Word Search

A Study on Subsequent Static Aging and Mechanical Properties of Hot-Rolled AA2017

Description
In this work, the effects of rolling parameters, cooling media, and deformation path on mechanical properties and aging behavior of hot-rolled AA2017 were studied. First, hot-rolling experiments were conducted under different working conditions, and
Categories
Published
of 11
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  A Study on Subsequent Static Aging and MechanicalProperties of Hot-Rolled AA2017 L. Khalili and S. Serajzadeh  (Submitted April 26, 2013; in revised form February 14, 2014; published online May 9, 2014) In this work, the effects of rolling parameters, cooling media, and deformation path on mechanicalproperties and aging behavior of hot-rolled AA2017 were studied. First, hot-rolling experiments wereconducted under different working conditions, and the rolled strips were then aged at room temperature forup to 57 days during which hardness and tensile tests were carried out to record the changes in themechanical properties of the alloy. Furthermore, due to the importance of static recrystallization on sub-sequent aging behavior, the rate of recrystallization was also computed. To this end, a mathematical modelwas developed to predict thermomechanical responses during hot rolling using the finite element software,Abaqus/Explicit. Then, a physically-based model was employed for the determination of the kinetics of static recrystallization using the predicted thermomechanical parameters. Finally, the effects of rollingschedule on the mechanical properties and the aging behavior of rolled alloy were evaluated by means of theexperimental results and the predictions. The results indicate that natural aging occurs in the hot-rolledalloy, while its influence on the mechanical properties is highly affected by the static recrystallizationoccurring in the interpass region and/or after rolling on the run-out table. Keywords  aging phenomena, aluminum alloys, deformation path,finite element modeling, hot rolling, static recrystal-lization 1. Introduction Aluminum alloys are widely used in various industrialapplications because of their high-strength-to weight ratio,resistance to fatigue crack propagation, and fracture toughness(Ref  1, 2). Theses alloys can be manufactured by deformation  processing under hot and warm working conditions, whichresults in different microstructures and mechanical properties(Ref  3). Due to the importance of these microstructural eventsincluding recovery and recrystallization, there are numerousstudies on these subjects (Ref  3-13). Moreover, deformation at  elevated temperature and subsequent rapid cooling may lead tothe occurrence of aging phenomena in hot-deformed alloy.Owing to the significant influence of this phenomenon on thefinal mechanical properties of aluminum alloys, several inves-tigations have also been carried out concentrating on this issue.For instance, Wang et al. (Ref  14) investigated the effect of pre-aging treatment on the tensile properties, formability, andmicrostructures of 2036 aluminum alloy. Wang et al. (Ref  15)showed that deformation-aging treatment caused a significant improvementinthetensilestrengthandthehardnessofAA2618.In another study, Liu et al. (Ref  16) studied the effects of aging parameters and chemical composition on the yield strength of different alloy systems including Al-Cu binary alloy, AA6061,andAl-Zn-Mgalloy.VandenBuekelandKocks(Ref 17)studiedthe dependence of both static and dynamic strains aging onapplied plastic strain. Carrera et al. (Ref  18) investigated theeffect of delay time between cooling and aging on the finalmechanical properties of cast aluminum alloys. It was found that the yield strength, total elongation, and to a lesser extent, theultimate tensile strength were affected by the delay time.Althoughpreviousinvestigationshaveprovidedvaluabledataontheagingbehaviorofaluminumalloys,onlya,limitednumber ofstudies areavailableontheeffectsofrollingparametersontheaging phenomena and the final mechanical properties of heat-treatable