Applied Mathematical Modelling 33 (2009) 4031–4048

Contents lists available at ScienceDirect

Applied Mathematical Modelling

journal homepage: www.elsevier.com/locate/apm

Development of a finite element model of metal powder compaction

process at elevated temperature
M.M. Rahman a,*, A.K. Ariffin b, S.S.M. Nor a
a
Department of Mechanical Engineering, Universiti Tenaga Nasional, Km 7, Jalan Kajang Selangor, 43009 Kajang, Selangor, Malaysia
b
Dept. of Mechanical and Materials Engineering, Universiti Kebangsaan Malaysia, 43600 Bangi, Selangor, Malaysia

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

Article history: This paper presents the finite element modelling of metal powder compaction process at
Received 3 March 2008 elevated temperature. In the modelling, the behaviour of powder is assumed to be rate
Received in revised form 28 January 2009 independent thermo-elastoplastic material where the material constitutive laws are
Accepted 16 February 2009
derived based on a continuum mechanics approach. The deformation process of metal
Available online 23 February 2009
powder has been described by a large displacement based finite element formulation.
The Elliptical Cap yield model has been used to represent the deformation behaviour of
the powder mass during the compaction process. This yield model was tested and found
Keywords:
Finite element modelling
to be appropriate to represent the compaction process. The staggered-incremental-itera-
Continuum mechanics tive solution strategy has been established to solve the non-linearity in the systems of
Staggered-incremental-iterative solution equations. Some numerical simulation results were validated through experimentation,
Yield criteria where a good agreement was found between the numerical simulation results and the
Experimentation experimental data.
Ó 2009 Elsevier Inc. All rights reserved.

1. Introduction

Powder metallurgy is the production of solid component from metal powder through compaction and sintering where the
manufacturing of solid components through this method has been in existence since early nineteen hundreds as a new gen-
eration of manufacturing process. In the last three decades, a wide range of structural components especially for automotive
industries have been developed for production using this method [1]. The development of manufacturing industries in
Malaysia provides greater market opportunities for this type of process. Statistics show the production of mechanical com-
ponents, using porous materials forming technology, in 1999 is 4.1 kt in Malaysia compared to 83.4 kt in Japan [2]. In order
to expand the market and give lowest total cost, efforts to improve this technology have focused on ways to enhance the
mechanical properties and tolerances of the finished parts. A major advance in this technology has been the warm forming
process, which can utilize traditional powder forming equipment. This method is applicable to most porous material/powder
systems but requires that both the powder mass and the die assembly are heated up to a temperature in the range of 100–
150 °C. The temperature range 100–150 °C roughly delimits the working temperature because at the temperature above
150 °C, lubricants begin to break down, and at the higher temperature iron powder oxidizes more rapidly. However, at tem-
perature below about 100 °C, sufficient loading effect cannot be achieved [3,4]. The process can produce components having
good surface finish and dimensional tolerance such that minimal further processing is required, therefore relatively less time
is taken to produce a mechanical component.
Warm powder compaction consists of three phases which are powder mixing with lubricant, powder heating and powder
forming in a die of desired shaped. Product produced or green compact generated through warm powder compaction route

* Corresponding author. Tel.: +60 389287269; fax: +60 389212116.

E-mail address: mujibur@uniten.edu.my (M.M. Rahman).

0307-904X/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.apm.2009.02.005
4032 M.M. Rahman et al. / Applied Mathematical Modelling 33 (2009) 4031–4048

faces several physical problems mainly arise from an inhomogeneous density distribution within the compact and internal
crack may be initiated on an internal shear plane. The density distribution depends on the combination of many factors such
as geometrical shape, mechanical properties of the materials to be compacted, powder-tool frictional behaviour, and press-
ing cycle. Deformation of metallic powder is significantly affected by the forming temperature [5]. The microscopic mech-
anisms that explain this behaviour are complex. Higher temperature gives lower deformation resistance, which is quantified
by yield strength. Therefore, analysis of the warm forming process is necessary to get a deeper understanding about the ef-
fects of process variables on the product. The optimization of process variables may be achieved by a time consuming trial-
and-error procedure or tooling development programme, which is expensive. Furthermore, it is relatively difficult to predict
the powder forming behaviour experimentally [6,7].
It becomes clear from the description above that the analysis of warm forming process is important because the compact
green density influences the performance of the final part. The variations of green density also causes shape distortion after
sintering along with regions of elevated stress and stress concentrations [8,9]. In this paper, a finite element model is devel-
oped to predict the stress, density and flow of powder mass during the warm forming process and some of the numerical
simulation results were validated through experimentation.

2. Literature review

Higher strength and better fatigue property are increasingly required to mechanical parts in line with the necessity for
making the components lighter in weight as well as for dealing with higher-performance engines. In powder metallurgy,
increasing sintered density is the most effective way to improve mechanical properties such as strength and fatigue prop-
erties. Various approaches have been proposed to increase density. In the warm compaction method, metallic powder is pre-
heated and compacted in a die heated to about 130 °C and then sintered [10–12]. However, the compaction and sintering is
repeated twice in the double-pressing and double-sintering method. Another approach is powder forging where a sintered
compact prepared by the single-pressing and single-sintering method is subjected to hot forging. In these methods, however,
production cost increases with attainable density. Hence, a new economic method for realizing high density is required.
In powder compaction modelling, the powder can be represented by a micromechanical scale [13–15] or by the basis of a
continuum [16–18]. The former approach focuses on powder particles, which practically have a variety of shape, size, hard-
ness and arrangement within the compact. This kind of approach may be useful in investigating the effect of particle char-
acteristics during the compaction and the compact [19]. The later approach considered the powder as a continuum rather
than an assemblage of individual particles. This approach can provide information on the global behaviour of the powder
mass such as the powder movement, density distribution, stress-state and the shape of the compact during and after the
compaction [20].
Powder compaction formulation can be done either based on load control or displacement control by considering the con-
tinuum framework [21]. The former approach is suitable for non-linear analysis such as reinforced concrete structures [22].
Some of the problems associated with load control approach include the inability to capture responses on the falling branch
of the load–displacement relationship. However, the displacement control approach is proven to be an efficient technique for
non-linear plasticity analysis such as to model the compaction behaviour of powder materials since at low density levels, a
large deformation can be resulted by a small amount of load [23].
In the case of powder forming, the simulation must capture the large displacements, which occur during the compaction
phase [24,25]. Several researchers have proposed and utilized finite element schemes for problems of large displacements
[26–31] which include diverse formulations of both the Eulerian and Lagrangian type. In Eulerian procedure, the physical
quantities are expressed as a function of time and position vector in the geometric space. Using this approach, McMeeking
and Rice [32] presented a finite element formulation for the problem of large elastic-plastic flow. The method is based on
Hill’s variational principle for incremental deformations and is ideally suited to isotropically Prandtl–Reuss materials and
high-speed deformation.
In Lagrangian strategy, the element geometry changes are determined by the deformation of the material. Brekelmans
et al. [33] modelled the complex shaped die compaction process using the total Lagrangian strategy. In this procedure,
the original element mesh was sufficiently accurate for any changes in physical quantity and material properties are cap-
tured accurately. However, instability may occur in the case of highly non-linear calculation where more than one material
properties are considered in one element [34]. In order to overcome the problems associated with the total Lagrangian strat-
egy, an updated Lagrangian strategy may be applied where the coordinates of the finite element mesh have to be referred to
the previous calculation step. Park et al. [35] developed a finite element model for the cold compaction of metal powders in
which the updated Lagrangian strategy has been utilized. However, difficulties still arise in analyses, which involve large dis-
placements because the element shapes can still be highly distorted leading to an unreliable solution due to mesh degener-
ation. In order to avoid the mesh distortion in Lagrangian strategy, several researchers adopted the adaptive finite element
modelling approach [36]. Adaptive finite element method, which has been studied for nearly 25 years, tries to automatically
refine, coarsen, or relocate a mesh and, or adjust the basis to achieve a solution having a specified accuracy in an optimal
fashion [37–39]. Another established formulation to the numerical simulation of metal forming process is the arbitrary
Lagrangian Eulerian (ALE) approach, which has been used by several researchers in modelling the sheet metal forming pro-
cess [40–42].
 M.M. Rahman et al. / Applied Mathematical Modelling 33 (2009) 4031–4048 4033

