A New Yield Function for Porous Materials

Embed Size (px)

Citation preview

  • 8/12/2019 A New Yield Function for Porous Materials

    1/8

    Journal of Materials Processing Technology 179 (2006) 3643

    A new yield function for porous materials

    Lus M.M. Alves, Paulo A.F. Martins, Jorge M.C. Rodrigues

    Institute Superior Tecnico, Departamento de Engenharia Mecanica, Av. Rovisco Pais, 1049-001 Lisboa, Portugal

    Abstract

    Densification of sintered copper under uniaxial compression is analyzed by means of three different plasticity criteriaShima and Oyane [S.

    Shima, M. Oyane, Plasticity theory for porous metals, Int. J. Mech. Sci. 18 (1976) 285291], Doraivelu et al. [S.M. Doraivelu, H.L. Gegel, J.S.

    Gunasekera, J.C. Malas, J.T. Morgan, J.F. Thomas, A new yield function for compressible P/M materials, Int. J. Mech. Sci. 26 (9/10) (1984)

    527535] and Lee and Kim [D.N. Lee, H.S. Kim, Plasticity yield behaviour of porous metals, Powder Metall. 35 (1992) 275279]. Theoretical

    predictions are compared with experimental data and served as a basis for developing a new yield function for compressible P/M materials. The

    proposed yield function is not only capable of providing a better agreement with the experiments as its implementation into an existing finiteelement computer program proves to be successful for the numerical modelling of a powder forged component.

    2006 Elsevier B.V. All rights reserved.

    Keywords: Powder forging; Finite element method (FEM); Experimentation

    1. Introduction

    Recent advances in computing technology and reduction in

    the associated costs are presently extending the availability of

    computer programs to simulate complex P/M forming processes.

    At present time computer programs have the potential to accu-rately model complex three dimensional P/M processes without

    the need to take advantage of possible geometrical and material

    flow simplifications.

    Most of the scientific and numeric ingredients which are

    necessary to develop and utilize these programs (discretization

    procedures, solution techniques, contact algorithms and remesh-

    ing schemes) are now nearly established but other important

    ingredients, probably even more relevant than the former, to

    the progress of numerical simulation, still deserve attention.

    The yield function and its associated constitutive equations for

    porous materials [1,2] is oneof these ingredients because it plays

    a crucial role in the overall performance of the computer pro-

    grams.

    This paper is based on a comprehensive investigation of the

    commonly used yield functions due to Shima and Oyane [3],

    Doraivelu et al.[4]and Lee and Kim [5].The main objective

    was to characterize and understand its limitations but also to

    develop a new more accurate one if necessary.

    Corresponding author. Fax: +351 21 841 90 58.E-mail address:[email protected](J.M.C. Rodrigues).

    The most important drawback in the utilization of the above-

    mentionedyieldfunctions is dueto thefact that theinitial relative

    density does not explicitly influence the geometric hardening of

    the base material and the plastic Poisson coefficient. Some of

    the yield functions cannot even be used when the initial relative

    density of the preforms is low.The paper deals with these problems and proposes a new

    yield function for powder materials that takes into account the

    initial relative density of thepreform. New testing procedures for

    characterizing the mechanical behaviour of the base material are

    also introduced in order to obtain a more accurate stressstrain

    curve of the powder material.

    The implementation of the new proposed yield function into

    an existing finite element computer program is also briefly con-

    sidered and assessment is provided by means a numerical and

    experimental study of a powder forging process. To this end, a

    laboratory controlled test comprising the forging of a sintered

    copper perform was designed in such a way that material flow

    behaviour, geometric profile, distribution of relative density and

    forging load, could be reproduced and compared against numer-

    ical estimates.

    2. Theoretical background

    The yield function for porous materials differs from the von

    Mises function for dense materials due to the changes produced

    in the volume of the porous bodies during plastic deformation.

