Comput Mech (2011) 48:307–318

DOI 10.1007/s00466-011-0617-2

ORIGINAL PAPER

Possibilities of the particle finite element method

for fluid–soil–structure interaction problems
Eugenio Oñate · Miguel Angel Celigueta ·
Sergio R. Idelsohn · Fernando Salazar ·
Benjamín Suárez

Received: 28 April 2011 / Accepted: 9 June 2011 / Published online: 8 July 2011
© Springer-Verlag 2011

Abstract We present some developments in the particle 1 Introduction

finite element method (PFEM) for analysis of complex cou-
pled problems in mechanics involving fluid–soil–structure The analysis of problems involving the interaction of flu-
interaction (FSSI). The PFEM uses an updated Lagrangian ids, soil/rocks and structures is of relevance in many areas
description to model the motion of nodes (particles) in both of engineering. Examples are common in the study of land-
the fluid and the solid domains (the later including soil/rock slides and their effect on reservoirs and adjacent structures,
and structures). A mesh connects the particles (nodes) defin- off-shore and harbour structures under large waves, construc-
ing the discretized domain where the governing equations for tions hit by floods and tsunamis, soil erosion and stability of
each of the constituent materials are solved as in the standard rock-fill dams in overspill situations, etc.
FEM. The stabilization for dealing with an incompressibil- These studies can be considered as an extension of the
ity continuum is introduced via the finite calculus method. so-called fluid–structure interaction (FSI) problems [46].
An incremental iterative scheme for the solution of the non Typical difficulties of FSI analysis in free surface flows using
linear transient coupled FSSI problem is described. The pro- the FEM both the Eulerian or ALE formulation include the
cedure to model frictional contact conditions and material treatment of the convective terms and the incompressibility
erosion at fluid–solid and solid–solid interfaces is described. constraint in the fluid equations, the modelling and tracking
We present several examples of application of the PFEM to of the free surface in the fluid, the transfer of information
solve FSSI problems such as the motion of rocks by water between the fluid and the moving solid domains via the con-
streams, the erosion of a river bed adjacent to a bridge founda- tact interfaces, the modeling of wave splashing, the possibil-
tion, the stability of breakwaters and constructions sea waves ity to deal with large motions of multi-bodies within the fluid
and the study of landslides. domain, the efficient updating of the finite element meshes
for both the structure and the fluid, etc. Examples of 3-D
Keywords Particle finite element method · analysis of FSI problems using ALE and space-time FEM
Fluid–soil–structure · Interaction are reported in [4,6,26,27,31,34,40].
Most of the above problems disappear if a Lagrangian
description is used to formulate the governing equations of
both the solid and the fluid domains. In the Lagrangian for-
S. R. Idelsohn is a ICREA Professor at CIMNE. mulation the motion of the individual particles are followed
and, consequently, nodes in a finite element mesh can be
E. Oñate (B) · M. A. Celigueta · S. R. Idelsohn · F. Salazar
viewed as moving material points (hereforth called “parti-
International Center for Numerical Methods in Engineering
(CIMNE), Campus Norte UPC, Gran Capitán s/n, cles”). Hence, the motion of the mesh discretizing the total
08034 Barcelona, Spain domain (including both the fluid and solid parts) is followed
e-mail: onate@cimne.upc.edu during the transient solution.
URL: http://www.cimne.com
A powerful Lagrangian method for FSI analysis is the
E. Oñate · M. A. Celigueta · B. Suárez so-called Soboran Grid CIP technique, which has been suc-
Universitat Politècnica de Catalunya (UPC), Barcelona, Spain cessfully applied to different class of 3-D problems [44].