3. Governing equations

The warm compaction process has been modelled using a thermo-elastoplastic constitutive law where the axi-symmetric
simulations are used to represent the process. Elliptical Cap yield model is used to represent the deformation behaviour of
metal powder during the compaction phase.

3.1. Yield criteria

A material will be deformed when it is subjected to any external forces. In the case of particulate materials, if the mag-
nitude of the forces is low to cause only particle rearrangement, the material comes to its original dimensions once the load
is released. This recoverable deformation is called elastic deformation, and the associated stresses are elastic stresses. The
material is said to have undergone non-recoverable or inelastic deformation if it is not returned to its original dimensions
after the release of the load. The stress-state at which the inelastic deformation commences is the flow stress or yield stress.
Therefore, a criterion has to be proposed to predict the onset of inelastic yielding.
Slater [43] defined the yield criterion as a hypothesis concerning the limit of elastic deformation due to any possible
stress-state. Inside the yield surface, the behaviour of powder materials may be isotropic non-linear elastic since the elastic
parameters are dependent on density. For ideally plastic condition, the stress-state will not proceed beyond the yield surface.
During the compaction process, the density of the powder mass increases, which causes the powder mass becomes hard-
er. Therefore, the hardening effect must be included during the deformation of powder hence the yield surface must move
outward as the compaction proceeds [44,45]. The size and position of the surface depend on the initial yield surface and
hardening hypothesis, which specify the manner in which the surface changes during the inelastic deformation. Due to
the consideration of the powder mass as an isotropic material, the yield function should be independent of the selection
of coordinate system. Therefore, the yield criterion is a function of the invariants of the stress tensor [46] as
F ¼ f ðI1 ; I2 ; I3 ; h; TÞ: ð1Þ
In general, the deformation due to stresses consists of two components, i.e., a volumetric component, rm and a distortion
or deviatoric component, r
 which produces a change in the geometry of the body [47]. In the case of powder, the hydrostatic
stress makes an important contribution due to the increase of the density of the workpiece [48]. Therefore, the yield function
can be written as
F ¼ f ðJ 1 ; J 2D ; J 3D ; h; TÞ ð2Þ
or
F ¼ f ðrm ; r
 ; h; h; TÞ: ð3Þ
The behaviour of most particulate materials is different from that of solid metals and their strength is dependent on the
hydrostatic stress. The strength of materials often increases with mean pressure and exhibits frictional characteristics [49].
In this work, only Elliptical Cap yield model is considered.

3.1.1. Elliptical Cap yield model

The Elliptical Cap yield model represents the consolidation and ultimate collapse of granular materials under applied load
[50]. An element of soil undergoing shear deformation could pass through a yield point without collapse and continue to
deform until eventually a critical state is reached. The yielding continues to occur until the material reaches a critical void
ratio. In this state, the material will continuously deform without further change of void ratio or stress. This particular void
ratio is called the critical void ratio and it can be considered as the critical state of the material and the yield surface shows an
elliptical shape. The critical state line and Elliptical Cap yield surface in stress invariant is shown in Fig. 1. The critical state
line can be expressed as
pffiffiffiffiffiffiffiffiffi
3J 2D ¼ MJ1 ; ð4Þ

pffiffiffiffiffiffiffiffiffi M is the slope of the critical state line, the ellipse passes through the origin and is centered on the point J 1 ¼ rc at
where
3J 2D ¼ 0, rc being equal to half of the reconsolidation load. The equation representing the Elliptical Cap yield surface
can be written as
pffiffiffiffiffiffiffiffiffi
ðJ 1  rc Þ2 ð 3J 2D Þ2
þ ¼1 ð5Þ
r2c ðM rc Þ2
or rearranging Eq. (5) gives the Elliptical Cap yield surface as

F ¼ 3J 2D þ M 2 J 1 ðJ 1 þ 2rc Þ ¼ 0: ð6Þ

This yield surface is strain dependent, it expands when the material hardens and it contracts when the material softens.
Strain hardening is associated with a volume decrease whereas strain softening is associated with a volume increase. In the
case of powder compaction, the strain hardening phenomena is taken into account. Therefore, the initial hydrostatic stress,
4034 M.M. Rahman et al. / Applied Mathematical Modelling 33 (2009) 4031–4048

3J 2 D M
critical state line

εp

elliptical cap

b A

a σc - J1

Fig. 1. Elliptical Cap yield surface.

rco and an exponential function of volumetric plastic strain, epv governs the size of the ellipse [51]. The major axis of the el-
lipse can be represented by
rc ¼ f ðepv Þ ¼ rco expðvepv Þ; ð7Þ
where v is the plastic hardening coefficient, which can be expressed as
kj
v¼ ; ð8Þ
1 þ e
where  e is the initial void ratio, j is an elastic index and k is the compaction index. The material parameters for this model
are rco ; k; andj which have been derived from warm compaction experiments [52].

