Modeling of transverse welds formation during liquid–solid extrusion directly following vacuum infltration of magnesium matrix composite

Jin Liu,Lehu Qi*,Hiyn Zhng,Heping Hou

aFaculty of Printing,Packaging Engineering and Digital Media Technology,Xi’an University of Technology,Xi’an 710048,China

bSchool of Mechatronic Engineering,Northwestern Polytechnical University,Xi’an 710072,China

Modeling of transverse welds formation during liquid–solid extrusion directly following vacuum infltration of magnesium matrix composite

Jian Liua,Lehua Qib,*,Haiyan Zhanga,Heping Houa

aFaculty of Printing,Packaging Engineering and Digital Media Technology,Xi’an University of Technology,Xi’an 710048,China

bSchool of Mechatronic Engineering,Northwestern Polytechnical University,Xi’an 710072,China

Liquid–solid extrusion directly following vacuum infltration(LSEVI)is an infltration–extrusion integrated forming technique,and transverse weld between upper residual magnesium alloy and magnesium matrix composites is a common internal defect,which can severely reduce the yield of composite products.To improve current understanding on the mechanism of transverse welding phenomenon,a thermo-mechanical numerical model of LSEVI for magnesium matrix composites was developed.The formation of transverse weld during extrusion was visualized using fnite element simulation method,and the formation mechanism was discussed from the aspect of velocity feld using a point tracking technique.The simulation results were verifed by the experimental results in term of weld shape.

Liquid–solid extrusion;Magnesium matrix composite;Transverse weld;Finite element simulation

1.Introduction

Magnesium matrix composite is a light-weight structural material with high performance,and has a wide application prospect in high-precision aerospace system,automotive industry and sports equipment due to their superior mechanical properties[1–3].However,their applications are usually limited by poor formability and relatively high fabrication cost.The development of cost-effective fabrication techniques,therefore,is an essential element for expanding their applications.Liquid–solid extrusion following vacuum infltration(LSEVI)is a special forming technique that integrates vacuum infltration,squeeze casting and semi-solid extrusion[4].It has a series of merits such as densifcation of matrix microstructure,uniform dispersion of reinforcement,perfect interfacial bonding,low deformation resistance in the liquid–solid state,and near-net forming.

Besides surface cracks and central cracks,inhomogeneous metal fow in liquid–solid extrusion process always leads to the formation of a cone-shaped transition region between the residual magnesium alloy and the composites(see Fig.1).The interface(transverse weld)region is always porous due to the different fow velocities and solidifcation shrinkage levels between the composites and residual magnesium alloy.Hence, the transition region was usually cut out to obtain the pure composite section.However,it is hard to determine the position and length of weld seam by non-destructive method.In extrusion production,the transverse welding patterns and length are generally estimated by trial-and-error.By comparison,FEM technique is an effective and low-cost defect prediction method [5],which has been successfully applied on studies of transverse weld formation in hot extrusion process of aluminum alloy[6–8],trying to understand welding behavior concentrated upon the fact that die modifcation may infuence the weld length and quality[9].In this process,the new billet is welded onto the back surface of the old billet under extrusion force to maintain continuous production.However,the discontinuity caused by a weld interface in the microstructure of the extruded product can severely reduce the strength of the product,so it is important to minimize the transverse weld interface length while at the same time providing a high-quality weld.

For hot extrusion,the welding process is achieved in solid state between two same billets.The LSEVI is carried out in the semi-solid state containing a small amount of liquid phase,and the transverse weld is formed between composites and matrixmetal.In addition,before extrusion,the composites and matrix metalbondtogetherduringsolidifcation.Atpresent,simulation studiesaboutLSEVIaremainlyconcentratedontheload-stroke curve,temperature history,deformation pattern or macroscopic material fow,e.g.Qi et al.[10]used coupled thermal–mechanical rigid-viscoplastic FEM to analyze the solidifcation and subsequent extrusion process of composites,and gained the distribution of the stress,strain,fow velocity and temperature evolution in the extrusion process.Wang et al.[11]applied three-dimensional thermo-mechanical fnite element method to reveal the formation mechanism of surface cracks based on temperature and velocity felds in the extrusion process. However,previous studies neglect the effect of residual magnesium alloy on the fow behavior of composites,and the transverse weld phenomenon has not been studied.In present paper,thethermo-mechanicalfniteelementmodelofLSEVIfor magnesium matrix composites reinforced by short carbon fbers wasdeveloped.Theaimofthisstudyistoprovideaninsightinto the transverse weld phenomenon in LSEVI.

