Journal of Manufacturing Processes 34 (2018) 374–388

Contents lists available at ScienceDirect

Journal of Manufacturing Processes

journal homepage: www.elsevier.com/locate/manpro

T
Modelling of guillotine cutting of multi-layered aluminum sheets
a b b,⁎ b c
D. Gasiorek , P. Baranowski , J. Malachowski , L. Mazurkiewicz , M. Wiercigroch
a
Institute of Theoretical and Applied Mechanics, Faculty of Mechanical Engineering, Silesian University of Technology, Akademicka 2A Street, 44-100 Gliwice, Poland
b
Department of Mechanics and Applied Computer Science, Faculty of Mechanical Engineering, Military University of Technology, Gen. Sylwestra Kaliskiego Street 2, 00-
908 Warsaw, Poland
c
Centre for Applied Dynamics Research, School of Engineering, University of Aberdeen, Aberdeen AB24 3UE, Scotland, UK

A R T I C LE I N FO A B S T R A C T

Keywords: The paper presents a numerical modelling of the guillotine cutting process of sheet aluminum bundles. A finite
Cutting element method with a smoothed particle hydrodynamics approach were coupled to simulate the cutting pro-
Guillotine cess. Moreover, arbitrary Lagrangian–Eulerian formulation was also proposed and applied to model the process.
Metal sheets Experimental results of the cutting were presented for the validation purposes. The measured force character-
Numerical modelling
istics in all numerical simulations and experiment were compared. The computational costs of the implemented
ALE
methods were also analyzed. Additionally, a mesh (particles) sensitivity study was performed, and the influence
FEM-SPH coupling
of the mesh on the obtained results was assessed. Based on the outcomes the arbitrary Lagrangian–Eulerian
model was selected as more suitable method of modelling the guillotine cutting process and it was validated in
different cutting conditions and the influence of technological parameters such as knife velocity and cutting-edge
angle was investigated.

1. Introduction
Nowadays, steel, or in general metals processing can be considered
as an indispensable part of many industries. Its role in aviation, auto-
motive or military fields is increasing and significant. The forming
process of metal sheets can be divided into several operations, which
includes slitting, blanking, piercing, trimming and orthogonal cutting.
Each of them from the technological point of view has a significant
degree of complexity, where shearing is the main cutting mechanism,
but ductile fracture is also important. During the process, a very large
plastic strain during the cutting operation initiates micro cracks, which
subsequently generate macro cracks. An insight into mechanisms which
are present during metal processing is crucial from the fundamental and
practical point of view. Such knowledge can be gained from experi-
mental tests or numerical modelling, which are today widely available
for optimizing and simulating metal processing operations. Recently,
numerical and experimental tools have been used to study trimming
process. The main aim of the paper by Golovashchenko [1] was to
modify the trimming process to eliminate slivers and burrs by looking at
the cause of their generation. The main learning was to limit bending of
the part being trimmed off and to create a preference for a crack pro-
pagation. Hilditch and Hodgson [2] investigated the influence of cut-
ting clearance on rollover and burr behavior of aluminum alloy and
magnesium alloy. They concluded that with a larger clearance, the burr
height and rollover depth increase. In their other paper [3], it was
pointed out that the strain-rate sensitivity is the most important ma-
terial property to control the distribution of strain in a trimmed sample.
The total elongation of trimmed edge of cut sheet is affected by material
anisotropy and trimming method according to [4]. Bohdal [5] used
smoothed particle hydrodynamics (SPH) method for simulating trim-
ming process, which was monitored using Digital Image Correlation
(DIC) technology during the experiment. Thanks to DIC it was possible
to quantify material nonlinearities and deformation of material struc-
ture.
A large number of papers is devoted to modelling and simulation of
orthogonal cutting and machining. One of the first results of cutting
modelling were presented in the paper by Marusich and Oritz [6], in
which they used Lagrangian Finite Element (FE) model for the ortho-
gonal cutting with continuous and adaptive remeshing implementation.
To prevent mesh distortion which accompanies Lagrangian modelling
other FE methods can be applied. Olovsson et al. [7] developed a FE
code with arbitrary Lagrangian–Eulerian (ALE) formulation to simulate
a two-dimensional metal cutting problem. They proved that ALE for-
mulation is numerically effective in such problems, however they
pointed out that a constitutive model with a thermal and strain-rate
dependency is crucial to obtain reliable results. The effectiveness of ALE

Corresponding author.
⁎

E-mail addresses: damian.gasiarek@polsl.pl (D. Gasiorek), pawel.baranowski@wat.edu.pl (P. Baranowski), jerzy.malachowski@wat.edu.pl (J. Malachowski),
lukasz.mazurkiewiczi@wat.edu.pl (L. Mazurkiewicz), m.wiercigroch@abdn.ac.uk (M. Wiercigroch).

