VOLUME 20

ISSUES 1-2
of Achievements in Materials January-February
and Manufacturing Engineering 2007

Numerical determination of the forming

limit diagrams
T. Pepelnjak*, K. Kuzman
Faculty of Mechanical Engineering, University of Ljubljana,
Aškerčeva 6, SI-1000 Ljubljana, Slovenia
* Corresponding author: E-mail address: [email protected]
Received 30.10.2006; accepted in revised form 15.11.2006

Analysis and modelling

Abstract
Purpose: At present the industrial practice demands a reliable determination of forming limits which assures
the prediction of properly selecting the forming process in a digital environment. Therefore, technological limits
defined with the forming limit diagrams (FLDs) have to be known. The experimental evaluation of FLDs for
sheet metal is time consuming and demands expensive equipment. The experimental work could be omitted by
predicting the FLD with numerical simulations.
Design/methodology/approach: The paper presents a methodology to determine the entire range of the FLD
for sheet metal in a digital environment. The Marciniak testing procedure simulated with the FEM program
ABAQUS was selected to determine the FLD. To assure the reliability of the developed method, different
materials were analysed: two types of deep drawing steel, an aluminium 3000 alloy, and a Ti-alloy. The selected
materials have different mechanical properties and sheet thicknesses ranging from 0.5 mm to 1.23 mm. For the
verification of numerically obtained results parallel experimental determinations of the FLDs were performed
showing a good correlation between the FLDs obtained by both approaches.
Findings: A specially developed method for the evaluation of the thickness strain as a function of time as well as
the first and the second time derivation of the thickness strain enable the determination of the onset of necking.
Research limitations/implications: The presented method of the digital evaluation of the FLDs is still in
a developmental phase and needs further improvements for industrial practice. However, in some cases the
numerical approach had already been used for a fast prediction of the FLD prior to performing the experiments.
At the current level the developed program still needs an expert to support it in some critical decisions.
Originality/value: Considering some methodological improvements and automation procedures the developed
method could be used in everyday practice.
Keywords: Numerical techniques; Plastic forming; Sheet metal; Forming limit diagram

1. Introduction
1. Introduction represents accurate material data of formed sheet metal. There are
several data describing the material behaviour of sheet metal, such
as its flow curve, coefficients of normal and planar anisotropy,
The processing of sheet metal is affected by numerous
yield stress and the formability limits, respectively.
parameters, such as material behaviour, tooling, lubrication,
necessary machinery, etc. All mentioned parameters have an
influence on technological limitations of sheet metal forming, its
wrinkling, necking and rupture. The last two are among the most 2. Forming
2. 
Forming limit diagram
limit diagram
critical defects appearing in sheet metal fabrication.
One of the most important input parameters for the numerical The development of the production of new parts, its set-up of
simulations of the forming process in a digital environment forming an operation or several progressive operations demands

© Copyright by International OCSCO World Press. All rights reserved. 2007 Short paper 375
 Journal of Achievements in Materials and Manufacturing Engineering Volume 20 Issues 1-2 January-February 2007

good knowledge of the attainable forming limits. It is necessary to analysing the material's thinning in order to determine the onset of
define the limit up to which the material can be formed. Above necking by the Nakazima method. The method is applicable to the
this limit the localisation with necking and fracture appears. entire range of deformation states of the FLD.
Nowadays it is particularly impossible to perform the analyses of Unfortunately, the evaluation of various materials showed that
the forming process in a digital environment when fracture data the leaping change of thinning acceleration (second thinning
are not known as an input parameter. derivation in time) is not at all times significant to the point that
The idea of a sheet metal evaluation with a diagram consisting the determination of the necking point could be doubtlessly
of the first main strain versus the second one which was named determined. This is the case at the right hand side of the FLD with
the forming limit diagram (FLD) was introduced in the late 1960s M2>0 in particular as well as with some materials on the entire
by Keeler and Goodwin [1]. With this so-called forming limit range of the FLD. Therefore, the research was focused on the
diagram it is possible to evaluate different strain conditions in the criterion which would allow a simple and reliable determination
same diagram and determine its fracture limits for a particular of the necking limit.
strain combination – Fig. 1.