aluminum alloys. Moreover, a few studies have beenconducted about the interconnection among rolling layout,metallurgical events, and subsequent aging processes. In thisregard,the main aim of the presentwork is to providethe data ontheeffectsofrollingparameters,coolingmedia,anddeformation path on the subsequent aging behavior of hot-rolled AA2017 bymeans of mathematical modeling and experimentation. A physically-based internal-state variable model is combined withafiniteelementanalysisandadditivityruleforpredictingtherateof static recrystallization after hot rolling of AA2017. Then, theeffects of rolling pass design and rolling parameters such asreduction, rolling speed, and cooling rate after hot rolling on theoccurrence of aging and mechanical properties of the alloy areevaluated employing the model predictions, the results of mechanical testing, and microstructural observations. 2. Macromodeling of Multipass Hot Rolling Mathematical modeling of a complex process such asmultipass hot rolling is a challenging task. For doing so, anefficient method is required for solving the governing thermaland mechanical problems at the same time. In this work, the L. Khalili  and  S. Serajzadeh , Department of Materials Science andEngineering, Sharif University ofTechnology,Azadi Ave., Tehran,Iran.Contact e-mails: khalili_laila@mehr.sharif.ir and serajzadeh@sharif.edu. JMEPEG (2014) 23:2894–2904   ASM InternationalDOI: 10.1007/s11665-014-0990-z 1059-9495/$19.002894—Volume 23(8) August 2014 Journal of Materials Engineering and Performance  finite element package, ABAQUS/Explicit was utilized tomodel multipass hot-rolling process under plane strain condi-tions. During hot rolling, temperature and displacement distri- butions within the strip being rolled were calculated byconcurrent solving of the following equations (Ref  19, 20): Z   A e @  W  T  @   x k   @  T  @   x  þ  @  W  T  @   y k   @  T  @   y   dA   Z  C  e W  T  k   @  T  @  ndC   Z   A e W  T  _ qdA   Z  C  e W  T  k  q c _ TdA  ¼  0 ð Eq 1 Þ Z   q € u d _ udV   þ Z   rd  DdV    Z   F  d _ udS   ¼  0 ;  ð Eq 2 Þ where,  q  is the density,  c  is the specific heat,  k   is the thermalconductivity, and  x  and  y  denote the longitudinal and thicknessdirections, respectively.  _ q  is the rate of heat generation owingto plastic deformation.  W   represents the weight function matrix,and  n  is the normal to the boundary.  r  and  D  are Cauchy stressand deformation rate tensors, respectively,  F   represents the vec-tor of surface boundary traction,  u  is the displacement vector,and  d _ u  denotes the virtual variation in the velocity field.In the deformation zone, at the contact region of strip/roll, adistributed surface flux  _ q fric , was generated owing to frictionalslidingthatmaybedeterminedbythefollowingequation(Ref 21): _ q fric  ¼  s f   A c D v  ;  ð Eq 3 Þ where  D v   is the relative velocity between the metal and thework-roll,  A c  is the contact area, and  s f   is the frictional stress.Accordingly, the boundary condition in deformation zone wasdefined as  k   @  T  @  n  ¼  h con  T     T  r  ð Þ   _ q fric ;  ð Eq 4 Þ where  T  r   indicates the work-roll surface temperature, and  h con is the interface heat transfer coefficient. Note that the conduc-tion-convection boundary conditions were applied for the freesurface boundaries as  k   @  T  @  n  ¼  h a  T     T  a ð Þ ð Eq 5 Þ In Eq 5,  T  a   represents surrounding temperature, and  h a   is theconvection heat-transfer coefficient to the surrounding. Physi-cal properties used in the thermal model are listed in Table 1.Also, in the mechanical part, the Coulomb friction model wasconsidered to determine the surface traction boundary condi-tion in the mechanical model as well as to compute the rateof heat generation due to friction. In addition, Von Misesyield function and the isotropic hardening rule were em- ployed in the mechanical model. Moreover, only one half of the rolling strip and the top work-roll were taken into account  because of the existing symmetry, and the work-rolls werealso assumed to behave as rigid