Fig.1.Longitudinal section of composite rod fabricated by extrusion directly following vacuum infltration technique.

2.Modeling

LSEVI technique consists of four steps.First,magnesium alloy is melted in a sealed melting furnace,which is flled with argon to prevent oxidation,and carbon fber preform is simultaneously preheated in extrusion die.Second,when the magnesium alloy and preform are preheated to the preset temperatures and held for a long time,the liquid meal is sucked into extrusion die via a stainless steel pipe which connects the melting unit and extrusion die,and then infltrated into the preform immediately under gas pressure.Third,after infltration,the magnesium alloy is forced to solidify under high squeezing pressure of punch.Finally,the infltrated composites containing a small fraction of liquid phase are extruded out via the die exit when the container is cooled to the preset temperature.The schematic diagram of experimental setup is shown in Fig.2.

In this study,the FE model of LSEVI consists of fve objects to be simulated:billet(composite and magnesium alloy), punch,container,forming die,and plug rod,as schematically shown in Fig.3.Due to the symmetry of billet,tools,boundary conditions and loads in whole forming process,half of the actual model was selected to establish the fnite element model in order to reduce the computational workload and storage.To ensure computing convergence and computing precision, remeshing is necessary.The maximum interference depth was selected to start a remeshing procedure.If any portion of a master object(tools)penetrates into a slave object(billet) beyond a critical depth(it takes 0.25 mm in present model), remeshing will be triggered.

The thermo-mechanical behavior of the magnesium matrix composite at both the elevated temperature and in the semi-solid state was described using a modifed viscoplastic law,which considers the effect of liquid phase[12],

Fig.2.Schematic diagram of liquid–solid extrusion following vacuum infltration.

Fig.3.Geometry models and meshes of the billet,die and other extrusion tooling.

whereσ,ε˙,T,andfldenote fow stress(MPa),strain rate(s?1), deformation temperature(K),and liquid fraction,respectively.A,a,andbare material constants.n,R,andQdenote stress exponent,universal gas constant(J/mol·K),and apparent activation energy(J/mol).These material parameters in Eq.(1) were determined using multiple linear regression method based on the compression test data of fow stress and strain:n=7.8351,A=3.0578,a=1.0044,b=4.2910,Q=2. 071× 105.The thermo-mechanical behavior of matrix AZ91D magnesium alloy was described using a viscoplastic constitutive model,which integrated the effect of temperature, strain rate,and strain[13],

The values for the material parameters in Eq.(2)were determined through hot compression tests:A=30.85,B=2.5×10?5,C=0.1051,D=5.8×10?5,E=?0.02,F=24.16,G=?2×10?4.The tooling was considered as thermo-rigid body.Both of these material models neglected the elastic behavior of the billet and tooling materials.The thermal conductivity and specifc heat capacity were calculated from the thermo physical properties of AZ91D magnesium alloy and carbon fber according to the mixing rule,and the effect of latent heat on the temperature feld during solidifcation was treated using equivalent specifc heat method[14].

The convection heat transfer coeffcient between tools, extrudate and environment takes 0.1,and the surface radiation coeffcients of extrudate and tooling take 0.12 and 0.7,respectively.Due to the fact that interfacial heat transfer coeffcient is always proportional to normal pressure,a linear relationship between the heat transfer coeffcient and the pressure levels used for squeeze casting was selected in this simulation[15]. The semi-solid material usually sticks to the die surface under the large contact pressure in the liquid–solid extrusion process, the shear friction model is more suitable for this situation.The friction between billet and container wall is a sticking friction, so the friction factor takes 1[16].The conical surface and bearing of the forming die were coated with oil-based graphite as lubricant,so the friction coeffcient between billet and forming die takes 0.3.

The simulation of LSEVI process consists of two steps.The frst step is heat transfer analysis of mushy composites and residual magnesium alloy during solidifcation process.The second step is thermo-mechanical analysis in the liquid–solid extrusion process.The obtained temperature feld by the frststep simulation was set as the initial temperature conditions of the second-step simulation.The dimensions of billet and die as well as the main process parameters adopted in the current numerical simulation are listed in Table 1.

