52nd ICFG Plenay Meeting, 15-18 September 2019, Donostia-SanSebastian-Basque Country-Spain

New 3D Adaptive Remeshing Techniques Applied to Piercing,

Trimming and Binding Simulations

Dr-Ing. Richard Ducloux1,a , Dr.-Ing José Alves1,b , Dr-Ing. Ugo Ripert1,c

1
TRANSVALOR, 950 Avenue Roumanille CS 40237 Biot, 06904 Sophia Antipolis - FRANCE
a) richard.ducloux@transvalor.com
b) jose.alves@transvalor.com
c) ugo.ripert@transvalor.com

Abstract

A so-called ‘kill element’ strategy is often used to incorporate piercing or trimming functionality in manufacturing
process simulation packages. A ‘damage’ function is evaluated at every node/element of the FEM mesh and when this
value reaches a given trigger value, then the affected element is deleted.
The main advantage of such strategy is its flexibility. Users can easily access to a list of ‘damage models’ as well as to
the trigger value. It also gives the possibility to effectively trim a workpiece in two separate parts which is difficult to
achieve within the frame of continuous mechanics without the help of dedicated models.
As usual, everything comes at a price and several drawbacks associated with this simplicity can be identified:
 Element deletion creates numerical material loss
 Topologies in the remaining mesh are often intricate and may be difficult to manage
 Characterization of damage behavior is mesh-dependent
 The surface shows rugosity after piercing and trimming, corresponding to the local edge element size
To overcome these difficulties, Transvalor has developed specialized techniques to improve the mesh in case of element
deletion.
 Automatic adaptive mesh refinement in the ‘killing zone’ to control the volume loss. It is possible to
significantly decrease/increase the local mesh size in order to foresee the arrival fracture using FORGE®’s
ability to efficiently manage local aspects of large meshes as well as to smooth-out surfaces just after “killing”
damaged elements and increase the mesh size to avoid unnecessary refinements.
 Specific work is done to only create good topologies, allowing remeshing algorithms to work on element shape
factors.
 A dedicated smoothing algorithm is applied to obtain good shape and a usable mesh to simulate subsequent
forming stages.
These new capabilities are demonstrated on various industrial applications, including
®
piercing, trimming and a new binding
example used to fasten individual metal sheets together by a process called FDS (Flow Drill Screw).
Keywords: Finite Elements; Fracture Mechanics; Automatic Mesh adaptation

1. Introduction

One of the biggest remaining challenges in the field of material forming simulation is simulation of trimming or
piercing processes. Most available material-forming simulation software is based on FEM methods applied to solve

International Cold Forging Group 1

52nd Plenary Meeting, San Sebastian
 52nd ICFG Plenay Meeting, 15-18 September 2019, Donostia-SanSebastian-Basque Country-Spain

the continuous mechanics equations. These do not take into account pre-existing or induced discontinuities
(fracture). Some specific algorithms have to be implemented in order to achieve cutting in such framework. A
possible idea is to use an additional variable (typically level-sets or phase-fields [1]) to define where the material is
or is not. The mesh remains continuous, but material can be split in one or several pieces. Another technique is to
use an XFEM type approach where the direction of cracks is evaluated, and additional DOF’s are added in the
elements along the crack line to properly describe it. According to some authors, the main advantages of such a
technique would be to avoid remeshing and subsequent interpolation phases as well as proper control of the added
degrees of freedom [2].
A third approach is based on the deletion of elements when an evaluation of the damage of such element becomes
greater than a given trigger value [3,4]. The strategy presented here refines this approach. FORGE® software has
been designed to efficiently manage the mix of high non-linearities, fine mesh, complex geometries and large
deformations so remeshing and reasonable increase of nodes number can be managed. The idea is to take advantage
of existing capabilities and refine them when needed and to add features only to remove remaining blocking points.
The first part of this paper is dedicated to specific improvements made to adapt FORGE® to cutting simulations.
The second part describes practical examples. They are not ‘real-world examples’, validated against experiments,
but can be seen as proof of concept. Their complexity is similar to that seen in real life. Their purpose is to show that
it is possible to go further than what has been possible before.

2. Recent improvements

2.1. Mesh adaptation and node duplication

Continuous mechanics does not account for the inclusion or apparition of discontinuities in matter such as
damage-based fracture in its intrinsic and basic concept. If a tensile test is simulated, a necking phenomenon will
appear but fracture propagation will not take place because it is not described by standard formulations. To
overcome this difficulty, several software packages, including FORGE®, use the concept of damage. Damage is
evaluated at each time step and at each element of the FE mesh. When the damage value of a given element reaches
a trigger value, the element is deleted. Cutting is achieved when all the elements along the rupture surface are
deleted. The drawback is that the rupture surface is very far from a smooth surface. The resulting irregular mesh
topologies may result in the fact that this mesh is not suitable for use in a subsequent simulation. The case
represented below (Fig 1) is a typical blanking application. The top (yellow) punch moves downwards while the
workpiece lies on the bottom (green) die. The mesh is refined where the cutting takes place. The zoomed picture
displays the kind of mesh which can be expected.

