Journal of Alloys and Compounds 831 (2020) 154851

Contents lists available at ScienceDirect

Journal of Alloys and Compounds

journal homepage: http://www.elsevier.com/locate/jalcom

Microstructure and damage based constitutive modelling of hot

deformation of titanium alloys
P.F. Gao a, b, J. Guo c, M. Zhan a, *, Z.N. Lei a, M.W. Fu b, **
a
State Key Laboratory of Solidification Processing, Shaanxi Key Laboratory of High-Performance Precision Forming Technology and Equipment, School of
Materials Science and Engineering, Northwestern Polytechnical University, Xi’an, 710072, China
b
Department of Mechanical Engineering, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong, China
c
Institute of Chemical Materials, China Academy of Engineering Physics, Mianyang, China

a r t i c l e i n f o a b s t r a c t

Article history: Development of a hybrid constitutive model for modelling of the flow behavior, microstructure evolu-
Received 14 December 2019 tion, damage initiation and fracture formation is crucial in study of hot deformation of titanium alloys. In
Received in revised form tandem with this, a microstructure and damage based constitutive model for modelling of hot working of
16 March 2020
Tie6Ale2Zre1Moe1V (TA15) alloy was developed. The hot tension deformation of the alloy was first
Accepted 18 March 2020
Available online 21 March 2020
conducted to analyze the microstructure evolution, damage and flow behaviors. It was found that
increasing b phase fraction and dynamic recrystallization (DRX) fraction suppress the damage initiation
and propagation in the way of void nucleation, growth and coalescence, and thus foster the increase of
Keywords:
Titanium alloys
fracture strain in an exponential form. Based on the experimental results, the microstructure evolution,
Hot deformation including phase transformation and DRX occurrence, was modelled by using the method represented by
Constitutive modelling physically-based internal state variables. The damage and fracture behaviors were modelled by
Microstructure evolution considering the effects of microstructure and stress state via introducing b phase fraction, DRX fraction
Damage and fracture and stress triaxiality into the classical Gurson-Tvergaard-Needleman damage model. The constitutive law
considering both the microstructure and damage evolution was then given and further the micro-
structure and damage based constitutive model was established. The parameters in the model were
calibrated by comparing the predicted and experimental results. Finally, the developed model was
successfully applied in finite element simulation of hot spinning of TA15 alloy tube for unified prediction
of macroscopic deformation, microstructure, damage and fracture. The research thus provides a basis for
tailoring and control of microstructure and damage in hot working of titanium alloys.
© 2020 Elsevier B.V. All rights reserved.

1. Introduction final mechanical properties [3,4]. Moreover, ductile damage and

fracture would also occur under the severe plastic deformation,
Titanium alloys possess superior specific strength, excellent leading to formation of defects in the manufactured components
thermal stability, biological compatibility and welding performance [4e6]. Therefore, optimal working process design is crucial to avoid
and thus are widely used to manufacture components in aero- the sensitive ductile fracture defects and obtain the desired
nautical, aerospace and biomedical industries [1,2]. Hot working, macroscopic shape and microstructure. Currently, numerical
such as hot spinning, rolling, and forging, is most frequently used to simulation, as an alternative and efficient approach for the tradi-
process titanium alloy. In hot working of titanium alloys, complex tional trial-and-error physical experiment, has become an indis-
hardening and softening flow behaviors and significant micro- pensable tool for optimization of hot working process. It needs an
structure evolution may take place, which affect forming of the accurate simulation models to predict macroscopic deformation,
requested geometries of components and tailoring of the needed microstructure evolution, as well as ductile damage and fracture
simultaneously in deformation process. In tandem with this, it is
necessary to establish a hybrid constitutive model to well represent
* Corresponding author. related physical mechanisms and processes of the above
** Corresponding author. mentioned deformation activities in hot working of titanium alloys.
E-mail addresses: [email protected] (M. Zhan), [email protected]. It is well known that the significant microstructure change,
hk (M.W. Fu).

https://doi.org/10.1016/j.jallcom.2020.154851
0925-8388/© 2020 Elsevier B.V. All rights reserved.
2 P.F. Gao et al. / Journal of Alloys and Compounds 831 (2020) 154851

