water impact on elastic obstacle

16
 (a) Time = 0.0 s (b) Time = 0.2 s

(c) Time = 0.3 s (d) Time = 0.4 s

(e) Time = 0.5 s (f) Time = 0.6 s

(g) Time = 0.7 s (h) Time = 0.8 s

(i) Time = 0.9 s (j) Time = 1.0 s

Figure 11: Pressure and stress distribution at different time steps for water impact on elastic obstacle

17
 Figure 12: Comparison of time histories of the deflection of the free end of the elastic obstacle

5.5. Damage and fracture of an elastic obstacle due to water impact

gravity
Elastic obstacle
d
Water
H
L
b
l
a
4W

Figure 13: Setup for water impact on an elastic obstacle with an initial notch

We examine the interaction between a brittle obstacle and water. The specific geometric ar-
rangement for this case is detailed in Fig. 13. Here, the initial crack/ notch is made by deleting
the particles of a row. The obstacle shown in the diagram has an initial crack measuring a = 0.008
m in length, positioned and l = 0.025 m above the ground. Simultaneously, water is in a state of
descent due to the gravitational force. The other dimensions are as follows: H = 0.3 m, W = 0.15
m, L = W, b = 0.09 m and d = 0.03 m. The experiment begins with the sudden release of the
water column, initiating its path towards a collision with the elastic obstacle. Initially, the den-
sity of the water and the flexible elastic barrier are established at 1000 Kg/m3 and 2500 Kg/m3 ,
18
respectively. Additionally, the deformable elastic obstacle exhibits an elastic modulus denoted as
E with a value of 106 N/m2 , and a Poisson ratio represented as ν with a value of 0. The fracture
strain, ϵ f , is set at 0.05. Therefore, the interaction between a pair of particles i and j is stopped
when the strain in the connecting pseudo-spring exceeds the value of 0.05 ( fi j = 0 i f ϵ f > 0.05).
The failure process is assumed to be permanent in this simulation.
Fig. 14 illustrates the progressive changes in the water’s free surface, pressure contour patterns,
and crack propagation throughout the simulation. As observed, when the fluid initially interacts
with the obstacle, there is a significant surge in pressure within the FSI zone. Subsequently,
due to this heightened pressure, the obstacle deforms, and the strains in the particles increase.
After a specific duration (t ≈ 17.5 s), the accumulated strain surpasses the material’s fracture
strain threshold ϵ f = 0.05, leading to the initiation of crack propagation of the obstacle and its
eventual detachment from the weak area, i.e., the preexisting crack tip. It’s worth noting that
following the complete detachment of the upper section of the obstacle, the lower part exerts a jet
effect on the fluid, causing the water to follow a predefined path. This phenomenon is depicted
in the final snapshots of Fig. 14. Similar observations were made in [42]. The overall process of
Fluid-Structure Interaction (FSI) and fracture is consistent when a fine discretization, i.e., inter-
particle spacing (∆p), is used. It may be observed from Fig. 15 that fine resolution yields a better
representation of crack initiation, propagation and fracturing process.
In order to highlight the efficacy of the pseudo-spring analogy in modelling material damage
and subsequent cracking, we perform a simulation of the same set-up without the failure strain
(i.e., even if the strains in the pseudo-springs are greater than ϵ f , the interaction coefficient is kept
same fi j = 1). It can be seen from Fig. 16 that the crack does not initiate, and the elastic obstacle
remains undamaged, i.e., does not suffer any failure.

