Numerical modelling of the gas-powder flow during the ......additive manufacturing(AM), laser metal...

6
Numerical modelling of the gas-powder flow during the laser metal deposition for additive manufacturing Quanren ZENG a , Yankang TIAN a , Zhenhai XU a, b , and Yi QIN a, 1 a Centre for Precision Manufacturing, Department of Design, Manufacture and Engineering Management, University of Strathclyde, Glasgow, G1 1XJ, UK b National Key Labfor Precision Hot Processing of Metals & School of Materials Science and Engineering, Harbin Institute of Technology, Harbin 150001, China Abstract. As one of the most popular additive manufacturing (AM) technologies in the aerospace industry, laser metal deposition (LMD) employs moving laser to melt the coaxially ejected metal powders near the laser focal point, forms a molten pool on the substrate and consequently traps the powders and solidifies the tracks to construct the components with complex geometry layer-by-layer. The mechanical properties and functionality-related performance of the deposited components by LMD depend on the factors such as metal powder’s material/shape, supply status of powders and gas, laser-related manufacturing parameters. According to these influencing factors, there are 4 sub-processes to be modelled in sequence to realize holistic LMD modelling: (1)CFD simulation of the gas-powder flow; (2)laser-powders interaction; (3)formation of molten pool due to laser irradiation with mass and heat addition; (4)solidification of molten pool with deposited metal powders and formed solid track. In this paper, gas-powder flow within the internal passages of laser deposition head and then ejecting from the nozzles’ tips were modelled and analyzed to give a well-depicted image of the related key physics during the LMD process. An in-depth study of the gas-powder flow in LMD via numerical simulation could give a better understanding of subsequent formation mechanism of molten pool and deposited tracks, which will eventually offer more controllable and optimized processing parameter sets to improve the functionality-related performance of LMDed parts. Keywords. additive manufacturing(AM), laser metal deposition (LMD), gas- powder flow, molten pool formation, solidification, deposited solid track 1. Introduction Laser metal deposition (LMD) is a promising additive manufacturing technology for the fabrication of the near net shape parts with streams of metal powder for some high performance applications in the aircraft & aerospace, high performance automotive, medical device, nuclear industry [1] . Many physical phenomena and influencing factors involve in the LMD additive manufacturing process and all of them will have direct or indirect effect on the microstructure and material properties of the final deposited parts [2-4] . A full control of these influencing factors and processing variables such as laser power, scanning speed, powder density, powder feed rate, size distribution and other processing variables, is required to achieve high-quality deposited parts. However, 1 Corresponding Author, Prof Yi Qin, DMEM, University of Strathclyde, UK; [email protected]

Transcript of Numerical modelling of the gas-powder flow during the ......additive manufacturing(AM), laser metal...

  • Numerical modelling of the gas-powder flow during the laser metal deposition for

    additive manufacturing Quanren ZENGa, Yankang TIANa, Zhenhai XUa, b, and Yi QINa, 1

    a Centre for Precision Manufacturing, Department of Design, Manufacture and Engineering Management, University of Strathclyde, Glasgow, G1 1XJ, UK

    b National Key Labfor Precision Hot Processing of Metals & School of Materials Science and Engineering, Harbin Institute of Technology, Harbin 150001, China

    Abstract. As one of the most popular additive manufacturing (AM) technologies in the aerospace industry, laser metal deposition (LMD) employs moving laser to melt the coaxially ejected metal powders near the laser focal point, forms a molten pool on the substrate and consequently traps the powders and solidifies the tracks to construct the components with complex geometry layer-by-layer. The mechanical properties and functionality-related performance of the deposited components by LMD depend on the factors such as metal powder’s material/shape, supply status of powders and gas, laser-related manufacturing parameters. According to these influencing factors, there are 4 sub-processes to be modelled in sequence to realize holistic LMD modelling: (1)CFD simulation of the gas-powder flow; (2)laser-powders interaction; (3)formation of molten pool due to laser irradiation with mass and heat addition; (4)solidification of molten pool with deposited metal powders and formed solid track. In this paper, gas-powder flow within the internal passages of laser deposition head and then ejecting from the nozzles’ tips were modelled and analyzed to give a well-depicted image of the related key physics during the LMD process. An in-depth study of the gas-powder flow in LMD via numerical simulation could give a better understanding of subsequent formation mechanism of molten pool and deposited tracks, which will eventually offer more controllable and optimized processing parameter sets to improve the functionality-related performance of LMDed parts.