3.2. Plastic work

The concept of plastic work plays an important role in the plastic stress–strain laws. The total work done per unit volume
of a deformable body during a strain increment can be written according to Mendelson [53] as
 
dW ¼ rij deeij þ dep;th
ij ð9Þ

or
e ine
dW ¼ dW þ dW ; ð10Þ
e p;th
where de is the elastic strain increment and de
ij is the combination of plastic strain and thermal strain increments. The
ij
e p;th
quantity dW is the elastic energy which is recoverable, whereas dW is the inelastic energy which can not be recovered.

3.3. Hardening hypothesis

When a particulate material is formed, it usually strain hardens, i.e., its resistance to further deformation increases. There-
fore, the stress required to maintain the deformation increases. A specification of the dependence of the yield criterion on the
internal variables is called a hardening hypothesis. Hardening is caused due to the elimination of voids, which causes the
decrease in the bulk volume. As shown previously, the yield criterion for a strain hardening material is the function of
stress-state, hardening parameter and temperature. The hardening parameter itself is a function of plastic strain and tem-
perature as
h ¼ f ðep ; TÞ: ð11Þ
According the hypothesis proposed by Hill [54], the progressive development of the yield surface at temperature T can be
defined by relating the yield stress to the plastic deformation by means of the hardening parameter, h. This can be done in
two ways, firstly the degree of work hardening can be postulated to be a function of the total inelastic work only, as

h ¼ W p;th ; ð12Þ
p;th
where W is the total inelastic work during a finite deformation which can be expressed as
Z
W p;th ¼ rij dep;th
ij : ð13Þ
 M.M. Rahman et al. / Applied Mathematical Modelling 33 (2009) 4031–4048 4035

Alternatively, h can be related to a measure of the total inelastic deformation. Due to the important contribution of hydro-
static stress during the compaction phase which causes the change of volume of the powder mass, therefore the hardening
parameter is considered as a function of volumetric plastic strain, epv and temperature, T.

3.4. Flow behaviour

The definition of plastic flow behaviour of materials is important in developing inelastic stress–strain relationship. When
the state of stress reaches the yield surface, the material undergoes inelastic deformation. The direction of plastic strain vec-
tors is defined through a flow rule by assuming the existence of plastic potential function to which the incremental strain
vectors are orthogonal. The incremental form of plastic strain, which is referred to as normality rule can be expressed as
@Q
depij ¼ dk ; ð14Þ
@ rij
where Q is the plastic potential function and k is a positive scalar factor of proportionality. For some materials, the plastic
potential, Q and the yield function, F can be assumed to be the same, i.e., Q  F. Such materials are considered to follow
the associated flow rule of plasticity. However, for geologic materials like soils, the plastic potential function, Q and the yield
function, F are often different. These materials are considered to follow the non-associated flow rule. In this work, an asso-
ciated flow rule is chosen, i.e., the plastic potential gradient, @Q =@ r is taken as the yield function gradient, @F=@ r.

3.5. Stress–strain relationship

In this section, the combination of elastic, plastic and thermal effects will be used to obtain the thermo-elastoplastic
stress–strain relationship. Combining the elastic strain, plastic strain, and thermal strain increments as

@Q 1
de ¼ D1
e dr þ dk þ madT: ð15Þ
@r 3
Differentiating the yield criterion gives the consistency equation at yielding as
 T  T  
@F @F @h p @F @h @F
dr þ de þ þ dT ¼ 0: ð16Þ
@r @h @ ep @h @T @T

Taking into account the flow rule in Eq. (14), Eq. (16) becomes
 T  T  
@F @F @h @Q @F @h @F
dr þ dk þ þ dT ¼ 0 ð17Þ
@r @h @ ep @r @h @T @T

or
@F
T  @h @F 
@r
dr þ @F
@h @T
þ @T dT
dk ¼ 
: ð18Þ
@F @h T @Q
@h @ ep @r

Substituting for dk into Eq. (14) gives the plastic strain increment as
@F
T  @h @F 
@r
dr þ @F þ @T dT @Q
dep ¼ 

@h @T
: ð19Þ
@F @h T @Q
p
@r
@h @e @r
@F
e
T
Multiplying Eq. (15) by @r
D and substituting for @@Fr dr by means of Eq. (16) leads to the following:
   T  T  T    
@F @F @F @Q @F @Q @F 1 @F @h @F
De de ¼  dk þ De dk þ De madT  þ dT: ð20Þ
@r @h @ ep @r @r @r @r 3 @h @T @T

The term dk can be obtained from Eq. (20) as

@F
e
T 1   
@r
D de  @@Fr De @Q
@r 3
madT þ @F @h
@h
@T
@F
þ @T dT
dk ¼ @F
T e @Q @F @F
T @Q : ð21Þ
@r
D @ r  @h @ep @ r

Multiplying Eq. (15) by De and substituting dk into the resulting equation gives
@F @h @F  !
1 De @Q
@ r @h @T
þ @T
dr ¼ Dep de  madT  D1
ep
T
T dT ; ð22Þ
3  @F @hp @Q þ @F De @Q @h @e @r @r @r
4036 M.M. Rahman et al. / Applied Mathematical Modelling 33 (2009) 4031–4048

where
@F
T e
e De @Q
@r @r
D
Dep ¼ D 
T @F
T e @Q : ð23Þ
@F @Q
 @ ep @ r þ @ r D @ r
In Eq. (23), Dep is the elastoplastic tangent modulus where the detail derivation of this matrix can be found in Ref. [55]. Rear-
ranging Eq. (22), the non-isothermal stress–strain relationship can be written as
"  ! #
@Q @F @h @F
1 þ @T
dr ¼ Dep de  ma  @r @h @T

dT ð24Þ
3 @F @h T @Q
@h p@e @r

or
dr ¼ Dep ðde  XdTÞ ð25Þ
where

@Q @F @h @F

1 þ @T
X¼ ma  @ r @h

@T

ð26Þ
3 @F @h T @Q
@hp @e @r

It is clear from Eq. (25) that not all amounts of stresses are caused by the mechanical deformation, but the temperature gra-
dient also causes some amount of stress which is defined as thermal stress.

3.6. Heat transfer analysis

As stated previously, in warm compaction process the powder mass and the tooling must be heated up to an elevated
temperature before compacting the powder. The workpiece may also become warm due to the plastic deformation and fric-
tion. The analysis of heat transfer associated with the compaction process is important because it can affect the process as
well as the quality of the green compact. In this work, the analysis of heat transfer has been coupled with the deformation
analysis.