123
308 Comput Mech (2011) 48:307–318

The authors have successfully developed in previous 2 The basis of the particle finite element method
works a particular class of Lagrangian formulation for solv-
ing problems involving complex interactions between (free Let us consider a domain containing both fluid and solid sub-
surface fluids) and solids. The method, called the particle domains (the solid subdomain may include soil/rock mate-
finite element method (PFEM, www.cimne.com/pfem), treats rials and/or structural elements). The moving fluid particles
the mesh nodes in the fluid and solid domains as particles interact with the solid boundaries thereby inducing the defor-
which can freely move and even separate from the main mation of the solid which in turn affects the flow motion and,
fluid domain representing, for instance, the effect of water therefore, the problem is fully coupled.
drops. A mesh connects the nodes discretizing the domain In the PFEM, both the fluid and the solid domains are mod-
where the governing equations are solved using a stabilized elled using an updated Lagrangian formulation [47]. That is,
FEM. all variables are assumed to be known in the current config-
An advantage of the Lagrangian formulation is that the uration at time t. The new set of variables in both domains
convective terms disappear from the fluid equations [11,48]. are sought for in the next or updated configuration at time
The difficulty is however transferred to the problem of ade- t + t. The finite element method (FEM) is used to solve
quately (and efficiently) moving the mesh nodes. We use a the equations of continuum mechanics for each of the sub-
mesh regeneration procedure blending elements of different domains. Hence a mesh discretizing these domains must be
shapes using an extended Delaunay tessellation with special generated in order to solve the governing equations for each
shape functions [17,19]. The theory and applications of the subdomain in the standard FEM fashion.
PFEM are reported in [2,7,10,18,20,21,23,26,32,34–39]. The quality of the numerical solution depends on the dis-
The FEM solution of (incompressible) fluid flow prob- cretization chosen as in the standard FEM. Adaptive mesh
lem implies solving the momentum and incompressibility refinement techniques can be used to improve the solution
equations. This is not a simple problem as the incompress- in zones where large motions of the fluid or the structure
ibility condition limits the choice of the FE approximations occur.
for the velocity and pressure to overcome the well known
div-stability condition [11,48]. In our work we use a stabi- 2.1 Basic steps of the PFEM
lized mixed FEM based on the finite calculus (FIC) approach
which allows for a linear approximation for the velocity and For clarity purposes, we will define the collection or cloud
pressure variables [15,29–31,33,34]. Among the other sta- of nodes (C) pertaining to the fluid and solid domains, the
bilized FEM with similar features we mention the PSPG volume (V) defining the analysis domain for the fluid and the
method [41], multiscale methods [3,6,8,9] and the CBS solid and the mesh (M) discretizing both domains.
method [9,48]. A typical solution with the PFEM involves the following
The aim of this paper is to describe recent advances of steps.
the PFEM for fluid–soil–structure interaction (FSSI) prob-
lems. These problems are of relevance in many areas of civil, 1. The starting point at each time step is the cloud of points
hydraulic, marine and environmental engineering, among in the fluid and solid domains. For instance n C denotes
others. It is shown that the PFEM provides a general analysis the cloud at time t = tn (Fig. 1).
methodology for treat such complex problems in a simple 2. Identify the boundaries for both the fluid and solid
and efficient manner. domains defining the analysis domain n V in the fluid
The layout of the paper is the following. In the next section, and the solid. This is an essential step as some bound-
the key ideas of the PFEM are outlined. Next the basic equa- aries (such as the free surface in fluids) may be severely
tions for a compressible/incompressible continuum using a distorted during the solution, including separation and
Lagrangian description and the FIC formulation are schemat- re-entering of nodes. The Alpha Shape method [12] is
ically presented. Then an algorithm for the transient solution used for the boundary definition.
is briefly described. The treatment of the coupled FSSI prob- 3. Discretize the fluid and solid domains with a finite ele-
lem and the methods for mesh generation and for identifi- ment mesh n M. In our work we use an innovative mesh
cation of the free surface nodes are outlined. The procedure generation scheme based on the extended Delaunay
for treating the frictional contact interaction between fluid, tessellation [17,19,20].
soil and structure interfaces is explained. We present several 4. Solve the coupled Lagrangian equations of motion for
examples of application of the PFEM to solve FSSI problems the fluid and the solid domains. Compute the state vari-
such as the motion of rocks by water streams, the erosion ables in both domains at the next (updated) configuration
of a river bed adjacent to a bridge foundation, the stability of for t + t: velocities, pressure and viscous stresses in
breakwaters and constructions under sea waves and the study the fluid and displacements, stresses and strains in the
of landslides falling into reservoirs. solid.

123
Comput Mech (2011) 48:307–318 309

Fig. 1 Sequence of steps to

Initial “cloud” ofnodes
update a “cloud” of nodes
representing a domain
containing a fluid and a solid
part from time n (t = tn ) to time → n
Flying Sub-domains

n + 2 (t = tn + 2t) Γ

→
Fixed Domain
Mesh
boundary

→ Cloud

n ,
.
n , n , n ,n ε , n ε , n σ
. →
nΓ

Mesh
→
Fixed
boundary Domain

→ n+1 ,
.
n+1 , n+1 , n+1 ,n+1ε , n+1ε , n+1σ
etc…
Cloud

5. Move the mesh nodes to a new position n+1 C where n+1 compression) ρ and K are the density and bulk modulus of
denotes the time tn + t, in terms of the time increment the material, respectively, bi and σi j are the body forces and
size. This step is typically a consequence of the solution the (Cauchy) stresses. Eqs. (1) and (2) are completed with
process of step 4. the constitutive relationships:
6. Go back to step 1 and repeat the solution process for the
next time step to obtain n+2 C (Fig. 1). Incompressible continuum

t+1
σi j = 2με̇i j − t+1 pδi j (3)
3 FIC/FEM formulation for a Lagrangian continuum
Compressible/quasi-incompressible continuum
3.1 Governing equations

