Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Materials Technology

Influence of Martensite Distribution on the Mechanical Properties of Dual Phase Steels:


Experiments and Simulation

C. Thomser, V. Uthaisangsuk, W. Bleck

Department of Ferrous Metallurgy, RWTH Aachen University, Intzestr.1, D-52072, Aachen, Germany

The application of multiphase steels in the automotive industry has been rapidly increased according to economic, environmental and
safety reasons. To determine an optimal combination of high strength and good formability of multiphase steels by using the FE modelling,
their complex microstructures have to be considered. Two-dimensional Representative Volume Elements (RVEs) were currently developed
based on real microstructures for dual phase (DP) steels. In general, the microstructure of DP steels contains hard martensite particles and
a soft ferritic matrix. The strain hardening behaviour of the individual phases was described in the model taking the microstructural
constituents and the carbon partitioning during intercritical annealing into account. Two dual phase microstructures with same martensite
content but different martensite distributions were investigated in experiment as well as in FEM simulation by means of the RVE. The
resulting mechanical properties of these steels are strongly influenced by the phase distribution and interaction. As validation, calculated
flow curves were compared with the experimental results from quasi-static tensile tests. In addition, the local stress and strain partitioning
between both phases depending on the spatial phase distribution and morphology is discussed.

Keywords: electromagnetic compression, joining, shot peening, milling

DOI: 10.2374/SRI09SP046; submitted on 20 January 2009, accepted on 20 February 2009

Introduction Material

Due to economic, environmental and safety reasons, the The investigated material is an industrial steel sheet with
application of high strength steels in car body design has a thickness of 1 mm. The chemical composition of this
been risen. An accurate material model is required for steel is presented in Table 1. The aluminium-killed steel
industrial simulations in order to utilise optimal was hot rolled on a conventional hot strip mill and further
combination of strength and formability of multiphase on cold rolled on a commercial tandem mill. The alloying
steels. In most FE simulations of sheet metal forming elements manganese, chromium and silicon were chosen
operations, the microstructure of multiphase steels is to adopt the required transformation behaviour and
currently not considered, although it is the most important strength level. The steel consists of an elongated
factor influencing mechanical properties of these steels. ferritic/pearlitic microstructure after cold rolling without
In international projects like the ULSAB project annealing. In order to obtain dual phase microstructures
(ultra light steel auto body); especially dual phase steels with the same martensite content but different martensite
play a decisive role for the automotive industry [1]. Their distribution (a fine and a coarse martensitic micro-
strain hardening behaviour is strongly influenced by the structure), two different types of heat treatments were
microstructure as well known from several experimental scheduled:
investigations [2-4]. Dual phase steels are composed of a (a) Intercritical annealing followed by fast cooling
soft ferritic matrix with hard inclusions of martensite, (b) Austenitic annealing followed by intercritical
which leads to a good formability and a high strength level annealing and fast cooling
at the same time. The description of microstructure Both applied annealing cycles (a) and (b) are shown in
evolution, strain hardening behaviour and formability will Figures 1 and 2, respectively. The resulting microstruc-
help to optimise the processing of multiphase steels by a tures from annealing cycle (a) and (b) are presented in
precise microstructure design. Figure 3 on the upper part. Both annealing cycles lead to a
Within these investigations, an approach is presented microstructure with approximately 42% martensite content.
which predicts the flow curve of dual phase steels by FE The accuracy of metallographic determination of marten-
simulation using 2D Representative Volume Elements site content is assumed to be +/-5%. A fine microstructure
(RVEs). Morphologies and phase distribution of real was achieved by the annealing cycle without austenitza-
microstructures was applied for the RVE simulations. In tion (a) and a coarse microstructure was produced by the
order to investigate the influence of martensite distribution cycle with austenitic annealing (b). The contiguities of the
on the mechanical properties, dual phase steels with coarse martensitic phase in the fine and coarse microstructure are
and fine distributed martensitic microstructure were 0.687 and 0.78, respectively. These values reveal that the
studied. The numerical results were compared with fine microstructure has explicitly more interface areas
experimental results from tensile tests. The stress and between ferrite and martensite than the coarse micro-
strain partitioning between martensite and ferrite, which structure. The ferrite grain size is similar in both micro-
have a significant effect on the deformation and fracture structures. A martensite island size cannot be specified since
behaviour, is depending on the microstructure morphology. both microstructures already have martensite networks.