https://doi.org/10.1016/j.jmapro.2018.06.014

1526-6125/ © 2018 The Society of Manufacturing Engineers. Published by Elsevier Ltd. All rights reserved.
Received 24 September 2017; Received in revised form 5 June 2018; Accepted 10 June 2018
D. Gasiorek et al.
Fig. 1. Schematic of cutting process investigated in study; (a) a knife contacting material, (b) formation of a plastic burr.
formulation in 3D problems was proved by Raczy et al. [8], who
modelled deformation of copper during the orthogonal cutting using an
Eulerian formulation for the workpiece. Mesh-free methods were also
used to simulate problems with high speed cutting, where Madaj and
Píškaa [9] investigated how to predict the chip shape and the cutting
force using SPH method. In the paper by Limido et al. [10] the ad-
vantages of SPH method implemented in LS-Dyna were discussed and
the meshless nature of the method with its natural separation possibility
was highlighted. However, it was found that the parameters of SPH
solver such as initial smoothing length, initial particle density and time
step have an effect of the machining variables [11].
Blanking was also investigated using numerical methods and for
example in the paper by Husson et al. [12], a new visco-plastic hard-
ening material model coupled with a new damage model was presented
and effectively applied in copper sheet blanking FE simulations. The
authors confirmed that the proposed constitutive model is suitable to
predict blanking process characteristic features. Mao et al. [13] in-
troduced a new discontinuous dot indenter blank holder and they in-
vestigated the effect of dot indenter parameters using FE simulations.
The parameters were also optimized and applied to the blanking pro-
cess. As a result, a perfect clean-cut surface was obtained.
In terms of turning metal machining, Duran and Nalbant [14]
pointed out the problem of bending of cutting tool. In their paper this
bending obtained by finite element method (FEM) was compared with
the bending calculated by Castigliano theorem. Saglam et al. [15] in-
vestigated an effect of factors, including the entering angle in tool
geometry on cutting force components and temperature generated on
the tool tip.
As discussed above a standard FEM is widely used in modelling
metal processing, however the following steps need to be taken into
account to obtain reliable results. Firstly, when using FEM, the material
behavior must be correctly modelled including crack initiation and
propagation, damage and failure. A proper and validated material data
with a suitable constitutive model is vital to obtain reliable numerical
results. Moreover, various numerical subroutines such as pre-crack
modelling and nodes disconnecting are also available for improving
discrete representation of the simulated process. For extremely large
deformations and to properly represent the crack propagation during
cutting tool-material interactions, a very fine mesh needs to be gener-
ated, which can be computational time prohibitive.
Continuously developed numerical methods give a possibility to
reduce the discussed above disadvantages, for example, by using
mentioned methods including meshless SPH or ALE approaches. Both of
them have been used in the past; however, it should be pointed out that
in the presented paper SPH and ALE formulations available in LS-Dyna
code are applied for modelling and simulation of the dynamic multi-
material cutting of several separate aluminum sheets bundles, not a
single one as has been done before [1,5,11,14,16,17]. FE modelling is
based on the experimental tests carried out using a novel, patented and
specially designed laboratory test stand, allowing measurements of
forces and applying a vertical, controlled and unique movement of the
cutting knife. In the FEM-SPH coupled model, it was decided to account
375
Journal of Manufacturing Processes 34 (2018) 374–388
for the metal sheets area within a direct interaction with a knife using
the particles, whereas the remaining part is represented by the La-
grangian elements. In the ALE formulation for a selected area where
largest deformations were expected, a finer mesh is used, which guar-
antees correctness of the coupling between the Eulerian and the La-
grangian formulations. Not only the Eulerian domain is remeshed but
also the Lagrangian sheet slices. In both cases a selected slice of the
aluminum sheets bundle with the knife is taken into account during
investigations. Such quantities as deformation, plasticity and force
characteristic are also analyzed. Additionally, a parametric study is
conducted to verify the influence of the applied mesh (number of par-
ticles) on the obtained results. The modelling was compared with the
actual tests proved effectiveness of the applied methods and showed
good correlation with the experiment. Results obtained are promising
in the case of perpendicular cutting.
The paper is organized as follows. In the next section the problem
will be introduced and formulated. In Section 3, numerical modelling
will be described. Results will be discussed in Section 4. Section 5
consists of final conclusions regarding the presented investigations.
2. Problem formulation
Generally, guillotining is realized with the two straight knives. The
upper knife moves downwards with a prescribed velocity, while the
lower knife is fixed. Usually, the metal to be cut sheet is clamped on the
lower knife. Initially, an interaction between the knife and cutting
surface can be described with Hertz contact theory. Subsequently,
plasticization of the sheet being cut, and formation of a burr occur. A
simple schematic of the described process is presented in Fig. 1, where a
sheet is cut progressively due to a small clearance between both the
sheets and upper knife.
Principles of operation of the proposed test stand are similar,
however, the lower knife is omitted and aluminum sheets are placed on
a flat surface. The test stand is installed on a MTS 858 Vertical Tensile
Testing Machine. Contrary to an ordinary guillotine where the knife
moves under 45 degrees towards the table, on the test stand, here the
knife moves vertically downwards and is driven directly by the hy-
draulic system. During testing a bundle of printing 1100-h12 aluminum
sheets (3 × 30 pcs) separated every 30th sheet with a cardboard sheet
were cut (Fig. 2). A more detailed description of the discussed test stand
and experiments can be found in the papers by Gasiorek [18] and Ga-
siorek et al. [19]. A schematic of the discussed experimental setup is
presented in Fig. 3.
3. Numerical modelling
3.1. Model definition
In all simulated cases, it was decided to model only a selected sec-
tion of the whole bundle of aluminum sheets with appropriate initial
boundary conditions. The assumptions for all cases were as follows:
D. Gasiorek et al.
Fig. 2. The close-up view of the test stand used in cutting process where the knife and the bundle of aluminum sheets seperated by cardboards can be seen.
- Simulations were carried using Multi Parallel Processing (MPP) LS-
Dyna explicit code with available SPH and ALE formulations.
- 1 mm section as shown in Fig. 4 was modelled, which drastically
reduced computational time. It should be noted that a different
approach was described and presented in the paper by Gasiorek
[18], where the whole model was considered.
- It was considered that such a local representation of the whole
model will allow to assume that the same phenomena can be ex-
pected over the entire width of aluminum sheets.
- The knife with the cutting-edge β1 = 28.5° was assumed as non-
deformable with steel properties implemented, where velocity,
v = 200 mm/s, was applied for both methods, and was taken from
experimental test.
- During the initial modelling of the cutting process using ALE method
an undesirable and non-physical leakage problem was observed in
the coupling between knife and aluminum sheets. This was the re-
sult of adopting a sharp edge, thus it was impossible to correctly
distribute the interaction forces. Therefore, the sharp cutting edge of
the knife was replaced with a rounded one with radius r = 0.01 mm.
Authors are aware that the radius value plays a considerable role in
FEA response. However, it was assumed that it is negligibly small
(five times smaller than the characteristic length in the finest mesh
and smaller gap in the FEM-SPH modelling) and is comparable with
blunting resulting from a wear process of the knife during cutting.
Fig. 3. The schematic of experimental set-up used in the cutting process with test conditions and cutting angles description; front and side view.
376
Journal of Manufacturing Processes 34 (2018) 374–388
The influence of radius was investigated in the parallel simulations
and the selected radius has the smallest value that allowed to avoid
the leakage. It should be also noted that radius value had no sig-
nificant effect on the FEM-SPH modelling, due to gaps between them
(minimum 0.05 mm) so that the knife movement causes particles to
initially move in lateral direction.
- First seven aluminum sheets with 0.4 mm width were considered to
minimize the simulation time. Each of them was described with a
regular and fine mesh (ALE) or grid of particles (FEM-SPH). In the
performed analyses all package of aluminum sheets was treated as
system of single and separate layers, where the interaction between
each layer was described. The mechanism of contact was im-
plemented using penalty formulation method.
- A number of simulations was carried out to assess the most appro-
priate length of aluminum area under the knife which should be
modelled accurately. It was noted, that there is no need for a precise
modelling of remaining parts.
- The friction mechanism plays a significant role in FE predictions of
cutting forces. In the presented ALE and FEM-SPH modelling the
interactions between knife and aluminum layers and were defined
using a penalty function approach with Coulomb’s friction model
[20]. The friction coefficient is often used in order to readjust the
cutting forces obtained from numerical simulations [9,10,21]. Thus,
a number of simulations were carried out with different friction
D. Gasiorek et al. Journal of Manufacturing Processes 34 (2018) 374–388

simulated
knife 1 mm section

0.4 mm
um seven aluminum sheets
(finer mesh)

la

Fig. 4. The knife and the bundle of aluminum sheets used in the cutting process with a representative section of model considered in all simulations (visible seven
aluminum sheets that were modeled using finer mesh).

coefficients and μ = 0.1 was selected for which the force predictions
were closest to the experimental results. It was also found that the
friction between layers did not significantly affect the results,
therefore frictionless interaction was assumed.
- The parametric study was conducted to verify the influence of the
applied mesh (number of particles) on the obtained results. The
adopted smallest value of the characteristic element length was
0.05 mm, which gave a very similar result as the experiment (it is
where: σ – stress, v – velocity, ij – indexes of components, E – internal
energy, m – mass of particle, W – kernel function, N – number of par-
ticles within smoothing length, p, q – indexes denoting different parti-
cles.
Eqs. (1) and (2) are solved by interpolation of a given function
value < f > (i.e., density, velocity, energy, etc.) in a given point based
on the known value of this function in the surrounding points (parti-
cles) using the following formula:

∫ f (x′) W (x − x′, h) dx′ ≃ ∑ f (xq) W (x − xq, h) ∆Vq

described later in the Results and Discussion section). Such a value
< f > (x ) =
N
seems to be appropriate and similar to the studies of other authors
[9–11], taking into account size of the model, its multi-layered Ω q=1 (3)
property and the level of complexity. It was assumed that this is a
compromise between simulations of cutting of the whole aluminum where: < f > – function interpolation, x – vector defining particle
bundle in microscopic and macroscopic level while considering position, h – maximum distance between particles (smoothing length),
layers layout and possibility to verify the results with available ex- W – kernel function.
perimental data. The majority of SPH implementations use a cubic B-spline as the
- In order to validate the model in different cutting conditions an kernel function (smoothing function), which is implemented in LS-Dyna
influence of technological parameters such as knife velocity code [20]:

⎧ 3 − R2 + 2 R3, 0 ≤ R < 1
(v = 100 mm/s) and cutting-edge angle (β2 = 23.5° and β3 = 33.5°)
⎪
2 1

W (R, h) = αd 1 (2 − R)3, 1≤R<2

were investigated. The results were compared with experimental
⎨6
⎪ 0,
outcomes.
⎩ R≥2 (4)

where: R = = αd =
3.2. SPH method r |x − x ′| 1 15 3
h h
, ; ;
for one-, two- and three-di-
h 7πh2 2πh3
mensional case respectively.
In the first modelling approach of the cutting process, FEM was The smoothing length varies through simulation. Therefore, a
coupled with SPH meshless method. SPH is a mesh-free particle method number of neighboring particles remains relatively constant. It is rea-
of the Lagrangian nature, where all computational information in- lized by recalculating the smoothing as follows:
cluding mass and velocity are associated with particles. The main dif-

h = h0 ⎛ 0 ⎞
1
ference between classical methods and SPH is the absence of a grid. d p

⎝ di ⎠
⎜ ⎟
This method is predominantly used to simulate problems were large
(5)
deformations [10,18,22] or damage (e.g. failure, erosion or wear)
[23,24] occur. It was developed in 1977 for astrophysics applications. A or by solving continuity equation:
1 h ∂d
= ⋅ ⋅ ,
detailed description of SPH is given in the papers [25,26]. In short, a
dh
d d ∂t
description of the SPH continuum uses the principles of conservation of
dt (6)
mass, momentum end internal energy as the governing equations in
Lagrangian description applied in FEM: where: d0 – initial density, h0 – initial smoothing length, p – is dimen-

∂v 1 ∂σij σij ∂vi

sion of space.
= −ρ i ; = =
dρ dvi dE
∂x i ρ ∂x j ρ ∂x j
; , In the presented investigations, the smoothing length parameters
dt dt dt (1) were set to default values and renormalized formulation which im-
proves the SPH method consistency was applied.
∂ρp ∂Wpq
= ∑q = 1 mq vpq
N
∂t ∂xp
,
3.3. Boundary conditions for FEM-SPH initial modelling
∑q = 1 mq ⎛⎜ 2 + 2 ⎞ ∂x ,
∂vp ∂Wpq
=
N σp σq
∂t
⎝ ⎠ p
⎟
ρ ρq
p During the movement of the cutting knife it was assumed that SPH
∂Ep ∂Wpq
= ∑q = 1 mq ρ ρ vpq ∂x ,
N σp σq particles will accurately represent the behavior of the bundle of alu-
∂t p q p (2) minum sheets in which large deformations and strong plasticization

377
D. Gasiorek et al. Journal of Manufacturing Processes 34 (2018) 374–388

occur. Having in mind that SPH methods are computationally ex-
pensive, only the area of direct interactions between the guillotine
cutter and the aluminum sheets was modelled using the particles. The
guillotine knife contact area was described with a regular grid of par-
ticles. The interaction between each of SPH aluminum sheets was si-
mulated using particle to particle coupling [20]. The remaining parts of
the aluminum bundle were modelled using fully integrated hexagonal
finite elements. This was not an object of interests, therefore each sheet
was modelled using only 556 elements, which gives a number of 3892
elements for all considered sheets. Other sheets with three pieces of the
cardboard after every thirty-aluminum sheet were described with a
coarser mesh using 391 finite elements. Such a number of elements
allowed for carrying out simulations in reasonable time. In order to
couple the particles within the area of direct interactions between the
guillotine cutter and neighboring elements, a kinematic constraint
method, where the particles are tied to the Lagrangian surface, was
applied allowing to maintain the continuity in each sheet. Using such a
way of modelling only seven aluminum sheets were considered to
minimize the simulation time. The boundary conditions corresponded
to the experiments were as follows. Aluminum sheets bundle was fixed
on the bottom and knife displacement and rotations were fixed for all
degrees of freedom (DOFs) except of Y – axis movement. Displacement
in X – axis of hexagonal elements perpendicular to the knife movement
was fixed using the symmetry, whereas for the particles SPH symmetry
plane was introduced. The discussed FEM-SPH coupled model with the
initial-boundary conditions is presented in Fig. 5.
It should be pointed out here that the bulk viscosity is a very sig-
b) area of direct interactions with knife modelled using 1941 particles
with 0.07 mm distance between each of them,
c) area of direct interactions with knife modelled using 945 particles
with 0.1 mm distance between each of them.
3.4. ALE modelling description
The arbitrary Lagrangian–Eulerian is a finite element formulation in
which the computational task is not fixed a priori in space (pure
Eulerian) or associated with material (pure Lagrangian). This approach
can alleviate many drawbacks of traditional Lagrangian-based or (and)
Eulerian based FE simulations. The ALE procedure can be applied to
simulate many engineering problems including fluid-structure interac-
tions [27,28] fluid flows [16] and metal cutting [7,8].
The ALE procedure consists of two major steps, namely, (i) the
classical Lagrangian and (ii) the advection Eulerian formulations.
During the first step, the deformation field is described typically using
the node displacements. For this configuration, the basic parameters of
the equation of state such as pressure, relative volume and internal
energy are estimated. The advection step is carried out under the as-
sumption that displacements of the nodes are very small in comparison
with characteristic dimensions of the surrounding elements. Moreover,
in this procedure, a constant topology of the mesh is assumed. The
advection is performed in a single step (suggested for a strongly-dy-
namic phenomena) or at each step [7,28,29].
The governing equations for the fluid domain (Euler domain) de-
scribe the conservation of mass, momentum and energy as [20]:

∫ ρdV = ∫ ρ (w − v )⋅ ndS ,
nificant parameter in SPH modelling, which prevents inter-particle
=
penetration and allows shocks to form and to damp post-shock oscil- dMA d
dt dt ∼ ∼ ∼
lations. Moreover, it was found that default values for the artificial bulk V (t ) S (t )

= ∫ ρ vdV = ∫ ρ v (w − v )⋅ ndS − ∫ ∇pdV + ∫ υ gdV ,

viscosity for the FE solid elements are not appropriate when SPH par- dQ
∼ d

V (t ) ∼ S (t ) ∼ ∼ ∼ ∼ V (t ) ∼ V (t ) ∼
ticles are used in such a kind of simulation, and which was highlighted dt dt