    In general, considering that the material is isotropic, and that the

    0924-0136/$ see front matter 2006 Elsevier B.V. All rights reserved.

    doi:10.1016/j.jmatprotec.2006.03.091

    mailto:[email protected]://localhost/var/www/apps/conversion/tmp/scratch_4/dx.doi.org/10.1016/j.jmatprotec.2006.03.091http://localhost/var/www/apps/conversion/tmp/scratch_4/dx.doi.org/10.1016/j.jmatprotec.2006.03.091mailto:[email protected]
  • 8/12/2019 A New Yield Function for Porous Materials

    2/8

    L.M.M. Alves et al. / Journal of Materials Processing Technology 179 (2006) 3 643 37

    Table 1

    The parametersA,B and for different porous yield functions

    A B

    Shima and Oyane[3] 31+(2.49/3)2(1R)1.028

    (2.49/3)2 (1R)1.0281+(2.49/3)2 (1R)1.028

    R5

    1+(2.49/3)2 (1R)1.028

    Doraivelu et al.[4] (2 +R2) (1R2 )

    3 (2R2 1)

    Lee and Kim[5] (2 +R2) (1R2 )

    3

    (RRc )2

    (1Rc)2

    deformation does not produce anisotropy, the yield function for

    porous materials must include the first invariant I1of the stress

    tensor and the second invariant of the deviatoric stress tensorJ2[1,2],

    AJ2 + BI21= 20= 2R (1)The yield function for porous materials represents in the prin-

    cipal stress space, the surface of an ellipsoid that is symmetrical

    with respect to the three principal stress axes and has a major

    axis that coincides with the mean stress axis. The parameters

    A, B and are to be determined from experimental work andmust be expressed as functions of the relative density R (ratio

    of the volume of the base material to the total volume of porous

    material). The symbol R denotes the apparent yield stress of

    the porous material which depends on the yield stress 0of the

    base material and the relative density R.

    Considering the porous material as a continuum medium,

    the strain-rate components can be obtained from the associated

    porous flow rule(1)as follows[7],

    ij=R

    R

    A

    2ij+ (3 A)mij

    (2)

    where m is the apparent mean stress and R is the apparenteffective strain rate of the porous material given by,

    R=

    2

    Aij

    2 + 13(3 A)

    2v

    1/2(3)

    The symbol v= R/R refers to the volumetric strain-rate(ratio of the rate of variation of the relative density to the relative

    density) and the apparent effective stress Ris determined by,

    R=

    (AJ2 + BI21 ) (4)The plastic deformation of a porous material includes two

    different forms of hardening; (i) geometric hardening due to

    material densification (parameter in Eq.(1))and (ii) materialhardening due to changes in the mechanical properties of the

    base material (the yield stress 0 of the base material). The

    relation between the apparent strain and strain-rate and those of

    the base material can be obtained by combining the plastic work

    rate,

    w = R RVR= 0 0V0= 0 0RVR (5)with the geometric hardening parameter resulting from mate-

    rial densification,

    0

    =

    RR 0

    = t

    0

    0dt

    = t

    0

    R0dt (6)

    Table 1summarizes the parameters A, B and of the yield

    functionsthat were proposed by Shima andOyane[3], Doraivelu

    et al.[4] and Lee and Kim[5].These functions are commonly

    utilized in thenumerical simulation of porousmetal forming pro-

    cesses and turn into the conventional von Mises yield function

    for dense materials as the relative density R approaches unity.

    In these limiting conditions ( = 1), the total apparent energy

    of deformation becomes equal to the elastic energy of distortion

    (becauseA =3andB = 0). Itis alsoimportantto notefromTable 1

    that the yield function proposed by Doraivelu et al. cannot be

    applied for values of the relative density Rc 1/2 = 0.71and that the yield function due to Lee and Kim makes use of

    a critical relative density Rc that needs to be determined from

    experiments.