578 steel research int. 80 (2009) No. 8


Materials Technology

Salt bath, T = 780°C, t = 5 min Salt bath, T = 880°C, t = 5 min


A3
A3 Salt bath, T = 710°C, t = 5 min
A1
A1
Temperature

Temperature
Water cooling Water cooling

Time Time

Figure 1. Schematic of annealing cycle (a), intercritical Figure 2. Schematic of annealing cycle (b), austenitic annealing
annealing followed by fast cooling. followed by intercritical annealing and fast cooling.

Fine microstructure Coarse microstructure


Contiguity C(α´α´) = 0.687 Contiguity C(α´α´) = 0.78
Vα´ = 42+/-5%, dα = 5.3µm Vα´ = 42+/-5%, dα = 6.7µm

Ferrite

Martensite

conversion in FE mesh

Ferrite

Martensite

(a) (b)

Figure 3. Microstructures and their corresponding RVE models after two annealing cycles. (a) Fine microstructure, (b) Coarse
microstructure.

Table 1. Chemical composition of the examined steel DP600, mass%.

C Si Mn P S Cr
0.072 0.246 1.577 0.015 0.001 0.552

Table 2. Sample dimensions for quasi-static tensile tests.

a b L0 Lc Lt B h R
Sample dimensions, mm <3 20 80 120 250 30 50 20

Experiment perature with a crosshead velocity of 4mm/min. The geo-


metry of the used tensile specimens is shown in Figure 4
Tensile tests at room temperature were carried out and its details in Table 2. The resulting mechanical
according to DIN EN 10002 [5]. The tests were performed properties for both microstructures are represented in
on the tensile testing machine Zwick 100 at room tem- Table 3. The steel with the fine microstructure exhibits a

steel research int. 80 (2009) No. 8 579


Materials Technology

1800

1600

1400 Martensite - fine and coarse microstructure


a
1200

True stress [MPa]


R

1000

b
800

B
Ferrite – fine microstructure
600
L0
400
h Lc h
200 Ferrite – coarse microstructure

Lt
0
0.00 0.02 0.04 0.06 0.08 0.10
True strain [-]

Figure 4. Sample geometry of quasi static tensile tests. Figure 5. Modelled strain hardening curves for individual phases
in both investigated microstructures

Table 3. Mechanical properties of the investigated steel. Table 4. Constants for the flow curve modelling for the ferritic phase.

Microstructure Rp0.2 Rm Ag A80 n-value Δσ, MPa L, m k


type MPa MPa % % -5
Ferrite 5000*Css dα 10 /dα
Fine 462 911 7.8 9.2 0.15
-8
Coarse 424 815 9.0 11.2 0.16 Martensite 3065[C]-161 3.8*10 41

higher yield strength as well as tensile strength than the The strain hardening behaviour of individual phases was
steel with the coarse microstructure. However, the uniform defined by a dislocation based constitutive model [7-9].
elongation and elongation at fracture for the fine The stress-strain characteristic described in this model is
microstructure is lower. Both steels represent continuous given in Equation (1).
elastic-plastic transition behaviour. No temper rolling was
applied. 1 − exp(− M ⋅ k ⋅ ε )
σ = σ 0 + Δσ + α ⋅ M ⋅ μ ⋅ b ⋅ (1)
k⋅L

Modelling M: Taylor factor (M = 3)