The equations to be solved are the standard ones in continuum t+1

si j = t σ̂i j + 2με̇i j + λε̇ii δi j (4a)
mechanics, written in the Lagrangian frame of reference:
where σ̂i j are the component of the stress tensor [σ̂ ]
Momentum
1 T
[σ̂ ] = F SF (4b)
J
∂vi ∂σi j
ρ = + bi in V (1) where S is the second Piola–Kirchhoff stress tensor, F is the
∂t ∂x j deformation gradient tensor and J = det F [22,47]. Param-
eters μ and λ take the following values for a fluid or solid
Pressure–velocity relationship material:

1 ∂p ∂vi Fluid: μ: viscosity; λ = t K − 2μ3

− =0 in V (2) t G 2Gνt
K ∂t ∂ xi Solid: μ = ;λ= , where ν is the Poisson
J J (1 − 2ν)
In above equations vi is the velocity along the ith global ration, G is the shear modulus and t the time
(cartesian) axis, p is the pressure (assumed to be positive in increment.

123
310 Comput Mech (2011) 48:307–318

In Eqs. (3) and (4a), (4b), si j are the deviatoric stresses, Box 1 Basic PFEM algorithm for a Lagrangian continuum
ε̇i j is the rate of deformation, μ is the viscosity and δi j is the 1. LOOP OVER TIME STEPS, t = 1, NTIME Known values
Kronecker delta. t (·) denotes values at time t. π , t T̄ , t μ, t f, t q, t C, t V, t M
t x̄, t v̄, t p̄, t π̄
Indexes in Eqs. (1)–(4a), (4b) range from i, j = 1, n d ,
2. LOOP OVER NUMBER OF ITERATIONS, i = 1, NITER
where n d is the number of space dimensions of the problem
• Compute nodal velocities by solving Eq. (8)
(i.e. n d = 2 for 2-D problems). These equations are com-  
1 1 t
pleted with the standard boundary conditions of prescribed M + K t+1 v̄i+1 = t+1 f + G t+1 p̄i + M v̄
t t
velocities and surface tractions in the mechanical problem • Compute nodal pressures from Eq. (9)
 
[11,36,47,48]. 1 1 t
πi +
− LM̄ t+1 p̄i+1 = GT t+1 v̄ i+1 + Qt+1π̄ M̄ p̄
t t
3.2 Discretization of the equations • Compute nodal pressure gradient projections from Eq. (10)
 T t+1 i+1
π i+1 = −M̂−1
n+1π̄
D Q p̄ , M̂ D = diag M̂ D
A key problem in the numerical solution of Eqs. (1)–(4a), (4b) • Update position of analysis domain nodes:
is the satisfaction of the mass balance condition for the t+t x̄i+1 = t xi + t+t vi+1 t
incompressible case (i.e. K = ∞ in Eq. (2)). A number Define new “cloud” of nodes t+1 C i+1
of procedures to solve his problem exist in the finite ele- • Update strain rate and strain values
ment literature [11,48]. In our approach we use a stabilized • Update stress values
formulation based in the so-called finite calculus procedure Check convergence → NO → Next iteration i → i + 1
[15,29–31,33,34]. The essence of this method is the solution ↓ YES
of a modified mass balance equation which is written as
Next time step t → t + 1
 
∂vi  ∂q ∂ p • Identify new analysis domain boundary: t+1 V
3
1 ∂p
− − τ + πi = 0 (5) • Generate mesh:t+1 M
K ∂t ∂ xi ∂ xi ∂ xi
i=1
Go to 1
where q are weighting functions, τ is a stabilization param-
eter given by [34]
 
