Prod. Eng. Res. Devel.

(2007) 1:79–84
DOI 10.1007/s11740-007-0034-8

COMPUTER AIDED ENGINEERING

Mesoscopic model: advanced simulation of microforming

processes
Manfred Geiger Æ Stefan Geißdörfer Æ
Ulf Engel

Received: 9 February 2007 / Accepted: 9 March 2007 / Published online: 25 May 2007
Ó German Academic Society for Production Engineering (WGP) 2007

Abstract Micro technology, one of the future key tech- market volume for smallest parts is obvious [1] which has
nologies, is gaining increasing interest in forming com- to be satisfied by production industry. Depending on the
munity. Up to now, most of the smallest parts are being required functionality of the parts and the production
manufactured by machining technologies well suited for numbers different technologies are available for manufac-
the production of small series. Forming technology instead turing the parts like machining, moulding and forming. In
seems to be more suited to meet the demands of serial case of smallest metallic parts, up to now the preponderant
production due to its high production output. In the field of numbers of these parts are produced using machining
microforming technology, an important aspect concerning technologies like turning, grinding or milling. For small
the process layout as well as its optimisation is the ability batch production numbers these technologies may be jus-
to predict the forming processes by simulation methods. tifiable, if large quantities are requested, forming technol-
When scaling down process dimensions to microscale, ogy is more adequate due to its high efficiency and
size-effects appear and have to be considered. In this study, remarkable precision. However, investigations on micro-
fundamentals have been developed to built up a so-called forming processes have shown significant differences in the
mesoscopic model taking into account the occurring size- forming behaviour of smallest parts compared to conven-
effects at microscale. The influence of grain size and its tional length scale forming, preventing forming technology
position are taken into account by creating a synthetic from being frequently used at micro scale. Research
material structure and the calculation of individual material activities in microforming during the last decade have
properties for each of the synthetic grains. analyzed and described the occurring size-effects and
identified two main influences to be the size-effect on the
Keywords Production process
Finite element material flow and on friction. The first can be explained by
simulation
Mesoscopic model a dependency of the material flow on the grain structure, or
more precisely on the ratio between mean grain size and
part dimension [2]. This effect has been identified by
1 Introduction several microforming processes [3] and a first effort to
simulate such processes has been done by [4] which show
Micro technology is gaining an increasing relevance due to scale dependend results in a good agreement with the ones
the circumstance that many products of everyday life are obtained by experiments. This theory is well suited to de-
becoming smaller like consumer electronics, mobile scribe the forming behaviour from a general point of view.
phones and others. Following this trend, an expanding If additional aspects have to be considered, there is a
necessity to describe the material behaviour in a more
detailed way. This was realized by a so-called mesoscopic
M. Geiger (&)
S. Geißdörfer
U. Engel model [5, 6] where the so-far continuously described
Lehrstuhl für Fertigungstechnologie,
material within the finite element simulation model will be
Friedrich-Alexander-Universität of Erlangen-Nürnberg,
Egerlandstraße 11, 91058 Erlangen, Germany discretizised into individual objects which are representing
e-mail: geiger@lft.uni-erlangen.de the real material structure in an adequate way.

123
80 Prod. Eng. Res. Devel. (2007) 1:79–84

Fig. 1 a CuZn15 finegrained

with dG = 39 lm b CuZn15
heat treated with dG = 135 lm

2 Experimental approach

The flat upsetting test, suitable for investigations on the

influence of the material structure on microforming results,
is characterized by its robustness against changing friction
conditions and by a plain strain condition, if the geometry
is chosen adequately. The tests are carried out using a
universal testing machine UTS 5 k equipped with a HBM
5 kN force measurement system. The cylindrical speci-
mens for the experiments are made from CuZn15, as it is
frequently used in electronics production, with a diameter
of 0.5 mm and a length of 3 mm.
As it is necessary for a comparability of the tests, fric-
tion is kept constant by applying a solid lubricant MoS2. In
general, size-effects can be obtained by either scaling down
geometry according to similarity theorem or changing the
ratio between grain size and specimen dimension. In the
first case it is nearly impossible to ensure comparable
friction conditions between different length scales due to
the topography being very difficult to scale down.
In the chosen second case friction conditions are kept
Fig. 2 Size dependency of the process results
reproducible only varying the mean grain size applying
different heat treatment operations on the specimen
(Fig. 1). In order to get an adequately reliable statistical
basis for the experimental investigations, a number of 25 and Hall-Petch. In this approach, developed especially to
specimen per batch is chosen. The evaluation of the flat describe the phenomena in microforming, the material
upsetting tests is done in terms of mean load F as well as its within the simulation system is subdivided into different
scatter, characterized by standard deviation sF, respectively regions, each characterized by individual material proper-