including phase transformation, dislocation multiplication, dy- threshold was correlated with deformation temperature and strain
namic recovery (DRV), dynamic recrystallization (DRX), etc, occurs rate by introducing the Zener-Hollomon factor into the constitutive
in hot working of titanium alloys [7]. These microstructure evolu- model. However, these works are based on the uncoupled fracture
tions would greatly affect the flow, damage and fracture behaviors models, which do not consider the damage induced flow softening.
of the materials. For example, dislocation multiplication causes Recently, some primary works about damage-based constitutive
strain hardening, while DRV and DRX lead to flow softening. The modelling for hot deformation of metals have been reported. In this
formation of defect-free DRX grains and a/b phase transformation respect, Tang et al. [18] modified the classical Lemaitre damage
would change the micro-scale heterogeneous deformation phe- model to take the dependency of damage on temperature and
nomena and then further influence damage and fracture behaviors, strain rate into account, which can predict the damage induced
and affect process of void nucleation, growth and coalescence. softening and fracture during hot stamping of high strength steel.
Damage evolution, on the other hand, would in turn produce Yang et al. [19] and Li et al. [20] developed a set of damage-based
softening effect on the macroscopic flow behavior. These charac- constitutive equations to model the flow behavior and damage
teristics pose a great challenge to development of hybrid consti- evolution in hot deformation of titanium alloys. In their works, the
tutive model for representing the hot deformation of titanium damage evolution and its softening effect on the flow behavior are
alloys. Specifically, the coupling of the damage and microstructure characterized by a scalar damage variable. However, the effect of
evolutions would make the model development more difficult. hot deformation on damage behavior is modelled by incorporating
Currently, the flow behavior and damage fracture are modelled temperature and strain rate into the damage model parameters, the
respectively in most reported researches. For modelling the flow underlying effect mechanisms of microstructure evolution on
behavior of titanium alloy, there are three typical types of methods, damage and fracture behavior thus cannot be captured. Shang et al.
i.e., the empirical regression model, the statistical model and the [21,22] developed an extended ductile fracture model considering
physically-based internal state variable (ISV) method. Classical the microstructure evolution for the hot deformation of 316LN
Arrhenius-type equation is the most common regression model. steel. However, the microstructure evolution and its effect on
Mosleh et al. [8] obtained the flow behavior of Tie6Ale4V (TC4) damage evolution of titanium alloys (two-phase alloy) may be quite
alloy during hot deformation at lower strain rates, and developed different from those of 316LN steel with single austenite phase
the corresponding constitutive model by regressing the Arrhenius- during hot deformation. As mentioned above, there exist complex
type equation. Ghavam et al. [9] also established a constitutive microstructure evolutions, such as dislocation evolution, DRX and
model based on the Arrhenius-type equation for hot defamation of a/b phase transformation in hot deformation of titanium alloys, and
IMI834 alloy. For the statistical model, Reddy et al. [10] developed a which greatly affects the damage and flow behavior. To correlate
constitutive model for hot deformation of TC4 alloy using the the underlying relationship among microstructure evolution,
artificial neural network, which presents high prediction accuracy. damage evolution and flow behavior, it needs to develop a hybrid
Liu et al. [11] first analyzed the effect significances of deformation constitutive model to consider both the microstructure and dam-
conditions on the hot flow behavior of Ti17 alloy by the orthogonal age evolution for hot deformation of titanium alloys.
analysis. Then, they developed a multivariate nonlinear mathe- In this paper, hot tension deformation experiments were con-
matical model considering the significant factors. However, the ducted to obtain the flow behavior and fracture feature of TA15
empirical regression models and statistical model provide less alloy at different conditions. The microstructure evolution and its
physical insight and cannot predict the microstructure evolution at effect on damage and fracture behaviors were analyzed. Based on
the same time. The physically-based internal state variables (ISV) is the experimental results, a microstructure and damage based
a feasible method to model the constitutive behavior of titanium constitutive model for hot deformation of TA15 alloy was estab-
alloy considering the underlying contribution of microstructure lished. The model and its implementation are described in four
evolution. Luo et al. [12] developed an ISV model considering the aspects, viz., microstructure modelling based on ISV method,
evolutions of dislocation density, grain size and phase content to damage modelling based on (Gurson-Tvergaard-Needleman) GTN
predict the hot deformation behavior of TC4 alloy. Fan and Yang model, constitutive law and numerical implementation. The model
[13] incorporated the ISV method into self-consistent constitutive parameters were finally determined by comparing the predicted
model to predict the deformation behavior and microstructure and experimental results and the model was validated by applying
evolution of two-phase titanium alloys. In these prior arts, the so- to the simulation of hot spinning of TA15 alloy and corroborated
lution strengthening, dislocation interaction, DRX, grain growth with experiments. The developed model can achieve the unified
and their effects on the flow behavior were considered. Similarly, prediction of flow stress, microstructure, damage and fracture, and
Bai et al. [14], Babu and Lindgren [15] developed ISV-based unified thus provide a basis for the coordinated control of microstructure
constitutive equations for two-phase titanium alloys at elevated and damage in hot working of titanium alloys.
temperature, in which the flow softening caused by globularization
of lamellar a was mainly considered. However, these works do not
2. Material and experiment
consider the damage evolution, which cannot predict the ductile
fracture defect and are just applicable to the smaller deformation
The material used in this work is a wrought two-phase TA15
scenarios.
alloy in the form of round bar. Its chemical composition is given in
As for the damage and fracture modelling of hot deformation of
Table 1. The initial microstructure of the as-received material
metals, Johnson and Cook [16] extended the traditional J-C fracture
consists of equiaxed primary a phase within the transformed b
model to hot deformation by expressing the strain to fracture as a
matrix (secondary a phase and b phase), as shown in Fig. 1. Its b-
function of temperature, strain rate and pressure. He et al. [17]
transus temperature is 1263 K, which is determined by the classical
correlated the critical damage factor in classical Oyane fracture
model with deformation temperature, strain rate and activation
energy of DRX, which makes it applicable to predict the fracture Table 1
during hot working of 30Cr2Ni4MoV steel. Similarly, Zhan et al. [6] The chemical composition of TA15 alloy.
also adopted the traditional Oyane fracture model to develop a Element Ti Al Mo Zr V Fe Impurity
thermal damage model for hot deformation of
Content (wt.%) Matrix 6.1e6.5 0.9e1.2 1.9e2.2 0.8e1.3 0.25 <0.3
Tie6Ale2Zre1Moe1V (TA15) alloy. In this work, the damage
 P.F. Gao et al. / Journal of Alloys and Compounds 831 (2020) 154851 3

It is well known that the stress state affects the damage and
fracture behavior during the deformation of metals [25,26]. And,
there always exist various stress states during hot working process.
To investigate and incorporate the effect of stress state on damage
and fracture feature into the constitutive model, another three
types of tensile samples were also designed to achieve the defor-
mation with different stress states, as shown Fig. 3. Furthermore,
hot tension deformation testes were also conducted using these
samples with different conditions. To obtain the stress states of four
types samples, as shown in Figs. 2 and 3, during the deformation,
conventional finite element simulation was conducted. The simu-
lation results show that the stress triaxiality at the sample center is
0.33, 0.6, 0.26 and 0.1 for the uniaxial tension, notch tension, 45
shear and 0 shear tensile tests, respectively. Moreover, the local
strain and fracture strain of sample were identified by simulation,
which will provide a basis for the microstructure analysis and
constitutive modelling.
Fig. 1. Initial microstructure of the as-received TA15 alloy.
3. Experiment results

metallographic method. In this section, the flow behavior of TA15 alloy during hot
Uniaxial tension deformation samples were machined from the deformation is first studied and provide a basis for constitutive
initial material, whose shape and size are shown in Fig. 2 [23]. The modelling. The characteristics of microstructure evolution, damage
hot tension deformations were conducted on a SANS CMT 5205 and fracture, especially for the effect of microstructure evolution on
electrical test machine under isothermal condition. In the test, the damage and fracture, will then be analyzed. The details are artic-
samples were heated to deformation temperature, held for 10 min, ulated in the following sections.
stretched to fracture at a constant strain rate. In this work, the
uniaxial tensile tests were carried out at four temperatures (973,
3.1. Flow behavior
1023, 1073, and 1123 K) and three strain rates (10
3, 10
2, and 10
1
s
1). The repeated hot tension tests under the same deformation
Fig. 4 shows the stress-strain curves of the uniaxial tensile tests
conditions present very close result, presenting well repeatability.
at different conditions. It is found that the variation of stress with
After tension deformation, the samples were water quenched to
strain at any condition can be divided roughly into four stages: (1)
maintain the microstructure characteristics at high temperature. In
elastic deformation stage: the stress increases linearly with strain
addition, the force-displacement curves were captured, based on
in this stage. (2) work hardening stage: plastic deformation occurs
which the stress-strain curves of tension deformation were also
and the flow stress increases gradually with strain in this stage,
calculated.
which is mainly caused by dislocation multiplication. (3) flow
The deformed samples were also sectioned vertically to the
softening stage: the stress decreases after peak stress. DRV and DRX
thickness direction and prepared for metallographic examination.
are the main reasons for the softening, meanwhile damage evolu-
The microstructures at different conditions were observed by
tion also contributes to the softening to some extent. (4) fracture
scanning electron microscope (SEM) using standard techniques.
stage: the stress decreases quickly with strain and fracture occurs
Moreover, the microstructure was also examined by electron back-
due to the dramatic development of damage. It is further figured
scattered diffraction (EBSD) with the scanning step of 0.3 mm. The
out that the flow behavior is closely related to the microstructure
hexagonal close-packed a phase and body-centered cubic b phase
and damage evolution, which should be incorporated in the
were both identified in EBSD test. Before EBSD, the samples were
constitutive modelling.
electro-polished at 20  C with a solution (5% HClO4 þ 30%
CH3OH þ 65% C4H10O) with a voltage of 25 V. After EBSD, the
microstructure characteristics were quantitatively analyzed 3.2. Microstructure, damage and fracture
through the HKL-Channel 5 software. When analyzing DRX
behavior, the DRX grains are distinguished from the deformed ones As a typical example, the microstructure evolution at the tem-
based on the grain orientation spread index (GOS) proposed by perature of 973 K and strain rate of 10
3 s
1 is shown in Fig. 5 [23].
Hadadzadeh et al. [24]. Here, regions with GOS < 3 is considered as As DRX mainly occurs in a phase, the evolution of grain-boundary
recrystallized grains. After the determination of DRX grains, the and misorientation in a phase are concerned in Fig. 5. At the smaller
DRX fraction can be obtained by calculating their area fraction. strain of 0.03, few DRX grain is observed and the density of low-
angle boundaries (<15 ) (LAB) is relatively low (Fig. 5(c)). With