3. 
Numerical
3. Numerical andand experimental
experimental analyses
analyses of forming limit
of forming
diagramslimit diagrams

3.1 Methodology
3.1. 
Methodology for temporal and spatial
for temporal and
determination of material neckingofbymaterial
spatial determination the
necking by the Marciniak test
Marciniak test
Based on Brun’s idea, the presented paper introduces the
methodology of spatial and temporal identification of the necking of
critical nodes of the FEM model and the determination of their
corresponding strains M1 and M2, respectively. Using this method,
time and spatial necking of a particular specimen can be predicted.
Fig. 1. Forming limit diagram
Considering the numerical approach of determining the FLD it is
The first pioneer works of the experimental determination of necessary to determine at which point and where on the specimen the
the FLD by Keeler and Goodwin were followed by numerous material rupture occurs. To assure fast and reliable data processing a
methodology for an automated evaluation of all specimens' nodes and
research activities ranging from the improved methods for
their evaluated parameter were sought. It was deduced that by means
experimental determination of the FLD to analytical concepts
of the evaluation of thinning and its acceleration in time it is possible
allowing the calculation of the FLD up to numerical approaches
to determine the necking of the analysed material.
which are based on the simulations of various testing methods in a The improved method used for the strain evaluation in a digital
digital environment. environment was performed in a three-step analysis, mostly an
Among several developed experimental tests two of them automated one in order to shorten processing time needed to obtain
have shown exceptional suitability for the evaluation of the entire the major and minor plane strains of critical nodes [3].
range of the FLD combined with simple tooling and an Step 1: Analysis of sheet thickness in time, searching for
experimental procedure – these are the Nakazima and the critical nodes and the determination of their corresponding node
Marciniak test. A Polish scientist Marciniak introduced a method numbers. In this first evaluation step, a huge amount of analysed
for the determination of the FLD with a flat punch [1]. The test data is already drastically reduced. Nodes with minimal
tool consists of the drawing die, the blankholder and the punch. thicknesses were inserted into the Abaqus post processor to obtain
The punch has an even and a partially sunk forehead. Various their thinning as a function of forming simulation time.
strain conditions are achieved by different widths of the analysed Step 2: Calculation of first and second temporal derivation of
specimens which enable the determination of the entire range of thinning for selected critical nodes. Due to a fast local change of
the FLD with one tool geometry only. During the testing sheet thickness at the necking point the thickness strain leaping
procedure the even punch forehead causes the plane strain changes its value. The acceleration of thickness deformation
conditions in the analysed specimen area. (second time derivation) defined as
Parallel to the experimental approach a computer-aided
numerical approach in a digital environment was sought and (1)
introduced in the last years. The finite element method,
commonly used for the numerical solving of forming processes was stored for all analysed nodes which are obtained as the output
with the FLD as an input parameter, was selected to analyse the of the first step.
Nakazima and Marciniak testing procedure. Comparative studies Step 3: A signal processing procedure was developed to purify
were done in the area of numerical determination of the forming data from all the stored nodes in order to obtain the critical node
limit curve (FLC) where different criteria of necking and fracture (location on the specimen) and the corresponding time when the
were presented. One of the first in this research area was Brun [2] necking at this node occurs. The node at which the maximum of

376 Short paper T. Pepelnjak, K. Kuzman

 Analysis and modelling

Table 1.
Material input data for selected sheet metals
t0 E U Rp C n r0 r45 r90
Q
(mm) (MPa) (kg/m3) (MPa) (MPa) (1) (1) (1) (1)
hot
galvanised 0.64 207 000 7850 0.3 410 720 0.153 0.948 0.793 1.003
steel
DDQ steel 0.5 207 000 7850 0.3 266 630 0.187 1.04 0.88 1.49
Al 3003 1.23 70 000 2750 0.33 36.7 197 0.24 0.538 0.588 0.452
Ti-alloy 0.9 143 000 4505 0.32 313 570 0.115 2.69* 4.0* 4.5*
* Comment: All coefficients of anisotropy represent r20 values except by the Ti-alloy where they are r10 values.

first appears (at minimal time) was assumed as the node where The FEM model of the Marciniak test consists of three rigid
the onset of necking started. tool parts – the die, the holder plate and the punch – as well as the
Nodes selected from each specimen’s geometry in step 3 were specimen and the guiding plate which are considered elasto-
considered to be at the necking point of the analysed material. The plastic – Fig. 2.
When the simulation and an experimental set-up is planned
major and minor strains for these nodes were collected from the
the attention has to be put to higher ductility and thickness of the
Abaqus result file and put into the FLD diagram. The entire three-
guiding plate in comparison to the analysed specimen. All formed
step procedure was repeatedly performed for all simulated
materials, specimens and support plates, undergo the elasto-
specimen geometries of each analysed material.
plastic Hollomon potential material law. At the FEM simulations
the Coulomb friction law was assumed for all surfaces in contact.