    Keywords. additive manufacturing(AM), laser metal deposition (LMD), gas-powder flow, molten pool formation, solidification, deposited solid track

    1. Introduction

    Laser metal deposition (LMD) is a promising additive manufacturing technology for the fabrication of the near net shape parts with streams of metal powder for some high performance applications in the aircraft & aerospace, high performance automotive, medical device, nuclear industry[1]. Many physical phenomena and influencing factors involve in the LMD additive manufacturing process and all of them will have direct or indirect effect on the microstructure and material properties of the final deposited parts[2-4]. A full control of these influencing factors and processing variables such as laser power, scanning speed, powder density, powder feed rate, size distribution and other processing variables, is required to achieve high-quality deposited parts. However,

    1 Corresponding Author, Prof Yi Qin, DMEM, University of Strathclyde, UK; [email protected]

  • choosing ttime consfrom beindevelop avariable se

    2. Descrip

    In LMD, aand are acthe nozzleplane prioand formspowders klayer and beam movsub-procemodellinginteractionpowder mand forme

    This powder fldetails of

    3. Numer

    The gas-pnozzles’ tflow can turbulent i

    the reasonablesuming and exng wider useda reliable numet for producin

    ption/breakd

    as shown in Fccelerated andes’ tips; then tor to depositios a molten pkeep depositinthe molten z

    ves away. Accsses could be

    g: (1) CFD sn[6]; (3) form

    mass and heat;ed solid track.

    F

    research mailow in LMDgas-powder fl

    rical Modellin

    powder flows itips and the sube characteriinert gas and

    e and robust pxpensive[5]. Td in practical

    merical modelng crack free

    own of LMD

    ig.1, metal pod becomes a fthe powder jeon on the subpool trapping ng and adding one finally socording to themodelled in simulation of ation of molt(4) solidifica

    Fig. 1 Schematic

    inly focused; and a 3D Clow status of L

    ng of gas-pow

    in the nozzlesubstrate are nized as a twothe secondary

    processing varThis to some e

    l production.l which couldand reliable p

    D process

    owders firstly fully developeets interact witstrate, which the incident

    into the moltolidifies and fese physical phsequence and the gas-pow

    ten pool due ation of molte

    diagram of holist

    on the firstCFD model wLMD process

    wder flow for

    s’ ducts and wnot a simple oo-phase flow,y phase consis

    riable set via extent hinders

    A more expd serve to ideparts by using

    flow within thed flow beforeth the laser beis melted duepowder jets;

    ten pool on theforms the trachenomena andthen coupled

    wder flow[4]; (to laser irrad

    en pool with d

    tic LMD process

    sub-process owhich could was establish

    LMD

    within the interone-phase turb, in which thsts of the meta

    experiment iss the LMD Apecting alternaentify proper LMD.

    he deposition e being ejecteeam near the le to the laser the melted e surface of thcks/clads whend procedures, to realize hol

    (2) laser/metadiation with adeposited meta

    [1]

    of modelling accurately sim

    hed and analyz

    raction zone bbulent flow. Inhe primary phal powder par

    s obviously AM process native is to

    processing

    head ducts ed out from laser focus irradiation or trapped he growing en the laser

    there are 4 listic LMD al powders addition of al powders

    gas-metal mulate the zed.

    between the n fact, this hase is the rticles. The

  • behavior othe particlcan be hasize districoupling tkind of stainvolved turbulencedeveloped

    .