Fig. 2. Schematic of the uniaxial tension deformation specimen [23]. Unit: mm.
4 P.F. Gao et al. / Journal of Alloys and Compounds 831 (2020) 154851

Fig. 3. Schematic of samples for tensile tests with different stress states: (a) notch tension, (b) 45 shear (c) 0 shear. Unit: mm.

Fig. 4. Flow stress of uniaxial tensile tests with different strain rates and the temperatures [35].

the increase of strain to 0.15, the quantity of a DRX grains is concentration and then suppress the void nucleation, growth and
observed around the primary coarse grain, and the density of LABs coalescence during hot deformation of metallic materials. Thus, the
is increased dramatically (Fig. 5(d)). These suggest that DRX of a quantitative effect of DRX on damage evolution in hot deformation
phase is an important microstructure phenomenon in the hot of TA15 alloy is analyzed below for modelling of damage consid-
tensile deformation of TA15 alloy. In addition, DRX is a key factor for ering microstructure evolution.
flow softening, and thus needs to be considered in constitutive Prior arts suggested that the phase contents and DRX kinetics
modelling. In tandem with this, Shang et al. [22] has proposed that are strongly dependent on temperature during hot deformation of
DRX induced softening effect may relieve the local stress TA15 alloy [23]. The effect of microstructure on damage behavior,
 P.F. Gao et al. / Journal of Alloys and Compounds 831 (2020) 154851 5

Fig. 5. Evolutions of grain-boundary and misorientation at 973 K, 10
3 s
1: (a, c) strain of 0.03, (b, d) strain of 0.15 [23]. Blue lines indicate LABs (<15 ) and black lines indicate HABs
(>15 ). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

including void nucleation (Fig. 6), void growth (Fig. 7) and void concentration, then also suppress the void nucleation. The relief of
coalescence (Fig. 8), were analyzed by comparing the situations at stress concentration caused by DRX would also suppress void
different temperatures of 973, 1023 and 1073 K. From Fig. 6, it is growth, as shown in Fig. 7. With the lower DRX fraction (Fig. 7(a)
found that the void nucleation rate is decreased with the increase of and (b)), the voids grow towards the direction without DRX grains.
working temperature, which is mainly caused by two reasons. With With the higher DRX fraction (Fig. 7(c)), the voids are surrounded
the increase of working temperature, the b phase fraction would by DRX grains, whose size is obviously smaller than those in
increase and undertake more deformation due to its lower hard- Fig. 7(a) and (b). These suggest that DRX possesses an obstacle ef-
ness than a phase. This will reduce the deformation heterogeneity fect to void growth. As for void coalescence, it is mainly achieved by
between two phases, and further decrease the driving force for void the ligament necking driven by local stress concentration. The gap
nucleation. Another reason is the increase of DRX kinetics with between two voids to be coalesced (coalescence gap) is an impor-
temperature. This would cause flow softening and relieve the stress tant indicator evaluating the level of difficulty for the occurrence of

Fig. 6. Void nucleation at the strain rate of 10
2s
1 and different temperatures: (a) 973 K; (b) 1023 K; (c) 1073 K.
6 P.F. Gao et al. / Journal of Alloys and Compounds 831 (2020) 154851

Fig. 7. Void growth at the strain rate of 10
2s
1 and different temperatures: (a) 973 K; (b) 1023 K; (c) 1073 K.

Fig. 8. Void coalescence at the strain rate of 10
2s
1 and different temperatures: (a) 973 K; (b) 1023 K; (c) 1073 K.

void coalescence [27]. From Fig. 8, it can be seen that the coales- tension deformation experiments at 1023 K. It can be seen that the
cence gap is decreased gradually from Fig. 8(a)e(c). This suggests fracture strain is first increased and then decreased with stress
that the increase of DRX kinetic would relieve the stress concen- triaxiality. The maximum fracture strain occurs at the stress triax-
tration and then retard the occurrence of void coalescence. iality of 0.33 corresponding to the uniaxial tensile test. The mini-
The above analyses indicate that increasing b phase and DRX mum fracture strain, however, exists at the stress triaxiality of 0.6
fractions would suppress the whole damage evolution, including corresponding to the notch tensile test. The above variation of
void nucleation, growth and coalescence, which may further affect fracture strain with microstructure and stress triaxiality needs to be
the plasticity (fracture strain) of material. Fig. 9 shows the variation considered in modelling of the effect if represented microstructure
of fracture strain with b phase and DRX fraction at different con- and stress state on damage phenomenon and will be described in
ditions. It can be seen that the fracture strain is increased with both Section 4.2.
of b phase fraction and DRX fraction in an exponential form.
As mentioned above, in addition to microstructure evolution, 4. Constitutive model
the stress state would also affect the damage and fracture. Fig. 10
shows the variation of fracture strain with stress state in the Fig. 11 shows the framework for constitutive modelling, which

Fig. 9. Variation of fracture strain with b phase fraction (a) and DRX fraction (b) at different conditions.
 P.F. Gao et al. / Journal of Alloys and Compounds 831 (2020) 154851 7

evolution in hot deformation of titanium alloy. It was developed by

the authors in previous work [23], which is briefly described in the
following. The a phase fraction (fa) and b phase fraction (fb) can be
calculated by the following equations, respectively.

fa ¼ z1 f1
exp½z2 ðz3
TÞg
(1)
fb ¼ 1
fa

where T is temperature. z1 , z2 and z3 are material constant with the

value of 0.9,
0.012 and 1263 K, respectively, for TA15 alloy [23].
The variation of dislocation density is represented by KM model
in Eq. (2) [28], which considers the working hardening and DRV.