2ρ|v| 8μ −1
τ= + 2 (6) Momentum
h 3h
In the above, h is a characteristic length of each finite ele- Mv̄˙ + Kv̄ − Gp̄ = f (8)
ment and |v| is the modulus of the velocity vector. In Eq. (5)
πi are auxiliary pressure projection variables chosen so as to Pressure–velocity relationship
ensure that the second term in Eq. (5) can be interpreted as
weighted sum of the residuals of the momentum equations M̄p̄˙ − Gv̄ − Lp̄ − Qπ̄
π =0 (9)
and therefore it vanishes for the exact solution. The set of
governing equations is completed by adding the following Pressure gradient projection
constraint equation [32,36]
   π + QT p̄ = 0
M̂π̄ (10)
∂p
τ wi +πi d V = 0 i = 1, n d (no sum in )i (7)
∂ xi ¯ denotes nodal variables, (·)
In Eqs. (8)–(10) (·) ˙¯ = ∂ (·).
¯
V ∂t
The different matrices and vectors are given in [22,34,36].
where wi are arbitrary weighting functions. The solution in time of Eqs. (8)–(10) can be performed
The rest of the integral equations are obtained by apply- using any time integration scheme typical of the updated
ing the standard weighted residual technique to the governing Lagrangian FEM [36,47]. A basic algorithm following the
Eqs. (1)–(3) and (5) and the corresponding boundary condi- conceptual process described in Sect. 2 is presented in Box 1.
tions [11,22,48].
We interpolate next in the standard finite element fashion 4 Generation of a new mesh
the set of problem variables. For 3-D problems these are the
three velocities vi , the pressure p, the temperature T and the One of the key points for the success of the PFEM is the fast
three pressure gradient projections πi . In our work we use regeneration of a mesh at every time step on the basis of the
equal order linear interpolation for all variables over meshes position of the nodes in the space domain. Indeed, any fast
of 3-noded triangles (in 2-D) and 4-noded tetrahedra (in 3-D). meshing algorithm can be used for this purpose. In our work
The resulting set of discretized equations using the standard the mesh is generated at each time step using the so-called
Galerkin technique has the following form extended Delaunay tessellation (EDT) presented in [17,19].

123
Comput Mech (2011) 48:307–318 311

The CPU time required for meshing grows linearly with We emphasize that the main difference between the PFEM
the number of nodes. The CPU time for solving the equations and the classical FEM is just the remeshing technique and
exceeds that required for meshing as the number of nodes the identification of the domain boundary at each time step.
increases. This situation has been found in all the problems
solved with the PFEM. As a general rule for large 3D prob-
lems meshing consumes around 15% of the total CPU time
6 Treatment of contact conditions in the PFEM
for each time step, while the solution of the equations (with
typically 3 iterations to reach convergence within a time step)
6.1 Contact between the fluid and a fixed boundary
and the assembling of the system consume approximately 70
and 15% of the CPU time for each time step, respectively.
The condition of prescribed velocities at the fixed bound-
These figures refer to solutions obtained in a standard single
aries in the PFEM is applied in strong form to the boundary
processor Pentium IV PC for all the computations and prove
nodes. These nodes might belong to fixed external bound-
that the generation of the mesh has an acceptable cost in
aries or to moving boundaries linked to the interacting solids.
the PFEM. The cost of remeshing is similar to that reported
Contact between the fluid particles and the fixed boundaries
in [24]. Indeed considerable speed can be gained using par-
is accounted for by the incompressibility condition which
allel computation techniques.
naturally prevents the fluid nodes to penetrate into the solid
boundaries [32,36].

5 Identification of boundary surfaces 6.2 Contact between solid–solid interfaces

One of the main tasks in the PFEM is the correct definition of The contact between two solid interfaces is simply treated by
the boundary domain. Boundary nodes are sometimes explic- introducing a layer of contact elements between the two inter-
itly identified. In other cases, the total set of nodes is the only acting solid interfaces. This layer is automatically created
information available and the algorithm must recognize the during the mesh generation step by prescribing a minimum
boundary nodes. distance (h c ) between two solid boundaries. If the distance
In our work we use an extended Delaunay partition for exceeds the minimum value (h c ) then the generated ele-
recognizing boundary nodes [19]. Considering that the nodes ments are treated as fluid elements. Otherwise the elements
follow a variable h(x) distribution, where h(x) is typically are treated as contact elements where a relationship between
the minimum distance between two nodes. All nodes on an the tangential and normal forces and the corresponding
empty sphere with a radius greater than αh, are considered displacement is introduced (Fig. 2).
as boundary nodes. In practice α is a parameter close to, but This algorithm has proven to be very effective and it allows
>one. Values of α ranging between 1.3 and 1.5 have been to identifying and modeling complex frictional contact con-
found to be optimal in all examples analyzed. This criterion ditions between two or more interacting bodies moving in
is coincident with the Alpha Shape concept [12]. water in an extremely simple manner.
Once a decision has been made concerning which nodes
are on the boundaries, the boundary surface is defined by all
the polyhedral surfaces (or polygons in 2-D) having all their
nodes on the boundary and belonging to just one polyhedron.
The method described also allows one to identify isolated
fluid particles outside the main fluid domain. These particles
are treated as part of the external boundary where the pressure
is fixed to the atmospheric value. We recall that each particle
is a material point characterized by the density of the solid
or fluid domain to which it belongs. The mass which is lost
when a boundary element is eliminated due to departure of a
node from the main analysis domain is again regained when
the “flying” node falls down and a new boundary element is
created by the Alpha Shape algorithm.
The boundary recognition method is also useful for detect-
ing contact conditions between the fluid domain and a fixed
boundary, as well as between different solids interacting with Fig. 2 Modelling of contact conditions at a solid–solid interface with
each other as detailed in the next section. the PFEM

123
312 Comput Mech (2011) 48:307–318

This algorithm can also be used effectively to model

frictional contact conditions between rigid or elastic solids
in structural mechanics applications [7,36].