= ∫ ρedV = ∫ ρe (w − v )⋅ ndS − ∫ p v⋅ ndS + ∫ ρ g ⋅vdV ,

in [10,24]. Thus, in the presented investigations, the values re- dE d
commended in the mentioned paper for SPH formulation: Q1 = 1.5 and dt dt
V (t ) S (t ) ∼ ∼ ∼ S (t ) ∼ ∼ V (t ) ∼ ∼
Q2 = 1.0, were used.
As it was mentioned before, one of the main tasks was to perform a (7)
sensitivity study of mesh density (particles number), which varied from where: ρ is fluid mass density, p is pressure, g is acceleration of gravity
the finer to the coarser one. It was decided to choose three different and e is total specific energy. The quantities MA, Q and E are total mass,
cases shown in Fig. 6: momentum and energy, respectively, in the control volume V(t),
bounded by surface S, which moves in the fluid (gas-air) with an ar-
a) area of direct interactions with knife modelled using 6511 particles bitrary velocity w which may be zero in Eulerian coordinates or v in
with 0.05 mm distance between each of them, ͠
Lagrangian coordinates. The vector n ‘n is the outwards normal to the

knife

cutting knife
movement
Lagrangian elements
Y
1 mm SPH particles Z

symmetry conditions

Y
cardboard
X

other aluminum
sheets Y
Z X tied contact
fixed

Fig. 5. The representative section of FEM-SPH coupled model with initial – boundary conditions used for simulating the cutting process of sheet aluminium bundles
(visible SPH particles representing area of direct interaction with knife).

378
D. Gasiorek et al. Journal of Manufacturing Processes 34 (2018) 374–388

0.07 mm 0.1 mm
0.05 mm

r = 0.01 mm r = 0.01 mm r = 0.01 mm

Fig. 6. Three different particles densities considered in modelling of SPH cutting zone (are of direct interactions with knife); (a) case with 6511 particles, (b) case
with 1941 particles, (c) case with 945 particles.

surface S. elements perpendicular to the knife movement was fixed (symmetry

Modelling of the coupling in the ALE formulation is one of the most
important problems to be dealt with. To simulate the process of cou-
pling, a penalty method is mostly used, however, a constraint-based
method can also be implemented. In case of the penalty method, a
fictional elastic element is defined between the interacting fluid ele-
ment and the node of the Lagrangian structure (Fig. 7). The value of its
stiffness is computed based on the penetration distance between the
two interacting domains. Moreover, the coupling stiffness parameters
are crucial for eliminating the penetration, which if occur, can influence
the force (pressure) distribution within the Lagrangian domain. In the
simulated problem using ALE method such the pressure distribution
(coupling) is between Lagrangian knife and Eulerian aluminium sheets.
It should be also noted, that the same friction properties were applied as
in the FEM-SPH modelling.
3.5. ALE method initial – boundary definition
conditions), whereas on the outer walls of the Eulerian domain, a non-
reflecting boundary condition was implemented to prevent artificial
stress wave reflections generated at the model boundaries. The dis-
cussed ALE model with initial-boundary conditions is presented in
Fig. 8.
Similarly to previous cases, three different mesh densities (within
the selected area) were also used and compared (see Fig. 9):
a) the area of direct interactions with the knife modelled using 6511
elements (characteristic length of element 0.05 mm) with the total
number of elements in the model: 215 948,
b) the area of direct interactions with the knife modelled using 1848
elements (characteristic length of element 0.07 mm) with the total
number of elements in the model: 172 391,
c) the area of direct interactions with the knife modelled using 640
elements (characteristic length of element 0.10 mm) with the total
number of elements in the model: 131 124.

The initial boundary conditions in the ALE simulations correspond

to the FEM-SPH modelling and the experiments. The guillotine knife 3.6. Numerical procedure
and other sheets with cardboards were described with a coarser mesh
using Lagrangian elements and were the same as in the FEM-SPH Numerical studies of the cutting process described in Section 3 were
model. The interactions between the Eulerian aluminum sheets, the carried out using FE application of ALE and SPH methods with an ex-
vacuum domain and the Lagrangian parts were described using the plicit integration procedure. In general, the explicit dynamics proce-
penalty coupling procedure with the Van Leer second-order advection dure uses explicit central difference time integration and the equation
method as the most accurate in the considered case [20,29]. The area of solved has the following form, which was adopted in the study for both
direct interactions between the guillotine cutter and the aluminum methods [20]:
sheets were modelled using a finer mesh, which guaranteed that the Mẍn = F ext
n − Fn − Cẋ n ,
int
(8)
coupling between the parts was correct. The boundary conditions ap-
int
plied were as follows: the aluminum sheets bundle was fixed at the whereM is a diagonal mass matrix; F ext
n is external and body forces; Fn
bottom, the knife displacement and rotations were fixed for all DOFs is a stress divergence vector;C is a damping matrix; x , ẋ , ẍ are nodal
except of Y – axis movement. Displacement in X – axis of hexagonal displacement, velocity and acceleration vectors.

n+1
t
n
t

xi xi
Fj=kAj

dn+1
d - penetration vector

Fig. 7. Coupling penalty based method scheme [29] implemented in the current study.

379
D. Gasiorek et al. Journal of Manufacturing Processes 34 (2018) 374–388

Fig. 8. The representative section of ALE model with initial – boundary conditions used for simulating the cutting process of sheet aluminium bundles (visible seven
sheets with Eulerian description emedded in air representing area of direct interaction with knife).

C c + C0 LE |ε˙kk | ε˙kk < 0

Q=⎧ 1
A solution of the above equation is obtained with numerical in- for
tegration of acceleration ẍn under the assumption that ẋ ≈ ẋn− 1 ⎨ ε˙kk ≥ 0
⎩
,
2
0 for (13)

xn = M−1 F ext
n − Cx n− − Fn . where LE = υE / AE max is the characteristic length of element, υE is a
.. .
( 1 int
2 ) (9)
volume of element, AE max is the largest side of element area and c is the
which after implementation of the central difference equations for the adiabatic speed of sound.
velocity and displacement results in: In the FEM-SPH case, the solution time step is directly influenced by
the contact stiffness, time step size related to the chosen FE dimensions,
xn = xn + − xn − 1 ) ⇒x = xn − 1 + ∆tn xn,
.. 1 . 1 . . . ..

∆tn( 2 2
n+ 1
2 2 (10)
selected from all elements in the model, and the time step size depen-
dent on the smoothing length and velocity of the selected particle

∆t = C⋅min (∆tcon, ∆tFEM , ∆tSPH ),

xn+ 1 = (xn +1 − xn) ⇒ xn +1 = xn + ∆tn + 1 xn+ 1 .
. 1 . (14)
2 ∆tn + 1 2 2
(11) where C is the scale factor related to (CFL) condition, ∆tcon is the time
2
step determined by contact stiffness, ∆tFEM is the time step based on
The major advantage of this method is a lack of time-consuming finite elements size, ∆tSPH is the time step based on SPH particles de-
operations involving the stiffness matrix inversion. Instead, only a di- termined by the following equation
agonal matrix of mass is inverted. However, the main disadvantage is
∆tSPH = C⋅min ⎛
that the method is conditionally stable requiring a time step to be hi ⎞
⎝ Ci + vi ⎠
⎜ ,
⎟

limited according to the Courant-Friedrichs-Lewy (CFL) stability con- (15)

dition [20],
where i is the particle number.
∆t =
LE In the ALE simulations, the solution time step is influenced by the
Q+ Q2 + c 2
,
(12) coupling stiffness and the time step size based on the element dimen-
sions and on the selection from all elements in the model [20]:

∆t = C⋅min (∆tcoupling , ∆tFEM , ∆tEuler ),

where Q is the function of viscous coefficients C0 and C1 formulated as
follows: (16)

0.05 mm 0.07 mm 0.1 mm

r = 0.01 mm r = 0.01 mm r = 0.01 mm

Fig. 9. Three different mesh densities considered in ALE modelling showing the cutting zone (area of direct interactions with the knife); (a) case with 6511 elements,
(b) case with 1848 elements, (c) case with 640 elements.