3.Results and discussion

Figure 4 shows the simulation results about the formation of a transverse weld between the residual Mg alloy and composite. It can be seen that the interface between residual Mg alloy and composite does not remain fat when fowing toward the forming die,and the central part of the billets moves much faster than the billet surface.When the punch moved down 30 mm,the front of the magnesium alloy has entered into die land,and the distance between the upper surface and fow front has reached 36.68 mm.When punch stroke reached 40 mm, which exceeds the composite billet height,the composite billet has still not been completely extruded out of the forming die, and the residual part sticks to container wall,forming a cladlayer.The magnesium alloy fowed into the center of composite extrudate,exhibiting an inverted cone.The interface between the fow front of magnesium alloy and the composite clad layer is just the transverse weld.

Table 1Simulation and experimental parameters.

The formation of the transverse weld is usually ascribed to the inhomogeneous metal fow in hot extrusion process[6].In order to reveal the formation mechanism of the transverse weld in LSEVI,four tracking points(P1,P2,P3 and P4)evenly spaced at the welding interface between the magnesium alloy and composite were selected,and the fow velocities of these points along the extrusion direction as function of punch stroke are shown in Fig.5.It can be seen from Fig.5 that the velocity curves of P1,P2,and P3 can be divided into four stages:in the frst starting stage,the fow velocity increases sharply to a certain value,and then enters a long stable stage(the second stage).In this stage,the velocity difference is relatively smaller. When the punch stroke reaches about 25 mm(the third stage), the fow velocities sharply increase again,and deformation non-uniformity is obviously aggravated,this may be attributed to that P1,P2,and P3 gradually approach deformation zone, where the billet is in the semi-solid state with low deformation resistance and better fowability.Figure 6 shows the position changes of the tracking points crossing the deformation zone as function of punch stroke.It can be seen from Fig.6 that when punch stroke reaches 25 mm,P1,P2,and P3 enter the semisolid zone one by one,in which the liquid fraction increases from outer zone to inner zone at the die exit.The higher the liquid fraction,the better the fowability is for magnesium alloy. Hence,the presence of liquid phase further aggravates the deformation inhomogeneity,which can be verifed by our previous experimental results[17].In the fnal stage,a new stable stage appears,and the fow velocities were almost same. However,the fow velocity of P4 is always almost constant,and even lower than the punch velocity,which can be ascribed to the sticking friction between billet and container.

To validate the model,the same process parameters were adopted to fabricate the composite rod using LSEVI technique, and the remaining materials in the forming die were sectioned using electrical discharge machining technique,as shown in Fig.7.It can be seen from Fig.7 that the remaining material can be divided into two parts in term of color,i.e.dark clad layer and central light region.According the observation using a microscope,they are composite and Mg alloy,respectively.By comparing Fig.7 and Fig.4d,it can be found that the simulation results basically agree with the experimental one in term of weld shape.

To minimize the weld length,optimization of the metal fow during extrusion processes is an important means.There are many factors infuencing the metal fow,among which the die structure is closely related to non-homogeneity of metal fow.In order to reveal the infuences of die structure on the shape and length of weld seam,another cone-shaped dies with die semiangle 45°and 60°were selected for comparison.Figure 8 shows the simulated weld shape under different die semi-angles.It can be seen that the corresponding weld lengths are 57.59,75.17, and 90.72 mm,which indicates that weld length increases withincreasing die semi-angle in the range of 45°–60°.However,the smaller the die semi-angle,the smaller the dead metal zone is. Considering that the outside surface of the extrudate originates from the material moving along the dead metal zone[18], which is necessary to remove some impurities at the outer layer of the billet,otherwise,it may be extruded out from the forming die and reduce the surface quality of extrudate,hence a moderate die semi-angle 55°was selected.

Fig.4.Evolution of transverse weld during liquid–solid extrusion,S denotes punch stroke.

Fig.5.Flow velocity of the tracking points along extrusion direction.

4.Conclusions

A 2-D numerical model of liquid–solid extrusion directly following vacuum infltration has been developed using the commercial FEM package DEFORM,and the formation process of transverse weld between residual magnesium alloy after infltration and composite was simulated,and the main conclusions can be summarized as follows:

Fig.7.Longitudinal section of remaining material in the forming die showing the shape of transverse weld.