3.6.1. Model of heat conduction

Generally, warm compaction process is conducted under closed environment and there is no fluid flow or radiation heat
transfer mode during the process. Therefore, only conduction heat transfer is involved during the warm compaction process.
The coupling between thermal and mechanical phenomena for two-dimensional axi-symmetric system with heat generation
can be written as
 
1 @qr @qz @Tðr; tÞ
 þ þ q_ ¼ qc ; ð27Þ
r @r @z @t
where qr and qz are the components of heat flow rate. For isotropic material, the heat flow rate can be expressed using Fou-
rier’s law as
@T @T
qr ¼ rk and qz ¼ k : ð28Þ
@r @z
Substituting qr and qz into Eq. (27), the transient heat conduction equation can be written as
   
1 @ @T @ @T @Tðr; tÞ
rk þ k þ q_ ¼ qc : ð29Þ
r @r @r @z @z @t

In Eq. (28), the thermal conductivity, k and the specific heat, c are two important material parameters, which are the function
of density.Heat transfer through a loosely packed powder is less rapid than through a consolidated body [56]. Therefore,
there is a need for expressing the relationship between the thermal conductivity and relative density. Argento et al. [57] pro-
posed an equation, which relates the thermal conductivity and relative density at any state of compaction. The relationship
can be expressed mathematically as
 
k q  q0 cð1q0 Þ
¼ ; ð30Þ
ks 1  q0
where ks is the thermal conductivity of the fully dense solid, k and q are the instantaneous thermal conductivity and relative
density, respectively, q0 is the initial density of the powder and c is a constant.

3.6.2. Sources of heat

During the warm compaction process, the powder mass and the tooling should be heated up before commencing the
compaction process. A considerable heat generation is also occurred due to the deformation of powder mass and friction be-
 M.M. Rahman et al. / Applied Mathematical Modelling 33 (2009) 4031–4048 4037

tween the workpiece and the die wall. In this work, heating up the powder mass and the tools are considered as external heat
input whereas heat generation due to plastic deformation and friction is considered as internal heat generation. Tabata et al.
[58] proposed the following relationship to estimate the amount of heat generated due to plastic deformation. The relation-
ship can be expressed as
q_ p ¼ nrep ; ð31Þ
where n is the remainder of the plastic deformation energy for the change in internal energy due to dislocation, density
change and change of grain boundaries [59]. The second contribution of internal heat generation is the amount of heat gen-
erated by friction along the interface between the powder mass and the tools. The heat generated by friction can be calcu-
lated using the following equation [60]:
q_ fr ¼ ff v r ; ð32Þ
where f, f and vr are the friction dissipation coefficient, friction force and relative velocity between die and workpiece,
respectively.

3.7. Hardening coefficient

The evolution of yield surface depends on the hardening coefficient where the yield surface expands or shrinks from its
original position [61]. Fig. 2 illustrates the compaction and relaxation path when portrayed in a void ratio and mean stresses
framework. The void ratio change by an elastic component can be written as
 
rA
Dee ¼ j ln ð33Þ
rAo
and the plastic component as
 
rA
Dep ¼ ðk  jÞ ln : ð34Þ
rAo
Since the critical state line is parallel with the compaction line, the ratio of the stress value of point A to that of point Ao is
equivalent to the ratio of stress value at point C and Co. The volumetric strain is related to the void ratio increase from its
initial value, which can be written as
De
ev ¼  ; ð35Þ
1 þ e
where e is the average void ratio during the change. For the plastic component, the volumetric strain can be written as
De p
epv ¼ h ¼  : ð36Þ
1 þ e
Substituting Eq. (35) into Eq. (33) gives the hardening parameter as
 
ðk  jÞ ln rrAA
o
h¼ ð37Þ
1 þ e

Description:
(i) Compaction Line
(ii) Critical State Line
(iii) Relaxation Line
(iv) Initial Relaxation Line
e

λ
Ao
Co (iv)
(i)
(ii)

κ
C
(iii)

σco σc ln J1

Fig. 2. Compaction and relaxation curves in terms of void ratio-mean pressure.

4038 M.M. Rahman et al. / Applied Mathematical Modelling 33 (2009) 4031–4048

or finally
rC
h ¼ v ln ; ð38Þ
rC O
where v ¼ kj is a dimensionless measure of plastic compressibility which has been established experimentally [52].
1þe

4. Finite element implementation

In practical design and analysis, the most important steps are the proper idealization of the actual problem and the cor-
rect interpretation of the result. Considerable knowledge regarding the characteristics of the system and its mechanical
behaviour are required in establishing an appropriate idealization depending on the complexity of the actual system to
be analysed. In order to construct the numerical solution algorithm of a continuum problem, it is basically necessary to
establish the algebraic equations that govern the response of the system. A most important formulation approach, which
is widely used for the establishment of solution algorithm for practical problems, is the displacement based finite element
method [62].

4.1. Deformation of powder mass

During the powder compaction process, the material domain transforms from loose powder state to solid state due to
densification. Therefore, the formulation must capture both these states of deformation. The starting point for this formula-
tion is the principle of virtual work [63]. This principle arises from consideration of a body subjected to two entirely unre-
lated loading states. Under the application of some external loads, internal stresses are generated which result in equilibrium
condition is being satisfied. By introducing the thermo-elastoplastic relationship [64], the residual force vector can be writ-
ten as
Z Z Z Z
Dep BadX  Dep XTdX  qNTe bdX  NTe tdCt ¼ re ð39Þ
X X X C

or
Z Z Z Z 
BTe Dep Be adX  Dep XTdX  qNTe bdX þ NTe tdCt ¼ re : ð40Þ
X X X C

Eq. (39) can be written in linear form as

Z Z 
@r @r
BTe Be dX  BTe XTdX fDae g ¼ fre g: ð41Þ
Xe @e Xe @e
Introducing the element stiffness matrix as
Z Z
@r @r
½ke  ¼ BTe Be dX  BTe XTdX: ð42Þ
Xe @e Xe @e
Eq. (41) can be written as
½ke fDae g ¼ fre g; ð43Þ
where ½ke  is the element stiffness matrix, fDag is the prescribed incremental displacement at the boundary condition or the
unknown incremental nodal displacement vector and frg is the force vector including the body force. The elastoplastic tan-
gent modulus is set as

@r ¼ De if F < 0
ð44Þ
@e ¼ Dep if F ¼ 0
The global stiffness matrix, ½K and force vector, fRg can be written as
X
n X
n
½K ¼ Ke and fRg ¼ re ð45Þ
e¼1 e¼1

where n is the total number of element. Therefore, the global displacement equation becomes
½KfDag ¼ fRg ð46Þ

5. Solution procedure

The thermo-elastoplastic constitutive laws lead to the Eq. (46) becomes non-linear. Therefore, it cannot be solved directly,
and requires an incremental iterative technique. An approximate solution can be obtained using an updated Lagrangian
 M.M. Rahman et al. / Applied Mathematical Modelling 33 (2009) 4031–4048 4039

strategy where all the variables must be referred to the previously calculated equilibrium configurations. In order to com-
plete the solution procedure for the deformation of powder mass, a staggered-incremental-iterative solution procedure is
applied. The total prescribed displacement is divided into n-numbers of small increments where each increment requires
an iterative procedure to obtain its solution. After the incremental displacements are obtained from Eq. (46), the incremental
strain vector can be calculated. In this solution procedure, the mechanical calculation is conducted using elasticity matrix, D
at first time step. The output data from mechanical calculation is input to the thermal calculation at the same time step. The
output data from thermal calculation is input to the mechanical calculation at the second time step where D is replaced by
Dep . This consecutive procedure is applied until the end of calculation.
The stress increment due to the thermo-elastoplastic deformation of powder mass during the warm compaction process
should follow the yield criteria set up previously. Setting up the computational procedure is required to bring back the
stress-state, which exceeds elastic limit of yield surface. The stress increment at each displacement increment can be
expressed as
Z Dein Z T
Drin ¼ Dep de  Dep XdT; ð47Þ
0 T ref

where

Dein ¼ BDain : ð48Þ

(1) The total prescribed displacement, a is divided equally into n small

increments, Δan.

For each increment:

(2) Calculate the tangential stiffness matrix, [K]n. At the first increment
use [De]n, otherwise use [Dep]n.

i
(3) Compute Δa ni = ∑ δa
k =0
k
n

(4) Compute Δε n = BΔa ni

(5) Calculate the current stress, σ n as

σ nr = σ nr−−11 + Δσ nr .

(6) Calculate the invariants, σ m , σ , J 3 and φ corresponding to σ nr

(7) Find, F (σ ) = F corresponding to the yield surface for a particular

yield model.

(8) Check if F ≤ 0 go to step (10)

If F > 0 continue step (9)

(9) Refine and correct the stress-state and calculate current stress in step (5)
until satisfied.

(10) Check convergence for each iteration using local convergence criteria at
each gaussian point as
Δσ nr − Δσ nr −1
≤ TOLER
Δσ n1
If not satisfied, calculate stress in step (5) then calculate the body force
using
T

∫
f r = B σ r dV
V
(11) Take result data from thermal calculation.

(12) Update stresses, strains, coordinates, density, reaction forces, etc.

(13). Continue for n = n + 1 at the next time step.

Fig. 3. Computation algorithm for mechanical problem.

4040 M.M. Rahman et al. / Applied Mathematical Modelling 33 (2009) 4031–4048

is the total strain change, and

X
i
Dain ¼ dakn : ð49Þ
k¼1

is the displacement increment. Eq. (47) is solved by subdividing the strain increment, Dein into k intervals. Therefore Eq. (48)
is replaced by
X
k l1
Dep De X T
l1
Dr ¼  Dep X DT; ð50Þ
n¼0
k T ref

(1) The total prescribed shear displacement, u is divided equally into n small
increments, Δun.

For each increment:

(2) Calculate the tangential stiffness matrix, [K] using appropriate GF

modulus. At the first step and first iteration i, set GF = GA. For
subsequent iteration, GF is defined in computational step 6.

(3) Compute Δun = K −1Ψni −1 increment of nodal displacement.

⎧ Δg ⎫ ⎧Δu p ⎫
(4) Compute ⎨ t ⎬ = B f ⎨ Δv ⎬ .
⎩Δg n ⎭ ⎩ p⎭

(5) Evaluate τ at n + 1 according to the slip criterion.

(6) Evaluate the frictional shear stress, τ

τE
τ ni +1 = τ ni −1 + GF Δg ni +1 where GF =
Δgt

(7) If F ≤ 0 , go to step 9, otherwise continue.

(8) Refine and correct the stress-state and calculate current stress in step (5)
until satisfied.

(9) Check convergence for each iteration using convergence criteria at each
gaussian point as
Δτ nr − Δτ nr −1
≤ TOLER
Δτ 1n
If not satisfied, calculate frictional stress in step (6).

(10). Continue for n = n + 1 at the next time step.

Fig. 4. Computation algorithm for frictional problem.

(1) Assume reference temperature, T0.

For each time step, Δt n = t n+1 − t n .

(2) Take results data from mechanical calculation.

(3) Evaluate stiffness matrix, K n+1 and load matrix, Fn+1 .

(4) Solve the global equation.

(5) Update the temperature.

(6) Send the calculated results to mechanical calculation.

Fig. 5. Computation algorithm for thermal problem.

 M.M. Rahman et al. / Applied Mathematical Modelling 33 (2009) 4031–4048 4041

where l  1 implies the evolution of Dep to be explicit and is taken from the previous increment. The total thermo-elastoplas-
tic computational algorithm for the calculation of mechanical problem can be seen in Fig. 3 where the calculations of fric-
tional and thermal problems are depicted in Figs. 4 and 5, respectively. The calculation of coupled thermo-mechanical
computational flowchart using staggered-incremental-iterative solution procedure can be seen in Fig. 6.

6. Results and discussions

This section is mainly concerned with the numerical results and experimental validation in order to illustrate the appli-
cability of governing equations, formulation, and solution strategy developed in previous section. The numerical analysis of
warm powder compaction process was modelled to generate a green compact of plain bush component of iron powder (Ta-
ble 1). The axi-symmetric representation of the compaction step has been presented in Fig. 7. The established model is used
to predict the relative density of a T-shape component (Fig. 8).
As stated previously, a displacement based formulation is used. Therefore, the implementation of loading in the finite ele-
ment code, developed by the author [65], is achieved by the use of prescribed nodal displacements. The direction of these
displacements is always in a vertical plane, which represents the top punch load. In order to represent the fixed bottom
punch, fixed nodal values were employed, which means that there is no relative movement between the punch and the pow-
der mass. The interface elements were used to represent the shear movement at the interface of powder and tool during the
generation of a green compact. In this study, a fill depth of 20 mm was used and the mesh consists of 38 elements and 101
nodes represent the powder material domain whereas 20 elements and 84 nodes represent the interface.

Mechanical Calculation Thermal Calculation

Read Data Read Data

& Initialize & Initialize

Loop Over
Time Step

Elastoplastic Evaluate Stiffness

Material Matrix
Evaluate
Stiffness Matrix &
Frictional
Subroutines
Time step loop

Solve Global Solve Global

System of Equations System of Equations

Solve Stress
Limit
Update
Variables
Update
Variables

Send Data to Send Temperature

Thermal Calculation Data to Mechanical
Calculation

Time Step

End

Fig. 6. Summary of staggered-incremental-iterative solution strategy.

4042 M.M. Rahman et al. / Applied Mathematical Modelling 33 (2009) 4031–4048

Table 1
Summary of material parameters in the model.

Name of parameter Value

Initial Young’s modulus, E (MPa) 32.00
Poisson’s ratio, m 0.35
Cohession, c (MPa) 2.50
Initial relative density, qo (relative to solid) 0.327
Elastic index, j 0.0433
Plastic index, k 0.3789
Initial stress, rco (MPa) 0.55
Thermal expansion coefficient, a (mm/°C) 24  103

top punch load core rod

powder

spring
back

sleeve

(i) (ii) (iii) (iv) (v)

bottom punch load

(i) initial mesh
(ii) deformed mesh - end of compaction phase
interface element (iii) relaxation
(iv) ejection
(v) emergence

Fig. 7. Initial and deformed mesh for a complete cycle of green compact generation.

During the simulation of compaction phase, the movements of the top punch were divided into 20 steps to give a 50%
height reduction of the original fill depth. The relaxation phase was modelled over seven incremental displacement steps,
which represent the top punch unloading to zero at the top surface nodes. The ejection where the bush is contained within
the tooling and the final emergence where the bush exists from the die were modelled using 10 incremental steps, respec-
tively. During these last two phases, the prescribed displacement was applied to the bottom punch. In order to represent the
Updated Lagrangian strategy, the nodal coordinates, stresses, strains and densities were updated at each time step.
As stated previously that the effective temperature of warm compaction is about 150 °C, therefore the compaction pro-
cess has been simulated for the temperature of 150 °C. Fig. 9 shows the force balance results presented as stress–strain
curves for the compaction phase where the stress predicted by Elliptical Cap yield model is compared with the experimental
data. A relatively close agreement with experimental data is achieved by the Elliptical Cap yield model. It is evident in Fig. 9
that the stress predicted by the model at relatively lower strain is higher than the stress obtained during experiment at the
same temperature. The similar trend is also found for the higher strain value.
Stress development at the axial punch during the warm compaction has been compared to cold compaction (at room
temperature) using Elliptical Cap yield model (Fig. 10). It is evident from Fig. 10 that the warm compaction process generates
a slightly lower stress than the cold compaction process. This is due to the shrinkage of Elliptical Cap yield surface at the
elevated temperature. The important material parameter for Elliptical Cap yield model, i.e., plastic hardening coefficient,
v has been derived experimentally and shown that it is strongly dependent on temperature [65]. The plastic hardening coef-
ficient is inversely proportional to the compaction temperature which manifested into the shrinkage of elliptical cap as com-
paction temperature increases (Fig. 11).
Fig. 12 shows the predicted results of the relaxation phase for iron powder compacted at room temperature (30 °C) and at
150 °C. The maximum spring-back obtained from experiment for these two conditions are also shown. The amount of spring-
back is higher in warm compaction compared to the cold compaction. The results are also comparable with the findings by
Bocchini [66], who reported that the radial pressure at the end of compaction is higher, whereas the residual radial pressure
at the start of ejection is lower. Consequently, the spring-back is higher in warm compaction than that attained by room tem-
perature compaction.
 M.M. Rahman et al. / Applied Mathematical Modelling 33 (2009) 4031–4048 4043

a b

Fig. 8. (a) Downward compaction (first stage of compaction). (b) Upward compaction (second stage of compaction). (c) Ejection and emergence.

120

100
Top Punch Stress (MPa)

80 Elliptical Cap
Experiment

60

40

20

0
0 0.1 0.2 0.3 0.4 0.5 0.6
Strain

Fig. 9. Top punch stress predicted by the model compared to experiment.

Fig. 13 shows the predicted ejection force for iron powder compacted at different temperatures. It is found that at the
beginning of ejection, a higher force is required to initiate the ejection process for the compaction at elevated temperature.
This is due to the sticking phenomenon of metal powder compact with the die at elevated temperature. The compact may
4044 M.M. Rahman et al. / Applied Mathematical Modelling 33 (2009) 4031–4048

160

140

120 30ºC

Top Punch Stress (MPa)

150ºC
100

80

60

40

20

0
0 0.1 0.2 0.3 0.4 0.5 0.6
Strain

Fig. 10. Predicted compaction stress at two different temperature levels.

0.16
Plastic Hardening Coefficient (χ)

0.14

0.12

0.10

0.08

0.06
0 50 100 150 200
Temperature (ºC)

Fig. 11. Variation of plastic hardening coefficient (v) with [65].

0.5
maximum spring-back (warm
compaction experiment)
0.4
maximum spring-back (cold
Spring-back (mm)

compaction experiment)

0.3
Warm

0.2

0.1
Cold

0
0 10 20 30 40 50
Top Punch Load (kN)

Fig. 12. Spring-back during unloading of axial punch.

 M.M. Rahman et al. / Applied Mathematical Modelling 33 (2009) 4031–4048 4045

8
7
6 150ºC

Ejection Force (kN)

90ºC
5
30ºC
4
3
2
1
0
0 2 4 6 8 10 12 14 16 18 20
Bottom Punch Displacement (mm)

Fig. 13. Predicted results for ejection and emergence.

Fig. 14. Relative density at different forming temperature; compaction by top punch.

Fig. 15. Relative density at different forming temperature; compaction by bottom punch.
4046 M.M. Rahman et al. / Applied Mathematical Modelling 33 (2009) 4031–4048

0.96

0.94

0.92

Relative Density
0.90

0.88

0.86

0.84

0.82

0.80
30 50 70 90 110 130 150 170
Temperature (ºC)

Fig. 16. Relative density of T-shape green compact simulated at different forming temperature.

expand during the relaxation phase which gives a higher radial force. Therefore, in order to overcome this higher radial force,
a relatively higher force is required to eject the powder compact from the die cavity.
The increase of relative density with respect to compaction pressure or axial stress in generating a T-shape green compact
by using the established model has been depicted in Figs. 14 and 15. The increment in displacement for second stage of com-
paction (compaction to upward direction) is relatively higher than the first stage of compaction (compaction to downward
direction). This is due to the high load exerted on smaller cross section area at the bottom of the die. The large displacement
increment in second stage of compaction leads to higher relative density increment. Therefore, compaction process at every
forming temperature also shows higher increment in relative density at second stage of compaction compare to first stage of
compaction. Furthermore, relative density obtained in this experiment supports the advantage of warm compaction process
which can offer higher green density (Fig. 16).