6. Conclusion
A computational framework for modelling large deformation and material damage and fail-
ure is proposed for fluid-structure interaction problems. In the integrated numerical approach, we
utilize a two-pronged strategy. Firstly, the fluid phase is simulated by employing the WCSPH
method, which includes a density diffusion term to enhance accuracy. The interaction between the
fluid phase and the rigid walls is modelled through specialized boundary particles designed to ex-
trapolate relevant variables. Secondly, for the solid phase, we implement a pseudo-spring analogy
in SPH, where the immediate neighbour particles are used for approximation. The pseudo-springs
help in modelling the material damage and subsequent crack propagation without requiring any
computationally intensive processes such as visibility criteria, particle splitting, etc. The interac-
tion between the moving fluid phase and the deformable solid structure/ obstacle is modelled by
a soft repulsive particle contact model. This approach results in the establishment of a cohesive
framework for effectively managing rigid wall boundaries and fluid-structure interactions.
The numerical results obtained in our study have been subjected to thorough comparisons with
analytical solutions, experimental data, and other existing numerical findings from the literature.
Our findings demonstrate the capability of accurately modelling free surface flow and dynamic
elastic problems without encountering instability. The validation of the approach was carried out
through the examination of different FSI scenarios involving deformable structures. Our numer-
19
 (a) Time = 0.15 s (b) Time = 0.15 s

(c) Time = 0.20 s (d) Time = 0.20 s

(e) Time = 0.22 s (f) Time = 0.22 s

(g) Time = 0.26 s (h) Time = 0.26 s

Figure 14: Pressure and Damage distribution at different time steps for water impact on an elastic obstacle
(∆p = 0.0025 m)

20
 (a) Time = 0.2 s (b) Time = 0.2 s

(c) Time = 0.26 s (d) Time = 0.26 s

Figure 15: Pressure and Damage distribution at different time steps for water impact on an elastic obstacle
(∆p = 0.001 m)

(a) Time = 0.2 s (b) Time = 0.6 s

Figure 16: Pressure and Damage distribution at different time steps for water impact on an elastic obstacle without
considering damage and fracturing (∆p = 0.0025 m)

21
ical outcomes exhibit good agreement with existing experimental, numerical and analytical data
from the literature, reaffirming the reliability of our method. We have also demonstrated in the
last numerical example that the proposed framework is capable of modelling material damage and
subsequent fracture under extreme hydrodynamic events. While our proposed method has shown
promising accuracy in preliminary assessments, it is important to acknowledge the limited avail-
ability of experiments in the existing literature that specifically address FSI problems involving
deformable structures exhibiting material damage and fracture. To ensure a comprehensive evalu-
ation of the precision and reliability of our approach, further investigations are necessary. In light
of this, we plan to conduct dedicated laboratory experiments as part of our validation process in fu-
ture works. These experiments will provide valuable real-world data and insights that can help us
refine and enhance the accuracy of our method in modelling FSI scenarios involving deformable
structures experiencing material damage and fracture. However, the proposed framework has
shown great stability and efficiency (when compared with the existing numerical schemes) and
holds the potential to become a widely employed method for modelling finite deformation and
material damage and failure in FSI problems.

7. Acknowledgments
The author acknowledges the computational support provided as a part of the IIT Delhi NFS
grant, on which the simulations have been run.