7 Modeling of bed erosion

Prediction of bed erosion and sediment transport in open

channel flows are important tasks in many areas of river and
environmental engineering. Bed erosion can lead to insta-
bilities of the river basin slopes. It can also undermine the
foundation of bridge piles thereby favouring structural fail-
ure. Modeling of bed erosion is also relevant for predicting
the evolution of surface material dragged in earth dams in
overspill situations. Bed erosion is one of the main causes of
environmental damage in floods.
Fig. 3 Modeling of bed erosion with the PFEM by dragging of bed
Bed erosion models are traditionally based on a relation- material
ship between the rate of erosion and the shear stress level [25].
In a recent work we have proposed an extension of the PFEM
to model bed erosion [35,36]. The erosion model is based on
the frictional work at the bed surface originated by the shear
n
W f = n−1 W f + τt γt t (13)
stresses in the fluid. The resulting erosion model resembles
Archard law typically used for modeling abrasive wear in 3. The onset of erosion at a bed point occurs when n W f
surfaces under frictional contact conditions [1]. exceeds a critical threshold value Wc .
The algorithm for modeling the erosion of soil/rock par- 4. If n W f > Wc at a bed node, then the node is detached
ticles at the fluid bed is the following: from the bed region and it is allowed to move with the
fluid flow. As a consequence, the mass of the patch of
bed elements surrounding the bed node vanishes in the
1. Compute at every point of the bed surface the resultant
bed domain and it is transferred to the new fluid node.
tangential stress τ induced by the fluid motion. In 3-D
This mass is subsequently transported with the fluid.
problems τ = (τs2 + τt )2 where τs and τt are the tangen-
5. Sediment deposition can be modeled by an inverse pro-
tial stresses in the plane defined by the normal direction
cess to that described in the previous step. Hence, a sus-
n at the bed node. The value of τ for 2-D problems can
pended node adjacent to the bed surface with a velocity
be estimated as follows:
below a threshold value is attached to the bed surface.

1 ∂vt vk Figure 3 shows an schematic view of the bed erosion algo-

τt = μγt with γt = = t (11)
2 ∂n 2h k rithm described.

where vtk is the modulus of the tangential velocity at the

node k and h k is a prescribed distance along the normal 8 Examples
of the bed node k. Typically h k is of the order of mag-
nitude of the smallest fluid element adjacent to node k 8.1 Dragging of rocks by a water stream
(Fig. 3).
2. Compute the frictional work originated by the tangential Predicting the critical speed at which a rock will be dragged
stresses at the bed surface as by a water stream is of great importance in many problems
in hydraulic, harbour, civil and environmental engineering.
The PFEM has been successfully applied to the study
t t  2
μ vtk of the motion of a 1Tn quasi-spherical rock due to a water
Wf = τt γt dt = dt (12)
4 hk stream. The rock lays on a collection of rocks that are kept
◦ ◦ rigid. Frictional conditions between the analyzed rock and the
rest of the rocks have been assumed. Figure 4a shows that a
Eq. (12) is integrated in time as water stream of 1 m/s is not able to displace the individual

123
Comput Mech (2011) 48:307–318 313

8.2 Impact of sea waves on piers and breakwaters

Figures 5 and 6 show the analysis of the effect of breaking

waves on two different sites of a breakwater containing rein-
forced concrete blocks (each one of 4 × 4 m). The figures
correspond to the study of Langosteira harbour in A Coruña,
Spain using PFEM.

8.3 Soil erosion problems

Figure 7 shows the capacity of the PFEM for modelling soil

erosion, sediment transport and material deposition in a river
bed. The soil particles are first detached from the bed surface
under the action of the jet stream. Then they are transported
by the flow and eventually fall down due to gravity forces
into the bed surface at a downstream point.
Figure 8 shows the progressive erosion of the unprotected
part of a breakwater slope in the Langosteira harbour in
A Coruña, Spain. The non protected upper shoulder zone
is progressively eroded under the sea waves.
Figure 9 displays the progressive erosion and dragging of
soil particles in a river bed adjacent to the foot of bridge pile
due to the water stream (water is not shown in the figure).
Note the disclosure of the bridge foundation as the adjacent
soil particles are removed due to erosion.
Other applications of the PFEM to bed erosion problems
can be found in [35,36].