variation coefficient sF =F: ties. In such a case of modelling where the material
In case of fine grains the total grain structure can be structure used for simulation is similar to real material
regarded as almost homogeneous with almost reproducible structure, it will even be possible to model local forming
results. However, for coarse grain, the grain structure might behaviour controlled by local grain structure. This requires
be more different from specimen to specimen even if the an algorithm capable to generate synthetic 2D as well as
nominal mean grain size is kept constant, causing an 3D grain structures with predefined parameters. Compre-
increasing scatter in material forming behaviour (Fig. 2). hensive analysis of available methods for synthetic grain
structure generation in terms of the reached mean grain size
and its standard deviation have confirmed the Monte-Car-
3 Theoretical background lo–Potts algorithm [9] to be well suited for the present
studies. Based on a minimization of the total system en-
The previously introduced mesoscopic model [5–7] is ergy, grain growth is controlled until the predefined
mainly based on metal physics theory given by Ashby [8] parameter of mean grain size is reached.

123
Prod. Eng. Res. Devel. (2007) 1:79–84 81

As it is given by the theory of Hall-Petch and Ashby [8],

the materials plastical response to external forces is mainly
controlled by grain size and the position of the grain within
the specimen. The first can be described by the correlation
between grain size and flow stress, as given by (1) where s
is the shear stress, s0 the critical shear stress, K the Hall-
Petch slope and dG the mean grain size.

K
s ¼ s0 þ pffiffiffiffiffiffi ð1Þ
dG

If forming behaviour on scale of some few grains is

to be considered, (1) is changing to (2) taking into account Fig. 3 Synthetic material structure and material properties distribu-
the influence of neighbouring grains on the mechanical tion in case of coarse grained and fine grained material
properties of a single grain.
pffiffiffi
m1;2 si d knG represents the mean value of yield stresses knG of
s ¼ s0 þ pffiffiffi ð2Þ
dG the adjoining grains, related to the real contact area fraction
ac/aG of the grains in case of 3D studies.
In (2), m1,2 represents a transformation matrix consid- In case of a free surface grain boundary it can easily be
ering the different sliding systems in adjoining grains, s0 seen by (6) that there kn*Gac is zero and thus decreasing
is the critical shear stress in the considered grain and the the Hall-Petch slope Ki for the considered grain. Fur-
factor d represents the distance between pile-up source thermore, this approach even reflects the influence of the
and the dislocation. In this case, dislocations can be as- free surface on grains being located not directly at the
sumed to pile-up until the stress concentration at the grain surface due to the consideration of the ratio between kG
boundary exceeds the stress si. From this point on, dis- and knG. Applying this model to macroscopic dimensions
location sources will be activated in neighbouring grains. there is a large number of smallest grains within the
Following the fact, that (2) is valid for only a single considered specimen and thus the influence of the free
grain, the Hall-Petch factor Ki (valid for a single grain) is surface on the integral material properties is rather low
given by (Fig. 3).
pffiffiffi A further enhancement of the mesoscopic model is done
K i ¼ m1;2 si d ð3Þ based on the theory of Meyers and Ashworth [10] con-
sidering the dislocation pile-up in the region of a grain
The correlation between the integral Hall-Petch factor K boundary in a more detailed way. This is achieved by a
(cf. 1) and Ki is given by: subdivision of each single grain into two main volumes,
first the grain boundary volume fraction aGB with a higher
K i ¼ Kni ð4Þ flow stress kf,GB due to the occurring dislocation pile-up
compared to the rather low flow stress kf,I within the
Thus, in case of a single grain (2) can be written as:
material volume fraction aI containing no pile-up. The
Kn integral material behaviour of a single grain is expressed by
s ¼ s0 þ pffiffiffiffiffii ffi ð5Þ linear superposition of the yield stresses of these two vol-
dG
umes.
The newly introduced factor ni describes the amount of
dislocation pile-up at the grain boundary region. Consid- kf ¼ aI kf;I þ aGB kf;GB ð7Þ
ering the influence of the free surface on the forming
In order to simplify the analytical model, the shape of a
behaviour of the specimen and thus the non-capability of
single grain is considered to be spherical. Thus, the above
dislocations to pile up in this region, a more detailed view
mentioned two diametrical volume fractions aI and aGB can
on the influence of the grain boundary on the properties is
be expressed by:
given in (6).
"    2  3 #
knG 1 1 X aGB ¼2 3
t
6
t
þ4
t
ð8Þ
ni ¼ ¼ ac knG ð6Þ dG dG dG
kG k G aG