References
[1] C. Hu, M. Kashiwagi, Two-dimensional numerical simulation and experiment on strongly nonlinear wave–body
interactions, Journal of marine science and technology 14 (2009) 200–213.
[2] A. Slone, K. Pericleous, C. Bailey, M. Cross, Dynamic fluid–structure interaction using finite volume unstruc-
tured mesh procedures, Computers & structures 80 (5-6) (2002) 371–390.
[3] M. Heil, An efficient solver for the fully coupled solution of large-displacement fluid–structure interaction prob-
lems, Computer Methods in Applied Mechanics and Engineering 193 (1-2) (2004) 1–23.
[4] T. Belytschko, On difficulty levels in non linear finite element analysis of solids, Iacm Expressions 2 (1996) 6–8.
[5] T.-P. Fries, T. Belytschko, The extended/generalized finite element method: an overview of the method and its
applications, International journal for numerical methods in engineering 84 (3) (2010) 253–304.
[6] R. A. Gingold, J. J. Monaghan, Smoothed particle hydrodynamics: theory and application to non-spherical stars,
Monthly notices of the royal astronomical society 181 (3) (1977) 375–389.
[7] L. B. Lucy, A numerical approach to the testing of the fission hypothesis, Astronomical Journal, vol. 82, Dec.
1977, p. 1013-1024. 82 (1977) 1013–1024.
[8] L. D. Libersky, A. G. Petschek, Smooth particle hydrodynamics with strength of materials, in: Advances in the
Free-Lagrange Method Including Contributions on Adaptive Gridding and the Smooth Particle Hydrodynamics
Method: Proceedings of the Next Free-Lagrange Conference Held at Jackson Lake Lodge, Moran, WY, USA
3–7 June 1990, Springer, 2005, pp. 248–257.
[9] M. Liu, G. Liu, Smoothed particle hydrodynamics (sph): an overview and recent developments, Archives of
computational methods in engineering 17 (2010) 25–76.
[10] J. J. Monaghan, Simulating free surface flows with sph, Journal of computational physics 110 (2) (1994) 399–
406.
[11] S. Adami, X. Y. Hu, N. A. Adams, A generalized wall boundary condition for smoothed particle hydrodynamics,
Journal of Computational Physics 231 (21) (2012) 7057–7075.

22
[12] C. A. D. Fraga Filho, C. Peng, M. R. Ibne Islam, C. McCabe, S. Baig, G. V. Durga Prasad, Implementation
of three-dimensional physical reflective boundary conditions in mesh-free particle methods for continuum fluid
dynamics: Validation tests and case studies, Physics of Fluids 31 (10) (2019).
[13] H. H. Bui, R. Fukagawa, K. Sako, S. Ohno, Lagrangian meshfree particles method (sph) for large deformation
and failure flows of geomaterial using elastic–plastic soil constitutive model, International journal for numerical
and analytical methods in geomechanics 32 (12) (2008) 1537–1570.
[14] J. Chen, Y. Xin, Practical method of conical cam outline expansion, Chinese Journal of Mechanical Engineering-
English Edition 24 (1) (2011) 127.
[15] C. Peng, W. Wu, H.-s. Yu, C. Wang, A sph approach for large deformation analysis with hypoplastic constitutive
model, Acta Geotechnica 10 (2015) 703–717.
[16] C. Peng, X. Guo, W. Wu, Y. Wang, Unified modelling of granular media with smoothed particle hydrodynamics,
Acta Geotechnica 11 (2016) 1231–1247.
[17] M. Liu, G. Liu, K. Lam, Z. Zong, Smoothed particle hydrodynamics for numerical simulation of underwater
explosion, Computational mechanics 30 (2003) 106–118.