7. Conclusion

A finite element model of metal powder under mechanical coupled with thermal load was developed. The Elliptical Cap
yield model was used to represent the deformation behaviour of powder mass during the warm forming process. Some re-
sults were validated through experimentation of iron powder compaction at elevated temperature. The study found that an
Elliptical Cap yield model is suitable to represent the compaction behaviour of metal powder mass at elevated temperature.
The study also found that compaction of metal powder at elevated temperature can increase the relative density due to the
softening of powder mass at above room temperature.

Acknowledgement

This research is financially supported by the Ministry of Science Technology and Innovation-Malaysia through the Re-
search Projects IRPA 09-99-03-0032 EA001 and 03-02-03-SF0146.

References

[1] D. Whittaker, Powder metallurgy applications in the automotive industry, in: Powder Metallurgy (PM90): Proceedings of the World Conference, 1990,
pp. 109–116.
[2] Y. Fujiwara, The present and future of powder metallurgy in Asia and Oceania, in: Powder Metallurgy (PM2000): Proceedings of the World Congress,
2000, pp. XXXII–XLIV.
[3] M.M. Rahman, F. Tarlochan, A.K. Ariffin, S.S.M. Nor, M.R. Jamli, Near-net-shape manufacturing of mechanical components through warm compaction
route, in: Brunei International Conference on Engineering and Technology, Bandar Seri Begawan – Brunei Darussalam, 4–7 July 2005.
[4] A.K. Ariffn, M.M. Rahman, Aidah Jumahat, An experimental investigation of warm powder compaction process, in: BSME-ASME International
Conference on Thermal Engineering, Dhaka, 31st December 2002–2nd January 2003.
[5] A. Jagota, P.R. Dawson, J.T. Jenkins, An anisotropic continuum model for the sintering and compaction of powder packing, Mech. Mater. 7 (1988) 255–
269.
[6] T. Nakagawa, S. Masaaki, Simulation of powder densification in die compaction process, Adv. Powder Metall. Partic. Mater. 2 (1992) 43–57.
[7] D. Bouvard, M. McMeeking, Estimation of the densification kinetics of particle aggregates through the simulation of the deformation of an average
inter-particle neck, Sinter. Technol. 4 (1996) 37–44.
 M.M. Rahman et al. / Applied Mathematical Modelling 33 (2009) 4031–4048 4047