380
D. Gasiorek et al. Journal of Manufacturing Processes 34 (2018) 374–388

Table 1 Table 3
1100-h12 aluminum parameters for Johnson – Cook model Parameters for bilinear model for cardboards [32,33].
[30].
Density, ρ [kg/m3] 2850
Density, ρ [kg/m3] 2787 Young’s Modulus, E [MPa] 3626
Yield stress, A [MPa] 148 Poissons’s ratio, ν [-] 0.3
B [MPa] 345 Yield strength, Re [MPa] 26.4
C [-] 0.001
n [-] 0.18

σy = σo + β Ep εeffp ,
Reference strain rate, ε̇0 [1/s] 1.00 3
m [-] 0.85 2 (19)
Tmelt [K] 893
T0 [K] 293 where σo is the yield point, β is the hardening parameter ranging from 0
Specific heat, Cp [J/kgK] 920 to 1 and Epis the tangent modulus. The effective plastic strain is given
by

εeffp = ∫ ⎛⎝ 23 ε˙ijp ε˙ijp ⎞⎠

where ∆tcoupling is the time step determined by coupling stiffness, ∆tEuler t 1/2

is the time step based on Eulerian elements size. dt ,

0 (20)

3.7. Constitutive modelling where ∈˙ p is

The elastic-plastic material model with isotropic hardening in-
cluding a strain rate effect was applied to describe the aluminum
properties. The Johnson-Cook (JC) model provides a satisfactory pre-
diction of flow stress σflow for large strains and high strain rates when its
dependence on strain rate is linear in a semi-logarithmic scale. The
material properties for JC aluminum model were taken from the lit-
erature [30] and they are listed in Table 1. The mathematical formula
which describes this model is as follows [20]:
the plastic strain rate. The material properties for bilinear
model were taken from the papers by Giampieri et al. [32] and Mur-
ayama et al. [33], and they are listed in Table 3.
4. Results and discussion
The experimental tests were carried out using a specially designed
laboratory test stand, allowing measurements of forces and applying a
vertical, controlled and unique movement of the cutting knife. Surfaces
of edges of the samples were examined with the use of a Scanning

T − Troom ⎞ ⎤
σflow = [A + B (ε p)n](1 + C ln ε˙ p ) ⎡1 − ⎛
m Electronic Microscope (SEM). Overall good quality of the cut edges was
* ⎢ ⎝ − Troom ⎠ ⎥
obtained (Fig. 10) as demonstrated in papers by Gasiorek and Gasiorek
⎣ ⎦
⎜ ,
⎟

Tmelt
where A, B, C, n and m are material constants, is the plastic strain, ε̇ p
εp
* found, that edge of a surface after cutting on a laboratory guillotine test
is the effective plastic strain rate, T is the actual temperature based on
the plastic deformation, Troom is the room temperature and Tmelt is the
melting point temperature.
The SPH allows a material separation and the JC failure parameters
normally are not needed. However, due to increased bulk viscosity
residual velocities of particles can fall to unexpectedly low values.
Therefore, it is advisable to include them in the SPH simulations so that
the residual particles velocity change to more realistic [9,24]. There-
fore, failure process or loss of material integrity was also taken into
consideration in the FEM-SPH modelling. It is worth noting that in ALE
simulations failure was not implemented similarly to the papers
[7,8,31]. An extended damage accumulation model has been proposed,
which is based upon equivalent value of the effective failure strain. It
considers the tri-axial stress state in material, strain rate effect and
temperature dependence. The parameters for the aluminum failure
model are listed in Table 2 and were taken from the aforementioned
paper [30]. The strain at the fracture state is given by equation [20,30]:
(17) et al. [18,19]. It can be seen that scratches in the vicinity of the cutting
line are vertical - perpendicular to the surface of the plate. It was also
stand was significantly less disturbed than surfaces investigated after
cutting in industrial conditions - with rotary knives or on a guillotine.
More details of the experimental investigations can be found in the
paper by [19].
4.1. Sensitivity study and calibration of FEM-SPH and ALE models
One of the main aspect of our investigations was to compare the
results obtained from different modelling methods and to analyze the
influence of mesh and particles densities. The measured force char-
acteristics in all numerical simulations were examined and compared
with experimental data to validate the models. Due to the overall re-
petitive character of obtained curves the maximum displacement was
set to 2 mm in the presented figures, which corresponds to the cutting of
only first five aluminum layers. It also allowed a better presentation of
the results

T − Troom ⎞ ⎤
In the FEM-SPH modelling resultant forces were taken from the
ef = [D1 + D2 e D3 p / σeff ](1 + D4 ln ε˙ p ) ⎡1 − D5 ⎛
* ⎢ ⎝ Tmelt − Troom ⎠ ⎥
interface between the knife and the SPH particles. It is possible to ob-
⎣ ⎦
⎜ ⎟ ,
(18) serve that a different number of particles influenced the results, namely,
the smaller number of particles, the higher force value was obtained, as
where: D1 - D5 are the material constants, p is the pressure and σeff is the can be seen in Fig. 11. The best correlation with the experiment can be
effective stress. noticed for the case with 6511 particles (0.05 mm distance). The dif-
An elastic – plastic constitutive model (or bilinear model) was ap- ferences in results are also related to the nature of the method, in which
plied to simulate behavior of the cardboard layers. The value of stress in the radius of interactions and the shape of the smoothing function are
the material is determined as follows, decisive factors.
In Fig. 12, a similar comparison of results is presented, however for
Table 2 the simulations with the ALE formulation for the aluminum sheets. The
Parameters for Johnson – Cook failure resultant force was estimated from the coupling interface between the
model for 1100-h12 aluminum [30].
aluminum sheets and the knife. In these cases, the same correlation
D1 [-] 0.07 between the mesh density and obtained results can be noticed. For the
D2 [-] 1.24 coarsest mesh (0.1 mm) force values seems to be overestimated. The
D3 [-] −1.14
obtained force values decreased as the characteristic length of the ele-
D4 [-] 0.14
D5 [-] 0.00 ment was reduced and the model with 0.05 mm mesh gave the force
predictions closest to the experimental results.