bodies. In this regard, zerodisplacement boundary conditions were imposed at the stripcenterline along the  y -axis and the roll/metal interface alongthe radial direction of the work-roll. The schematic geometryof deformation zone as well as thermal and mechanical boundary conditions is shown in Fig. 1(a). For solving thegoverning thermal and deformation problems, the finite ele-ment software Abaqus/Explicit was then utilized in which thedeveloped working model was capable of considering bothsingle-pass as well as multipass rolling schedules. Forwardrolling layout was taken in the multipass rolling operations asthe strip passes through successive stands. The mesh genera-tion was carried out using four-node bilinear temperature-dis- placement elements, while the mesh system included 100elements in length and 5 elements along thickness direction,and the Coulomb friction model with a constant friction coef-ficient of 0.4 was considered at the contact region (Ref  22). It should be mentioned that a mesh-sensitivity analysis was first carried out to obtain the optimum mesh size which givesaccurate results with reasonable computation time. Figure 1(b)shows the deformed meshes during hot rolling with an initialtemperature of 480   C, a rolling speed 50 rpm, and a reductionof 30%. 3. Static Recrystallization Model In the case of high-temperature rolling, various restoration phenomena such as recovery and recrystallization might beoperative. However, due to the high stacking fault energy (SFE)of the examined alloy, the occurrence of discontinuous dynamic Table 1 Material properties used for modeling of therolling process in this work  Parameter Value q  2790 kg m  3 c  p  880 J kg  1  C  1 k   134 Wm  1  C  1 h con  57000 kW/m 2  C T  r   60   C T  a   20   C h a   20 W m  2  C  1 Fig. 1  (a) Illustration of the geometry and boundary conditionsused in the modeling; (b) deformed elements during hot rolling of a plate with initial temperature of 480   C, rolling speed of 50 rpm,and reduction of 30% Journal of Materials Engineering and Performance Volume 23(8) August 2014—2895  recrystallization is insignificant (Ref  23). Furthermore, thecontinuous dynamic recrystallization such as geometricdynamic recrystallization may occur in aluminum alloys, but this type of recrystallization would be operative at large plasticstrains. Therefore, it is expected that the main restorationmechanism during deformation would be dynamic recovery.On the other hand, because of high rate of dynamic recoveryduring deformation, the effect of subsequent static recovery isexpected to be trivial, and thus, in this work, the major restoration mechanism after hot rolling was assumed to be staticrecrystallization. The relationship between the volume fractionof statically recrystallized volume fraction as a function of timemay be expressed by Avrami equation as follows (Ref  23):  X  v  ¼  1    exp   0 : 693  t t  50   n st    ;  ð Eq 6 Þ where,  t   is recrystallization time,  n st   is the Avrami exponent witha commonly reported value of 2, and  t  50  is the time required for 50% recrystallization. Empirical and physically-based modelsare usually utilized for the determination of   t  50 . In the present work, a physically-based model developed in Ref  24, 25 was uti- lized for determining recrystallization kinetics. According to thismodel the parameter   t  50  could be computed as t  0 : 5  ¼  A M  GB  P  D 1  N  v   1 = 3 ð Eq 7 Þ where  P  D  denotes the stored energy due to dislocation densityintroduced during deformation,  A /   M  GB  is a calibration con-stant, and  N  v  is a factor relating to nucleation process that isdefined as  N  v  ¼  2 c d d  0 f 2   ½ exp ð  Þ þ  exp ð  Þ þ  1  ;  ð Eq 8 Þ where  C  d  is another calibration constant,  f  is the size of sub-grains, and  d  0  represents initial grain size. The stored energy mayalso be approximated using the following equation (Ref  25):  P  D  ¼  Gb 2 10  q i ð 1    ln ð 10 q 0 : 5i  ÞÞ þ  2 h b f  1  þ  ln  h C h     ; ð Eq 9 Þ where  G   is the shear modulus,  b  is the Burgers vector, and  q i is the internal dislocation density that consists of two partsincluding the density of geometrically necessary dislocations( q g ) and random dislocation density ( q r  ) i.e.  q i  =  q r   +  q g .  