(1)The formation process of transverse weld can be mainly ascribed to the nonuniform fow of materials,which was caused by the sticking friction between billet and container as well as the presence of liquid phase in the deformation zone.

(2)Simulation results are basically consistent with the experimental one in term of weld shape.Weld length increases with increasing die semi-angle in the range of 45°–60°,and it can be shortened by employing smaller die semi-angle.

Acknowledgements

The authors would like to gratefully acknowledge the fnancial support of National Natural Science Foundation of China (Grant No.51305345),and Natural Science Basic Research Plan in Shaanxi Province of China(Grant No.2014JQ6228).

Fig.6.Position of the tracking points as function of punch stroke,liquid fraction denotes the volume fraction of liquid phase in the semi-solid material.

Fig.8.Weld length(mm)under different die semi-angle.

[1]M.Gupta,N.M.L.Sharon,Magnesium,Magnesium Alloys,and Magnesium Composites,John Wiley&Sons,Inc.,Hoboken,2011,pp. 113–231.

[2]M.J.Shen,M.F.Zhang,W.F.Ying,J.Magn,Alloy 3(2015)162–167.

[3]M.Balakrishnan,I.Dinaharan,R.Palanivel,R.Sivaprakasam,J.Magn, Alloy 3(2015)76–78.

[4]J.Liu,L.H.Qi,J.T.Guan,Y.Q.Ma,J.M.Zhou,Mater.Sci.Eng.A 531 (2012)164–170.

[5]H.J.Hu,D.F.Zhang,F.S.Pan,M.B.Yang,Acta Metall.Sin.22(2009) 353–364.

[6]Q.Li,C.Harris,M.R.Jolly,Mater.Des.24(2003)493–496.

[7]J.Liu,G.Y.Lin,D.Feng,Y.M.Zou,L.P.Sun,Special Cast.Nonferrous Alloys 29(2009)1105–1108(in Chinese).

[8]A.J.den Bakker,R.J.Werkhoven,W.H.Sillekens,L.Katgerman,J.Mater. Process.Technol.214(2014)2349–2358.

[9]Y.Mahmoodkhani,M.A.Wells,N.Parson,W.J.Poole,J.Mater.Process. Technol.214(2014)688–700.

[10]L.H.Qi,L.Z.Su,C.X.Jiang,J.M.Zhou,H.J.Li,Mater.Sci.Eng.A 454–455(2007)608–613.

[11]Z.J.Wang,L.H.Qi,J.M.Zhou,J.Mater.Process.Technol.209(2009) 2068–2076.

[12]L.H.Qi,Z.J.Wang,J.M.Zhou,L.Z.Su,H.J.Li,Compos.Sci.Technol.71 (2011)955–961.

[13]S.S.Xie,X.G.Li,X.X.Jiang,J.Plast.Eng.12(2005)89–93(in Chinese).

[14]L.H.Qi,J.Liu,J.T.Guan,L.Z.Su,J.M.Zhou,T.Nonferrous Metal.Soc. 20(2010)1737–1742.

[15]H.Hu,A.Yu,Modell.Simul.Mater.Sci.Eng.10(2002)1–11.

[16]G.Liu,J.Zhou,J.Duszczyk,J.Mater.Process.Technol.186(2007) 191–199.

[17]L.Z.Su,L.H.Qi,J.M.Zhou,H.J.Li,J.Alloys Compd.509(2011) 775–781.

[18]I.Flitta,T.Sheppard,Mater.Sci.Technol.21(2005)648–656.

Received 15 February 2015;revised 12 July 2015;accepted 20 July 2015 Available online 1 October 2015

*Corresponding author.School of Mechatronic Engineering,Northwestern Polytechnical University,Xi’an 710072,China.Tel.:+8602988460447;fax: +86 29 88491982.

E-mail address:qilehua@nwpu.edu.cn(L.Qi).

http://dx.doi.org/10.1016/j.jma.2015.07.001

2213-9567/?2015 Production and hosting by Elsevier B.V.on behalf of Chongqing University.

?2015 Production and hosting by Elsevier B.V.on behalf of Chongqing University.