381
D. Gasiorek et al.
Fig. 10. Cutting surface on test stand for selected strip in the package: a) general view, b) upper side, c) lower side [19].
In Fig. 13, the results from each FE case that were closest to ex-
periment were compared with the actual tests. For the FEM-SPH
modeling the result from 0.05 mm case was selected, whereas for the
ALE modeling force history was taken from the simulation with
0.05 mm characteristic length of element. Despite the different model-
ling methods, a similar behavior of the model can be observed. How-
ever, rise times slightly differ from each other, what is caused by dis-
continuity in particle modelling (SPH case) of the aluminum sheets. In
other words, it took time for the knife to fully interact with the particles
due to gaps between them. On the other hand, the stiffness of the
material is correctly reflected in both methods. In the figure some os-
cillations appear, which are related with different aluminum layers and
their interaction with the knife. This phenomenon is reproduced better
in ALE modelling than in the FEM-SPH method. In the FEM-SPH
method declines in force are more evident, which is also related to the
discontinuity in particle modelling. Reduction of the space between
particles resulted in a smaller force drop (see Fig. 11). Based on the
characteristics presented in Figs. 10 and 11 work of the force and
average force were calculated, compared and juxtaposed with each
other (Table 4). It also reveals that the ALE modelling (more particular
0.05 mm case) slightly better reproduce aluminum behavior in com-
parison with the experiment. However, in other cases of FEM-SPH
modelling the values of work of force and average force are closer to the
experiment.
All simulations were carried using 192 Intel Xeon cores and Multi
Parallel Processing (MPP) LS-Dyna explicit. The ALE formulation, due
to its complex mathematical background and a large number of ele-
ments used, is computationally expensive. Hence the simulations were
carried out for approximately 18.5, 15 and 8 hours for the smallest to
the largest gap, respectively. Longer computations were also caused by
short time steps which were basically dependent on the element size
and the coupling and contact stiffness.
7000
6000
5000
4000
3000
2000
1000
SPH 0.05 mm SPH 0.07 mm
0
0 0.4 0.8
Fig. 11. Resultant force vs. displacement of the cutting knife obtained from experimental test and FEM-SPH simulations of the cutting process.
382
Journal of Manufacturing Processes 34 (2018) 374–388
For the FEM-SPH modelling, a time step varied due to the large
elements deformation as well as the contact procedure in which the
stiffness affects the integration step. The longest simulation taken ap-
proximately 30 hours, whereas, in the case of the coarsest mesh, a
“normal termination” prompt displayed after ∼10 hours. The time
steps are of a similar order as in the ALE simulations. However, shorter
integration steps are reflected in computational time as can be seen in
Table 5. Therefore, the SPH also requires a high computing power.
The results of numerical simulations and experimental tests at the
selected stages of the cutting processes are shown in Figs. 14–16. For
the FEM-SPH modelling the result from 0.05 mm case was selected,
whereas for the ALE modelling force history was taken from the si-
mulation with 0.05 mm characteristic length of element. The behavior
of the aluminum sheets during the cutting process is quite similar in all
cases and it can be noticed that discrete models behave correctly based
on physical assumptions presented in Fig. 1 and results presented by
other authors [5,7,8,10,18]. During the cutting of the first layer, both
bending and bulging in the bottom side of cut sheet directly under the
knife can be observed in experiment. This phenomenon is reflected in
numerical modelling using both the methods. In the next phases
forming of the plastic burr occurs, which is also well relatively reflected
in both the methods, it is more evident in the ALE modelling, where a
fluid-like material behavior is visible.
Figs. 17 and 18 show the plastic strain distribution during the same
stages as previously. In all time moments the plastic strain value equals
to 1.0 was set as a maximum value for plastic strain fringe displaying
for both methods. The obtained results slightly differ for each method
used, however overall material behavior in the local area within the
knife is similar. In all moments of time in the FEM-SPH modelling the
interaction with the knife resulted in more global response (the plastic
strain distribution can be observed on a wider area). In the ALE mod-
elling, at the beginning, the plasticization occurs in the contact area
SPH 0.1 mm Experiment
1.2 1.6
D. Gasiorek et al. Journal of Manufacturing Processes 34 (2018) 374–388

8000

7000

6000
5000

4000

3000
ALE 0.05 mm
2000 ALE 0.07 mm
1000 ALE 0.1 mm
Experiment
0
0 0.4 0.8 1.2 1.6

Fig. 12. Resultant force vs. displacement of the cutting knife obtained from experimental test and ALE simulations of the cutting process.

6000

ALE 0.05 mm
SPH 0.05 mm
Experiment
0
0 0.4

Fig. 13. Resultant force vs. displacement of the cutting knife obtained from experimental test and simulations of the cutting process.

Table 4 with the guillotine knife. In the FEM-SPH case the same can be ob-
Comparison table of two implemented methods. served, however the highest values of plastic strain are distributed over
Work of force [kJ] Average force [N]
entire surface of the knife which is in the contact with aluminum ma-
terial. In the next stages of the cutting process, the plastic zone pro-
Experiment 8.10 4048 pagates in the material: the plastic strain distribution is mainly accu-
FEM-SPH mulated in the first aluminum layer within the knife. In the t3 = 2.8 ms
0.05 mm 7.86 3992
0.07 mm 8.34 4187
stage the plasticization is evident in both sheets under the knife. This
0.1 mm 8.89 4326 phenomenon can be noticed for both methods of modelling, however
ALE once again the discontinuity property and the nature of the SPH method
0.05 mm 8.03 4023 are clearly visible. Based on the results presented it can be concluded
0.07 mm 10.41 5289
that the ALE model best represented the actual tests due to its reason-
0.1 mm 11.48 5793
able reproduction of material behavior and force during guillotine
cutting. Therefore, it was selected for further modelling and simulations
Table 5 aimed at investigating the influence of technological parameters such as
Computational cost of two implemented methods. knife velocity and cutting-edge angle.

Time step [s] Computational time [min]

4.2. Influence of selected technological parameters
FEM-SPH
0.05 mm 3.19e−9 ∼1760 In the next stage of the study, numerical simulations were per-
0.07 mm 3.65e−9 ∼1030 formed for ALE model with the finest mesh (0.05 mm) in order to
0.1 mm 7.21e−9 ∼610
analyze the influence of selected technological parameters and to va-
ALE
0.05 mm 6.40e−9 ∼1120 lidate the proposed method of modelling under different operating
0.07 mm 8.30e−9 ∼900 conditions. Therefore, three additional FE analyses were carried out:
0.1 mm 1.02e−8 ∼430
- with the cutting-edge angle β2 = 23.5°,
- with the cutting-edge angle β3 = 33.5°,
- with the knife velocity changed to v = 100 mm/s.