[18] M. R. I. Islam, S. Chakraborty, A. Shaw, S. Reid, A computational model for failure of ductile material under
impact, International Journal of Impact Engineering 108 (2017) 334–347.
[19] M. R. I. Islam, A. Shaw, Pseudo-spring sph simulations on the perforation of metal targets with different damage
models, Engineering Analysis with Boundary Elements 111 (2020) 55–77.
[20] M. R. I. Islam, C. Peng, P. K. Patra, A comparison of numerical stability for esph and tlsph for dynamic brittle
fracture, Theoretical and Applied Fracture Mechanics (2023) 104052.
[21] C. Antoci, M. Gallati, S. Sibilla, Numerical simulation of fluid–structure interaction by sph, Computers &
structures 85 (11-14) (2007) 879–890.
[22] A. Rafiee, K. P. Thiagarajan, An sph projection method for simulating fluid-hypoelastic structure interaction,
Computer Methods in Applied Mechanics and Engineering 198 (33-36) (2009) 2785–2795.
[23] A. Salehizadeh, A. Shafiei, A coupled isph-tlsph method for simulating fluid-elastic structure interaction prob-
lems, Journal of Marine Science and Application 21 (1) (2022) 15–36.
[24] S. Marrone, A. Colagrossi, A. Di Mascio, D. Le Touzé, Prediction of energy losses in water impacts using
incompressible and weakly compressible models, Journal of Fluids and Structures 54 (2015) 802–822.
[25] D. Meringolo, A. Colagrossi, S. Marrone, F. Aristodemo, On the filtering of acoustic components in weakly-
compressible sph simulations, Journal of Fluids and Structures 70 (2017) 1–23.
[26] G. Pahar, A. Dhar, Modeling free-surface flow in porous media with modified incompressible sph, Engineering
Analysis with Boundary Elements 68 (2016) 75–85.
[27] G. Pahar, A. Dhar, Coupled incompressible smoothed particle hydrodynamics model for continuum-based mod-
elling sediment transport, Advances in water resources 102 (2017) 84–98.
[28] N. Trask, K. Kim, A. Tartakovsky, M. Perego, M. L. Parks, A highly-scalable implicit sph code for simu-
lating single-and multi-phase flows in geometrically complex bounded domains., Tech. rep., Sandia National
Lab.(SNL-NM), Albuquerque, NM (United States) (2015).
[29] X. Guoa, B. D. Rogersb, Developing highly scalable 3-d incompressible sph, ARCHER Community, ISPH
embedded CSE Report (2015).
[30] X. Guo, B. D. Rogers, S. Lind, P. K. Stansby, New massively parallel scheme for incompressible smoothed
particle hydrodynamics (isph) for highly nonlinear and distorted flow, Computer Physics Communications 233
(2018) 16–28.
[31] D. Molteni, A. Colagrossi, A simple procedure to improve the pressure evaluation in hydrodynamic context
using the sph, Computer Physics Communications 180 (6) (2009) 861–872.
[32] S. Marrone, M. Antuono, A. Colagrossi, G. Colicchio, D. Le Touzé, G. Graziani, δ-sph model for simulating
violent impact flows, Computer Methods in Applied Mechanics and Engineering 200 (13-16) (2011) 1526–1542.
[33] A. Ferrari, M. Dumbser, E. F. Toro, A. Armanini, A new 3d parallel sph scheme for free surface flows, Computers
& Fluids 38 (6) (2009) 1203–1217.
[34] J. W. Swegle, D. L. Hicks, S. W. Attaway, Smoothed particle hydrodynamics stability analysis, Journal of
computational physics 116 (1) (1995) 123–134.
[35] J. J. Monaghan, Sph without a tensile instability, Journal of computational physics 159 (2) (2000) 290–311.