dr pffiffiffi
p ¼ k1 r
k2 r (2)
dεm
p
where r is dislocation density, εm is the plastic strain of matrix, k1
Fig. 10. Variation of fracture strain with stress triaxiality at 1023 K. describes the athermal dislocation storage. k2 evaluates the ther-
mally activated dynamic recovery through dislocation cross-slip or
climb, which is related to deformation temperature and strain rate:
mainly consists of three parts: microstructure modelling, damage
modelling and constitutive law. For microstructure modelling, an
 p 
1=n
ISV based model considering the variation of phase fraction, k2 ¼ k20 ε_ m expðQact =RTÞ (3)
dislocation multiplication, DRV, DRX and loss of Hall-Petch
strengthening was applied. The damage model was derived based where Qact is the activation energy, R is the gas constant (8.314 J/
on the classical void-based GTN model and considers the effects of (mol$K)), k20 and n are material constants.
microstructure and stress state on damage evolution (including For the important microstructure phenomenon DRX, its fraction
void nucleation, growth and coalescence) by introducing b phase evolution during hot deformation can be expressed by:
fraction, DRX fraction and stress triaxiality into the GTN model. The
constitutive law was then developed based on the microstructure  p  c1  
and damage models. The flow stress only considering the micro- S_ ¼ c0 ð1
SÞ ε_ m q εp
εc Mb P = d (4)
structure evolution was first calculated as the matrix stress of
material. Based on the matrix stress, the final equivalent stress where S is the DRX fraction, q is step function, c0 and c1 are material
considering the softening effect of damage evolution would be constants, εc , Mb , P and d are the critical strain for occurrence of
obtained. As a result, the microstructure and damage based DRX, grain-boundary mobility, driving force per unit area and
constitutive model was developed for the unified prediction of flow average grain diameter, respectively. Their detailed formulations
stress, microstructure, damage and fracture during hot deformation are given in the following. The critical strain εc is a function of
of titanium alloy. temperature and strain rate:


4.1. Microstructure modelling  p f1 Qact f2
εc ¼ f0 ε_ m exp (5)
RT
An ISV based microstructure model considering the variation of
phase fraction, dislocation density variation, DRX and loss of Hall- where f0 , f1 , and f2 are material constants. The grain-boundary
Petch strengthening was applied to describe the microstructure mobility Mb in Eq. (4) is denoted as:

Fig. 11. Microstructure and damage based constitutive modelling.

8 P.F. Gao et al. / Journal of Alloys and Compounds 831 (2020) 154851

Mises equivalent stress, flow stress of the undamaged matrix ma-

dD0 b Qpd terial and hydrostatic stress, respectively. q1 and q2 are fitting co-
Mb ¼ exp
(6)
kT RT efficients. Tvergaard [32,33] has proposed that taking q1 as 1.5 and
q2 as 1 can promote the predictive ability for the material with
where d is characteristic grain-boundary thickness, D0 is self- lower hardening exponent. Since titanium alloy is right a material
diffusion constant, k is Boltzmann’s constant, and Qpd is with lower hardening exponent, thus they are used in this study. f *
boundary-diffusion activation energy. The driving force per unit is equivalent void volume fraction evaluating the damage degree,
area P takes the form: which is a function of void volume fraction f :


P ¼ rGb2 (7) f f  fc
f* ¼ (15)
fc þ kðf
fc Þ f > fc
where Gb2 is the linear energy of the dislocation. Considering the
static growth and DRX-induced grain refinement, the evolution of where, k is accelerating factor for void growth and represented by
average grain diameter can be expressed as:
.
k ¼ ðfu
fc Þ ff
fc (16)
d_ ¼ a1 d
g1
a2 S_ 2 dg3
g
(8)

where d is average grain size, a1, a2, g1, g2, and g3 are material where fu ¼ 1=q1 , which is the critical void volume fraction when
constants. the stress-carrying capacity vanishes, fc and ff are the critical void
It should be noted that DRX will also cause dislocation annihi- volume fraction for occurrence of void coalescence, and the critical
lation, which should also be considered in the variation of dislo- void volume fraction for fracture.
cation density. Thus, the evolution of density can be expressed in The description of void volume fraction variation, i.e., damage
the form: evolution, is the key part in GTN model. It is derived from the void
nucleation, void growth and shear effect as the following equation:
pffiffiffi
r_ ¼ ε_ m k1 r
k2 r
rS_ = ð1
SÞ
p
(9)
df ¼ dfg þ dfn þ dfs (17)
In addition, Semiatin and Bieler [29] reported that the loss of
where the first term dfg evaluates the void growth and is related to
grain-boundary (a/b interface) strengthening, i.e., Hall-Petch
the plastic volumetric strain:
strengthening, is also a critical microstructure phenomenon and
plays a great role in flow softening. Thus, the loss of Hall-Petch
dfg ¼ ð1
f Þdεpkk (18)
strengthening has also been considered in the microstructure
modelling. The Hall-Petch effect was evaluated by
where dεpkk is the plastic volumetric strain increment induced by
pffiffiffi hydrostatic stress. The second term in eq. (17) dfn represents the
KHP ¼ HG b (10)
void nucleation and is expressed by the formulation proposed by
Chu and Needleman [34]:
where KHP is Hall-Petch coefficient, H takes the form of
p
H ¼ lH0 (11) dfn ¼ Adεm (19)
εpm
εN
where l is a coefficient with an initial value of 1 and varies with where A ¼ p fNffiffiffiffiffi
exp
12 sN , dεpm is equivalent plastic strain
s N 2p
strain according to the following equation: increment of matrix, fN is volume fraction of nucleated voids, εN is
nucleation strain, sN is standard deviation of nucleation strain. The
 p
dl dεm ¼ cl ðlss
lÞ (12) third term in eq. (17) dfs estimates the contribution of shear effect
to the variation of void volume fraction, through which the effect of
where cl and lss are material constants. H0 in eq. (11) is a function of stress state on damage evolution can be considered. The NH model
temperature, strain rate, and phase volume fraction: proposed by Nahshon and Hutchinson [25] is used here to repre-
sent this effect:
 p j1 Qact j2
H0 ¼ Href ε_ m exp ðexpðfa ÞÞj3 (13) sij dεij
p
RT
dfs ¼ kw wf (20)
s
where Href, j1, j2, and j3 are material constants.
where kw is the magnitude of damage growth rate in pure shear
state, sij is deviatoric stress tensor, dεpij is plastic strain increment, w
4.2. Damage modelling is a function describing the stress state:

2
GTN model developed by Gurson, Tvergaard and Needleman 27J3
w¼1
(21)
[30e32] is a well-accepted damage-mechanism-based model to 2s3
predict the damage and fracture of ductile metals. It is very suitable
to describe the ductile fracture through void nucleation, growth where J3 is the third invariant of deviator stress. w locates in the
and coalescence, thus which is applied in this work. The GTN model range between 0 and 1. At compression and tensile state, w equals
is formulated as: 0. At pure shear stress state, w equals 1.
As mentioned in Section 3.2, b phase and DRX fractions both
s 2
3 q2 sH have great effects on the void nucleation, growth and coalescence
F¼ þ 2q1 f * cosh

1 þ q21 f *2 ¼ 0 (14)
sm 2 sm in the hot deformation of TA15 alloy. However, the classical equa-
tions describing void evolution (eq. (15-19)) does not consider the
where F is the yield surface of material. s, sm , and sH are the von effects of microstructure evolution. To overcome this problem, they
 P.F. Gao et al. / Journal of Alloys and Compounds 831 (2020) 154851 9

are improved by introducing b phase fraction and DRX fraction in where t0 is the mechanical threshold stress, DG is the activation
an exponential form according to the exponential variation law of energy for deformation, ε_ ref is the reference strain rate. p and q are
fracture strain in Fig. 9. The detail improvements for equations material constants. The later athermal stress mainly considers the
describing damage evolution are as follows. Nucleation strain εN is strain-hardening and grain-boundary effects, which can be
the critical factor determining development of void nucleation, expressed by
which is updated by incorporating b phase and DRX fractions:
pffiffiffi
   tm ¼ aGb r þ KHP d
1=2 (29)
εN ¼ εN0 exp Cs S þ Cb fb* (22)
where G is the shear modulus. For TA15 alloy, it is a function of
For void growth, a growth factor c depending on b phase and temperature [23]:
DRX fractions is introduced in eq. (18) to consider the microstruc-
ture effect. The updated formulation for void growth rate is: 5:821
G ¼ 49:02
(30)
   expð181=TÞ
1
dfg ¼ cð1
f Þdεpkk ¼ c0 exp Ds S þ Db fb* ð1
f Þdεpkk (23)
a is material constant, and b is the magnitude of Burgers vector
Similarly, the critical void volume fraction for void coalescence fc (2.95  10
10 m). KHP and d are the Hall-Petch coefficient and
is also improved to consider the effects of b phase and DRX average grain size, respectively, which have been described in
fractions: microstructure modelling.
   Combining eq. (26-30) and microstructure model in Section 4.1,
fc ¼ fc0 exp Es S þ Eb fb* (24) the flow stress of TA15 alloy considering microstructure evolution
can be calculated. It should be noted that this is the matrix stress of
In eq. (22-24), εN0 , c0 , fc0 are reference values of nucleation
material without any damage, which is just applicable to predict
strain, growth factor and critical void volume fraction for coales-
the flow stress of titanium alloy in early deformation stage, as
cence, respectively. c0 is taken as 1 in this work. While, εN0 and fc0
described in Refs. [12e15,23]. It is still needed to substitute the
will be determined in Section 5.1. fb* is a parameter related to b
matrix stress in eq. (26) into eq. (14) and calculate the equivalent
phase fraction, which is expressed by
stress considering damage-induced softening based on damage
fb
fbr model in Section 4.3. This equivalent stress is right the desirable
fb* ¼ (25) combined result considering both the microstructure and damage
1
fbr evolution.

where fbr is reference value of b phase fraction. According to eq. (1),

the b phase fraction varies little below 973 K, while varies greatly 4.4. Numerical implementation of the model
above 973 K. Thus, the b phase fraction at 973 K is taken as the
reference value. Besides, Cb ; Db ; Eb and Cs ; Ds ; Es are b phase fraction The proposed microstructure and damage based constitutive
and DRX fraction related coefficients. model cannot be solved analytically. The return mapping algorithm
Substituting the improved equations (22-24) into classical GTN combined with implicit Euler backward method was used to realize
model, the effects of microstructure (b phase and DRX fractions) on the numerical calculation, which will be described in this section.
void nucleation, growth and coalescence can be considered in The detailed numerical calculation procedure is given in Fig. 12.
damage modelling.
p
Step 1 Initialize the processing parameters (T; ε_ m ), microstructural
4.3. Constitutive law parameters (r; d), equivalent plastic strain of matrix (εpm ),
volume fraction of void (f ), stress tensor and strain tensor at
In this section, the constitutive law considering microstructure t ¼ 0 time.
and damage evolution will be developed. First, the matrix stress p p
Step 2 Read the variable values of st , εt , εm;t , Dεm;t , rt , dt , ft and Tt at
only considering microstructure evolution for a polycrystal metal t moment.
can be represented by Ref. [23]: p
Step 3 Calculate elastic trial stress sT , sT , sT
tþDt tþDt and ε_
H;tþDt m;tþDt
at tþDt moment.
sm ¼ Mt (26)
Step 4 Calculate the microstructural parameters StþDt , fa;tþDt and
fb;tþDt according to equations in Section 4.1 and sm;tþDt ac-
where M is the Taylor factor with the value of 3.06. According to the
cording to equations in Section 4.3.
Taylor assumption, t is the shear stress consisted of two parts, as
Step 5 Calculate yield function:
shown below:

F2 ¼ F sTtþDt ; sm;tþDt ; sTH;tþDt ; εm;t ; ft*

p
t ¼ t* þ tm (27)

where t* is the thermally activated stress for moving dislocations If F2 < 0, the current state is pure elastic state, and stþDt ¼ sTtþDt .
through lattice and pass short-range obstacles. tm is the athermal Then, go back to Step 2 and start the next incremental step.
stress caused by long-range effects (such as the grain boundaries If F2  0, the current state is elastic-plastic state. Go to Step 6 and
and dislocation forests). The former thermally activated stress de- start the plastic correction.
pends on the temperature and strain rate, which can be calculated
by Step 6 In terms of the associated flow law and yield condition, the
following two equations are given.
" #
1=q 1=p
* 0 RT ε_ ref vF vF
t ¼t 1
ln (28) F1 ¼ Dεpv þ Dεpd ¼0 (31)
DG ε_ pm vs vsH
10 P.F. Gao et al. / Journal of Alloys and Compounds 831 (2020) 154851

Fig. 12. Numerical calculation procedure of the microstructure and damage based constitutive model.