    (a) deposit

    As shconical mcentral nointernal pdiagram ogas-powdreaches a the deposinozzles’ tithen intermaximus domain arof overall complicattaken for steady-stainlets; (2)coupled dradiation with an av

    For thinner gas the melt p(argon) vi

    of particles sules and the flo

    andled by the ibution. The ithe discrete anable balance[7in the powdee phenomenod based on the

    tion head

    Fig.2 The prac

    hown in Fig.2etal structure

    ozzle for gas aassages and d

    of how the laser mixture isfully developition head andips, 4 gas-pow

    rsect with eacpowder mas

    re discretized calculation sp

    ted geometry modelling of

    ate turbulent f)the powder discrete phaseis temporarily

    verage diamethe deposition nozzle to pro

    pool. The TiAia 4 nozzles w

    uspended in aow. Turbulentconcept of ed

    interaction of nd continuous7-9]. To realizeer feeding syson for the ge nozzle setup

    (b) geome

    ctical laser deposi

    2(a), the typicawith 4 copperand laser passducts are giveser deposition s firstly injecped flow whild nozzles. Onwder mixed jech other to fos concentratiby different lpeed and the(such as the nf the gas-powflow with a cparticles volue modeling (y neglected; (er of 100um. simulation, a

    otect the laser Al powders wewhich are spac

    a turbulent flot dispersion ofddy diffusion the particle a

    s phase until te the simulatiostem of the Lgas-powder flshown in Fig

    etry of internal pa

    ition head and its

    al constructior powder nozzsing through en in Fig.2(b)head deliver

    cted into the le travelling thnce the gas-poets will firstly orm a small cion of the galevel of meshrequired accunozzle tip andwder flow: (1constant veloume fraction (DPM) is ado(4)the particle. 25L/min argooptics from t

    ere delivered ced equally a

    ow depends of both the partenergy in som

    and gas couldthe solutions ion and analyzLMD depositiflow, a nume.2(a).

    assages

    internal geometr

    on of a real lazles facing to coaxially. its ). Fig.2(b) alspowders withinlets of lase

    hrough the paowder flow isflying withouircular focuseas-powder flohes to both couracy at some d focus plane))gas-powder

    ocity and presis less than

    opted; (3)the e size is assum

    on jet is supplthe reboundinto the melt pround the inn

    n the propertiticles and the me range of td be solved byin both phasese powder flowion head, esperical model

    (c) meshes of C

    ry with passages

    aser depositiona focal point adetailed geom

    so shows the h assistance oer deposition assages and du

    sprayed out ut interactive aed spot whichow. Fig.2(c) nsider the imlocal critical

    ) Some assummixture is tr

    ssure distribut10% and theheat transfe

    med to be sph

    ied and ejecteg particles anool by using

    ner gas nozzle

    ies of both carrier gas

    the particle y two-way s achieve a w behavior

    pecially the has been

    CFD domain

    n head is a and 1 main metry with schematic

    of gas.. The n head and ducts inside

    from the 4 actions and h show the shows the

    mprovement areas with

    mptions are reated as a tion at the e one-way

    er by laser herical and

    ed from the nd to shield

    carrier gas e. The total

  • flow rate around thcarrier gatip/outlet iThe initiaTable 1.

    Tab

    Inputs for

    pressure oInner gas dInner gas fInner gas fpressure ocarrier gascarrier gascarrier gas

    4. Results

    There areparticle mbeing ejecaffect the If powdermolten povariable m

    (a) velocity

    The velocvelocity ooutlet of constantlyof the velacceleratebeing ejecmaximumThe collisdue to the

    of the carrierhe main inner as-powder jetis 1.17mm. E

    al conditions f

    ble 1 Material pro

    r CFD

    f inner gas at inledensity (Ar) flow velocity (Arflow rate (Ar) f carrier gas at in

    s density (Ar) s flow(Ar) s flow rate (Ar)

    s and discussi

    e many outpumass concentra

    cted from thefurther sub-pr

    rs have sufficiool on the subsmay need to pa

    of gas-powder (t

    Fig.3

    city of gas-poof the gas-pownozzles. It is

    y while the selocity for the ed in the passcted out from

    m magnitude (asion of particl blown gas. A

    gas for the 4gas nozzle a

    ts be trappedach of the nozfor CFD simu

    operties & proces

    Value

    et 300 MPa1.67 kg/m

    r) 0.118 m/416.7 mm

    nlet 300 MPa1.67 kg/m0.588 m/83.3 mm

    ion of CFD si

    ut variables fration, DPM nue nozzles, arerocesses of deiently large vestrate, then theay attention to

    wo phase) (b) ve

    3 CFD simulation

    owder flow iwder flow is

    s because the ectional area n

    dispersed posages and dum the nozzlesaround 40m/s)es means the

    As shown in Fi

    4 nozzles is 5 and point to td to the meltzzles is titled ulation of gas

    ssing parameters u

    Inputs f

    a average m3 powder /s powder m3/s powder a dia of cem3 number /s dia of po

    m3/s Stand-o

    imulation

    from CFD simnumber of colle the main vaeposition and selocity and cae rebounding

    o.