383
D. Gasiorek et al.
Fig. 14. The close-up view of the cutting process in t1 = 0.0012 s where the first aluminum sheet is deformed due to interaction with the knife.
The initial boundary conditions FEA corresponded to the experi-
ments and were identical to those in the mesh sensitivity studies. Only
the above-mentioned elements were changed in the model for each
case. The relationships between force and displacement obtained from
numerical simulations were compared with the experimental results. In
all figures, results from FEA with the basic cutting-edge angle
β1 = 28.5° are also included for comparison purposes.
In Fig. 19 the results for the cutting-edge angle β2 = 23.5° are pre-
sented. The oscillations related to different aluminum layer and inter-
action with the knife are also present similarly to the observations in
the sensitivity studies. The numbers of these rises and drops are iden-
tical between the actual and numerical tests. Although this character-
istic is reproduced quite well, the obtained force value in the initial
stage of interaction (first aluminum layer) is slightly lower than the
experimentally determined value. The maximum values obtained from
the modelling and the actual tests are approximately 5 kN observed in a
peak corresponded to the fifth aluminum layer. Reducing the cutting-
edge angle had an impact on the results: the obtained force values are
lower than the values from experiment with the basic cutting-edge
angle.
In the second case, the cutting process was simulated using the
Fig. 15. The close-up view of the cutting process in t2 = 0.0021 s where the first aluminum sheet is nearly cut due to interaction with the knife.
384
Journal of Manufacturing Processes 34 (2018) 374–388
cutting-edge angle β2 = 33.5°. Fig. 20 presents the force versus dis-
placement for the numerical simulation and experimental test. The
behavior of the model is well reproduced with visible oscillations cor-
responded to the cutting of subsequent aluminum layer. Moreover, in
this case the FEA maps an initial peak of force equal to ∼6.0 kN. The
maximum value of the force (∼6.5 kN) can be observed in the second
peak which corresponds to the second aluminum layer. Due to larger
cutting-edge angle the ratio between shear and normal force changed.
Thus, the effect of friction is more visible, resulting in larger cutting
force values compared to the cases with other cutting-edge angles, i.e.
23.5° and 28.5°.
The characteristics of the force vs. displacement curve for the last
case with cutting knife velocity v = 100 mm/s are presented in Fig. 21.
The oscillations and force values are smaller, in contrast to the other
two cases with knife velocity v = 200 mm/s. Despite underestimating
the force values compared to the experimental results, the model
properly reproduced characteristic. However, the analysis did not map
an initial peak of force equal to ∼4.0 kN. The maximum value of the
force (∼4.6 kN) can be observed in the fifth force peak. The source of
differences between the experimental test and the FE modelling may
result from the following: in the FEA there were gaps between each
D. Gasiorek et al. Journal of Manufacturing Processes 34 (2018) 374–388

Fig. 16. The close-up view of the cutting process in t3 = 0.0028 s where the first aluminum sheet is completely seperated and the knife strarts to interact with the
second aluminum sheet.

Fig. 17. Plastic strain distribution in FEM-SPH modelling of the cutting process of aluminum sheets for the previously selected and presented moments of time.

Fig. 18. Plastic strain distribution in ALE modelling of the cutting process of the aluminum sheets for the previously selected and presented moments of time.

385
D. Gasiorek et al. Journal of Manufacturing Processes 34 (2018) 374–388

0
Experiment - 28.5°
6000

5000

4000

3000

2000

1000 ALE 0.05 mm - 23.5°

Experiment - 23.5°
0
0 0.4 0.8 1.2 1.6

Fig. 19. Resultant force vs. displacement of the cutting knife obtained from experimental test and FEM-ALE simulation of the cutting process with cutting-edge angle
β2 = 23.5°.

Fig. 20. Resultant force vs. displacement of the cutting knife obtained from experimental test and FEM-ALE simulation of the cutting process with cuting-edge angle
β3 = 33.5.

Experiment - 28.5°, v = 200 mm/s

6000

5000

4000

3000

2000

1000 ALE 0.05 mm - 28.5°, v = 100 mm/s

Experiment - 28.5°, v = 100 mm/s
0
0 0.4 0.8 1.2 1.6

Fig. 21. Resultant force vs. displacement of the cutting knife obtained from experimental test and FEM-ALE simulation of the cutting process with knife velocity
v = 100 mm/s.

sheet, whereas in the actual testing the bundle of 90 aluminum sheets deviations). This applies to all investigated cases.
were pressed against the table using screws; in the FEA there was an Based on the characteristics presented in Figs. 19, 20 and 21 work of
ideal representation of the sheets, while in the experimental tests each the force and average force were calculated, compared and presented in
sheet could have slightly different geometry (with acceptable Table 6. In all presented cases, the ALE model produced satisfactory