Step 7. Calculate the increment values of εpm and f according to the

 p *
 p
following equations at the current iteration. Then, the εm and f are
F2 ¼ F s; sm ; sH ; εm ; f ¼0 (32)
updated.
where Dεpv and Dεpd are plastic strain increment induced by hy-
drostatic stress and equivalent stress, respectively. Their correction
values can be obtained through solving eq. (31-32) using Newton-
p p
Raphson iteration. Then the Dεv and Dεd are updated.
 P.F. Gao et al. / Journal of Alloys and Compounds 831 (2020) 154851 11

related to stress state (kw ). They will be calibrated in sequence

p p

sH Dεv þ sDεd below.
Dεpm ¼ (33)
ð1
f Þsm The parameters in classical GTN model (f0 ; fN ; fc ; ff ; εN0 ; SN0 ) are
not relevant to the microstructure evolution, thus they are first
Df ¼ Dfg þ Dfn þ Dfs (34) calibrated by the uniaxial tensile test at room temperature. The
parameter calibration was conducted by a careful trial-and-error
method to make the simulated load-displacement curve closely
Step 8 Calculate the increment values of r and d according to match the experimental one. Table 3 lists the determined param-
equations in Section 4.1 and update them. eters in classical GTN model. From Fig. 13, it can be seen that the
Step 9 Check the yield condition and convergence condition, i.e., simulated load-displacement curves and fracture point (marked
eq. (31-32). If |F1| < 10
10 and |F2| < 10
10, the iteration with star) agree very well with the experimental results.
converges. Output the stress, strain and internal state vari- For the new introduced parameters related to microstructure
ables with the values derived from the last iteration and features (Cb ;Cs ;Db ;Ds ;Eb ; Es ), they are calibrated in two sub-groups.
start the next incremental step. Otherwise, go back to Step 6 The DRX related parameters Cs ; Ds ; ES were first determined
and start a new iteration. through the uniaxial tensile tests with various DRX kinetics. The
detailed deformation conditions of tensile tests are 973 K and
10
3s
1, 10
2 s
1. At these conditions, the b phase fraction is very
5. Model validation
small and changes little, thus the b phase fraction related param-
eters Cb ; Db ; Eb make little sense to the calculation results. The
5.1. Determination of model parameters
determined Cs ; Ds ; ES are 1.75,
0.375 and 0.625, respectively. Using
these parameters, good agreement between the simulated and
As mentioned in Section 4, the numerical calculation of consti-
experimental load-displacement curves can be obtained, as shown
tutive model includes two main steps: (1) calculating the matrix
in Fig. 14. Then, the b phase fraction related parameters Cb ; Db ; Eb
flow stress based on microstructure model and constitutive law, (2)
are determined by the uniaxial tensile tests with strain rate of 10
2
calculating the final equivalent flow stress based on matrix stress
s
1 and temperatures of 1023 K and 1123 K. These samples present
and damage model. Correspondingly, the model parameters were
different b phase fraction and DRX kinetics during deformation. The
also determined in two parts: parameters in microstructure and
determined Cb ; Db ; Eb are 0.25,
0.125, and 0.125, respectively.
constitutive models, and parameters in damage model.
Fig. 15 shows the simulated results under these parameters, which
The first part of parameters involved in microstructure and
are very close to the experimental results. Based on the above
constitutive models (Section 4.1 and 4.3) were determined by ge-
analysis, the determined new introduced parameters related to
netic algorithm objective optimization method. The results of
microstructure features in damage model are summarized in
uniaxial tensile experiments at temperatures of 973, 1023, 1073,
Table 4.
1123 K and strain rates of 10
3, 10
2, 10
1 s
1 were used to calibrate
At last, the stress state related parameter kw was determined by
the parameters. The objective functions were defined as the square
tensile tests with various stress states and deformation parameters.
of difference between experimental and numerical calculated data
The detailed tensile tests include 0 shear tensile test at room
for stress, DRX fraction and grain size. The model parameters were
temperature, 45 shear tensile test at 1023 K and notch tensile test
determined by minimizing the sum of above objective functions. It
at 1073 K. The calibration indicates that kw should be 1.5 to obtain
should be noted that the later-stage stress-strain curve presenting
well simulated load-displacement curves close to the experimental
significant flow softening cannot be used here, because which is
curves, as shown in Fig. 16.
closely related to the damage development. The detailed determi-
nation process for parameters in microstructure and constitutive
law can be found in the authors’ previous work [23]. The deter- 5.2. Model application
mined parameters are given in Table 2, which have been verified by
the comparisons between predicted and experimental results of In this section, the proposed microstructure and damage based
stress, DRX fraction and grain size. constitutive model was applied in a hot spinning process of TA15
On the basis of determined parameters in Table 2, the estab- alloy tube to verify its reliability. During hot spinning, the material
lished microstructure and damage based constitutive model was deforms under complex temperature, strain rate and stress states,
integrated into ABAQUS/Explicit via user subroutine to simulate the which will cause phase transformation, DRX, damage and fracture.
hot tensile tests. The simulated load-displacement results were
compared with the experimental ones to calibrate the parameters
in damage model (Section 4.2). The parameters in damage model Table 3
can be classified into three groups: the parameters in classical GTN Determined parameters in the classical GTN damage model.
model (f0 ;fN ;fc ;ff ;εN0 ;SN0 ), the new introduced parameters related to Parameters f0 fN εN0 SN0 fc0 ff
microstructure features (Cb ; Cs ; Db ; Ds ; Eb ; Es ) and the parameters
Values 0.0015 0.03 0.25 0.1 0.08 0.15

Table 2
Determined model parameters in the microstructure and constitutive law [23].

Parameters t0 DG q p a k1 k20

Value 352.2 419.2 0.083 0.424 0.500 4.0  109 1.2  107
Parameters Qact n c0 c1 f0 f1 f2
Value 602.1 6.0 1.510 0.453 5.85  10
5 0.102 0.093
Parameters dDo Qb a1 g1 a2 g2 g3
Value 0.550 295.0 1.1  10
7 3.9  10
6 0.026 0.9 3.9  10
6
Parameters cll lss j1 j2 j3 Href
Value 0.8 0.05 0.105 0.497 5.004 0.0217
12 P.F. Gao et al. / Journal of Alloys and Compounds 831 (2020) 154851

Table 4
Determined new introduced parameters related to microstructure features in
damage model.

Parameters CS Cb DS Db ES Db

Value 1.75 0.25
0.375
0.125 0.625 0.125

Fig. 13. Comparison between the simulated and experimental load-displacement

curves of uniaxial tensile test at room temperature.

Fig. 16. Comparisons between the simulated and experimental load-displacement of

tensile tests with different stress states and temperatures.

This makes it very suitable to valid the proposed constitutive model

in this study.
The FE model for hot spinning of TA15 alloy tube considering the
thermal and deformation events was established, as shown in
Fig. 17. The tie constraint is employed at the bottom of blank and
mandrel to limit the movement of tube blank. Three rollers coun-
terclockwise revolve and feed along the axial direction at the same
time. The mandrel and roller are assumed to be isothermal
analytical rigid bodies. The tube blank is set as a deformable body
and discretized through C3D8RT element. To avoid mesh distortion,
the ALE adaptive grid technique was applied. The proposed
Fig. 14. Comparisons between the simulated and experimental load-displacement microstructure and damage based constitutive model was inte-
curve of the hot uniaxial tensile tests at 973 K and 10
3s
1, 10
2s
1.
grated into ABAQUS via user subroutine to describe the material
behavior of tube. It should be noted that the damage just accu-
mulates when the stress triaxiality is larger than 0. The friction
between tube blank and tools was described by Coulomb’s friction
model with the coefficient of 0.1. For the thermal events, 90%
deformation energy was set to be converted into thermal energy.

Fig. 15. Comparisons between the simulated and experimental load-displacement

curves of the hot uniaxial tensile tests at 1023 K, 1123 K and 10
2s
1.

Fig. 17. FE model of hot spinning of TA15 alloy tube.

 P.F. Gao et al. / Journal of Alloys and Compounds 831 (2020) 154851 13

Table 5 comparisons between simulated and experimental forming results.

Processing parameters of spinning. It can be seen from Fig. 18(a) that the simulated thickness distri-
Parameters Value bution along circumferential direction is very close to experimental
Outer diameter/inner diameter/height of tube billet (mm) 346/308/120
result with the maximum relative error of 13.2%. As for the pre-
Mandrel diameter/height(mm) 307.5/330 diction accuracy of microstructure evolution, the microstructure at
Roller diameter/(mm) 116 mid-thickness point was observed and compared with the simu-
Roller radius(mm) 10 lated results, as shown in Fig. 18(b) and (c). The predicted and
Attack angle/ 22.5
experimental DRX fractions are 0.612 and 0.581, respectively, pre-
Mandrel revolution speed(rad/s) 9.32
Roller feed rate(mm/r) 1 senting the relative error of 5.3%. As for b phase fractions, the
Spinning thinning ratio 50% predicted and experimental values are 0.228 and 0.212, respec-
Initial spinning mandrel temperature (K) 623 tively, with the relative error of 7.5%. Besides, the prediction of
Initial spinning temperature (K) 1073
damage and fracture was also verified by a set of spinnability ex-
periments. In spinnability experiments, the spinning thinning ratio
increases gradually to produce fracture. In simulation, the void
The heat conduction, heat film and heat radiation were also volume fraction (VVF) reaches the critical value of fracture when
considered to simulate the temperature change of blank. Table 5 the thinning ratio increases to 68.9%, as shown in Fig. 19(a). While,
shows the processing parameters in hot spinning. the fracture produces when the thinning ratio reaches 71.2% in
To validate the FE model, hot spinning experiment was con- experiment. The relative error of predicted limiting thinning ratio is
ducted at the same conditions shown in Table 5. Fig. 18 shows the 3.2%.

Fig. 18. Comparisons between the simulated and experimental results: (a) thickness distribution along circumferential direction; (b) DRX fraction; (c) b phase fraction.

Fig. 19. Comparisons between the simulated damage and fracture at thinning ratio of 68.9% (a) and experimental fracture results at thinning ration of 71.2% (b).
14 P.F. Gao et al. / Journal of Alloys and Compounds 831 (2020) 154851

The above comparisons suggest that the microstructure and National Science Fund for Distinguished Young Scholars of China
damage based constitutive model proposed in this work is valid (No. 51625505) and the Key Program Project of the Joint Fund of
and can be applied in the FE modelling of hot working of TA15 alloy Astronomy and National Natural Science Foundation of China (No.
to predict the macro-shape, DRX evolution, damage and fracture U1537203) National Natural Science Foundation of China (No.
phenomena at the same time. It is helpful to reveal the micro- 51875467), the Hong Kong Scholar Program (No. XJ2018010), the
structure evolution, damage and fracture behaviour of titanium Young Elite Scientists Sponsorship Program by CAST (No.
alloy during hot deformation, and can provide important guidance 2018QNRC001) and the Research Fund of the State Key Laboratory
for the coordinated control of microstructure and damage in the of Solidification Processing (NPU), China (Grant No. 2019-TS-10).)
hot working of titanium alloy components.