    elocity of powder

    n result of the velo

    is shown in t still within inlet pressur

    near the nozzlowders is showucts of nozzles, their veloc) before they possiblity thaig.3(c), the co

    L/min. Thesethe melt poolt pool. The d65 degree fro-powder flow

    used for numerica

    for CFD

    diameter of powdensity (TiAl) flow rate (inlet)flow velocity (inentral nozzle (outof powder nozzleowder nozzles(ou

    ff distance

    mulation, in wlision and trajariables of insolidification

    annot been absvelocity of po

    r(discrete phase) (

    ocities of gas & p

    the Fig.3(a). the nozzle dure and flow rle is at the miwn in the Figes stage-by-stcities keep incollide with e

    at powder willollision are mo

    e 4 nozzles ar which coulddiameter of t

    om the substraw are given an

    al simulation of L

    Value

    ders 100um 3910kg/m2.55 (g/m

    let) 0.118m/tlet) 6.4mm es 4 utlet) 1.17mm

    9.5 mm

    which particlejectory of pow

    nterest that wof molten poosorbed or trapowders is anot

    (c) DPM number

    powder

    We could seucts especiallyrate of gas arinimum. The g.3(b). The paage. After th

    ncreasing and each other neal collide with ost likely to be

    re installed d make the the nozzle ate surface. nd listed in

    LMD

    /m3 min) /s

    m

    e velocity, wders after

    will directly ol in LMD. pped by the ther output

    r of collisions

    ee the max ly near the re supplied magnitude

    articles are he particles d reach the ar the focus.

    each other e happened

  • within thechange. Bcollide unThis meancollision

    (a) PM

    (c) PM

    F

    Particle mspecified where thevariation ejecting otravelling 4(b)), the jets travelblurred. Wparticle mIf the focaheat couldfocal poinextremelyconsequenshows a la(c). It is mand even b

    e internal ductBut after the pntil four jets ons that the pountil they are

    MC at different di

    MC at the level y=

    Fig. 4 Particle ma

    mass concentrplane or in a

    e particles areof particle m

    out from the nto and concenparticle mass

    ls to the levelWhen the 4 j

    mass concentraal point of a d be employednt and the poy important fntly generatioarger influencmainly becausbounce back o

    ts especially apowders are ejf the powders

    owders ejectedintersected w

    stances away from

    =-6.5mm

    ass concentration

    ration is a vaspecified vol

    e more likely mass concentra

    nozzles, therentrating on a ps concentrationl of y=-6.5mmjets travels toation of the 4 jlaser beam isd to melt the dsition of the for full fusion of crack-fre

    cing area of pase that it is neonce comes in

    at the positionjected out fros from the 4 nd from the no

    with each other

    m nozzle tip (b

    (d)

    at different locat

    ariable that delume. It couldto gather. Fig

    ation along the is a clear trpoint of intersn of the 4 jetsm (Fig. 4(b)),o the level ojets seem to c

    s just at this lodensest cloud convergent h

    on of the deee and high-qarticle mass cear the focal

    nto contact wit

    n where its secom the nozzlesnozzles concenozzles will flyr near the foca

    b) PMC at the lev

    )PMC around lase

    tions away from t

    escribe the ded give the strag. 4(a) showshe stand-off drend that the section. At ths are separate, the boundarf y=-9.5mm

    converge at thocation, then of particles. M

    highest particlepoisted substquality depoisoncentration tpoint and somth the substrat

    ctional area has, they are lesntrate on the f

    y with less poal point.