h  isthe misorientation, and  h C  is the critical misorientation for ahigh-angle boundary, e.g., 15  . The amounts of   d ,  q r  , and  h can be directly computed by solving the following differentialequations that were derived based on the classical theories of work hardening and dynamic recovery (Ref  24, 25). d  f d   ¼  f  f f ss ð f ss    f Þ ð Eq 10 Þ d  h d   ¼  1  h ð h ss    h Þ ð Eq 11 Þ d  q r  d   ¼  C  1 q 0 : 5r     C  2 r f   Z   q r    ;  ð Eq 12 Þ where,  d ss  and  h ss  are the subgrain size and misorientation at steady-state deformation conditions, respectively;  e d  and  e f are characteristic strains;  r f   (in Pa) is the frictional stress;  C  1 and  C  2  are the material constants.  Z   is the Zener-Hollomon parameter defined as  Z   ¼  _  exp ð Q =  RT  Þ  in which  _   is theequivalent strain rate,  Q  is the apparent activation energy of hot deformation, and  T   is the deformation temperature. It isworth mentioning that the above system of differential equa-tions was solved using finite difference method (Ref  26),while the initial conditions for the above equations are givenin Table 2, i.e.,  q 0 ,  h 0 , and  f 0 . In this regard, a code in MAT-LAB was generated to solve the above system of equationsfor each element, while the microparameters such as plasticstrain and temperature were determined using the results of the finite element analysis performed in Abaqus/explicit. Inother words, the model was handled sequentially. The macro- parameters were first predicted by the thermomechanicalmodel, and then, the microstructural events were simulated by means of the above procedure.It should be note that, for the determination of the internaldislocation density, it was required to calculate the density of geometrically necessary dislocations at the same time. Accord-ingly, this parameter was calculated as follows (Ref  24): q g  ¼  1 b 1  R g   hf    ð Eq 13 Þ where, 1/   R g  is the local lattice curvature that was taken as aconstant according to Ref  24. Note that for the case of site-saturated nucleation conditions, the recrystallized grain sizemay be estimated as d  rex  ¼  D ð  N  v Þ  13 ;  ð Eq 14 Þ where  D  is a material constant. Finally, it is possible to com- pute the parameter,  t  0.5 , under different deformation conditionsutilizing Eq 7 to 14. It should be noted that Eq 6 is applicable under isothermal recrystallization conditions and cannot be em- ployed under continuous cooling conditions as it occurs duringrolling. Accordingly, the concept of temperature-compensatedtime parameter ( W)  was used for the determination of the rateof recrystallization under nonisothermal cooling conditions(Ref  27). If transformation at any given temperature isassumed to be a function of the amount of transformation andthe temperature, i.e., the additivity condition, then the transfor-mation behavior can be divided into a set of isothermal events.In this regard, Avrami equation can be generalized as Table 2 Parameters used in the physically-based micro-structure model (Ref  25) Parameter Value  A /   M  GB  5 9 10 9 C  d  2.6 9 10  6  D 0  50  l m b  2.86 9 10  10 m q 0  10 11 m  2 1 0  10  6 m h 0  0  e f  7 9 10  10 Z e h  5 9 10  10 Z 1/4 1/   R g  5 9 10  4 m  1 C  1  3 9 10 7 m  1 C  2  2.09 9 10 12 Pa  1 s  1  D  2.347 2896—Volume 23(8) August 2014 Journal of Materials Engineering and Performance   X  v  ¼  1    exp   0 : 693  ww 50   n st     ð Eq 15 Þ w 50  ¼  t  0 : 5  exp   Q rex  RT    ;  w  ¼ X D t  i  exp   Q rex  RT  i   ; ð Eq 16 Þ where,  t  i  is the time interval during which the slab has a tem- perature of   T  i . Obviously, the above model should be coupledwith the thermal model for determining the time-temperature behavior of cooling strip at the same time. It should be notedthat the rate of cooling after hot rolling and the parameter   t  50 at different temperatures were calculated employing