6. Conclusions References

[1] D. Banerjee, J.C. Williams, Perspectives on titanium science and technology,

In this paper, the microstructure, damage and flow behavior of Acta Mater. 61 (2013) 844e879.
TA15 alloy during hot tensile deformation were experimentally [2] G. Lütjering, J.C. Williams, Titanium, second ed., Springer-Verlag, Berlin Hei-
analyzed. Based on the experimental results, a microstructure and delberg, 2007.
[3] P.F. Gao, M.W. Fu, M. Zhan, Z.N. Lei, Y.X. Li, Deformation behavior and
damage based constitutive model was developed. The following microstructure evolution of titanium alloys with lamellar microstructure in
conclusions can be drawn: hot working process: a review, J. Mater. Sci. Technol. 39 (2020) 56e73.
[4] S. Semiatin, V. Seetharaman, A. Ghosh, Plastic flow, microstructure evolution,
and defect formation during primary hot working of titanium and titanium
(1) The increasing b phase fraction and DRX fraction would aluminide alloys with lamellar colony microstructures, Phil. Trans. Roy. Soc.
cause softening effect, relieve local stress concentration, and Lond. Math. Phys. Eng. Sci. 357 (1999) 1487e1512.
thus suppress damage evolution (void nucleation, growth [5] S. Tamirisakandala, B.V. Vedam, R.B. Bhat, Recent advances in the deformation
processing of titanium alloys, J. Mater. Eng. Perform. 12 (2003), 661-173.
and coalescence). As a result, fracture strain is increased with [6] M. Zhan, T. Zhang, H. Yang, L. Li, Establishment of a thermal damage model for
both b phase fraction and DRX fraction in an exponential Ti-6Al-2Zr-1Mo-1V titanium alloy and its application in the tube rolling-
form. In addition, the stress state also presents great effect on spinning process, Int. J. Adv. Manuf. Technol. 87 (2016) 1345e1357.
[7] P.M. Souza, H. Beladi, R. Singh, B. Rolfe, P.D. Hodgson, Constitutive analysis of
the damage and fracture strain. hot deformation behavior of a Ti6Al4V alloy using physical based model,
(2) A microstructure and damage based constitutive model for Mater. Sci. Eng. A 648 (2015) 265e273.
hot deformation of titanium alloy was developed. The model [8] A.O. Mosleh, A.V. Mikhaylovskaya, A.D. Kotov, J.S. Kwame, S.A. Aksenov, Su-
perplasticity of Ti-6Al-4V titanium alloy: microstructure evolution and
is mainly composed of three parts: microstructure model,
constitutive modelling, Mater. 12 (2019) 1756.
damage model and constitutive law. The microstructure [9] M.H. Ghavam, M. Morakabati, S.M. Abbasi, H. Badri, Flow behavior modeling
model is developed using the internal variable method, of IMI834 titanium alloy during hot tensile deformation, Trans. Nonferrous
which considers phase transformation, variation of disloca- Metals Soc. China 25 (3) (2015) 748e758.
[10] N.S. Reddy, C.H. Park, Y.H. Lee, C.S. Lee, Neural network modelling of flow
tion density, DRX and loss of Hall-Petch strengthening. The stress in Tie6Ale4V alloy with equiaxed and Widmansta €tten microstructures,
damage model considers the effects of microstructure and Mater. Sci. Technol. 24 (2013) 294e301.
stress state on damage evolution by introducing b phase [11] J. Liu, W. Zeng, Y. Lai, Z. Jia, Constitutive model of Ti17 titanium alloy with
lamellar-type initial microstructure during hot deformation based on
fraction, DRX fraction and stress triaxiality into the classical orthogonal analysis, Mater. Sci. Eng. A 597 (2014) 387e394.
GTN model. The constitutive law considers both effects of [12] J. Luo, M. Li, X. Li, Y. Shi, Constitutive model for high temperature deformation
microstructure and damage evolution on flow behavior. of titanium alloys using internal state variables, Mech. Mater. 42 (2010)
157e165.
(3) The parameters of the microstructure and damage based [13] X.G. Fan, H. Yang, Internal-state-variable based self-consistent constitutive
constitutive model were calibrated by comparing the pre- modelling for hot working of two-phase titanium alloys coupling micro-
dicted and experimental results. The developed model was structure evolution, Int. J. Plast. 27 (2011) 1833e1852.
[14] Q. Bai, J. Lin, T.A. Dean, D.S. Balint, T. Gao, Z. Zhang, Modelling of dominant
successfully applied in the numerical simulation of hot
softening mechanisms for Ti-6Al-4V in steady state hot forming conditions,
spinning of TA15 alloy tube. It can accurately predict the Mater. Sci. Eng. A 559 (2013) 352e358.
macroscopic deformation, microstructure, damage and [15] B. Babu, L.-E. Lindgren, Dislocation density based model for plastic deforma-
tion and globularization of Ti-6Al-4V, Int. J. Plast. 50 (2013) 94e108.
fracture during hot spinning, which verifies the reliability of
[16] G.R. Johnson, W.H. Cook, Fracture characteristics of three metals subjected to
the developed constitutive model. The developed model fa- various strains, strain rates, temperatures and pressures, Eng. Fract. Mech. 21
cilitates the coordinated control of microstructure and (1985) 31e48.
damage in hot working of titanium alloys. [17] J. He, Z. Cui, F. Chen, Y. Xiao, L. Ruan, The new ductile fracture criterion for
30Cr2Ni4MoV ultra-super-critical rotor steel at elevated temperatures, Mater.
Des. 52 (2013) 547e555, 1980-2015.
[18] B.T. Tang, S. Bruschi, A. Ghiotti, P.F. Bariani, An improved damage evolution
Declaration of competing interest model to predict fracture of steel sheet at elevated temperature, J. Mater.
Process. Technol. 228 (2016) 76e87.
[19] L. Yang, B. Wang, G. Liu, H. Zhao, W. Xiao, Behavior and modelling of flow
The authors declare that they have no known competing softening and ductile damage evolution in hot forming of TA15 alloy sheets,
financial interests or personal relationships that could have Mater. Des. 85 (2015) 135e148.
[20] J. Li, B. Wang, H. Huang, S. Fang, P. Chen, J. Zhao, Y. Qin, Behaviour and
appeared to influence the work reported in this paper. constitutive modelling of ductile damage of Ti-6Al-1.5Cr-2.5Mo-0.5Fe-0.3Si
alloy under hot tensile deformation, J. Alloys Compd. 780 (2019) 284e292.
CRediT authorship contribution statement [21] X. Shang, Z. Cui, M.W. Fu, A ductile fracture model considering stress state and
ZenereHollomon parameter for hot deformation of metallic materials, Int. J.
Mech. Sci. 144 (2018) 800e812.
P.F. Gao: Writing - original draft, Methodology, Investigation. J. [22] X. Shang, Z. Cui, M.W. Fu, Dynamic recrystallization based ductile fracture
Guo: Data curation, Software, Validation. M. Zhan: Supervision, modelling in hot working of metallic materials, Int. J. Plast. 95 (2017)
105e122.
Project administration. Z.N. Lei: Visualization. M.W. Fu: Concep-
[23] J. Guo, M. Zhan, Y.Y. Wang, P.F. Gao, Unified modelling of work hardening and
tualization, Writing - review & editing. flow softening in two-phase titanium alloys considering microstructure
evolution in thermomechanical processes, J. Alloys Compd. 767 (2018) 34e45.
Acknowledgments [24] A. Hadadzadeh, F. Mokdad, M.A. Wells, D.L. Chen, A new grain orientation
spread approach to analyze the dynamic recrystallization behavior of a cast-
homogenized Mg-Zn-Zr alloy using electron backscattered diffraction,
The authors would acknowledge the funding support from the Mater. Sci. Eng. A 709 (2018) 285e289.
 P.F. Gao et al. / Journal of Alloys and Compounds 831 (2020) 154851 15

[25] K. Nahshon, J.W. Hutchinson, Modification of the Gurson model for shear Mater. Technol. 99 (1977) 2e15.
failure, Eur. J. Mech. Solid. 27 (2008) 1e17. [31] V. Tvergaard, A. Needleman, Analysis of the cup-cone fracture in a round
[26] L. Xue, Constitutive modelling of void shearing effect in ductile fracture of tensile bar, Acta Metall. 32 (1984) 157e169.
porous materials, Eng. Fract. Mech. 75 (2008) 3343e3366. [32] V. Tvergaard, Influence of voids on shear band instabilities under plane strain
[27] M.A. Rahman, C. Butcher, Z. Chen, Void evolution and coalescence in porous conditions, Int. J. Fract. 17 (1981) 389e407.
ductile materials in simple shear, Int. J. Fract. 177 (2012) 129e139. [33] V. Tvergaard, On localization in ductile materials containing spherical voids,
[28] U.F. Kocks, H. Mecking, Physics and phenomenology of strain hardening: the Int. J. Fract. 18 (1982) 237e252.
FCC case, Prog. Mater. Sci. 48 (2003) 171e273. [34] C.C. Chu, A. Needleman, Void nucleation effects in biaxially stretched sheets,
[29] S.L. Semiatin, T.R. Bieler, The effect of alpha platelet thickness on plastic flow J. Eng. Mater. Technol. 102 (1980) 249e256.
during hot working of TIe6Ale4V with a transformed microstructure, Acta [35] J. Guo, M. Zhan, M.W. Fu, P.F. Gao, F. Ma, Extrapolation based constitutive
Mater. 49 (2001) 3565e3573. modeling of flow stress of titanium alloy sheet under hot-working condition,
[30] A.L. Gurson, Continuum theory of ductile rupture by void nucleation and Mater. Des. 154 (2018) 96e107.
growth: Part Idyield criteria and flow rules for porous ductile media, J. Eng.