μ: Shear modulus (μ = 80000 MPa)
Micrographs of the investigated dual phase micro- b: Burger’s vector (b = 2.5·10-10 m)
structures obtained by light optical microscopy were taken α: Constant (0.33)
to generate the model for the FE simulation. The real Δσ: Strengthening by precipitation and carbon in solution
microstructure was converted into a two-dimensional RVE L: Dislocation mean free path
model (representative volume element) with regard to the k: Recovery rate
colour difference of martensite and ferrite after etching. σ0 describes the effect of the Peierls stress and of the
Since the creation of a three-dimensional FE model from elements in solid solution as:
real microstructure is very complex, this work was based
on 2D simulation. Phase distribution and phase fraction of σ 0 = 77 + 80%Mn + 750%P + 60%Si + 80%Cu
martensite and ferrite can be properly described in the 2D (2)
+ 45%Ni + 60%Cr + 11%Mo + 5000N ss
RVE model. It is assumed that these models are able to
represent the overall macroscopic properties of the
investigated materials and therefore they are called repre- The parameters in this model are dependent on the local
chemical composition of material, grain size, the disloca-
senttative volume elements. The lower part of Figure 3
tion mean free path and the recovery rate. The constants of
shows the RVE models of the fine and coarse dual phase
Rodriguez [10, 11] were taken into account as shown for
microstructures. Both microstructures have approximately
ferritic and martensitic phase in Table 4. Css is the carbon
the same martensite content of 42%. For the simulations in
in solution; [C] is the martensite carbon content and dα is
2D, plane strain element type was applied in the RVE
the ferrite mean linear intercept. To obtain the parameter
model. The influence of mesh size was previously
dα, the grain size of ferrite was experimentally measured
examined. The number of element in the RVE was
by software image analysis. Grain sizes of 5.3μm and
increased step by step until no changing in the resulting
6.7μm were determined for the fine and coarse micro-
stress-strain response was observed. Then the considered
structure, respectively. The carbon content in ferrite was
mesh size (260 elements per sides) was used for further estimated to be 0.007 mass% for both microstructures. In
investigation of both microstructures [6]. The area of martensite, the carbon content was calculated according to
microstructures taken for the RVE had a width of 135 μm the mass balance and the phase fraction with 0.162 mass%.
and a length of 108 μm. Other elements were assumed to be homogeneously

580 steel research int. 80 (2009) No. 8


Materials Technology

distributed in ferrite and martensite. With all 1200


data, input flow curves of the single phases
were calculated for RVE simulation. The 1000 fine – experiment coarse – experiment
curves are presented in Figure 5. The flow
stress of the harder martensite is

True stress [MPa]


800
considerably higher than the flow curve of coarse – simulation
ferrite, and a nearly constant strain fine – simulation
hardening at higher deformation is 600
recognised. The flow curve of ferrite of the
coarse microstructure is a bit lower than the 400
one of the fine microstructure because of
different grain sizes.
Uniform tensile deformation was applied 200
to the RVE under plane strain condition.
Degrees of freedom of all nodes on the right 0
hand vertical plane and upper horizontal 0 0.02 0.04 0.06 0.08 0.1 0.12
plane were restricted so that these planes True strain [-]
remain in parallel to the opposite planes
during the whole calculation. Averaged Figure 6. Comparison between stress-strain curves in experiments and simulations
stress-strain responses of the RVE were and calculated stress distribution in the fine and coarse microstructures.
determined and compared to the ex-
perimental results. Remarkable local dif-
50000
ferences in simulation results are anticipated
due to distribution of different phases. 45000
Fine microstructure - simulation
40000
Coarse microstructure - simulation
Strain hardening d /d

35000
Results Fine microstructure - experiment
30000
Coarse microstructure - experiment
RVE simulations were carried out until the 25000
equivalent plastic strain of 0.25 was reached 20000
for both microstructures. Since no failure 15000
criterion was defined in the simulations, no
stress carrying capacity loss was detected. 10000
However, the main objective of this work 5000
was to study the influences of different dual 0
phase microstructures with the same 0.00 0.02 0.04 0.06 0.08 0.10
martensite content on their stress-strain True strain [-]
response as well as distribution of
partitioned stress and strain between hard Figure 7. Comparison between strain hardening curves in experiments and
and soft constituents. In Figure 6, the simulation.
determined equi-valent plastic true stress-
strain curves from the RVE simulations were
compared with the experimental results of tensile tests. coarse microstructure presents slightly higher maximum
The calculated stress-strain curves are all underestimated, local stress than the fine microstructure, even though its
but the tendencies of strain hardening can be correctly overall stress-strain response is lower. In particular, the
predicted in both cases. This underestimation can be narrow martensitic phase and the sharp edge of martensite-
explained due to the application of the plane strain ferrite interface are the risk zone for high stress
elements in the simulations. Based on micromechanical concentration.
modelling, different cell models of dual phase micro- The strain hardening rates as a function of true strain
structures with an approximate volume fraction of from simulations and experiment are compared in Figure 7
martensite of 32-36% were investigated by Al-Abbasi [12]. for both microstructures. Differences in strain hardening
The results showed that stress-strain response for between coarse and fine martensite structures are observed
axisymmetric model is consistent with experimental at small strains (<0.02). The strain hardening behaviour at
behaviour. The plane strain models appeared to under- higher strains is very similar for all microstructures. Byun
predict the strain hardening. This is in agreement with the analysed tensile properties and inhomogeneous deforma-
outcome of Figure 6. At small deformation stages, the tion of ferrite-martensite dual phase steels experimentally
flow curve of the finer distributed microstructure exhibits [14]. It was evidenced that the volume fraction depen-
higher hardening rate than the flow curve of the coarse dence of the stress and the characteristics of the strain-
microstructure. This effect is approved by the ex- hardening rate were influenced by the plastic deformation
perimental flow curves. The simulated RVEs in Figure 6 of martensite. Average partitioned strains of ferrite and
demonstrate the stress distribution at the uniaxial deforma- martensite were determined from the RVE simulations,
tion of approximately 10% in both microstructures. The illustrated in Figure 8a. The strain progress in martensite