theresults of the finite element model and the above-mentioneddislocation model, and then, another code in MATLAB wasgenerated to evaluate the kinetics of recrystallization basedon the additivity condition described by Eq 15 and 16. Further- more, the constants in the computations of microstructuralmodeling are given in Table 2 4. Materials and Experiments The aluminum alloy, AA2017, with the chemical compo-sition of 0.68%Si, 4.18%Cu, 0.79%Mn, 0.26%Fe, 0.52%Mg,and balance Al (in wt%), was studied in this work. The flowstress behavior of this alloy was defined as a function of thechemical compositions, strain, strain rate, and temperature based on the equation developed in Ref  28. A series of hot-rolling experiments were performed under the rolling parameters summarized in Table 3, and the work-rolldiameter was 150 mm. Specimens with the dimensions of with 5 9 60 9 250(mm 3 ) were heated at 480   C for 1 h andthen rolled. In some samples, a 1.5-mm-diameter hole wasdrilled on one side of the test specimens, while a K-typethermocouple was embedded in the drilled hole to recordtemperature history during and after hot rolling. Additionalhot-rolling experiments, the so-called rex1 samples, were alsoconducted to determine recrystallization progress after hot rolling in which the hot-rolled strips with a rolling speed of 50 rpm and a reduction of 15% were water quenched after 10,17, and 30 s of air-cooling.Optical microscopy and scanning electron microscopy(SEM) were performed on the hot-rolled samples. The speci-mens were mechanically grounded and chemically polishedusing modified Keller   s Reagent containing 6 mL HF, 3 mLHNO 3 , and 150 mL water. In addition, microanalysis wasconducted using energy-dispersive spectroscopy (EDX). Theaverage recrystallized grain size was measured using a linear intercept method according to ASTM E112-96 at two regions,i.e., center and surface of the rolled strips. For each sample, fivemicrographs were taken along the rolling direction while at least 500 grains were analyzed in each set of experiments. Sincethe grain structure is not equiaxed but elongated, the grain sizehas been evaluated on two of the three principal planes, i.e.,longitudinal and transverse surfaces. Then, an average valuewas reported as the mean grain size, and the sample standarddeviation was reported as the amount of error in the measure-ments. Furthermore, hardness testing was selected to examinethe kinetics of softening within the rolled samples according tothe following equation (Ref  29):  X   ¼  H  m    H  t   H  m    H  0 ð Eq 17 Þ where,  X   represents the softening fraction;  H  m  and  H  0  are theas-rolled and fully annealed hardness values of the material,respectively; and  H  t   is the hardness value of the aged samplefor aging duration of ‘‘ t  ’’.Tensile and hardness tests were performed on the as-rolledsamples to examine the mechanical behavior of the rolledsamples. The rolled samples were then aged at room temper-ature, and hardness evaluations were repeatedly conductedthroughout the aging period. Tensile tests were also carried out after 57 days. The tensile samples were prepared from thecentral part of the rolled strips according to ASTM-E8M.Tensile tests were carried out at room temperature at constant cross-head speed of 2 mm/min. Hardness values were alsomeasured using Vickers hardness testing with 30 kg for 15 s. Note that in the case of hardness tests, at least six indentationswere conducted for each sample, and an average value wasdefined for each set of data, and the sample  s standard deviationwas reported as the associated error. 