[8] H.-Å. Häggblad, K. McEwan, Explicit versus implicit finite element simulation of metal powder compaction, in: Numerical Methods in Industrial
Forming Processes: Proceedings of the International Conference, 1992, pp. 875–881.
[9] D.G. Wang, Y.C. Wu, Research and simulation of the influence of three-axial compaction on powder metallurgic product properties, Acta Metall. Sin.
(Engl. Lett.) 21 (2) (2008) 116–124.
[10] U. Shigeru, O. Yukiko, U. Satoshi, Steel powder and die-lubricated warm compaction for automotive sintered parts with high density and excellent
fatigue property, JFE Technical Report, No. 4, November 2004.
[11] H. Rutz, F. Hanejko, The application of warm compaction to high density powder metallurgy parts, in: International Conference on Powder Metallurgy
& Particulate Materials (PM2TEC ‘97), Chicago, June 29–July 2, 1997.
[12] C.L. Martin, Elasticity, fracture and yielding of cold compacted metal powders, J. Mech. Phys. Solids 52 (2004) 1691–1717.
[13] C.L. Martin, D. Bouvard, S. Shima, Study of particle rearrangement during powder compaction by the discrete element method, J. Mech. Phys. Solids 51
(2003) 667–693.
[14] E. Kontou, Micromechanics model for particulate composites, Mech. Mater. 39 (2007) 702–709.
[15] R.J. Henderson, H.W. Chandler, A.R. Akisanya, C.M. Chandler, S.A. Nixon, Micro-mechanical modelling of powder compaction, J. Mech. Phys. Solids 49
(2001) 739–759.
[16] S.B. Brown, G.G.A. Weber, A constitutive model for the compaction of metal powders, Modern Develop. Powder Metall. 18 (21) (1988) 465–476.
[17] H.-Å. Häggblad, M. Oldenburg, Modelling and simulation of metal powder die pressing with the use of explicit time integration, Modell. Simul. Mater.
Sci. Eng. 2 (1994) 893–911.
[18] A.C.F. Cocks, I.C. Sinka, Constitutive modelling of powder compaction – I: theoretical concepts, Mech. Mater. 39 (2007) 392–403.
[19] T. Hisatsune, T. Tabata, S. Masaaki, A yield criterion of porous materials with anisotropy caused by geometry or distribution of pores, J. Eng. Mater.
Technol. 113 (1991) 425–427.
[20] R. Cytermann, R. Geva, Development of new model for compaction of powders, Powder Metall. 30 (4) (1987) 256–260.
[21] J.C. Cante, J. Oliver, C. González, J.A. Calero, F. Benítez, On numerical simulation of powder compaction processes: powder transfer modelling and
characterization, Powder Metall. 48 (1) (2005) 85–92.
[22] I.M. May, T.H. Ganaba, A full range analysis of reinforced concrete slabs using finite element, Int. J. Numer. Meth. Eng. 26 (1988) 973–985.
[23] A.R. Khoei, A. Shamloo, A.R. Azami, Extended finite element method in plasticity forming of powder compaction with contact friction, Int. J. Solids
Struct. 43 (2006) 5421–5448.
[24] A.R. Khoei, R.W. Lewis, Finite element simulation for dynamic large elastoplastic deformation in metal powder forming, Finite Elem. Anal. Des. 30
(1998) 335–352.
[25] R.W. Lewis, A.R. Khoei, Numerical modelling of large deformation in metal powder forming, Comput. Meth. Appl. Mech. Eng. 159 (1998) 291–328.
[26] A.R. Khoei, S. Azizi, Numerical simulation of 3D powder compaction processes using cone-cap plasticity theory, Mater. Des. 26 (2005) 137–147.
[27] H. Chtourou, M. Guillot, A. Gakwaya, Modeling of the metal powder compaction process using the cap model. Part I. Experimental material
characterization, Int. J. Solids Struct. 39 (2002) 1059–1075.
[28] H. Chtourou, A. Gakwaya, M. Guillot, Modeling of the metal powder compaction process using the cap model. Part II: Numerical implementation and
practical applications, Int. J. Solids Struct. 39 (2002) 1077–1096.
[29] A.R. Khoei, A.R. Azami, S. Azizi, Computational modeling of 3D powder compaction process, J. Mater. Process. Technol. 185 (2007) 166–172.
[30] A. Michrafy, D. Ringenbacher, P. Tchoreloff, Modelling the compaction behaviour of powders: application to pharmaceutical powders, Powder Technol.
127 (2002) 257–266.
[31] R. Rossi, M.K. Alves, H.A. Al-Qureshi, A model for the simulation of powder compaction process, J. Mater. Process. Technol. 182 (2007) 286–296.
[32] R.M. McMeeking, J.R. Rice, Finite-element formulation for problems of large elastic-plastic deformation, Int. J. Solid Struct. 11 (1975) 601–616.
[33] W.A.M. Brekelmans, J.D. Janssen, A.A.F.V. Ven, G. With, An Eulerian approach for die compaction process, Int. J. Numer. Meth. Eng. 31 (1991) 509–524.
[34] D.V. Tran, R.W. Lewis, D.T. Gethin, A.K. Ariffin, Numerical modelling of powder compaction processes: displacement based finite element method,
Powder Metall. 36 (1993) 257–266.
[35] S.J. Park, H.N. Han, K.H. Oh, D.N. Lee, Model for compaction of metal powders, Int. J. Mech. Sci. 41 (1999) 121–141.
[36] A.R. Khoei, R.W. Lewis, Adaptive finite element remeshing in a large deformation analysis of metal powder forming, Int. J. Numer. Meth. Eng. 45 (1999)
801–820.
[37] I. Babuska, J. Chandra, J.E. Flaherty, Adaptive Computational Methods for Partial Differential Equations, SIAM, Philadelphia, 1983.
[38] M.J. Berger, J. Oliger, Adaptive mesh refinement for hyperbolic partial differential equations, J. Comput. Phys. 53 (1984) 484–512.
[39] O.C. Zienkiewicz, J.Z. Zhu, Adaptive techniques in the finite element method, Commun. Appl. Numer. Meth. 4 (1988) 197–204.
[40] R. Boman, L. Papeleux, Q.V. Bui, J.P. Ponthot, Application of the arbitrary Lagrangian Eulerian formulation to the numerical simulation of cold roll
forming process, J. Mater. Process. Technol. 177 (2006) 621–625.
[41] F. Martinet, P. Chabrand, Application of ALE finite elements method to a lubricated friction model in sheet metal forming, Int. J. Solids Struct. 37 (2000)
4005–4031.
[42] K. Davey, M.J. Ward, A practical method for finite element ring rolling simulation using the ALE flow formulation, Int. J. Mech. Sci. 44 (2002) 165–190.
[43] R.A.C. Slatter, Engineering Plasticity, Theory and Application to Metal Forming Process, The Macmillan Press Limited, London, 1977.
[44] S.J. Lacy, J.H. Prevost, Constitutive model for geo-materials, in: Proceedings of the Second International Conference on Constitutive Laws for
Engineering Materials: Theory and Applications, Springer-Verlag, Arizona, 1987, pp. 149–160.
[45] A.R. Akisanya, A.C.F. Cocks, N.A. Fleck, The yield behaviour of metal powders, Int. J. Mech. Sci. 39 (12) (1997) 1315–1324.
[46] S.K. Ray, S. Utku, A numerical model for the thermo-elasto-plastic behaviour of a material, Int. J. Numer. Meth. Eng. 28 (1989) 1103–1114.
[47] J. Lubliner, Plasticity Theory, Macmillan Publishing Company, New York, 1990.
[48] C.S. Desai, H.J. Sriwardhane, Constitutive Laws for Engineering Materials with Emphasis on Geologic Materials, Prentice-Hall Inc., New Jersey, 1984.
[49] R.W. Lewis, A.R. Khoi, A plasticity model for metal powder forming processes, Int. J. Plast. 17 (2001) 1659–1692.
[50] R.M. Haythornthwaite, A more rational approach to strain-hardening data, Engineering Plasticity, University Press, Cambridge, 1968. pp. 201–218.
[51] D.R.J. Owen, E. Hinton, Finite Elements in Plasticity: Theory and Practice, Pineridge Press Limited, Swansea, UK, 1980.
[52] M.M. Rahman, A.K. Ariffin, S.S.M. Nor, Analysis of warm metal powder compaction process – an experimental investigation, in: Powder Metallurgy
2006 (PM’2006), Busan-Korea, September 2006.
[53] A. Mendelson, Plasticity: Theory and Applications, Macmillan Publishing Co., Inc., New York, 1968.
[54] R. Hill, The Mathematical Theory of Plasticity, Clarendon Press, England, 1950.
[55] R.W. Lewis, B.A. Schrefler, The Finite Element Method in the Deformation and Consolidation of Porous Media, Wiley & Sons, New York, 1987.
[56] W.B. Li, M.F. Ashby, K.E. Esterling, On densification and shape change during hot isostatic pressing, Acta Metall. 35 (12) (1987) 2831–2842.
[57] C. Argento, D. Bouvard, P. Stutz, R. Cherlier, A.M. Habraken, Numerical modelling of hot isostatic pressing of metal powder: influence of constitutive
equations, in: Powder Metallurgy World Congress, 1994, pp. 725–731.
[58] T. Tabata, S. Massaki, K. Kamata, Coefficient of friction between metal powder and die-wall during compaction, Powder Metall. Int. 13 (4) (1987) 79–
81.
[59] S. Glaser, B. Kröplin, Thermomechanical coupling in elasto-plastic analysis, Numerical Methods in Industrial Forming Processes, Springer-Verlag,
Balkema, Rotterdam, 1991. pp. 249–255.
[60] S. Keshavarz, A.R. Khoei, A.R. Khaloo, Contact friction simulation in powder compaction process based on the penalty approach, Mater. Des. 29 (2008)
1199–1211.
[61] W. Prager, Recent developments in mathematical theory of plasticity, J. Appl. Phys. 20 (3) (1949) 235–241.
[62] J.K. Bathe, Finite Element Procedures in Engineering Analysis, Prentice-Hall Inc., New Jersey, 1982.
4048 M.M. Rahman et al. / Applied Mathematical Modelling 33 (2009) 4031–4048

[63] O.C. Zienkiewicz, R.L. Taylor, The Finite Element Method, McGraw-Hill Book Company, United Kingdom, 1989.
[64] M.M. Rahman, A.K. Ariffin, A. Anuar, Finite element method for the analysis of warm metal powder compaction process, in: Proceedings of the 2nd
World Engineering Congress, 2002, pp. 258–262.
[65] M.M. Rahman, S. Ramesh, A.K. Ariffin, S.S.M. Nor, M.R. Jamli, Manufacturing of mechanical components through warm compaction process – a finite
element analysis and experimental investigation, in: 9th Japan International SAMPE Symposium & Exhibition, Tokyo Big Sight, 2005, 29th November–
2nd December 2005.
[66] G.F. Bocchini, Warm compaction of metal powders: why it works, why it requires a sophisticated engineering approach, Powder Metall. 42 (2) (1999)
171–180.