steel research int. 80 (2009) No. 8 581


Materials Technology

0.4 1500
there are more interface areas between
Ferrite - fine
0.35
harder and softer phases. Therefore,
Martensite - fine
martensite phase in fine microstructures can
average partitioned true strain [-]

Ferrite - coarse 1200 Ferrite - fine

average partitioned stess [MPa]


0.3 Martensite - coarse Martensite - fine be easier deformed as they have higher local
Ferrite - coarse strain than in coarse microstructure. The
0.25 plastic deformation in the martensite
900 Martensite - coarse

0.2
particles takes place due to the strain and
load being transferred through the interface
total
0.15 DP material
600 areas [13].
Figure 8b represents variations of
0.1
average partitioned stress with equivalent
300
0.05
local strain in ferrite and martensite of both
microstructures. The fine structure exhibits
0 0 higher stress values than the coarse
0 0.05 0.1 0.15 0.2 0.25 0.3 0 0.05 0.1 0.15 0.2 0.25 0.3 microstructure in ferrite as well as in
average local true strain [-] average local true strain [-] martensite. This result is the reason of the
higher overall stress-strain response in the
(a) (b)
fine microstructure. Local deformation of
Figure 8. Strain partitioning (a) and stress partitioning (b) between different phases ferrite is approximately three times higher
depending on the equivalent local plastic strain. than deformation of martensite in these
investigated dual phase steels. Figure 9a
0.25 3 and 9b illustrate the strain and stress ratio
between martensite and ferrite as a function
ratio of average partitioned stess ΠM/ΠF [-]

of average local strain. The strain ratio


[-]

2.5
M/ F

0.2 values, which represent the degree of


uniformity of straining between two phases,
ratio of partitioned true strain

2
are always less than 1. It can be seen that
0.15
during the initial stage the ratio values are
1.5 zero, and then they increase rapidly and
0.1 approach a constant value at higher
1 deformation. From the average strain of 0.02,
fine microstrcuture fine microstructure the strain ratio in the fine microstructure is
0.05
coarse microstructure 0.5 coarse microstructure
much higher than the ratio in the coarse
microstructure. The coarse and fine
microstructures exhibit a small difference in
0 0
0 0.05 0.1 0.15 0.2 0.25 0.3
the stress ratio with respect to different grain
0 0.05 0.1 0.15 0.2 0.25 0.3
average local true strain [-]
sizes. Figure 10 shows the plastic strain
average local true strain [-]
distribution in the fine and coarse micro-
(a) (b) structure at the uniaxial deformation of 10%.
Bands of the localized plastic strain can be
Figure 9. The ratio of εM/εF (a) and σM/σF (b) as function of equivalent local plastic observed in the ferrite on 45° to the tensile
strain.
direction in both cases. In the coarse
microstructure a localization band with very
localization band with very high plastic strain high plastic strain was detected on the left-
upper side of the RVE. The coarse DP
microstructure is the morphology which
causes a highly local strain concentration.

Conclusions

Within these investigations, an approach


for prediction of the flow curve of dual
fine microstructure coarse microstructure
phase steels by FE simulation using 2D RVE
was presented. The morphologies and phase
distribution of real microstructures were
Figure 10. Plastic strain distribution in fine and coarse DP microstructure.
taken into account. In order to investigate
the influence of martensite distribution on
is definitely lower than in ferrite. This linear dependency the mechanical properties, dual phase steels with coarse
of strain was also established in in-situ measurement of and fine martensitic structures were in-vestigated.
local strain partitioning in dual phase steel regarding to the The results of the RVE simulations show the strain
research of Ososkov et al. [15]. In finer microstructures, hardening similar to the results from experimental tensile

582 steel research int. 80 (2009) No. 8


Materials Technology

tests, but the absolute values of the simulation results are [3] Bag, A.; Ray K. K.; Dwarakadasa, E. S.: Influence of martensite
lower. That is likely because of the usage of plane strain content and morphology on tensile and impact properties of high-
martensite dual-phase steels, Metallurgical and Materials
elements in the 2D simulations. Especially at the begin-
Transactions A, 30A (1999), p. 1193-1202
ning of the flow curve, the hardening is higher in the finer [4] Rashid, M. S.: Dual phase steels, Annual Review of Materials
distributed microstructure than in the coarse micro- Science, 11 (1981), p. 245-266.
structure in the experiments as well as in the simulations. [5] DIN EN 10002, Teil 1: Zugversuch, Normenausschuß,
The local strain and stress distribution in the dual phase Materialprüfung (MVP) in DIN, Deutsches Institut für Normung
micro-structures with the same martensitic fraction but e.V., 1992.
different sizes were analysed. Higher plastic deformation [6] Imada, Y.: Modelling of mechanical properties of dual phase steel
based on microstructure, Master thesis, RWTH Aachen, 2007.
in martensite was observed in the fine microstructure due
[7] Bergström, Y.: A dislocation model for the stress-strain behaviour
to the higher number of interface areas. The strain
of polycrystalline α-Fe with special emphasis on the variation of
proportion between martensite and ferrite in the fine the densities of mobile dislocations, Material Science Engineering,
microstructure is also explicitly higher than in the coarse 5 (1969/1970), p. 193-200.
microstructure. The development of local stress and strain [8] Gil-Sevillano, J.: Flow stress and work hardening, Materials
in dual phase microstructures and the resulting crack Science and Technology, 6 (1993), p. 19-28.
initiation are strongly influenced by the morphology and [9] Estrin, Y.; Mecking, H.: A unified phenomenological description
distribution of martensite. of work hardening and creep based one-parameter models. Acta
Metallurgica, 32 (1984), p. 57-70.
[10] Rodriguez, R.; Gutierrez, I.: Unified formulation to predict the
Acknowledgements tensile curves of steels with different microstructures. Materials
Science Forum, 426-432 (2003), p. 4525-4230.
[11] Rodriguez, R.; Gutierrez, I.: Mechanical behaviour of steels with
This research was carried out under the project mixed microstructure. In: Proceeding of TMP’04, Liege, 2004, p.
MC5.03170 „Multiphase steel plasticity“ in the framework 356-363.
of the Research Program of the Materials Innovation [12] Al-Abbasi, F. M.; Nemes J. A.: Micromechanical modelling of
Institute M2i (www.m2i.nl), the former Netherlands dual phase steels, International Journal of Mechanical Sciences, 45
Institute for Metals Research. The authors would like to (2003), p. 1449-1465.
thank the Materials Innovation Institute for the scientific [13] Al-Abbasi, F. M.; Nemes J. A.: Micromechanical modelling of the
and financial support of the project. effect of particle size difference in dual phase steels, 40 (2003), p.
3379-3391.
[14] Byun, T. S.; Kim I. S.: Tensile properties and inhomogeneous
References deformation of ferrite-martensite dual phase steels, Journal of
Material Science, 28 (1993), p. 2923-2932.
[1] http://www.worldautosteel.org/ulsab-avc, 2006. [15] Ososkov, Y.; Wilkinson, D. S.; Jain, M.; Simpson, T.: In-situ
[2] Kim, N. J.; Thomas G.: Effects of morphology on the mechanical measurement of local strain partitioning in a commercial dual-
behaviour of a dual phase Fe/2Si/0.1C steel, Metallurgical phase steel, International Journal of Material Research, 98 (2007),
Transactions A, 12A (1981), p. 483-489. No. 8, p. 664-673.

steel research int. 80 (2009) No. 8 583

You might also like