8.4 Falling of a lorry into the sea by sea wave erosion

Fig. 4 Study of the drag of an individual rock of 1Tn under a water of the road slope
stream at speeds of a water speed of 1 m/s. The individual rock cannot
be dragged by the stream and b water speed of 2 m/s. The individual
rock is dragged by the stream Figure 10 shows a representative example of the progressive
erosion of a soil mass adjacent to the shore due to sea waves
and the subsequent falling into the sea of a 2-D object repre-
rock. An increase of the water speed to 2 m/s induces the senting the section of a lorry. The object has been modeled
motion of the rock as shown in Figure 4b. as a rigid solid.

Fig. 5 Breaking waves on breakwater slope containing reinforced concrete blocks. Mesh of 4-noded tetrahedra near the slope

123
314 Comput Mech (2011) 48:307–318

Fig. 6 Study of breaking waves on the edge of a breakwater structure formed by reinforced concrete blocks

Fig. 7 Erosion, transport and deposition of soil particles at a river bed due to an impacting jet stream

This example, although still quite simple and schematic,

shows the possibility of the PFEM for modeling complex
FSSI problems involving soil erosion, free surface waves and
rigid/deformable structures.

8.5 Simulation of landslides

The PFEM is particularly suited for modelling and simulation

of landslides and their effect in the surrounding structures.
Figure 11 shows an schematic 2-D simulation of a landslide
falling on two adjacent constructions. The landslide material
has been modelled as a viscous incompressible fluid.

8.6 The landslide in Lituya Bay

A case of much interest is when the landslide occurs in the

vicinity of a reservoir [43]. The fall of debris material into
the reservoir typically induces large waves that can overtop
the dam originating an unexpected flooding that can cause
Fig. 8 Erosion of an unprotected shoulder of a breakwater due to sea severe damage to the constructions and population in the
waves downstream area.

123
Comput Mech (2011) 48:307–318 315

Fig. 9 Erosion and dragging of

soil particles in a river bed
adjacent to the foot of a bridge
pile due to a water stream. Water
is not shown

Fig. 10 Erosion of a soil mass due to sea waves and the subsequent falling into the sea of an adjacent lorry

In the example, we present some results of the 3-D analy- incompressible continuum with a prescribed shear modulus.
sis of the landslide produced in Lituya Bay (Alaska) on July No frictional effect between the sliding mass and the under-
9, 1958 (Fig. 12). The landslide was originated by an earth- neath soil has been considered. Also the analysis has not
quake and movilized 90 millions tons of rocks that fell on taken into account the erosion and dragging of soil material
the bay originating a large wave that reached a hight on the induced by the landslide mass during motion.
opposed slope of 524 m. PFEM results have been compared with observed values
Figure 13 show images of the simulation of the landslide of the maximum water level in the north hill adjacent to
with PFEM. The sliding mass has been modelled as a quasi- the reservoir. The maximum water level in this hill obtained

123
316 Comput Mech (2011) 48:307–318

Fig. 11 Landslide falling on

two constructions 2-D
simulation using PFEM

Fig. 12 Lituya Bay landslide. Left geometry for the simulation. Right landslide direction and maximum wave level [13,14]

with PFEM was 551 m. This is 5% higher than the value particles, surface waves, water splashing, frictional contact
of 524 m. observed experimental by [13,14]. The maximum situations between fluid–solid and solid–solid interfaces and
height location differs in 300 m from the observed value [13, bed erosion, among other complex phenomena. The success
14]. In the south slope the maximum water height observed of the PFEM lies in the accurate and efficient solution of the
was 208 m, while the PFEM result (not shown here) was equations of an incompressible continuum using an updated
195 m (6% error). Lagrangian formulation and a stabilized finite element
More information on the PFEM solutions of this example method allowing the use of low order elements with equal
can be found in [38,39]. order interpolation for all the variables. Other essential solu-
tion ingredients are the efficient regeneration of the finite
element mesh, the identification of the boundary nodes using
the Alpha-Shape technique and the simple algorithm to treat
9 Conclusions frictional contact conditions and erosion/wear at fluid–solid
and solid–solid interfaces via mesh generation. The exam-
The particle finite element method (PFEM) is a promising ples presented have shown the potential of the PFEM for
numerical technique for solving fluid–soil–structure interac- solving a wide class of practical FSSI problems in
tion (FSSI) problems involving large motion of fluid and solid engineering.