International Cold Forging Group 2

52nd Plenary Meeting, San Sebastian
 52nd ICFG Plenay Meeting, 15-18 September 2019, Donostia-SanSebastian-Basque Country-Spain

Figure 1 – Blanking simulation

The mesh is automatically refined when the damage value approaches the trigger value in order to improve the
situation. In so doing, the ‘numerical rugosity’ dimension is reduced, which can be seen in Figure 1. Also, nodal
singularities can be created in such a way that a group of elements is connected by a single node. This situation is
critical to the mechanical solver as well as the remeshing algorithm. Specific work has been performed on the mesh
topology to duplicate such singular nodes, which then create a natural split in the mesh. A benefit of the combined
effect of pre-refinement and split of critical nodes is that the produced mesh topology is cleaner, and it is easier to
work on these nodes to improve element shape factors. Figure 2 shows a regular mesh (left side) and an adapted and
improved mesh (right side) during a blanking operation.

Figure 2 – automatic refinement and topology improvement

2.2. Removal of separated chips and debris

Chips are created and later are separated from the main body during machining and drilling processes. Depending
on the purpose of the simulation, one might want to keep or discard them. The picture below shows how chips are
created and then broken. In some other cases, the removed material is made of small particles and chips are never
really created. After separation, they move away from the main body due to inertia. By default, remeshing after
remeshing, FORGE® will reduce the volume of this small debris until it completely disappears. This saves the
number of nodes and elements and simplifies contact analysis. Thus computation is faster. One may also keep them
if desired.

International Cold Forging Group 3

52nd Plenary Meeting, San Sebastian
 52nd ICFG Plenay Meeting, 15-18 September 2019, Donostia-SanSebastian-Basque Country-Spain

Figure 3 – chip separation

2.3. Local geometry smoothing

One of the new frontiers is the possibility to chain all the manufacturing process stages simulations and then to
continue with dimensioning simulation. It is then important that cutting or piercing simulations output a good
quality mesh. Some local smoothing algorithms have been developed to improve mesh in the cutting area. It is very
important to ensure that these algorithms work locally in order to avoid any unnecessary geometrical changes in
areas which have not been affected by the cutting process. A side benefit of this smoothing is that one can reduce the
number of nodes in the model. Figure 4 illustrates how a smooth geometry can be obtained from the blanking
simulation with a relatively coarse mesh in the case presented in paragraph 2.1.

Figure 4 – Smoothing of the cut area

3. Practical examples

3.1. Drilling

The first application is simulating drilling a hole in a metal sheet. This kind of situation is the most widely used
to demonstrate the ability of various software to take into account ‘rupture effects’. Figure 5 shows two different

International Cold Forging Group 4

52nd Plenary Meeting, San Sebastian
 52nd ICFG Plenay Meeting, 15-18 September 2019, Donostia-SanSebastian-Basque Country-Spain

instants of the drilling operation near the beginning of the process. It is already possible to appreciate how new mesh
adaptation developments enable such complex simulation, even preserving geometrical representation of the
resulting debris.

Figure 5 - Drilling simulation at the very beginning and slightly after

This kind of simulation can be used to compare the effect of different drill types. The first difference is the shape
of the chip (Figure 6).

Figure 6 - Chip shape using different drills

But global observables vastly change. As it can be seen on the following picture (Figure 7), drilling force and
torque may also be very different.

International Cold Forging Group 5

52nd Plenary Meeting, San Sebastian
 52nd ICFG Plenay Meeting, 15-18 September 2019, Donostia-SanSebastian-Basque Country-Spain

Figure 7 - Force and torque vs drill displacement in both cases

Heating effects are also interesting as it may affect the metallurgy of the part. In our case, it is known that process
parameters such as ratio between displacement and rotation velocity, lubrication, etc., are more significant.

Figure 8 - Temperature distribution

Following a standard approach to the simulation and depending on the sheet thickness, this is expensive, even for
drilling a small percentage of the part. This makes it nearly impossible to answer some questions of interest. One
question is the shape of the hole in the exit area. In order to answer that, some clever approaches are possible. These
can include starting from a pre-drilled geometry. In this case, however, temperature distribution is unknown and not
taken into account. It is difficult to know in advance the effect of such an approximation on the final result as the
heating of the material is important and may affect the material behavior. Thanks to the improvement presented
earlier in this text, a complete drilling simulation can be achieved in several hours: 14 hours to reach the situation
corresponding to the lower right area of the picture below as opposed to several weeks without the automatic local
mesh adaptation.