    vel y=-1mm

    er focus level y=-

    the outlet of nozz

    ensity of partaightforward is the simulatedistance of LM

    4 jets of pae level of y=- and discrete.y of 4 jets is (Fig. 4(c)), t

    he middle of ththe highest la

    Matching the e mass concetrate and parted tracks. Ththan that of Fime particles wte.

    as a abrupt ss likely to focus leve.

    ossibility of

    -9.5mm

    zle

    ticles on a impression

    ed result of MD. After articles are -1mm (Fig. . When the s becoming the highest he domain. aser power laser beam

    entration is rticles and he Fig.4(d) ig.4(b) and will collide

  • 5. Conclusions

    LMD, based on blowing a powder stream into a moving laser-induced melt pool, is widely employed in the field of advanced manufacturing, such as rapid manufacturing, surface enhancement, tooling and repair. In LMDS, metal powders firstly travel in the laser deposition head, then eject out from the 4 nozzles, gradually converge into the focal point and get molten near or within the molten pool, and finally deposited as a track layer on the substrate. Modelling of LMD is difficult as it is characterized by multiple phase changes, mass and heat flows. In this research, the whole LMD process is analyzed and the gas-powder flow including powder conveyance and dispersion are detailedly investigate due to its direct influence on the subsequent track formation and residual stress on the substrate. The proposed CFD numerical model of gas-powder flow could be used to gain full insight into the powder deposition process and to analyze the influence of the geometrical & processing parameters such as the standoff distance, volumetric gas flow rate, and powder mass flow rate on the quality of the LMD. Also, the developed model provides important parameters for the calculation of the heat transfer boundary condition for the holistic LMD process once the laser radiation is added into the model.

    Acknowledgement

    The authors would like to acknowledge the support from EU H2020 project MMTECH (GA 633776) for the research reported in this paper.

    References

    [1] Zeng, Q., Xu, Z., Tian, Y. & Qin, Y. Advancement in additive manufacturing & numerical modelling

    considerations of direct energy deposition process, in: Proceeding of the 14th International Conference on Manufacturing Research: Advances in Manufacturing Technology XXX. Goy, Y. M. & Case, K. (eds.). Amsterdam: IOS Press, p. 104-109 (ICMR, 6 Sep 2016)

    [2] V. Manvatkar, A. De, T. DebRoy, Heat transfer and material flow during laser assisted multi-layer additive manufacturing, Journal of Applied Physics 116 (2014), 124905

    [3] H. Qi, J. Mazumder, and H. Ki, Numerical simulation of heat transfer and fluid flow in coaxial laser cladding process for direct metal deposition, Journal of Applied Physics 100 (2006), 024903

    [4] S. Zekovic, R. Dwivedi, R. Kovacevic, Numerical simulation and experimental investigation of gas-powder flow from radially symmetrical nozzles in laser-based direct metal deposition, Int. J. Mach. Tools Manuf 47 (2007), 112–123.

    [5] M. Grujicic, Y. Hu, G.M. Fadel, D.M. Keicher, Optimization of the LENS rapid fabrication process for in-flight melting of feed powder, J. Mater. Synth. Process. 9(5) (2001), 223–233

    [6] J.C. Liu, L.J. Li, Y.Z. Zhang, X.Z. Xie, Attenuation of laser power of a focused gaussian beam during interaction between a laser and powder in coaxial laser cladding, J. Phys. D Appl. Phys. 38(10) (2005), 1546–1550.

    [7] H. Pan, F. Liou, Numerical simulation of metallic powder flow in a coaxial nozzle for the laser aided deposition process, J. Mater. Process. Technol. 168 (2) (2005), 230–244.

    [8] H. Pan, T. Sparks, Y.D. Thakar, F. Liou, The investigation of gravity driven metal powder flow in coaxial nozzle for laser-aided direct metal deposition process, Trans. ASME J. Manuf. Sci. Eng. 128(2) (2006), 541–553.

    [9] A.J. Pinkerton, L. Li, Modelling powder concentration distribution from a coaxial deposition from a coaxial deposition nozzle for laser-based rapid tooling, Trans. ASME J. Manuf. Sci. Eng. 126(1) (2004), 33–41.