123
82 Prod. Eng. Res. Devel. (2007) 1:79–84

aI ¼ 1  aGB ð9Þ
with a thickness of the grain boundary layer t and grain size
dG. Inserting (8) and (9) into (7) leads to
   
kf ¼ kf;I þ 6 kf;GB  kf;I td 1
G  12 k f;GB  kf;I
  3 3 ð10Þ
t2 d2
G þ 8 kf;GB  kf;I t d G

As it is clear, different cross sections of a single grain will

produce different area fractions of aG and aGB. Hence Me-
yers and Ashworth recommended using mean values of t and
dG ; t and dG ; with values according to following approach.
p
t ¼ 1:57t; dG ¼ dG ð11Þ
4

Considering the variation of the thickness t of the work

hardening layer to be

1=2
t ¼ ðk1 k2 dG Þ1=2 ¼ kMA d G ð12Þ
in order to fullfil two effects: the fluctuation of the stress Fig. 4 Verification of mesoscopic model by 2D simulation
field varies with dG if the grain size is decreased, leading to
a dependency t = k1dG and the dislocation spacing to be
unchanged and the dislocation interactions will dictate a
constancy in t, thus t = k2dG0. Assuming the term td–1 g to be
flat upsetting test are compared with results gained from
approximately equal to 2td–1 g , (10) can be written as experiments. The parameters of interest for the comparison
are the mean forming force and its standard deviation,
  1=2 respectively.
kf ¼ kf;I þ 12kMA kf;GB  kf;I d G  24k2MA
    3=2 ð13Þ As it is shown in Fig. 4 the simulation model is well
kf;GB  kf;I d1 3
G þ 16k MA k f;GB  kf;I d G suited to describe the dependency of the forming behaviour
when changing the ratio between mean grain size and part
In case of grains being in the micrometer range the d–1/2
G dimension. The experimentally observed decreasing mean
term dominates and thus the Hall-Petch relation is ob- forming force when increasing the grain size (as it is
tained, where the Hall-Petch slope K is equal to similar to scaling down specimen dimension) as well as the
  increasing scatter of the process factors can be considered
K ¼ 8kMA kf;GB  kf;I ð14Þ in finite element simulation when applying the mesoscopic
model.
Finally, inserting (3) and (5) in (7), the flow stress of the In bulk metal forming at conventional length scale,
boundary region kf,GB and the inner region kf,I of the grain simulation methods are frequently used to get detailed
can be calculated to information on the material flow, die filling and shape
evolution. As shown above, at microscale there is a sig-
Kni
kf;GB ¼ kf;0 þ pffiffiffiffiffiffi and ð15Þ nificant influence of the material structure on the material
aGB dG flow and the shape evolution. As the material structure is
kf;I ¼ kf;0 ð16Þ considered in FE-simulation using the described meso-
scopic model, the influence of the material structure on
The thickness of the grain boundary region t can be the process results can be predicted in terms of shape
calculated by inserting (14) into (12). For the investigations evolution (Fig. 5). In dependency of the position and size
in this paper, t was assumed to be of single grains within the specimen, the material prop-
erties are calculated individually yielding different local
t ¼ 0:133d 0:7
G ð17Þ material properties and thus different influence on the
forming process. Especially in case of backward extrusion
For purpose of verification, 2D simulation results of the processes, shape evolution of the can is an decisive factor
advanced mesoscopic model using the previously described for the quality of the production process.