International Cold Forging Group 6

52nd Plenary Meeting, San Sebastian
 52nd ICFG Plenay Meeting, 15-18 September 2019, Donostia-SanSebastian-Basque Country-Spain

Figure 9 – Drilling through the whole sheet

3.2. Screwing

New processes are coming on the market in the field of mechanical joining. Because these processes are new, a
new potential user’s knowledge may be limited. Therefore, the simulation should be used to improve the level of
expertise. One of these simulations is called FDS (Flow Drill Screw).
At a first glance, we could be tempting to do the simulation in 2D, but the actual filling of the threads would not
be properly taken into account. Figure 10 displays a typical situation including in the head area of the screw, some
room for material.

Figure 10 – FDS typical situation

This case is associated with many difficulties. It is a 3D case, with damage effects, several deformable bodies,
and large temperature variations. Figure 11 displays the situation at different stages of the process. A relatively
strong friction assumption has been used; this is why the thick sheet heat is quite high. As temperature increases,
thick sheet material flows while the thin sheet remains more rigid in comparison. The resulting geometry certainly
affects the quality of the binding and is probably difficult to guess without the help of the simulation.

International Cold Forging Group 7

52nd Plenary Meeting, San Sebastian
 52nd ICFG Plenay Meeting, 15-18 September 2019, Donostia-SanSebastian-Basque Country-Spain

Figure 11 – Temperature and shape evolution

3.3. Progressive Stamping

Progressive stamping includes various deformation operations such as bending, punching and even coining. The
process relies on an automatic feeding system that pushes a metal strip through several consecutive deformation
stages. At each stage, one or several operations are completed at the same time. Hence, the part is progressively
conveyed by the feeding system to the next stage until the final trimming operation where the finished part is
separated from the metal strip.

This means that not only one hole, but many holes have to be trimmed in this simulation. The simplest way to
simulate such a process would be to compute one station after the other. We have decided, however, to simulate four
stations in one single simulation in order to remain as close to the real process as possible.

Figure 12- Initial situation and situation after 1, 2, 3, 4, 5, 6, 7 seconds

The picture in the top left corner in Figure 12 corresponds to the initial situation. Each picture thereafter displays
the situation after the incremental displacement of the sheet from left to right in the first line followed by the second
line. The continuous increase of part complexity tends to generate a similar increase in mesh size. FORGE®’s
ability to keep a clean trimmed surface is a mandatory requirement to control the number of elements (Figure 13).

International Cold Forging Group 8

52nd Plenary Meeting, San Sebastian
 52nd ICFG Plenay Meeting, 15-18 September 2019, Donostia-SanSebastian-Basque Country-Spain

Figure 13 – Mesh adaptation and surface smoothing

Even if mesh is derefined after the trimming, it remains fine enough to keep the geometry and a decent description
of the residual stresses. Figure 14 shows the Von Mises equivalent stress distribution. Relatively smooth distribution
of the value indicates that a finer mesh would probably improve the quality of the result. But the fact that results in
different holes are quite similar leads us to conclude that we have been able to catch the right trend.

Figure 14 – Equivalent Von Mises stress distribution

4. Conclusion

Simulating trimming, piercing, cutting or drilling processes in real 3D industrial simulation has been very
challenging for packages using a traditional implicit FEM approach coupled with a ‘kill element’ strategy. Recent
progress has been made in meshing techniques mixing automatic adaptation, smoothing and ‘debris’ cleaning. This
makes many simulations which had been impossible or excessively time-consuming into those which can be
completed during a single day. All simulations have been computed using a Latham and Cockroft damage criterion
mainly because it is the simplest one. The next stage is now to select the most relevant criteria and parameters. Since
such simulations are now stable and relatively fast, one can use automatic inverse analysis approaches. These should
make it easier to identify the values of damage criterion parameters.

International Cold Forging Group 9

52nd Plenary Meeting, San Sebastian
 52nd ICFG Plenay Meeting, 15-18 September 2019, Donostia-SanSebastian-Basque Country-Spain

References

[1] R. Alessi, M. Ambati, T. Gerasimov, S. Vidoli, and L. DeLorenzis, “Comparison of phase-field models of fracture coupled with plasticity,” in
Advances in Computational Plasticity, pp. 1–21,Springer,2018.
[2] D. Koschier, J. Bender, N. Thuerey, Robust extended finite element for complex cutting of deformables, SIGGRAPH 2017.
[3] R. El Khaoulani, P.O. Bouchard, An anisotropic mesh adaptation strategy for damage and failure in ductile materials, Finite Elements in
analysis and design, Elsevier 2012.
[4] T.S Cao, Numerical simulation of 3D ductile cracks formation using recent improved Lode-dependent plasticity and damage models
combined with remeshing, International journal of solids and structures 2014

International Cold Forging Group 10

52nd Plenary Meeting, San Sebastian