5. Results and Discussion In the first stage, the validation of the thermomechanicalmodel was performed by comparing the predicted and therecorded temperature variations. Figure 2 compares the pre-dicted temperature variations and the measured temperaturehistory for sample 5 at the depth of 1/4th of thickness. As can be seen, there is a good agreement between the two sets of data,which indicates the validity of the employed thermomechanicalmodel. Moreover, Fig. 3 shows the predicted temperaturevariations in samples 1, 3, and 5 that were hot rolled under different rolling conditions, i.e., single-pass rolling withdifferent rolling speeds of 30 and 50 rpm and reductions of 30 and 40%. As seen in this figure, in all the samples,significant temperature drop occurs in the roll/strip interface,while for the rolling speed of 50 rpm, a trivial temperature riseis observed at the central region owing to heat of deformation.However, after hot rolling, heat conduction from center tosurface takes place to reduce the temperature gradient alongthickness direction. In addition, Fig. 4 shows the predictedeffective plastic strain distributions for the above mentionedsamples. It can be seen that the equivalent plastic strain Table 3 Rolling conditions used in the hot-rolling experi-ments Specimenno.Reductionin 1stpass (%)Reductionin 2ndpass (%)Rollingspeed(rpm)Inter-passtime (s)Quenchingmedia 1 30  …  50  …  Air 2 30  …  50  …  Water 3 30  …  30  …  Air 4 30  …  30  …  Water 5 40  …  50  …  Air 6 15 15 50 17 Air 7 15 15 50 17 Water 8 15 15 30 17 Air 9 15 15 30 17 Water  Journal of Materials Engineering and Performance Volume 23(8) August 2014—2897  increases from the center to the surface of the strip due to theshear strains at the surface region of the strip (Ref  30).As is observed in Fig. 3 and 4, process parameters can change both temperature and strain distributions significantly.For instance, rolling speed is an important parameter that changes the strain rate, flow stress, roll force, and heat of deformation. Figure 5(a) shows the effect of this parameter onthe surface temperature at the exit position of deformation zone.Rolling speed affects flow stress of deforming alloy and contact time between the work-rolls and the strip, while adiabaticheating may occur at high rolling speeds, e.g., rolling speed of 50 rpm in this study. The consequence of these factors reducesthe temperature drop in the deformation zone in high-speedrolling as shown in Fig. 5(a) and 3(a), (b). The amount of  reduction also affects the temperature distribution within thestrip rolling. The effect of reduction on surface temperature of the rolled strip is given in Fig. 5(b) in which the initialtemperature of 480   C and rolling speed of 50 rpm were takenin the modeling. As seen, higher reduction leads to larger temperature drop at the surface region due to longer contact time with the work-rolls; on the other hand, heat of deformationis also increased for higher reductions. However, the contact time and cooling effect of work-rolls seem to be the dominant factors (Ref  31). Furthermore, Fig. 6 shows the predicted strain distribution along the thickness direction of the rolled sample.As the rolling speed increases from 30 to 50 rpm, thedistribution of effective strain is altered. This may be attributedto the change in frictional stress at the contact region.Furthermore, the shorter contact time in high-speed rollingmay lead to adiabatic heating in central region of rolling metalas shown in Fig. 3. Accordingly, it can affect the flow stress behavior of deforming alloy as well as the strain field asdisplayed in Fig. 4(a), (b).Progress of static recrystallization and recrystallized grainsize after hot rolling can also be predicted by the model.Figure 7(a) compares the predicted and the measured recrys-tallized fractions at the surface of samples ‘‘rex1.’’ Figure 7(b)shows the predicted recrystallized grain size and the measuredmean grain size for this sample. In addition, Fig. 8 shows grainsize images taken from center and surface of the sample. It can Fig. 2  Comparison between the predicted and the measured tem- perature for sample 5 Fig. 3  Predicted temperature distributions for different hot-rollingconditions: (a) sample 1, (b) sample 3, (c) sample 5 Fig. 4  Predicted effective strain distributions under different hot-rolling conditions: (a) sample 1, (b) sample 3, (c) sample 5 2898—Volume 23(8) August 2014 Journal of Materials Engineering and Performance
Search
Similar documents
View more...
Tags
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks