Transactions on Engineering Sciences vol 6, © 1994 WIT Press, www.witpress.

com, ISSN 1743-3533

Phenomenological modelling of adiabatic

shear band localization fracture of solids in
dynamic loading processes
P. Perzyna, M.K. Duszek-Perzyna
Centre of Mechanics, Institute of Fundamental Technological
Research, Polish Academy of Science, Swietokrzyska 21,
00-049 Warsaw, Poland

ABSTRACT

The main objective of the present paper is the application of a recently developed
viscoplastic-damage type constitutive theory for high strain rate flow process
and ductile fracture to the problem of shear band localization and fracture of
dynamically loaded inelastic bodies experiencing strain rates ranging between
10^ — 10^ s~*. In the first part of the paper an adiabatic inelastic flow process is
formulated and investigated. The Cauchy problem is examined and the conditions
for well-posedness are discussed. The relaxation time is used as a regularization
parameter. The viscoplastic regularization procedure assures the unconditionally
stable integration algorithm by using thefiniteelement method. The second part
of the paper is devoted to the numerical investigation of the three-dimensional
dynamic adiabatic deformations of a steel thin tube twisted in a split Hopkinson
bar at nominal strain rates ranging 10"* - 10^ s~"\

INTRODUCTION

In technological dynamical processes fracture can occur as a result of an

adiabatic shear band localization generally attributed to a plastic instability
generated by thermal softening during plastic deformation.
Hartley, Duffy and Hawley [4], Marchand and Duffy [11], Marchand, Cho
and Duffy [10] and Cho, Chi and Duffy [2] made microscopic observations
of the shear band localization on the thin-welled steel tubes in a split Hop-
kinson torsion bar Three different steels were tested. Dynamic deformation
in shear was imposed to produce shear bands. It was found whenever the
shear band led to fracture of the specimen, the fracture occurred by a pro-
cess of void nucleation, growth and coalescence. No cleavage was observed
 Transactions on Engineering Sciences vol 6, © 1994 WIT Press, www.witpress.com, ISSN 1743-3533

580 Localized Damage

on any fracture surface, including the most brittle of the steel tested.
This is presumably due to the thermal softening of the shear band ma-
terial that results from the local temperature rise occurring during defor-
mation process.
In recent years Zbib and Jurban [23] have investigated numerically a
three-dimensional problem involving the development of shear bands in a
steel bar pulled in tension and Batra and Zhang [1] the three-dimensional
dynamic thermomechanical deformations of a 4340 steel thin tube twisted
in a split Hopkinson bar at nominal strain rate of 1000, 2500 and 25000 s~*.
The main objective of the present paper is the application of a recently
developed viscoplastic-damage type constitutive theory for high strain rate
flow process and ductile fracture to the problem of shear band localization
and fracture of dynamically loaded thin-walled tubes experiencing strain
rates ranging between 10^ — 10^ s~*.
In chapter 2 a constitutive model is developed within a thermodynamic
framework of the rate type material structure with internal state variables.
Such important effects as the micro-damage mechanism and thermome-
chanical coupling are taken into consideration. It has been assumed that
the intrinsic micro-damage mechanism consists of the nucleation, growth
and coalescence of microvoids. The rate dependent evolution equation for
the porosity parameter has been postulated.
In chapter 3 the formulation of an adiabatic inelastic flow process is
given. The Cauchy problem is investigated and the conditions which guar-
antee its well-posedness are examined. Main feature of rate dependent
plastic model have been discussed. Particular attention has been focused on
the viscoplastic regularization procedure for the solution of the dynamical
initial-boundary value problems with localization of plastic deformation.
Some simplifications are introduced and a particular elastic-viscoplastic
constitutive model for damaged solids is developed.
Chapter 4 is devoted to the numerical investigation of the three-di-
mensional dynamic adiabatic deformations of a steel thin tube twisted in
a split Hopkinson bar at nominal strain rates ranging 10^ — 10^ s~*. A
thin shear band region of finite width along the circumference of the tube
which undergoes significant deformations and temperature rise has been
determined. Its evolution until occurrence of fracture has been simulated.
Numerical results are discussed and compared with available experimental
observation data.

CONSTITUTIVE STRUCTURE

Rate type constitutive structure for an elastic-viscoplastic material

The main objective is to develop the rate type constitutive structure for
an elastic-viscoplastic material in which the effects of the micro-damage
mechanism and thermomechanical coupling are taken into consideration.
 Transactions on Engineering Sciences vol 6, © 1994 WIT Press, www.witpress.com, ISSN 1743-3533

Localized Damage 581

Let us introduce the axioms as follows:

(i) Axiom of the existence of the energy function in the form

V» = ,Xe,F,,);,4), (1)

where e is the Eulerian strain tensor, F the deformation gradient, $

a temperature field and fi denotes the internal state variable vector.

(ii) Axiom of objectivity (spatial co variance). The constitutive structure

should be invariant with respect to any diffeomorphism £ : S — > S
in].

(iii) The axiom of entropy production. For any regular process <^, i),, fi^ of
a body B the constitutive functions are assumed to satisfy the reduced
dissipation inequality

- r : d - (777? + </>) - —q • gradtf > 0, (2)

pRef PV

where p and p^f denote the mass density in the actual and reference
configuration, respectively, r is the Kirchhoff stress tensor, d = d"+ <P
the rate of total deformation, 77 denotes the specific (per unit mass)
entropy and q is the heat vector field.

Let us postulate fi = (C,£)» where £ denotes the new internal state vector
which describes the dissipation effects generated by viscoplastic flow phe-
nomena and £ is the volume fraction porosity parameter and takes account
for micro-damage mechanism.
Let us introduce the plastic potential function for damaged material in
the form

/ = J, 4- nf J ^n; J, = r*, ^ = *^2w, (3)

n = n(tf) is the temperature dependent material function and g denotes the

metric tensor in S.
Let us postulate the evolution equations as follows (L%, defines the Lie
derivative with respect to the velocity field and the dot denotes the material
derivative)
d' = AP, Wf = AZ, ( = 5, (4)
where for the elastic-viscoplastic model of a material we assume (cf. Perzy-
na [14,15,17,18])

Tm denotes the relaxation time for mechanical disturbances and K is the

isotropic work-hardening parameter, $ is the empirical overstress function
 Transactions on Engineering Sciences vol 6, © 1994 WIT Press, www.witpress.com, ISSN 1743-3533

582 Localized Damage

and the bracket {•} defines the ramp function, P = ^4j=^, the material
function Z is intrinsically determined by the constitutive assumptions pos-
tulated and the scalar valued evolution function El has to be determined.
Thus, we have

fob = —7=T^gca9db + Agab, A = —=n£T^gab- (6)

2VJ2 W2
The isotropic hardening-softening material function AC is assumed in the
form as follows

(1 - W), (7)

where q* — ^-,KQ ACQ and ACI denote the yield and saturation stress of the
matrix material, respectively, k = /i(i9) is the temperature dependent strain
hardening function for the matrix material, &— /J(|d^ : d^)^df is the
equivalent plastic deformation, ^ denotes the value of porosity at which
the incipient fracture occurs and 6 is a material coefficient; the overstress
viscoplastic function $ is postulated in the form (cf. Perzyna [14,15])

$(/-/c)= (£-1) , where m = 1,3,5,... (8)

The axioms (i)-(iii) and the evolutions equations (4) lead to the rate
equations as follows

(9)
CppRef
where

X* and x** are the irreversibility coefficients.

To make possible numerical investigation of the three-dimensional dy-
namic adiabatic deformations of a body for different ranges of strain rate
we introduce some simplifications of the constitutive model.
(i) By analogy with the infinitesimal theory of elasticity we postulate
linear elastic properties of the material, i.e.

(jC ) =G

where G and K denote the shear and bulk modulus of damaged ma-
terial, respectively.
 Transactions on Engineering Sciences vol 6, © 1994 WIT Press, www.witpress.com, ISSN 1743-3533

Localized Damage 583

(ii) It is assumed that
£'•' : £'* = %, (12)
where 0 is the thermal expansion coefficient in elastic range.
Because of the presence of microvoids, the elastic shear and bulk moduli
G and A', respectively, are assumed to be degraded according to the model
proposed by Mac Kenzie [8]

where G and K are the elastic moduli of unvoided material.

Intrinsic micro-damage process
The intrinsic micro-damage process consists of nucleation, growth and co-
alescence of microvoids (microcracks). Recent experimental observation re-
sults (cf. Shockey et al. [21]) have shown that coalescence mechanism can be
treated as nucleation and growth process on a smaller scale. This conjecture
simplifies very much the description of the intrinsic micro-damage process
by taking account only of the nucleation and growth mechanisms. Then
the porosity or the void volume fraction parameter £ can be determined by
\ / nucl \ /grow
Physical considerations (cf. Curran et al. [3]) and Perzyna [16]) have
shown that the nucleation of microvoids in dynamic loading processes which
are characterized by very short time duration is governed by the thermally-
activated mechanism. Based on this heuristic suggestion we postulate for
rate dependent plastic flow

(14)

where k denotes the Boltzmann constant, /i*(£,i9) represents a void nu-

cleation material function which is introduced to take account of the ef-
fect of mi cr ovoid interaction, m*(t9) is a temperature dependent coefficient,
a = (1/3) Ji is the mean stress and cryv^,^, &) is the porosity, tempera-
ture and equivalent plastic strain dependent threshold stress for microvoid
nucleation.
For the growth mechanism we postulate (cf. Johnson [6], Perzyna [16],
Perzyna and Drabik [19,20] and Nemes et al. [13])

where T^>//c denotes the dynamic viscosity of a material, g*(£,$) repre-

sents a void growth material function and takes account for void interaction
and &eq(£, $,£**) is the porosity, temperature and equivalent plastic strain
dependent void growth threshold mean stress.
 Transactions on Engineering Sciences vol 6, © 1994 WIT Press, www.witpress.com, ISSN 1743-3533
584 Localized Damage

ADIABATIC INELASTIC FLOW PROCESS

Formulation of an adiabatic inelastic flow process

Let us define an adiabatic inelastic flow process as follows (cf. Perzyna
[17,18]). Find </>, t>, pM, T, £ and d as function of t and x such that

(i) the field equations

) = V,

divr -

[a-.

: sym^t, + 2sym r :
V 9X
1 /f
:P —
/?M(! -£K
X"£'* 1

(16)

,d, €')]},

(ii) the boundary conditions

(a) displacement <f> is prescribed on a part d$ of d(f>(B) and tractions

(r • n)* are prescribed on part dr of d<f>(B), where 9^ fl dr = 0
and dtUdr = d<f>(B);
(b) heat flux q • n = 0 is prescribed on d<f>(B)]
 Transactions on Engineering Sciences vol 6, © 1994 WIT Press, www.witpress.com, ISSN 1743-3533
Localized Damage 585

(iii) the initial conditions

</>, -U, PM, $, (, and r are given at each particle X £ B at £ = 0;
are satisfied.
In thefieldequations (16) a mapping x = </>(X,/) represents a motion
of a body, pM and p^f denote the actual and reference mass density of the
matrix material, respectively, £o is the initial porosity of a material and Dv
denotes the spatial velocity gradient.
The Cauchy problem
Let us consider the Cauchy problem

<f> = A(t,<p)<p + f(t,<p), fe[<M,], y(0) = </, (17)

where A is a spatial differential operator and f is a nonlinear function, both
defined by the governing equations (16), (cf. Perzyna [17,18]).
In order to examine the existence, uniqueness and well-posedness of the
Cauchy problem (17) let us assume that the spatial differential operator A
has domain V(A) and range H(A), both contained in a real Banach space
E and the nonlinear function f is as follows f : E — » E. To investigate
the existence as well as the stability of solutions to (17) it is necessary to
characterize their properties without actually constructing the solutions.
This can be done by considering the properties of a nonlinear semi-group
because if the operator A + f(-) generates a nonlinear semi-group [1F^ t >
0}, then a solution to (17) starting at t — 0 from any element <p® £ %)(.A)
is given by
¥>(*,x) = F,v°(x) for <€[(),*,]. (18)
We say the problem (17) is well posed if Ft is continuous (in the topology
on V(A) and H(A) assumed) for each t € [0,*/].
Let us postulate as follows:
(i) the strong ellipticity condition in the form:

is strongly elliptic (at a particular deformation <^>) if there is an 6 > 0

such that

for all vectors £ and f G #?;

(ii) for positive numbers Aj- and A£ and for Tm > 0

(21)
and
t -> f (t, <p) G E is continuous. (22)
 Transactions on Engineering Sciences vol 6, © 1994 WIT Press, www.witpress.com, ISSN 1743-3533
586 Localized Damage

Using the results presented by Hughes et al. [5] and Marsden and Hughes
[11] it is possible to show (cf. Perzyna [17,18]) that the conditions (i) and (ii)
guarantee the existence of (locally defined) evolution operators JF^ : E — > E
that are continuous in all variables. In other words the solution of the
Cauchy problem (17) in the form (18) exists, is unique and well-posed.
Fundamental features of rate dependent plastic model
It has been proved that the localization of plastic deformation phenomenon
in an elastic-viscoplastic solid body can arise only as the result of the re-
flection and interaction of waves. It has different character then that which
occurs in a rate independent elasto-plastic solid body (cf. Perzyna [17,18]).
Rate dependency (viscosity) allows the spatial difference operator in the
governing equations to retain its ellipticity and the initial value problem is
well-posed. Viscosity introduces implicitly a length-scale parameter into
the dynamical initial-boundary value problem and hence it implies that
the localization region is diffused when compared with an inviscid plastic
material. In the dynamical initial-boundary value problem the stress and
deformation due to wave reflections and interactions are not uniformly dis-
tributed, and this kind of heterogeneity can lead to strain localization in the
absence of geometrical or material irregularities. This kind of phenomenon
has been recently noticed by Nemes and Eft is [12] (cf. also the results by
Sluys et al. [22]).
The theory of viscoplasticity gives the possibility to obtain mesh-insen-
sitive results in localization problems with respect to the width of the shear
band and the wave reflection and interaction patterns (cf. Sluys et al. [22]).
Since the rate independent plastic response is obtained as the limit case
when the relaxation time Tm tends to zero (cf. Perzyna [17,18]) hence the
theory of viscoplasticity offers the regularization procedure for the solution
of the dynamical initial-boundary value problems with localization of plastic
deformation.

SHEAR BAND LOCALIZATION FRACTURE

Formulation of the initial-boundary value problem for a thin steel tube

Cho, Chi and Duffy [2] tested the specimens machined in the shape of thin-
walled tubes with integral hexagonal flanges for gripping. Torsional loading
at high strain rates was applied in a torsional Kolsky bar (split-Hopkinson
bar).
We idealize the initial-boundary value problem (cf. Batra and Zhang
[1]) by assuming the specimen in the shape of thin-walled tube.
The initial conditions are taken in the form

<^(x,o) = o, %,(x,o) = o, x*,o)

r(x, 0) = 0, £ (x, 0) = &, tf(x, 0) =tfo= constant in B. (23)
 Transactions on Engineering Sciences vol 6, © 1994 WIT Press, www.witpress.com, ISSN 1743-3533

Localized Damage 587

That is, the body is initial at rest, is stress free at a uniform temperature
i?o and the initial porosity at every material point is fg.
For the boundary conditions, we assume
T -n = 0 on the inner and outer surfaces of the tube,
q.n = 0 => gradtf • n = 0 on all bounding surfaces,
%,(%!, %2,0,Z) = 0, r(zi,Z2,L,f) = w'M^ + ^yn*, (24)
where n is a unit outward normal to the respective surfaces, w*(Z) is the
angular speed of the end surface £3 = L of the tube, and n* is a unit vector
tangent to the surface £3 = L. It is assumed that
0 < f < 20//S,

The rise time of 20 //s is typical for torsional tests done in a split Hopkinson
bar (cf. Batra and Zhang [1]).
Computation and discussion of the results
The aforestated initial-boundary value problem has been solved by using
the wide spectrum of ABAQUS possibilities (cf. Lodygowski et al. [7]).
The finite element mesh consisted by 8-noded brick elements with 400
uniform elements along the gage length of the tube, 5 uniform across the
thickness, and 100 uniform elements along the circumference.
It has been assumed following values to various material parameters
(HY-100 steel)
PM = 7860 kg/nf, G = 80 GPa, i^ = 20 %, & = 0.001 ,
Cp = 473 J/kg(>C, A" = 210 GPa, ^ = 5 • 10^ s, (^ = 0.25,
6 = 6*(^ - &), %o = 580 MPa %, = 1.2 • %o, %* = 0.85,
6* -0.01, A = 5.15, m = 7, n = 1.25,
The tube has been twisted at nominal strain rates ranging 10^ — 10^ s~*.
Particular forms of the material functions /&*, ^*, m*, a^ and cr^ which affect
the micro-damage mechanism and have the influence on thefinalfracture
of the tube have been postulated and discussed during the computation
process.
A thin shear band region offinitewidth along the circumference of the
tube which undergoes significant deformations and temperature rise has
been determined. Its evolution until occurrence of fracture has been simu-
lated.
It has been found that the width of the shear band region and the tem-
perature rise vary with the nominal strain rate as well as with the relaxation
time assumed.
The numerical results obtained are in good agreement with experimental
observation data of Cho, Chi and Duffy [2]. An exhaustive discussion of the
results obtained and the comparison with the experimental observation data
will be published elsewhere.
 Transactions on Engineering Sciences vol 6, © 1994 WIT Press, www.witpress.com, ISSN 1743-3533

588 Localized Damage

ACKNOWLEDGMENT
The paper has been prepared within the research programme sponsored by
the Committee of Scientific Research under Grant 3 1112 91 01.

REFERENCES

1. Batra R.C. and Zhang X. Int. J. Plasticity, (in print), 1993.

2. Cho K., Chi Y.C. and Duffy J. Brown University Report, September 1988.
3. Curran D.R., Seaman L. and Schockey D.A. Physics Reports, Vol.147, pp.
253-388, 1987.
4. Hartley K.A., Duffy J. and Hawley R.H J. Mech. Phys. Solids, Vol.35, pp.
283-301, 1987.
5. Hughes T.J.R., Kato T. and Marsden J.E. Arch. Rat. Mech. Anal, Vol.63,
pp. 273-294,1977.
6. Johnson J.N. J. Appl Phys., Vol.53, pp. 2812-2825, 1981.
7. Lodygowski T., Lengnick M., Perzyna P. and Stein E. Archive of Mechan-
ics, (in print), 1994.
8. Mac Kenzie J.H. Proc. Phys. Soc., Vol.63B, pp. 2-11, 1950.
9. March and A., Cho K. and Duffy J. Brown University Report, September
1988.
10. Marchand A. and Duffy J. J. Mech. Phys. Solids, Vol.36, pp. 251-283,
1988.
11. Marsden J.E. and Hughes T.J.R. Mathematical Foundations of Elasticity,
Prentice-Hall, Englewood Cliffs, New York, 1983.
12. Nemes J.A. and Eftis J. Int. J. Plasticity, Vol.9, pp. 243-270, 1993.
13. Nemes J.A., Eftis J. and Randies P.W. J. Appl. Mech., Vol.57, pp. 282-291,
1990.
14. Perzyna P. Quart. Appl. Math., Vol.20, pp. 321-332, 1963.
15. Perzyna P. Advances in Applied Mechanics, Vol.11, pp. 313-354, 1971.
16. Perzyna P. Int. J. Solids Structures, Vol.22, pp. 797-818, 1986.
17. Perzyna P. Acta Mechanica, (in print), 1993.
18. Perzyna P. IUTAM Symposium on Nonlinear Waves in Solids, Victoria
(Canada) 15-20 August 1993; Appl. Mech. Rev., (submitted for publica-
tion), 1993.
19. Perzyna P. and Drabik A. Arch. Mech., Vol.41, pp. 985-908, 1989.
20. Perzyna P. and Drabik A. Int. J. Plasticity, (in print), 1993.
21. Schockey D.A., Seaman L. and Curran D.R. Int. J. Fracture, Vol.27, pp.
145-157, 1985.
22. Sluys L.J., Block J. and de BORST R. Proc. Third International Con-
ference on Computational Plasticity, Fundamentals and Applications, Bar-
celona, 6-10 April 1992. (Ed. Owen D.R.J., Onate E. and Hinton E.),
Pineridge Press, Swansea, pp. 539-550, 1992.
23. Zbib H.M. and Jurban J.S. Int. J. Plasticity, Vol.8, pp. 619-641, 1992.