    The numerical simulation of P/M metal forming processes

    starts from the following weak variational form expressed

    entirely in terms of the arbitrary variation in the velocity u,

    =

    V

    R RdV

    S

    FiuidS= 0 (7)

    where V is the control volume of the porous body limited by

    surfacesSu and SF, where velocity u i and traction Fi are pre-

    scribed, respectively. The solution of Eq. (7) by thefiniteelementmethod requires subdivision of the control volume Vby means

    ofMelements linked through Nnodal points. In each element

    the velocity distribution can be represented in the vectorial form

    as,

    u = Nv (8)

    whereN is the matrix containing the shape functions Ni of the

    element and v is the nodal velocity vector. The strain rate vector

    can be obtained from,

    = Bv (9)whereB is the velocity strain rate matrix. Substituting Eqs.(8)

    and(9)into Eq.(7)at the elemental level and assembling the

    element equations with global constraints, it follows,

    Mm

    V(m)

    RR

    Kv dV

    S(m)NF dS

    = 0 (10)

    whereK = BTDBandDis a matrix resulting from the inversion

    of the porous constitutive Eq.(2).Further details on the imple-

    mentation of the finite element method for P/M metal forming

    processes can be found elsewhere[6,7].

  • 8/12/2019 A New Yield Function for Porous Materials

    3/8

    38 L.M.M. Alves et al. / Journal of Materials Processing Technology 179 (2006) 3643

    Fig. 1. Scanning electron micrographs of the copper powder utilized in the experiments: (a) undeformed copper grains (magnification of 1000); (b) compacted andsintered copper grains (magnification of 4000).

    3. Experimental work

    3.1. Uniaxial compression test

    Thematerialemployed in theexperimentswas a purecopperpowder(99.5%)

    withan apparentdensityof 2.86g/cm3 andthe followingdistributionof grainsize

    (obtained by means of sieve analysis): 94.92% 45m, 5.08% 45m and0% 75m.Fig. 1presents two photomicrographys of the material showingthe original (Fig. 1(a)) and the deformed powder grains (Fig. 1(b)) after being

    compacted and sintered at 825 C, during 30 min in a vacuum atmosphere of0.55 mbar.

    The experimental work was performed in a 500kN computer-controlled

    hydraulicpress under a veryslow speed. The cylindrical compactedand sintered

    specimens utilized in the uniaxial compression tests were produced in a double

    effect tool, lubricated with zinc stearate, under different applied pressures in

    order to ensure a good homogeneity and to obtain test specimens with different

    values of the initial relative densityR0.Fig. 2presents some of these specimens

    before and after being deformed.

    In order to allow load recording during the test, force was applied to the

    upper compression plate through one load cell based on traditional strain-gauge

    technology in a full Wheatstone bridge. Displacements were measured using a

    laser-based optical displacement transducer and a PC-baseddata logging systemwasused to recordand store loads anddisplacements. In addition to this, dimen-

    sions and density of the compression test specimens were also measured at each

    stepof deformation. Density wasmeasuredby means of a submersion technique,

    using a precisionbalance with an accuracy of0.0001 g, andthe specimens hadto be previously waterproof sealed with a varnish of known density.

    The compression tests were initially performed with the objective of inves-

    tigating the consistency and accuracy of the aforementioned porous plasticity

    criteria (Table 1)and of identifying a criterion to subsequent implementation

    into an existing finite element computer program that is currently being devel-

    oped within the research team. However, the study of the main features of these

    criteria combined with some discrepancies that were found during its exper-

    imental assessment stimulated the authors to develop a new porous plasticity

    criterion. The criterion, entitled the initial density criterionIDC, considers,

    Fig. 2. Compression test specimens made from compacted and sintered copper

    powder before and after deformation. The labels 15 refer to specimens with

    different values of the initial relative densityR0.

    as the name indicates, the influence of initial relative densityR0 of the sintered

    parts and will be presented in detail in the following sections.

    3.2. Mechanical characterisation of the base material

    The approach followed by the authors for the mechanical characteriza-

    tion of the base material diverges from the commonly utilized methodology ofemploying a wrought material with identical chemical composition because the

    metallurgical conditions play a crucial role in the overall mechanical response

    of the materials. As shown inFig. 3the grain size of the wrought electrolytic

    copper is much bigger than that of the compacted and sintered copper powder

    with the same level of purity. Actually, this is one of the main advantages that

    are usually attributed to the structural components produced by means of P/M

    processes, because the reduced size of the powder grains is capable of ensuring

    better mechanical properties whenever the overall level of porosity is reduced.