123
Comput Mech (2011) 48:307–318 317

Fig. 13 Lituya Bay landslide. Evolution of the landslide into the reservoir obtained with the PFEM. Maximum level of generated wave (551 m) in
the north slope

Acknowledgments This research was partially supported by pro- 3. Badia S, Codina R (2009) On a multiscale approach to the tran-
ject SEDUREC of the Consolider Programme of the Ministerio de sient Stokes problem transient subscales and anisotropic space–
Educación y Ciencia (MEC) of Spain and the projects SAFECON time discretizations. Appl Math Comput 207:415–423
and REALTIME of the European Research Council of the European 4. Baiges J, Codina R (2010) The fixed-mesh ALE approach applied
Commission (EC). Thanks are also given to the Spanish construction to solid mechanics and fluid-structure interaction problems. Int
company Dragados for financial support for the study of harbour engi- J Numer Methods Eng 81:1529–1557
neering problems with the PFEM. 5. Bazilevs Y, Calo VM, Cottrell JA, Hughes TJR, Reali A,
Scovazzi G (2007) Variational multiscale residual-based turbu-
lence modeling for large eddy simulation of incompressible flows.
Comp Methods Appl Mech Eng 197:173–201
References 6. Bazilevs Y, Hsu M-C, Zhang Y, Wang W, Liang X, Kvamsdal T,
Brekken R, Isaksen J (2010) A fully coupled fluid–structure inter-
1. Archard JF (1953) Contact and rubbing of flat surfaces. J Appl action simulation of cerebral aneurysms. Comput Mechan 46:
Phys 24(8):981–988 3–16
2. Aubry R, Idelsohn SR, Oñate E (2005) Particle finite element 7. Carbonell JM, Oñate E, Suárez B (2010) Modeling of ground exca-
method in fluid mechanics including thermal convection-diffusion. vation with the particle finite element method. J Eng Mechan
Comput Struct 83(17–18):1459–1475 (ASCE) 136(4):455–463