386
D. Gasiorek et al. Journal of Manufacturing Processes 34 (2018) 374–388

Table 6
Comparison table of ALE model and experiment (different knife velocities and
cutting-edge angles).
Cutting edge Knife velocity Work of Average force
angle [°] [mm/s] force [kJ] [N]
23.5 200 Experiment 6.72 3045
FEA (ALE) 6.45 3146
28.5 200 Experiment 8.10 4048
FEA (ALE) 8.03 4023
33.5 200 Experiment 9.32 4690
FEA (ALE) 9.54 4801
28.5 100 Experiment 7.79 3623
FEA (ALE) 7.54 3456
outcomes. The influence of cutting-edge angle is evident: the lowest
value of average force was obtained with smallest angle of knife cut-
ting-edge. The same relationship can be observed for the values of work
of force. As expected, the values of average force and work of force for
the case with the lower velocity of the knife are smaller than for the
analysis with velocity v = 200 mm/s.
5. Conclusions
One of the main objectives of the paper was to demonstrate the
capabilities of the numerical modelling in terms of cutting process of
metal sheets. The FE analyses were motivated by the experimental
testing where the cutting operation was carried out using the patented
and original test stand developed by one of the authors (see [19]).
Based on the actual tests modelling and simulation of the multi-material
cutting of several separate aluminum sheets bundles was performed.
Numerical simulations were carried out using the LS-Dyna commercial
explicit code with two available techniques for modelling interactions
occurring during the cutting process: coupling of FEM and SPH methods
and by using ALE formulation. One of the main aims of the study was
also to perform a mesh (particles) sensitivity study to assess the influ-
ence of the mesh size on the obtained results.
The computed force characteristics in all numerical simulations
were examined and compared with each other and with the experi-
mental data. The influence of mesh density and particles number was
noticed: for the smaller number of particles the highest values of force
were obtained (Fig. 11). In the ALE modelling the same correlation
between the mesh density and the obtained results was noticed
(Fig. 12).
The SPH method is vulnerable to the regularity of particles dis-
tribution, its formulation and also bulk viscosity parameter (with a
specific SPH formulation). In the case of the ALE modelling the cou-
pling procedure is one of the most important problems to be dealt with
and the proper fluid - structure interaction parameters have to be ad-
justed for a specific case. Both methods were found to be adequate for
the presented guillotine cutting simulations but simultaneously they
have some advantages and disadvantages. However, all in all, the ALE
modelling seems to be the most suitable and efficient method due to its
reasonable reproduction of material behavior during guillotine cutting.
Therefore, it was selected for further modelling and simulations aimed
at investigating the influence of technological parameters such as knife
velocity and cutting-edge angle and to validate the proposed method of
modelling under different operating conditions. A reasonable agree-
ment between the numerical and experimental results were observed.
The adopted level of discretization gave the possibility to obtain
comparable and similar results as in the experiment tests. It was as-
sumed that this is a compromise between simulations of multi-layered
cutting process sheet aluminum bundle in microscopic and macroscopic
levels while taking into account layers layout and possibility to verify
the results with available experimental data.
The model after modifications (using a finer mesh) can be used in
more detailed investigations for optimizing and more thorough analysis
of the presented cutting process in micro level.
Acknowledgements
This research was carried out with the support of the
Interdisciplinary Centre for Mathematical and Computational Modeling
(ICM) University of Warsaw under grant no GB65-19.
References
[1] Golovashchenko SF. A study on trimming of aluminum autobody sheet and devel-
opment of a new robust process eliminating burrs and slivers. Int J Mech Sci
2006;48:1384–400. http://dx.doi.org/10.1016/j.ijmecsci.2006.07.007.
[2] Hilditch TB, Hodgson PD. Development of the sheared edge in the trimming of steel
and light metal sheet: part 1 - experimental observations. J Mater Process Technol
2005;169:184–91. http://dx.doi.org/10.1016/j.jmatprotec.2005.02.267.
[3] Hilditch TB, Hodgson PD. Development of the sheared edge in the trimming of steel
and light metal sheet: part 2 - mechanisms and modeling. J Mater Process Technol
2005;169:192–8. http://dx.doi.org/10.1016/j.jmatprotec.2005.02.267.
[4] Ilinich AM, Golovashchenko SF, Smith LM. Material anisotropy and trimming
method effects on total elongation in DP500 sheet steel. J Mater Process Technol
2011;211:441–9. http://dx.doi.org/10.1016/j.jmatprotec.2010.10.014.
[5] Bohdal Ł. Application of a SPH coupled FEM method for simulation of trimming of
aluminum autobody sheet. Acta Mech Autom 2016;10:56–61. http://dx.doi.org/10.
1515/ama-2016-0010.
[6] Marusich T, Ortiz M. Modelling and simulation of high‐speed machining. Int J
Numer Methods Eng 1995;38:3675–94. http://dx.doi.org/10.1002/nme.
1620382108.
[7] Olovsson L, Nilsson L, Simonsson K. ALE formulation for the solution of two-di-
mensional metal cutting problems. Comput Struct 1999;72:497–507. http://dx.doi.
org/10.1016/S0045-7949(98)00332-0.
[8] a Raczy, Elmadagli M, Altenhof W, Alpas A. An eulerian finite-element model for
determination of deformation state of a copper subjected to orthogonal cutting.
Metall Mater Trans A 2004;35:2393–400. http://dx.doi.org/10.1007/s11661-006-
0219-3.
[9] Madaj M, Píška M. On the SPH orthogonal cutting simulation of A2024-T351 alloy.
Procedia CIRP 2013;8:152–7. http://dx.doi.org/10.1016/j.procir.2013.06.081.
[10] Limido J, Espinosa C, Salaün M, Lacome JL. SPH method applied to high speed
cutting modelling. Int J Mech Sci 2007;49:898–908. http://dx.doi.org/10.1016/j.
ijmecsci.2006.11.005.
[11] Heisel U, Zaloga W, Krivoruchko D, Storchak M, Goloborodko L. Modelling of or-
thogonal cutting processes with the method of smoothed particle hydrodynamics.
Prod Eng 2013;7:639–45. http://dx.doi.org/10.1007/s11740-013-0484-0.
[12] Husson C, Correia JPM, Daridon L, Ahzi S. Finite elements simulations of thin
copper sheets blanking: study of blanking parameters on sheared edge quality. J
Mater Process Technol 2008;199:74–83. http://dx.doi.org/10.1016/j.jmatprotec.
2007.08.034.
[13] Mao H, Zhou F, Liu Y, Hua L. Numerical and experimental investigation of the
discontinuous dot indenter in the fine-blanking process. J Manuf Process
2016;24:90–9. http://dx.doi.org/10.1016/j.jmapro.2016.08.001.
[14] Duran A, Nalbant M. Finite element analysis of bending occurring while cutting
with high speed steel lathe cutting tools. Mater Des 2005;26:549–54. http://dx.doi.
org/10.1016/j.matdes.2004.07.028.
[15] Saglam H, Yaldiz S, Unsacar F. The effect of tool geometry and cutting speed on
main cutting force and tool tip temperature. Mater Des 2007;28:101–11. http://dx.
doi.org/10.1016/j.matdes.2005.05.015.
[16] Barlow AJ, Maire PH, Rider WJ, Rieben RN, Shashkov MJ. Arbitrary
Lagrangian–Eulerian methods for modeling high-speed compressible multimaterial
flows. J Comput Phys 2016;322:603–65. http://dx.doi.org/10.1016/j.jcp.2016.07.
001.
[17] Gurusamy MM, Rao BC. On the performance of modified Zerilli-Armstrong con-
stitutive model in simulating the metal-cutting process. J Manuf Process
2017;28:253–65. http://dx.doi.org/10.1016/j.jmapro.2017.06.011.
[18] Gąsiorek D. The application of the smoothed particle hydrodynamics (SPH) method
and the experimental verification of cutting of sheet metal bundles using a guillo-
tine. J Theor Appl Mech 2013;51:1053–65.
[19] Gąsiorek D. Metallographic investigations of metal plate edges after cutting. J Mater
Process Manuf Sci 2013;60:31–8.
[20] Hallquist J. LS-DYNA® theory manual. 2006.
[21] Bil H, Kiliç SE, Tekkaya AE. A comparison of orthogonal cutting data from ex-
periments with three different finite element models. Int J Mach Tools Manuf
2004;44:933–44. http://dx.doi.org/10.1016/j.ijmachtools.2004.01.016.
[22] Baranowski P, Janiszewski J, Malachowski J. Study on computational methods
applied to modeling of pulse shaper in split-Hopkinson bar. Arch Mech
2014;66:429–52.
[23] Baranowski P, Damaziak K, Malachowski J, Sergienko VP, Bukharov SN. Modeling
of abrasive wear by the meshless smoothed particle hydrodynamics method. J Frict
Wear 2016;37:94–9. http://dx.doi.org/10.3103/S1068366616010037.
[24] Schwer LE. Aluminum plate perforation : a comparative case study using lagrange
with erosion, multi-material ALE, and smooth particle hydrodynamics. 7th Eur LS-
DYNA Conf. 2009. p. 28.

387
D. Gasiorek et al.
[25] Liu MB, Liu GR. Smoothed particle hydrodynamics (SPH): an overview and recent
developments vol. 17. 2010. http://dx.doi.org/10.1007/s11831-010-9040-7.
[26] An Introduction to Meshfree Methods and Their Programming. n.d.
[27] Baranowski P, Malachowski J, Mazurkiewicz L. Numerical and experimental testing
of vehicle tyre under impulse loading conditions. Int J Mech Sci 2016;106:346–56.
http://dx.doi.org/10.1016/j.ijmecsci.2015.12.028.
[28] Malachowski J. Modelling and testing on interaction of solid body-gas at the
pressure impulse influence on elements of pipeline (in polish). Warsaw: BEL Studio;
2010.
[29] Haufe A, Weimar K. Fluid-structure-interaction effects in airbag out-of- position
load cases : an introduction to the ALE- framework in LS-DYNA. 5th Eur LS-DYNA
Users Conf. 2005. p. 5c–9.
388
Journal of Manufacturing Processes 34 (2018) 374–388
[30] Iqbal MA, Gupta NK. Energy absorption characteristics of aluminum plates sub-
jected to projectile impact. Lat Am J Solids Struct 2008;5:259–87.
[31] Raczy A, Altenhof WJ, Alpas AT. An eulerian finite element model of the metal
cutting process. 8 Th Intern LS-DYNA Users Conf Met Form. 2004. p. 11–26.
[32] Giampieri A, Perego U, Borsari R. A constitutive model for the mechanical response
of the folding of creased paperboard. Int J Solids Struct 2011;48:2275–87. http://
dx.doi.org/10.1016/j.ijsolstr.2011.04.002.
[33] Murayama M, Nagasawa S, Fukuzawa Y, Yamaguchi T, Katayama I. Orthotropic
effect and strain dependency of paperboard on load characteristic of center bevel
cutter indented on paperboard. J Mater Process Technol 2005;159:199–205. http://
dx.doi.org/10.1016/j.jmatprotec.2004.03.033.