    In order to overtake the inherent difficulties of obtaining the mechanical

    characteristicsof thebasematerialby meansof directmethods, it wasdecidedto

    develop a new indirect method based in compression tests, without friction, that

    employscylindricalspecimens withdifferent valuesof theinitialrelative density.

    Thecompression tests areperformed in such a waythateach incrementof stroke

    Fig. 3. Micrographs showing the structure of: (a) a wrought electrolytic copper and (b) a sintered copper powder with equal purity. The magnification is equal to

    400 in both cases.

  • 8/12/2019 A New Yield Function for Porous Materials

    4/8

  • 8/12/2019 A New Yield Function for Porous Materials

    5/8

  • 8/12/2019 A New Yield Function for Porous Materials

    6/8

    L.M.M. Alves et al. / Journal of Materials Processing Technology 179 (2006) 3 643 41

    Fig.7. Powder forging of a copper flanged component. (ac)Initial, intermediate and finalgeometryof the flanged P/M component. (d) Correlation between measured

    and computed geometrical profile of the cross section of the flange at three different stages of deformation.

    all the porous plasticity criteria under investigation. The investigation made

    use of test specimens with different initial densities and as it can be seen the

    initial density criterion (IDC) is the only model that accurately reproduces the

    mechanical behaviour of the sintered material for the test specimens with low

    values of the initial relative densityR0 (Fig. 6(d)). In the case of the Doraivelu

    et al. porous plasticity criterion it is not even possible to get an estimate for the

    experimental test case performed with R0=0.686 because the critical relative

    density of this criterion is higher Rc= 0.707> 0.686.

    4. Powder forging of a flanged component

    This section deals with the numerical modelling of the cold

    powder forging of a copper flanged component and it was

    included in the paper with the objective of validating the numer-

    ical predictions provided by the finite element computer pro-

    gram (modified in order to include the IDC porous criterion)

    with the experimental measurements obtained from laboratory-

    controlled conditions. Assessment is performed in terms of

    material flow, geometrical profile, relative density and forging

    load.

    On account of rotational symmetry it was possible to carry

    out the numerical simulation of the process under axisymmet-

    rical modelling conditions. The initial cross section of the sin-tered preform (with an initial relative density R0= 0.79) was

    discretized by means of a mesh consisting of 584 four-node

    axisymmetric quadrilateral elements.Larger elements were used

    to fill the body and the initial free regions of the preform whereas

    the surfaces of interest (where contact is expected to occur) were

    meshed with smaller elements in order to obtain more accurate

    results on the filling behaviour. The contour of the dies were

    approximated by means of contact-friction linear elements and

    the friction factor was found to be equal to m = 0.2 in accordance

    to a calibrationprocedurebased on the conventionalring-test [8].

    Fig. 7 shows the geometry of the compacted and sintered pre-

    forms that were used for producing the powder forged flanged

    components together with deformed parts corresponding to the

    intermediate and final stages of deformation. Geometry and rel-

    ative density was also measured at these stages of deformation

    and latter compared with the numerical estimates derived from

    Fig.8. Powder forging of a copper flanged component. (a) Computedand exper-

    imental evolution of the load vs. displacement; (b) computed and experimental

    evolution of the average relative density vs. displacement.

  • 8/12/2019 A New Yield Function for Porous Materials

    7/8

    42 L.M.M. Alves et al. / Journal of Materials Processing Technology 179 (2006) 3643

    finite elements. As seen fromFig. 7(d) the agreement between

    the theoretical and experimental profiles is a very good.

    Further assessment between computed and experimental

    results can be found in Fig. 8, where the estimates of the

    forging load and average relative density are plotted against

    the correspondent experimental values. In case of the forging

    load the agreement is excellent. Two different stages can be

    identified; an initial stage where the evolution of the load is

    Fig. 9. Computed and experimental distribution of the relative density. (ab) Computed distribution of the relative density at an intermediate stage of deformation

    and at the final stage of deformation. (c) Micrographies showing the porosity in selected regions of the cross section of a component at the final stage of deformation.