123
318 Comput Mech (2011) 48:307–318

8. Codina R (2002) Stabilized finite element approximation of tran- 29. Oñate E (1998) Derivation of stabilized equations for advective–
sient incompressible flows using subscales. Comput Methods Appl diffusive transport and fluid flow problems. Comput Methods Appl
Mech Eng 191:4295–4321 Mech Eng 151:233–267
9. Codina R, Coppola-Owen H, Nithiarasu P, Liu CB (2006) Numer- 30. Oñate E (2004) Possibilities of finite calculus in computational
ical comparison of CBS and SGS as stabilization techniques for mechanics. Int J Numer Methods Eng 60(1):255–281
the incompressible Navier–Stokes equations. Int J Numer Methods 31. Oñate E, García J (2001) A finite element method for fluid–
Eng 66:1672–1689 structure interaction with surface waves using a finite calculus for-
10. Del Pin F, Idelsohn SR, Oñate E, Aubry R (2007) The mulation. Comput Methods Appl Mech Eng 191:635–660
ALE/Lagrangian particle finite element method: a new approach 32. Oñate E, Idelsohn SR, Del Pin F, Aubry R (2004) The particle finite
to computation of free-surface flows and fluid–object interactions. element method: an overview. Int J Comput Methods 1(2):267–307
Comput Fluids 36:27–38 33. Oñate E, Valls A, García J (2006) FIC/FEM formulation with
11. Donea J, Huerta A (2003) Finite element method for flow prob- matrix stabilizing terms for incompressible flows at low and high
lems. Wiley, Chichester Reynold’s numbers. Comput Mechan 38(4–5):440–455
12. Edelsbrunner H, Mucke EP (1999) Three dimensional alpha 34. Oñate E, García J, Idelsohn SR, Del Pin F (2006) FIC formula-
shapes. ACM Trans Graphics 13:43–72 tions for finite element analysis of incompressible flows Eulerian,
13. Fritz HM, Hager WH, Minor HE (2001) Lituya Bay case: rockslide ALE and Lagrangian approaches.. Comput Methods Appl Mech
impact and wave run-up. Sci Tsunami Hazards 19(1):3–22 Eng 195(23–24):3001–3037
14. Fritz HM, Hager WH, Minor HE (2004) Near field characteris- 35. Oñate E, Celigueta MA, Idelsohn SR (2006) Modeling bed ero-
tics of landslide generated impulse waves. J Waterway Port Coast sion in free surface flows by the particle finite element method.
Ocean Eng ASCE 130(6):287–302 Acta Geotechnia 1(4):237–252
15. García J, Oñate E (2003) An unstructured finite element solver for 36. Oñate E, Idelsohn SR, Celigueta MA, Rossi R (2008) Advances
ship hydrodynamic problems. J Appl Mech 70:18–26 in the particle finite element method for the analysis of fluid–
16. Idelsohn SR, Oñate E, Del Pin F, Calvo N (2002) Lagrangian multibody interaction and bed erosion in free surface flows. Com-
formulation: the only way to solve some free-surface fluid mechan- put Methods Appl Mech Eng 197(19–20):1777–1800
ics problems. In: Mang HA, Rammerstorfer FG, Eberhardstein- 37. Oñate E, Rossi R, Idelsohn SR, Butler K (2010) Melting and spread
er J (eds) 5th World Congress on Comput Mechanics, July 7–12, of polymers in fire with the particle finite element method. Int
Vienna, Austria J Numer Methods Eng 81(8):1046–1072
17. Idelsohn SR, Oñate E, Calvo N, Del Pin F (2003) The meshless 38. Oñate E, Salazar F, Morán R (2011) Modeling of landslides into
finite element method. Int J Numer Methods Eng 58(6):893–912 reservoir with the particle finite element method. Research Report
18. Idelsohn SR, Oñate E, Del Pin F (2003) A lagrangian meshless CIMNE No. PI355. Submitted to Int J Numer Methods Geomechan
finite element method applied to fluid–structure interaction prob- 39. Salazar F, Oñate E, Morán R (2011) Modelación numérica de
lems. Comput Struct 81:655–671 deslizamientos de ladera en embalses mediante el método de
19. Idelsohn SR, Calvo N, Oñate E (2003) Polyhedrization of an partículas y elementos finitos (PFEM). Rev Int Mét Num Cálc Dis
arbitrary point set. Comput Methods Appl Mech Eng 192(22– Ing. Accepted for publication
24):2649–2668 40. Takizawa K, Tezduyar TE (2011) Multiscale space–time fluid–
20. Idelsohn SR, Oñate E, Del Pin F (2004) The particle finite ele- structure interaction techniques. Comput Mechan. doi:10.1007/
ment method: a powerful tool to solve incompressible flows with s00466-011-0571-z
free-surfaces and breaking waves. Int J Numer Methods Eng 61: 41. Tezduyar TE, Mittal S, Ray SE, Shih R (1992) Incompressible
964–989 flow computations with stabilized bilinear and linear equal-order-
21. Idelsohn SR, Oñate E, Del Pin F, Calvo N (2006) Fluid–struc- interpolation velocity–pressure elements. Comput Methods Appl
ture interaction using the particle finite element method. Comput Mech Eng 95:221–242
Methods Appl Mech Eng 195:2100–2113 42. Tezduyar TE (2007) Finite elements in fluids: special methods and
22. Idelsohn SR, Marti J, Limache A, Oñate E (2008) Unified enhanced solution techniques. Comput Fluids 36:207–223
Lagrangian formulation for elastic solids and incompressible flu- 43. Wan CF, Fell R (2004) Investigation of erosion of soils in embank-
ids: application to fluid–structure interaction problems via the ment dams. J Geotechn Geoenviron Eng 130:373–380
PFEM. Comput Methods Appl Mech Eng 197:1762–1776 44. Yabe T, Takizawa K, Tezduyar TE, Im H-N (2007) Computation of
23. Idelsohn SR, Mier-Torrecilla M, Oñate E (2009) Multi-fluid flows fluid–solid and fluid–fluid interfaces with the CIP method based on
with the particle finite element method. Comput Methods Appl adaptive Soroban Grids: an overview. Int J Numer Methods Fluids
Mech Eng 198:2750–2767 54:841–853
24. Johnson AA, Tezduyar TE (1999) Advanced mesh generation and 45. Zienkiewicz OC, Jain PC, Oñate E (1978) Flow of solids during
update methods for 3-D flow simulations. Comput Mech 23:130– forming and extrusion: some aspects of numerical solutions. Int
143 J Solids Struct 14:15–38
25. Kovacs A, Parker G (1994) A new vectorial bedload formulation 46. Zienkiewicz OC, Taylor RL, Zhu JZ (2005) The finite element
and its application to the time evolution of straight river channels. method. Its basis and fundamentals. Elsevier, Oxford
J Fluid Mech 267:153–183 47. Zienkiewicz OC, Taylor RL (2005) The finite element method for
26. Larese A, Rossi R, Oñate E, Idelsohn SR (2008) Validation of solid and structural mechanics. Elsevier, Oxford
the particle finite element method (PFEM) for simulation of free 48. Zienkiewicz OC, Taylor RL, Nithiarasu P (2006) The finite
surface flows. Eng Comput 25(4):385–425 element method for fluid dynamics. Elsevier, Oxford
27. Löhner R (2008) Applied CFD techniques. Wiley, Chichester
28. Löhner R, Yang C, Oñate E (2007) Simulation of flows with violent
free surface motion and moving objects using unstructured grids.
Int J Numer Methods Fluids 153:1315–1338

123