3.2 Materials
3.2. 
Materials

The presented three-step methodology for the determination

of the FLD was performed on several different materials in order
to verify the reliability of the proposed numerical approach. To
assure different material parameters and analyse the reliability of
the developed methodology for determining the FLD under
different conditions, the following materials were selected:
x Material 1: Hot galvanised deep drawing sheet metal with a
thickness of t0 = 0.64 mm.
x Material 2: Low carbon deep drawing steel with good
ductility and an initial thickness of t0 = 0.5 mm.
x Material 3: The aluminium alloy 3003 with a thickness of t0 =
1.23 mm is a soft aluminium alloy.
x Material 4: The last material is a Ti-alloy with an initial Fig. 2. FEM model of the Marciniak test
thickness of t0 = 0.9 mm.
On the other hand, in all cases of numerical determination of
the FLD the material data of the analysed material representing the 3.4 EExperimental
3.4.  design, results
xperimental design, results and
and
input data for the FEM simulation were obtained experimentally discussion
discussion
with standardized uni-axial tensile tests. The measured and
calculated material parameters are presented in Table 1. The experimental determination of the FLD underwent a
forming procedure for six selected specimen widths which had to
be identical with those selected at the numerical approach.
For the evaluation of the acquired frames obtained wth CCD
3.3 FEM
3.3. 
FEMmodel
model
camera with a frequency of 5 frames/sec a specialised program
was used to enable a precise determination of the tearing limit. A
For the evaluation of the forming limit diagram in a digital deformation state defining the rupture limit at a particular
environment with the experimental approach the Marciniak specimen shape was analysed from the last grabbed frame before
method was used. With this method it is possible to simplify the the material’s tearing. The used program also enables the
analysis of the processed data to the plain strain state of the evaluation of the deformation path which has to be linear in order
observed area. As already presented in the prior chapter, with the to properly determine the FLD.
Marciniak approach it is possible to determine the entire range of The experimental evaluations of all four materials have shown
the FLD with the same tooling geometry and different specimen different material behaviour from the necking limit onwards. While
geometries. The strip-shaped specimens with various widths both steels have shown good ductility the difference between the
define different strain paths of the observed FLD – Fig 1. necking and the tearing limit was very small at the Al3003 and Ti-

Numerical determination of the forming limit diagrams 377

 Journal of Achievements in Materials and Manufacturing Engineering Volume 20 Issues 1-2 January-February 2007

alloy. Once all specimens for a particular material were evaluated FLD by both steels. Unfortunately, there are still some problems
with both approaches the comparative FLDs were constructed. The detecting the proper course of the FLD at the right side of the
forming limit diagrams for all analysed materials are presented as numerically determined FLD of the Al-3000 alloy where the
follows: low carbon deep drawing steel - Fig 3, hot dip galvanised evaluations of round specimens resulted in excessive necking
steel – Fig. 4, Al3003 alloy – Fig 5, and Ti-alloy – Fig 6. values. The experimental determination of the FLD for the Ti-
alloy has shown an interesting fact that this material merely resists
0.8
a minor bi-axial load and has a very low level at the right hand
side of the FLD. The comparison of the numerical and
0.7
experimental approach is in this case very good as well.
0.6

0.5

0.4
4. Conclusions
4. Conclusions
M

0.3

0.2
The presented system for a numerical determination of the
0.1 FLD was proven on four materials with different chemical
0 compositions and mechanical properties. Comparative results of
-0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 steel materials and Ti-alloy have shown a good correlation
M
between the experimental and numerical results. In the case of the
Experimental results FEM approach Al-3003 alloy the right side of the FLD is not as accurate as it has
been expected. The improvements of the description of material
Fig. 3. Numerical and experimental FLD of low carbon deep models in the Abaqus for those materials are the first stage for
drawing steel better accuracy of the FLD determination. Furthermore, a wider
range of different material thicknesses and qualities shall be
performed to verify and consequently improve the reliability of
the system. Some works in this field were already have already
been performed with very thin tinplate materials and their
evaluation of the digitally determined FLD is already in progress.
The final objective of the developed system for a numerical
FLD determination is its wider application also for those
materials where the experimental approach with available
equipment cannot be applied.

0.8

0.6
M

Fig. 4. Numerical and experimental FLD of dip galvanised deep 0.4

drawing steel
0.2

0
-0.6 -0.4 -0.2 0 0.2
M
Experiments FEM Approach

Fig. 6. Numerical and experimental FLD of Ti-alloy

References
References
[1] D. Banabic, H.J. Bunge, K. Pöhland, A.E. Tekkaya,
Formability of Metallic Materials, Springer Verlag,
Germany 2000.
[2] R. Brun, A. Chambard, M. Lai P. de Luca, Actual and virtual
testing techniques for a numerical definition of materials,
Fig. 5. Numerical and experimental FLD of Al-3003 alloy Proc. NUMISHEET’99, Besançon, France, 1999, 393-398.
[3] A. Petek, T. Pepelnjak, K. Kuzman, New method for
The comparative evaluations have shown an excellent determination of FLD in digital environment, Journal of
correlation between numerically and experimentally obtained Mechanical Engineering, Vol. 51 (2005), 6, 330-345.

378 Short paper READING DIRECT: www.journalamme.org