    Labels 115 refer to positions indicated in (b).

  • 8/12/2019 A New Yield Function for Porous Materials

    8/8

    L.M.M. Alves et al. / Journal of Materials Processing Technology 179 (2006) 3 643 43

    smooth and similar to what is generally obtained for head-

    ing operations with flat punches (although some radial flow

    into the bottom of the preform begins at the early stages of

    deformation), and a final stage where the forging load increases

    sharply. The later starts when the die cavity commences to be

    filled.

    In what concerns the average relative density there is also

    a good agreement between theory and experimentation but the

    number of measurements and the average (integrator) charac-

    teristics of the parameter are not adequate to conclude about the

    overall quality of the finite element distribution of the relative

    densityRinside the workpiece.

    The above-mentioned limitations stimulated the authors to

    perform a set of micrographies in selected regions of the cross-

    section of the final flanged component and to compare theoreti-

    calpredictionswith experimental observations. A closer analysis

    ofFig. 9 enables to conclude that there is a good agreement

    between the theoretical and experimental distribution of density

    inside the workpiece and proves that the initial density criterion

    IDC can be successfully applied in the numerical modelling ofP/M metal forming processes.

    5. Conclusions

    The porous plasticity criteria of Shima and Oyane, Doraivelu

    et al. and Lee and Kim present difficulties in modelling the

    mechanical behaviour of sintered parts with low values of the

    initial relative density. This is due to the influence of the initial

    relative density in the geometric hardening of the base material

    and in the plastic Poisson coefficient.

    The initial density criterion (IDC) that was proposed by

    the authors takes this influence into account and proved tobe able to provide better estimates of the apparent field vari-

    ables during the uniaxial compression of sintered copper powder

    specimens.

    A new indirect method for determining the stressstrain

    behaviour of the base material was also proposed as an alter-

    native to the commonly utilised approximate procedures based

    on the utilization of wrought materials with similar chemical

    compositions.

    The implementation of the initial density criterion (IDC)

    in a finite element computer program that is currently being

    developed by the authors proved to be successful in the numer-

    ical modelling of a powder forged component. Assessment is

    provided by comparing finite element predictions of the mate-

    rial flow, geometrical profile, relative density and forging load

    against experimental measurements obtained from laboratory-

    controlled conditions.

    References

    [1] H.A. Kuhn, C.L. Downey, Deformation characteristics and plasticity the-

    ory of sintered powder materials, Int. J. Powder Metall. 7 (1971) 15

    25.

    [2] R.J. Green,A plasticity theoryfor porous solids, Int. J. Mech. Sci. 14 (1972)

    215224.[3] S. Shima, M. Oyane, Plasticity theory for porous metals, Int. J. Mech. Sci

    18 (1976) 285291.

    [4] S.M. Doraivelu, H.L. Gegel, J.S. Gunasekera, J.C. Malas, J.T. Morgan, J.F.

    Thomas, A newyield functionfor compressible P/M materials., Int. J. Mech.

    Sci. 26 (9/10) (1984) 527535.

    [5] D.N. Lee, H.S. Kim, Plasticity yield behaviour of porous metals, Powder

    Metall. 35 (1992) 275279.

    [6] M.L. Alves, J.M.C. Rodrigues, P.A.F. Martins, Simulation of three-

    dimensional bulk forming processes by finite element flow formulation,

    Model. Simul. Mater. Sci. Eng. 11 (2003) 803821.

    [7] P.A.F. Martins, M.J.M. Barata Marques, A general three-dimensional finite

    element approach for porous and dense metal-forming processes, Proc. Inst.

    Mech. Eng. 205 (1991) 257263.

    [8] L.M.M. Alves, Deformacao plastica na massa de sinterizados metalicos:

    modelacaonumericaeanaliseexperimental, Tese de Doutoramentoem Enga

    Mecanica, IST, 2004.

    [9] G.M. Zhdanovich,Theory of Compacting of Metal Powders(Teorize presso-

    vaniya metzllichaskikli poroshkov), 1969, pp. 1262.