123
Prod. Eng. Res. Devel. (2007) 1:79–84 83

results with respect to the evaluation and simulation of

local forming behaviour.

5 Conclusion

Simulation in the field of microforming processes close to

the results of experiments is winning a growing interest in
microforming. It is well recognized, that simulation rep-
resents an important presupposition for pushing forward
the application of microforming technology in practice. As
it has been shown for some selected cold forging processes,
a first step has been done successfully to reach this goal.
The instruments developed so far are capable to model
essential phenomena of microforming at least on a quali-
tative level. Additionally, on the basis of experimental and
simulative results, the most relevant gaps in modelling are
identified as well as the methods to fill them. Hence, pro-
gress can be expected in near future towards generalization
of the modelling approach and its applicability on other
processes relevant for microforming yielding precise and
reliable information on simulation results, simulation reli-
ability and process stability.

Acknowledgments The authors acknowledge the German Research

Foundation (DFG) for the finacial support of this work within the
framework of SPP 1138. Also, this work was carried out within the
Fig. 5 Three-dimensional Simulation of can backward extrusion framework of the EC Network of Excellence ‘‘Multi-Material Micro
process applying mesoscopic material model a material structure b Manufacture: Technologies and Applications (4M)’’.
equiv. plastic strain influenced by the material structure

References
4 Discussion 1. Wechsun R et al (2002) Nexus market analysis for microsystems,
2000–2005. Nexus task force market analysis. Wicht Technologie
Applying this model to various finite element simulations Consulting, München
of cold forging and even sheet metal forming processes 2. Engel U, Tiesler N, Eckstein R (2001) Microparts—a challenge
for forming technology. In: Kuzman K (ed) ICIT 2001 3rd
will finally allow more exact and reliable simulation results international conference on industrial tools, April 22–26, Celje
taking into account the specifics of microscale forming Slovenia, pp 31–39
processes. The mesoscopic model is a first approach to 3. Engel U, Eckstein R (2002) Microforming—from basic research
describe the influence of the material structure on the mi- to its realization. J Mater Process Technol 125–126:35–44
4. Engel U, Meßner A, Geiger M (1996) Advanced concept for the
croforming process with justifiable computational effort FE-simulation of metal forming processes for the production of
with simplifications done regarding anisotropic material microparts. In: Altan T (ed) Advanced technology of plasticity
forming behaviour. 1996, proc. of the 5th ICTP, 7–10 October 1996, Columbus, OH,
The results of these investigations show clearly, that for USA, vol II, pp 903–907
5. Geißdörfer S, Engel U (2004) Mesoscopic model—simulation of
basic tests a good agreement between simulation and size effects in microforming. In: Jan Kusiak et al (eds) Pro-
experimental results has been reached in terms of integral ceedings of the 10th international conference on metal forming
process parameters like forming force and process scatter. 2004, Krakow, 19–23 September 2004, Bad Harzburg 2004, pp
If local forming behaviour has to be considered to get more 699–703
6. Geißdörfer S, Engel U, Geiger M (2004) FE-simulation of mi-
precise information on shape evolution or die filling, future croforming process applying a mesoscopic model. In: Proceed-
work will focus on the description of the local forming ings of the 1st international conference on new forming
aspects in order to get an enhancement of the simulation technologies, 6–9 September 2004, Harbin, pp 141–146

123
84 Prod. Eng. Res. Devel. (2007) 1:79–84

7. Geißdörfer S, Diehl A, Eichenhüller B, Engel U (2006) Micro- 9. Mehnert K, Klimanek P (1997) Grain growth in metals with
forming—application and 3D simulation. In: Proceedings of the strong textures: three-dimensional Monte-Carlo simulations.
5th CIRP seminar on intelligent computation in manufacturing Comput Mater Sci 9:261–266
engineering, Ischia, 25–28 July 2006, pp 521–526 10. Meyers MA, Ashworth E (1982) Phil Mag A 46, p 737ff
8. Ashby MF (1970) The deformation of plastically non-homoge
neous materials. Phil Mag 21:254–255

123