23
[36] J. P. Gray, J. J. Monaghan, R. Swift, Sph elastic dynamics, Computer methods in applied mechanics and engi-
neering 190 (49-50) (2001) 6641–6662.
[37] R. Vignjevic, J. R. Reveles, J. Campbell, Sph in a total lagrangian formalism, CMC-Tech Science Press- 4 (3)
(2006) 181.
[38] T. Belytschko, Y. Guo, W. Kam Liu, S. Ping Xiao, A unified stability analysis of meshless particle methods,
International Journal for Numerical Methods in Engineering 48 (9) (2000) 1359–1400.
[39] Y. Vidal, J. Bonet, A. Huerta, Stabilized updated lagrangian corrected sph for explicit dynamic problems, Inter-
national journal for numerical methods in engineering 69 (13) (2007) 2687–2710.
[40] M. R. I. Islam, K. V. Ganesh, P. K. Patra, On the equivalence of eulerian smoothed particle hydrodynamics,
total lagrangian smoothed particle hydrodynamics and molecular dynamics simulations for solids, Computer
Methods in Applied Mechanics and Engineering 391 (2022) 114591.
[41] M. R. I. Islam, W. Zhang, C. Peng, Large deformation analysis of geomaterials using stabilized total lagrangian
smoothed particle hydrodynamics, Engineering Analysis with Boundary Elements 136 (2022) 252–265.
[42] M. N. Rahimi, G. Moutsanidis, An sph-based fsi framework for phase-field modeling of brittle fracture under
extreme hydrodynamic events, Engineering with Computers (2023) 1–35.
[43] M. N. Rahimi, G. Moutsanidis, Modeling dynamic brittle fracture in functionally graded materials using hyper-
bolic phase field and smoothed particle hydrodynamics, Computer Methods in Applied Mechanics and Engi-
neering 401 (2022) 115642.
[44] Y. Zhao, Z. Zhou, J. Bi, C. Wang, Simulation of brittle fractures using energy-bond-based smoothed particle
hydrodynamics, International Journal of Mechanical Sciences 248 (2023) 108236.
[45] X. Zhou, Y. Zhao, Q. Qian, A novel meshless numerical method for modeling progressive failure processes of
slopes, Engineering Geology 192 (2015) 139–153.
[46] S. Zhai, X. Zhou, J. Bi, N. Xiao, The effects of joints on rock fragmentation by tbm cutters using general particle
dynamics, Tunnelling and Underground Space Technology 57 (2016) 162–172.
[47] X. Zhou, X. Zhang, The 3d numerical simulation of damage localization of rocks using general particle dynam-
ics, Engineering Geology 224 (2017) 29–42.
[48] S. Chakraborty, A. Shaw, A pseudo-spring based fracture model for sph simulation of impact dynamics, Inter-
national Journal of Impact Engineering 58 (2013) 84–95.
[49] H. Wendland, Piecewise polynomial, positive definite and compactly supported radial functions of minimal
degree, Advances in computational Mathematics 4 (1995) 389–396.
[50] J. J. Monaghan, R. A. Gingold, Shock simulation by the particle method sph, Journal of computational physics
52 (2) (1983) 374–389.
[51] S. Eliezer, A. K. Ghatak, H. Hora, E. Teller, An introduction to equations of state: theory and applications,
Cambridge University Press Cambridge, 1986.
[52] J. Chen, J. Beraun, T. Carney, A corrective smoothed particle method for boundary value problems in heat
conduction, International Journal for Numerical Methods in Engineering 46 (2) (1999) 231–252.
[53] Z. Zhang, K. Walayat, J. Chang, M. Liu, Meshfree modeling of a fluid-particle two-phase flow with an improved
sph method, International Journal for Numerical Methods in Engineering 116 (8) (2018) 530–569.
[54] M. Liu, Z. Zhang, Smoothed particle hydrodynamics (sph) for modeling fluid-structure interactions, Science
China Physics, Mechanics & Astronomy 62 (2019) 1–38.
[55] J. Martin, W. Moyce, Part iv. an experimental study of the collapse of liquid columns on a rigid horizontal plane,
Philosophical Transactions of the Royal Society of London Series A 244 (882) (1952) 312–324.
[56] A. Knight, The propagation of water waves, Journal of the Association of German Engineers 36 (33) (1892)
947–954.
[57] A. Colagrossi, A meshless lagrangian method for free-surface and interface flows with fragmentation, These,
Universita di Roma (2005).
[58] W.-K. Sun, L.-W. Zhang, K. Liew, A smoothed particle hydrodynamics–peridynamics coupling strategy for
modeling fluid–structure interaction problems, Computer Methods in Applied Mechanics and Engineering 371
(2020) 113298.
[59] L. Zhan, C. Peng, B. Zhang, W. Wu, A stabilized tl–wc sph approach with gpu acceleration for three-dimensional
fluid–structure interaction, Journal of Fluids and Structures 86 (2019) 329–353.

24
[60] E. Walhorn, A. Kölke, B. Hübner, D. Dinkler, Fluid–structure coupling within a monolithic model involving
free surface flows, Computers & structures 83 (25-26) (2005) 2100–2111.
[61] S. R. Idelsohn, J. Marti, A. Limache, E. Oñate, Unified lagrangian formulation for elastic solids and incom-
pressible fluids: application to fluid–structure interaction problems via the pfem, Computer Methods in Applied
Mechanics and Engineering 197 (19-20) (2008) 1762–1776.

25