Materials Science Research International, Vol.10, No.1 pp.1-11 (2004) Review paper Macro-, Meso- and Micro-scopic Metallo-thermo-mechanics — Application to Phase Transformation Incorporating Process Simulation — Tatsuo INOUE Department of Mechanical Systems Engineering, Fukuyama University Gakuen-cho 1-Sanzo, Fukuyama 729-0292, Hiroshima, Japan Abstract. Three kinds of approach from macro-, meso- and micro-scopic viewpoints are summarized, inthis paper, relevant to simulating the phase transformation incorporating processes. Since the fields of material structure, ‘temperature and siress/strain induced in a body in such process are coupled each other, the ordinal way to solve {independent (or, uncoupled) governing equations for transformation kinetics, heat conduction and stress analysis is ‘insufficient, but thermodynamics-based consideration is needed to obtain the coupled equations among three fields. ‘A transformation parameter indicating. progressive phase change is introduced as one of internal parameters ‘representing volume fraction of each phase or phase field parameter, and governing three kinds of equation for the [Parameter, Fourier law and stress-strain constitutive equation are derived. Thus obtained equations are applied to ‘Formulate’ the macroscopic finite element scheme and also the phase field method in meso-scopic sense in the Framework of continuum thermodynamics, "The molecular dynamics approach is also carried out to evaluate ‘microscopic, of physical aspect of the fields. Some examples of the computer simulated processes with phase ‘transformation are illustrated based on such three kinds of approach, Keywords: Metalo-thermo-mechanics, Phase transformation, Finite element method, Phase field method, Molecular dynamics method. 1. INTRODUCTION Fields of metallic structure, temperature and stress and/or strain induced under phase transformation sometimes affect cach other in engineering processes such as quenching, welding, casting and so on, Outer triangle in Fig. 1 ‘lstrates the schematic representation of the nefallo-thermo-mechanical coupling with the induced phenomena [1-6]. When the temperature distribution in a material varies, thermal stress is caused inthe body, and the induced phase transformation affects the structural distribution, which is vell knowns during melting or solidification in sotid-iqui a transition and austenite-pearlite or -martensite transform- ation in solid phase. Local dilatation due to structural ‘changes in the body brings out the transformation stress and interrupts the stress or strain field in the body, and the effect of transformation plasticity [7-15] plays a dominant role in the stress/strain constitutive equation. In contrast to these phenomena, which are well known in ordinal analysis, arrows in the opposite direction indicate coupling in the following manner. A. part of the ‘mechanical work done by the existing stress in the material is converted into heat, which is sometimes predominant in the case of inelastic deformation, thus disturbing the temperature distribution, ‘STRUCTURE CHANGE. ‘Macro- and micro-scopic representation of metallo-thermo-mechanics, Fig.l. Received July 28, 2003 Accepted January 28, 2004Tatsuo INovE Macroscopic feld = Finite element method me (A Mesoscopic field = Phase feld method @ Liquid Solid Liquia - / ole MoM Microscopic field = Molecular dynamics method erode boundary (x and z drectons ) rice ayer opt ayer Meting region Fig2 Strategy of three kinds of metallo- thermo-mechanical approach. Free aura ‘The acceleration of phase transformation by stress or strain, which is called stress- or strain-induced trans- formation, has been investigated by metallurgists as one of leading parameters of transformation kinetics. The ‘opposite arrow corresponds to the latent heat due to phase transformation, which is essential in determining the temperature. When no coupling among the fields is taken into account, independent analyses of heat conduction, stress and transformation kinetics are possible to be carried out as shown in the outer circles. In such case of coupled issue, however, global formulation based on continuum thermodynamics [16] is needed, which will be basically solved by the finite ‘element method from the viewpoint of macroscopic metallo-thermo-mechanics. The top figure of Fig.2 indicates a summary of strategy of this macroscopic approach. On the contrary to the macroscopic approach, some successful results are obtained by use of the phase field method. As is reviewed by Sekerka [17], the method was initiated carly in 1990s [18-25], based on the diffusion interface model [26-28] of crystallization from melt with thermodynamic. consideration of interfacial free energy between solid and liquid. Since the method mainly treats the initiation of crystal nuclei and successive growth of dendrite in small region, the author regards it as a mesoscopic approach comparing to ‘macroscopic one stated above. Bearing in mind that few papers by the phase field ‘method discuss the mechanical field of stress and strain [29,30], this paper motivates to formulate fundamental equations demonstrating the coupled fields of Temperature, stress-strain and phase transformation Here, a general type of phase transformation parameter is defined to interpret the volume fraction of phases in ‘macroscopic sense and also phase field parameter appearing in the phase field model, called mesoscopic metallo-thermo-mechanics as is illustrated in the second figure in Fig.2. If we focus the motion of atoms or molecules composing the continuum, however, all three fields are identified by solving Newton equations of motion for each atom, since the structural change depends on the configuration, temperature is the vibration itself and stressistrain is related t0 the interatomic distance as is shown by the inner illustration in. Fig.l. Molecular dynamics method [31-34], or in the other case, Monte-Carlo method [35], is one of the most promising techniques to emit a light especially to such kind of coupled metallo-thermo-mechanical problems. In this stage of the development of computer hardware, there are a lot of restrictions on the number of atoms to be treated in the model and the time executed, and the identification of potential suitable for the material and temperature in spite of successful achievement of the finite element implementation. Nevertheless, the way to solve the problem is here called microscopic, or nanoscopic: metallo-thermo-mechanical approach, sche ‘matically illustrated in the bottom figure in Fig2 Some examples of the simulated results for typical ‘engineering processes will be presented in the order of ‘maero-, meso- and micro-scopic approaches with some discussions CONTINUUM THERMODYNAMICS FOR INELASTIC BODY INCORPORATING PHASE, ‘TRANSFORMATION A parameter representing the structural change due to phase transformation is generally denoted by y_ termed as transformation parameter, which will be respectively reduced to a volume fraction Z, of each constituent to represent global filed in macroscopic simulation and a phase field parameter_@ in the phase field method. ‘The thermodynamic state of @ material is assumed to be determined by stress g, » temperature T and temperature gradient g, (= glad 7) as external state variables, and a set of internal state variables s of inelastic strain {as well as other internal (scalar or tensor) variables fepresenting back stress and hardening parameter related to inelastic deformation together with the parameter y. Then, the general form of the constitutive equation for Gibbs free energy function G, entropy density 77, heat flux hand recoverable elastic strain ef denoted by @ is expressed as, = 0(6,,T.2,.5¥) + a ‘The evolution equation for the internal variables, such as inelastic strain ',, other variables s and_y, is defined in the equipresenting form as Eq. (1), ie 9=46,,1.2,.5.V38 108,050) @ ‘The second law of thermodynamics is nowMacro-, meso- and mic represented such that the total entropy production of the entire domain with volume V and surface S with ‘outer normal n is nonnegative: rs [nav [Meas av +f kewspar 20> @) where hy and 7 respectively denote the heat flux and heat generation, which is simply represented by the latent heat 1 for progressive change of yin the form y=. ‘The fourth term in Eq. (3) denotes an Additional entropy production on the interface between ‘wo phases [17], which is to be normally neglected in case of in-solid transformation using the volume fraction defined, in Sec. 3, and K(y) will be regarded as K(y)=mid?yl@xax,)_ with a positive constant m in the phase-field model in Sec-4. Then, we have the Clausius-Duhem inequality, of local form of the second law of thermodynamics, as toexonl2(4) er 20 @ Applying the Legendre transformation Go,.7.8,5¥)=Ulej.n8.5¥)-o,6,-Tn. (3) to the local form of energy balance equation indicating the first law of thermodynamics 5 ipyao sO Pl = poe, $+ pr =0 we have the alternative form of Eq. (4) such that [aD rlo pon “AE, (oi 22pe From the restriction that Eq, (7) holds forall possible processes as well as the energy balance equation (6), then the following series of equations are obtained with 6=G(o,.T.s): Suess-eiastie strain constitutive law; 7 &) o (00) with the heat conductivity k , and scopic metallo-thermo-mechanics Phase transformation kinetics; aG TK(y)-=— ay are ay with a positive constant and aG -o,¢,-2Gos20 12) o,44,-%2 ) Here, the symbol “e™ denotes the scalar product of any Scalar, vector or tensor. Equation (11) corresponds to the phase field equation, when the parameter ys chosen as the phase field 4, Following the conventional way of the phise-field model, the function x(q) usually takes mmd'y xa). Then, the phase-eld equation ay ) ac ‘= Tm Eee ee (13) ¥ (2x) ov . is derived. When K( is neglected, Eq, (11) becomes 7 co) hich indicates the direction of the phase transformation Such that increases when OG/Oy is negative (ue. Gibbs free energy decreases as y increases), while y decreases for positive aG/ay. This is consistent with the fundamental concept of the conventional thermodynamics Substting the Fourie law (10) and the defnion of the internal energy including an additional term into ‘energy balance equation (6) leads to prs a8 (1) core Kien: as) where K,(y) isa function in similar form as K(y). 3. FINITE ELEMENT SIMULATION = MACRO- ‘SCOPIC METALLO-THERMO-MECHANICS. 3.1. Mechanical Constitutive Equation Application of the derived fundamental equations are made in this section to the macroscopic simulation incorporating phase transformation. In most cases of phase transformation occurring in the process of, quenching and casting, several constituents are induced to compose a material point so as to assume that the ‘material point is a mixture of IV kinds of constituents, which is chosen in place of the transformation parameter ¥ . Denoting the volume fraction of the / th constituent as g,, a thermophysical or mechanical property z of the material are assumed to be governed by the mixture law [36], or a linear combination of the properties, of the /-th constituent as, 406, = EE), with EE, (16)Tatsuo IvouE where EM, = is the summation for suffix! from 1 to 'N. All material parameters appearing below are defined in the manner of Ea. (16). To obtain an explicit expression forthe elastic strain in Eq. (8), the Gibbs free-energy G is supposed to be determined by that of constituent G, in the same form of Eq. (16) as, Glo,,7,8,8,)=24,G,(0,,7,s) an When G; is assumed to be divided into the elastic and inclatie pats as G,(o,,7.¢),8)=Gi(o,,7)+G;(T,¢).s)) (18) we can derive the elastic strain from Eq. (8) by expanding elastic part Gf around the natural state, 0, =0 and T=T, , in terms of the representation theorem for an isotropic function; oes ezeia om +5, if Ea,§dT + 5,EB/(6, ~$0) Here, E; .¥) , a; and, correspond to Young's modulus, Poisson's ratio, thermal expansion coefficient and dilatation ofthe J th constituent, respectively. Evolution equation (2) for inelastic strain &¢ and transformation plastic strain rate é? is summarized to obtain total strain rate +S el sap, 20) where inelastic (plastic, creep and/or viscoplastic) strain rate is generally denoted by en with the potential function like F = F(o,,€%,x,7,é,) in the case of plasticity. The transformation plastic strain rate usually takes the form [7], af 30K, (1-8 Es, 22) 3.2, Heat Conduction Equation Applying the Legendre transformation to Gibbs free ‘energy function in the form of Eq. (5), energy balance ‘equation (15) by neglecting the term of K is reduced to the equation of heat conduction; ae, nr-A(0Z)> one 38. with emthelpy density 1 and latent heat /, due 10 the increase of the /-th phase a as, ‘The fifth term on the left-hand side of Eq. (23) denotes the heat generation by inelastic dissipation, which is significant when compared with the clastic ‘work represented by the fourth term, and the third term arises from the latent heat through phase change. Since the effect of heat generation due to elastic work and other terms related to the enthalpy appearing in Eq.(23) is trivial compared to those of latent heat and inelastic work, practical numerical calculation will be carried out by applying thus simplified equation | (24) 3.3. Kinetics of Phase Transformation During phase transformation, a given volume of material is assumed to be composed of several kinds of constituent ¢, as expressed in Eq. (16). We choose four kinds of volume fraction; liquid, , austenite ¢, , pearlite Ep and martensite y, and other structures induced by precipitation by recovery effect, say, during annealing process. When austenite is cooled in ‘equilibrium, bainite, ferite and carbide are produced in addition to’ pearlite, for simplicity, but for brevity all these structures resulting from a diffusion type of transformation are called as pearlite. The nucleation and growth of pearlite in an austenitic structure are phenomenologically governed by the mechanism for a diffusion process, and Johnson and Mehl [37] postulated a formula for volume fraction pas [4-6] Sealey.) 5) where V, means the extended volume of the pearlite structure generally given by a function of stress as well as temperature, v= [lr.,)e-2)' ar: 26) Here, the function /(7,0,) can be determined by fitting the popularly” used. temperature-time-transform- ation (TTT) diagram or continuous-cooling transform- ation (CCT) diagram depending on stress. [38-41]. ‘The empitical relationship forthe austenite-martensite transformation is. also obtainable by modifying the kinetic theory of Magee [42] to the form; Su =l-explalr-M,)+4(0))] 27) The function #4(@,) is identified by the data subjected to applied stress [43-45]‘Macro-, meso- and micro-scopie metallo-thermo-mechanies In the case of solidification, we employ the well-known lever rule, and the volume fraction of austenite is fe (-Tm, es) 5 Fan, TV Mm, where T, and Ts denote the liquidus and solidus temperature, respectively, and m, and ms are gradients of the liquidus and solidus temperature with respect to the carbon content in the phase diagram. 4, PHASE FIELD SIMULATION = MESOSCOPIC METALLO-THERMO-MECHANICS ‘The phase-field model has been used to simulate the complicated pattern formation processes during the crystal growth, and successfully obtained various patterns of dendrites. ‘The leading parameter used in the phase field model is the phase filed parameter 9, which varies from 0 for liquid to 1 for solid corresponding to the progressive solidification from liquid state. This value is not always the volume fraction of solid at given material point in previous section, but is related to & such as @=0(E), and is assumed to be regarded as the transformation parameter y_ defined in Sec. 2. Then the governing equations for mechanical, thermal and phase fields are respectively expressed by Eqs. (8), (15) and (13). The problem is now how to identify the free ‘energy G appearing in these equations. In the simple case of phase transformation from phase ‘Ato B in an clastic body, we assume that G takes the form G=Gle,,7.6)=[1- 20)KG.(0,.7}+ rl0XGe(0,.7) 1 + Q9) +h. 0(@) which means that the combination of first two terms indicate the same form of mixture law (16) if p(@)=0 ‘when neglecting the extra third term Substituting Eq. (29) info Eq (13), we finally have 612t7)-P oem en)" 016) G0) aan atr)= [i pat as a phase field equation. And the stress-strain constitutive equation for linear elastic body in this first tral simply reads, Ma, L oud, all To}, + 018)B9,- (31) where mechanical and thermophysical material para- meter 7 like Ev Ja is assumed to be expressed as, roll-nles + ree G2) Equations (30) and (31) as well as Eq. (15) govern the framework of phase field method considering. meso- scopic metallo-thermo-mechanical coupling, 5, MOLECULAR DYNAMICS SIMULATION =MICROSCOPIC METALLO-THERMO- MECHANICS So-called classical molecular dynamics method is based on solving the Newton equation of motion for each atoms with a force derived from an empirical interatomic potential ¥ . Then the position of the /-th atom x is. given by integrating the following equation of motion, mini = 0-2, 63) where mis the mass of the /-th atom, and /(? is the force working on the atom. When the location and velocity of each atom are identified, temperature T and stress, and pressure p as wel as internal energy U are written in the forms, 2 1 o a Tami, G4) aT Pts Oe Wee DTT ser") ay © i with Myf) a) p H[mareizaa”), (36) uaosStmiyingy + on and the evolution of phase change is represented by the configuration of atoms, Here, v;” is the velocity of the Fh atom, 1V total number of atoms in the system, ky Boltzmann constant and V volume of the system. Three kinds of theory and the governing equations developed in the previous sections 2, 3 and 4 for macro-, ‘meso- and micro-scopic metallo-thermo-mechanics are now applied to some processes incorporating phase transformation. Results of simulation carried out by the group of the present author are simply illustrated without detail explanation. Readers are recommended to refer the corresponding titerature.‘Tatsuo IvouR 6, RESULTS OF MACROSCOPIC SIMULATION BY THE FINITE ELEMENT METHOD The author and his group has accumulated finite clement programs to simulate some kinds of processes such as quenching and tempering with in-sold phase change and welding and casting incorporating solidification. The program HEARTS (HEAt tRetment Simulation) [46-48] was developed in 1991 mainly applied to quenching process to evaluate temperature, slresses and structure change as well as diffusion of carbon applied in advance to the quenching, and hardness evaluation by the cooperation with CRC Resarch Institute, Co. Recently, a new program COSMAP (COmputer System for MAterial Processing) with higher quality and wider field of application is developing by the author's group [49]. There are some other programs relevant to finite clement simulation of metallo-thermo-mechanical fields such as GRANTAS (Komatsu, Japan) [50], QUESS (Sumitomo Metals, Japan) [51], SYSWELD(SYSTUS, France) 52], DEFORM-HT(SFTC, USA)_ [53,54], DANTE(Deformation Control Tech. USA) [55-57] and 0 on, which the readers are recommended to visit their websites So many data are necessary to be propounded for the practical simulation of some machine parts made of different material depending on temperature, and the transfer of these data to simulation code is laborious and time consuming job. To reduce the cost and difficulty of measurement or collection and handling of necessary data, the Sub-committee on Material Database under the Committee on Engineering Plasticity, the Society of Material Science, Japan, has conducted a joint project to construct a database for the simulation for seven years. ‘The developed database tentatively called "MATEQ” -MATErial data for Quenching) [58] contains the stress -strain constitutive data in elastic and inelastic range as mechanical property, thermal expansion coefficient, heat conductivity and specific heat as thermephysical prop- erties, and isothermal and continuous cooling diagram as data related to phase transformation. More than 50 kinds of steels ate covered for each property. All data are recorded as digital values on spread sheets with the figures. 6.1. Quenching of a Gear Whee! By use of the data as well as the cooling characteristics data such as heat transfer coefficient of steels by many kind of coolants, some simulations are cartied out as a cooperative project of Japanese consortium of VHT (http:/wvww.ims.mste.or jplenglish! index3.html). One of the example of the task is to simulate a helical gear wheel of SCR420 steel under the condition of 930Cx200min for carbonization and 850°C x30min for diffusion followed by quenching into an oil quenchant of SOC. D.Y Ju, et al [49] carried out @ sophisticated simulation by a model with 8436 nodes and 5700 elements. Figures 3(a), (b) and (c) respectively illustrate simulated results of sintered carbon content, ‘martensite fraction and equivalent stress, where the left figures are on the surface of the gear and right ones on the middle section. Validity of the simulation will be confirmed through the project by comparing with the experimental data. 62. Quenching of Japanese Sword ‘The second example of the quenching process simulation is for Japanese sword [59-61]. As is known, quenching is applied atthe very end of forging operation, which produce the hard blade with martensite and soft ridge with pearlite. Before quenching. the Japanese sword into water, the yakiba-tsuchi clay is coated on the surface of the sword to control the cooling condition of the surface ofthe steel. Since the temperature disti- 070s eae osn ase Surface Section (a) Sintered carbon content oO Surface Section (b) Martensite fraction vee Surface (© Equivalent stress Fig3. A helical gear under carbonized quenching; Simulated results are shown on surface (left) and ‘middle section (right) -- by D.Y. Ju, etal‘Macro-, meso- and micro-scopic metallo-thermo-mechanies bution is tobe calculated in the body ofthe sword, use is made of the data of heat transfer coefficient depending an the thrickness of pasted clay [59] “The s word treated here is 300 mm in length with 7 mmm in raaximum width, The sword is uniformly heated up to 850, at which temperature ofthe whole region is changed into susteitic structure, and the sword is quencheed into the water of 40°C. Figure 4 (a) demonstrates the temperature distribution ofthe surface af the sword with successive time from the beginning of the querrching, and the mode of deformation is. also depicted in the figure. The part of blade with thin thickness: shrinks due to thermal contraction by cooing, which leads to the bending to the downvard termed as ‘gaku-sorri or reverse bending at 0.35s as is shown in the figure. Figures 4(b) and (c) represen the change in calculated distribution of martensite and pearlit faction sssociated with deformation. When martensitic trans- (a1) 05 (2.2)035s —_———_— (2:3) 1.308 (b-3) 1.308 ———__—> ee 4)2.308 (b-4) 2.305 we SS (2-5) 2008 (b-5) 200 peel (a) Terperature 7,°C _ (b) Longitudinal stress 6, MPs (c-1) 0s (d-1) 0s EE (c-2)0.35s (4-2) 0.358 SSS (63) 1.305 (4-3)1.308 EE (¢-4) 2,308 (4-4) 2.308 (e-5) 2008 (d-5) 200s | | (© Volume fraction of pearlite ¢p (@) Volume fraction of martensite fy Fig. Successive variation of temperature, fractions of ‘martensite and pearl, and longitudinal stress in Japanese sword. formation occurs in that part, however, normal bending called sori to the upper direction is observed due to the volumetric dilatation by martensite formation (see figure (a-3)). Gyaku-sori again appears at =2.3s, because of the pearlite transformation in the part of ridge seen in Fig (c-4). In the successive stage of cooling, hot ridge side shrinks gradually because of thermal contraction, and finally, the normal bending can be obtained. Thus simulated deformation gives good agreement with the actual bending mode of sori. Stress distribution in the longitudinal direction in the course of quenching is represented in Fig4 (4-5). This data of simulated residual stresses after complete cooling are compared with measured data by X-ray diffraction technique, and satisfactory coincidence is obtained. It is also noted that the maximum stress near the tip of the sword during. quenching reaches the fracture stress, which sometimes. leads to cracking or breakage of the sword during the ‘operation. 800 /< “| 600mm/e “yl ° (a) Temperature (b) Fraction of solid 16g omm/9 (@) Velocity field (©) Horizontal stress Fig 5.Thin slab continuous casting by twin roll method.‘Tatsuo INovE 6.3. Strip Casting by Twin Roll Method ‘A viscoplastic constitutive model available to describe the material behavior of inelastic solid and viscous fluid proposed by the authors [62,63] with the condition of melting is applied to the simulation for coupled solidification and viscoplastic stress in the strip continuous casting of SUS-304 stainless steel by the twin roll method [64,65] Figure 5 (a) represents the spatial temperature distribution simulated by the finite element method in the whole region of the strip and roll under steady operating, condition, in which temperature varies due to cooling by ‘water and heating by the strip. The casting speeds of the slab are chosen to be ¥=400-1500mmi/s. The initial temperatures. of molten ‘metal and rolls are 1470°C and 20°C, respectively. Simulated results of steady temperature both in the strip and roll are shown in Fig,5(a), and Figs. (b, (c) and (d) respectively show the fraction of solid, stress in horizontal direction and velocity field. 7. RESULTS OF MESOSCOPIC SIMULATION BY THE PHASE FIELD METHOD. Some examples of phase field simulation are presented in this chapter. The author is happy to succeed in the simulation of coupled issue of mechanical field with phase change and temperature based on the theory developed in Sec. 4 very recently in spite of the first step of elasticity, which we expect to extend to plasticity and also viscoplasticity 7.1. Asymmetric Phase Transformation of a Plate ‘A heated square plate with constrained boundaries in vertical direction and thermally insulated ones is cooled from the outer periphery. Due to the symmetry of the ‘model, a quarter part in the lower right is focused To make clear the coupling effects, a nucleus of the second phase is placed at the bottom center of the quarter part Distribution of the phase, temperature and horizontal normal stress at ¢ = 14s are respectively represented in Fig.6 [65], where the equi-phase field line of melting Tiguid phase is plotted in the figures. The molten state indicated by @= 1 is seen to propagate from the nuclei (sce Fig. (a)) depending on the temperature profile in Fig, (b). Due to the latent heat generation, temperature reveals to be violated near the lower right part of the plate, Compressive stress is induced around the location of the second phase nucleation owing to the phase change dilatation 7.2 Dendrite Growth of 2D and 3D Crystal Another example of phase field simulation is the pattern formation of dendrites in solidification process. Figures 7a) and (b) are two- and three-dimensional dendrites, respectively [67,68]. In both cases, a small nucleus of crystal is placed in a supercooled liquid, Since the temperature is below the melting point, the solid Ae ‘ TK o.MPa (a) Phases, _(b)Temperature. (c) Horizontal stress. Fig. 6 Two dimensional finite element simulation based ‘on the mesoscopic metallo-thermo-mechanics by using the phase field model latent heat generation, the temperature on the interface increases as the liquid solidifies, which brings out instability on the interface, and a small tip appear there. Once a tip is formed, it grows faster than other points, since the tip is surrounded by the supercooled liquid, and finally dendritic pattern is formed as shown in Fig.7. The shape of dendrite obtained depends on the parameter, which is a litte difficult to evaluate experimentally ‘Therefore, some approaches to evaluate the parameters by using molecular dynamics. simulations have been ‘carried out recently [69,70], In this simulation, we used the phase field equation coupled with only the temperature field, and the effects of stresvstrain are neglected, since the effect of the dilatation due to solidification is released toward liquid. However, some amount of stress is expected to occur in such a small region of dendrite branch, and stress or surface tension at the interface will affect the phase transformation behavior. This motivates that the formulation including stresses coupled with temperature ‘and phase change is treated in this paper, which will be applied to further simulations. 8, RESULTS OF MICROSCOPIC SIMULATION BY THE MOLECULAR DYNAMICS METHOD. The ability of computer being needed for the ‘molecular dynamics simulation forces to restrict the number of atoms focused and also the time period treated. Nevertheless, the results of MD simulation gives a lot of (20 (b)3-D Fig. 7 Dendrite growth in two and three dimensional phase field simulation,Macro-, meso- and micro-scopic metallo-thermo-mechanics characteristic information, especially in nanoscopic viewpoint of material science. Further development of powerful computer in future will possibly give a strong tool to us. Here, only two examples of MD simulation are presented being applied to phase transformation incorporated issues 8.1. A Bar Constrained at Both End and Cooled by the Center: A simple method of molecular dynamics is adopted to «bar illustrated in Fig.8 [71] (actually, periodic boundary condition is applied in vertical direction) where mechanically and thermally constrained at both end with heat input in the middle part. A two-body interatomic potential of Morse type is employed in place of ¥ Material parameters such as phase change temperature in heating and cooling process, heat conductivity, and volumetric dilatation are identified by also uniform MD simulation in advance. It_is. seen from Fig.8 (a) indicating the change in distribution of potential energy when heat is supplied at the central portion [71]. ‘The temperature and induced thermal stress are plotted in upper and lower charts in Fig.8 (b), respectively. Stress vanishes in molten area with compressive stress at the periphery, while tensile stress is observed neat the ends at lower temperature. 8.2. Shape Memory Effect ‘The final demonstration is on the characteristics of shape memory mechanism for Ni-Ti alloy. Figure 9 (a) shows the typical process of shear loading and unloading. Resonant (a) Configuration of atoms and (b) Temperature and change of potential energy thermal stress Fig. 8. Distribution of atomic position and potential energy, and temperature and thermal stress in a bar constrained at both ends with heat supply at the center. followed by heating and cooling as an illustrative example of MD simulation for small number of atoms [72]. Calculated stress-strain hysteresis by evaluating stress by Eq. (35) and strain evaluated from shear angle and volumetric dilatation is depicted in Fig. (b). Zig-zag lines are observed since the number of atoms is rather small. Nevertheless, a typical pattern of SM effect is found to be realized. 9. CONCLUDING REMARKS The summary ofthe macro-, meso~ and micro-scopic ‘metallo-hermo-mechanics, MMM-MTM, by the present author and his group is reviewed. Emphasis is placed fon the derivation of unified thermodynamics. theory relevant to describing the governing equations for finite ¢lement implementations and for phase field applications Some results of computer simulation of engineering processes incorporating phase transformation mainly coupled with mechanical and thermal field are iMlustrated. ‘There are some restrictions on the theory and simula- tion in this stage of development due to the capability of computer, but nevertheless this kind of approach will Initial Loading Unloading Heating Cooling Martensite Deformed Elastic ‘martensite recovery to bee ‘Transform. Transform. to martensite (a) Change of configuration (b) Hysteresis loop, Fig9. Configuration change due to shape memory effect and stress-strain hysteresis loop.‘Tatsuo INOUE open the windows for phase transformation incorporating technology. in the wide range from mucro- 10 micro-scopic viewpoints. Acknowledgment: The author expresses his sincere gratitude to Dr. T. Uehara, Kyoto University, for instructive discussions and ‘careful review on. the thermodynamics theory derived in this paper. The data of the simulated examples are quoted from the works of the cooperated group with the author, especially with Dr T, Uchara, Prof. D.Y Ju, Saitama Institue of Technology, and my graduate students from Kyoto University; Mr. T. ‘Ohtsuka, Mr, M. Fukuda, Mr. N. Masago, Mr. T. Tsujino and others. "Sincere gratitude is for Profs. K. Tanaka and S. Nagaki, S. Imatani, B. Raniecki and Dr. Z.G ‘Wang for long term cooperation with the author in the present field. REFERENCES. 1. Inoue, 8. Nagaki, T. Kishino and M. Monkawa, Ingenicur-Archiv, Vol. 50,No. 5 (1981) pp. 315-327 2. T. Inoue and Z.G. Wang, Material Science and ‘Technology, Vol. (1985-10) pp. 845-850 3. T Inoue, Berg und Huttenmannische Monatsheft, Vol.132, No.3, 1987, _pp.63-71 T. inoue, ‘Thermal’ Stresses, Vol3 (Ed. by Richard B.Hetnarski), 1988, North-Holland 5, T.Inoue, Handbook of Materials Behavior Modes (ed. J Lemaitre) Chap. 9-11, 2002, Academic Press) 6, G Totten, M. Howes and T. Inoue, Handbook of Residual Stress and Deformation in Steel, 2002, ASM International) 7. GW. Greenwood and RH. Johnson, Proc, Roy, Soc, London, 283A, 403 (1965), pp.403-422 8, JB. Leblond, Int. . Plast, 5, (1989), pp.573 9. FD. Fischer, Eur J. Mech,, A/Solids, 11, 2 (1992), pp.233-244 S! Denis, E. Gautier and A. Simon, Intl, Conf. on Residual Stresses, ICRS2 (1991) , pp.393-398 CL Magee, in HL" Aaronson (ed), Phase ‘Transformations, AMS, Metals Park (1968), pp. 115-156 PH. Shipway and H.K.DH. Bhadeshia, Matenal Science and Technology, A201 (1995),_pp.143-149 T. Inoue, Z.G. Wang and K. Miyao, Proc. IUTAM. ‘Symposium on Thermomechanical Coupling in. Solids, aris, pp.257-262 (1987-5), Elsevier Science Publishers, B.V., North-Holland ‘T. Otsuka and T. Inoue, Proc.13th Heat Treating Symposium, Intemational Federation of Heat Treatment Science and Engineering, Columbus, 2002-10 (to appear) ED, Fscher, QP. Sun and T. Tanaka, Appl. Mech, Reviews, 49, 317 (1996) GM, Maugin, The Termo-mechanics of Plasticity and Fracture, Press Syndicate ofthe Universuty of Cambridge, 1992 RE. Sekerka, Fundamentals of phase field theory, ‘Advancesin Ciystal Growth Research, ed. By K. Sato, Furukawa and K. Nakajima, Elsevier, Amsterdam (2001) , pp2I-4l R. Kobayashi, Bul. Japan Soc. Ind. Appl. Math, 1 (1991) 2. IR. Kobayashi, Physica D, 63(1993)__pp. 410-423. SIL. Wang and R-F. Sekerka, Phys. Rev., E53 (1996) 3760, | S.C. Wang and RF. Sekerka, J. Cono. Phys. 127(1996) 10, 10 2, 2 24, 25, 26, 2, 28, 2», 30 31 32. 33. 3s. 3s. 36. m1 38 39, 40. 4 2 8 46, 41, 48, 49, 50, SI 32 S.L, Wang, RLF Sekerka, A. A. Wheeler, B.T. Murray, S, R, Coriell, RJ. Braun and GB. McFadden, Physica D, 69 (1993), pp. 189-200. RE Sekerka, Fundamentals of phase field theory, ‘Advances in Crystal Growth Research, eds K. Sato, ¥. Furukawa and K: Nakajima, Elsevier, Amsterdam, (2001), pp. 21-41 ACA. Wheeler , WJ. Boettinger and GB, McFadden, Phys Rey. E47 (1993) 1893. 2. Bi and RF. Sekerka, Physica, A261 (1998) 95. A.W. Cahn, Acta Meta, 8 (1960) 554 Caginalp and PC. Fife, Phys. Rev., B 33(1986) 7792. G Caginalp, Arch. Rat. Mech. Anal., 92 (1986) 205 K. Kassner, C. Misbah, J. Muller, J. Kappey and P Kohlert, Phys. Rev. E, 63 (2001), 036117, pp.1-27 LO. Eastgate, JP. Seihna, M. Rauscher, 7. Cretegny, CS. ‘Chen and C.R. Myers, Phys. Rew. E, 65 (2002), 036117, 1-10. W.G Hoover, Molecular dynamies (Lecture notes in physies, Vol. 358), (1986), Springer-Verlag D.C. Rapaport, The art of molecular dynamics simulation, (1995), Cambridge University Press. B,J. Alder and T-E. Wainwright, J. Chem. Phys.,27 (1957), 1208 LL. Verlet, Computational Experiments on Classical Fluids. [.Thermodynamical Properties of Lennard-Jones Molecules, Phys. Rev. 159 (1967), pp.98-103, Y. Saito, Statistical physics of erystal growth, (1996), World Scientific. RM. Bowen, Theory of Mixture, A.C. Eringen, ed, Continuum Physics (Academic Press, New York, 1976, Vol. 3, pp. 2-129, W.A. Johnson and RIF. Mehl, Trans. AIME 135,1939,pp. 416-458 M. Fujita and M. Suzuki, Trans, ISI 14, (1974), pp. 44-53, S, Bhattacharyya and GL. Kehl, Trans. ASM 47 (1955), pp. 351-379 5. Sjostrom, Material Science and Technology, | (1985), pp823-429 Sis. Denis, A. Simon and G Beck, Transactions ISI, 22 (1982), pp-504-513 C.L, Magee, The Nucleation of Martensite (ASM, New York, 1968) Ch. 3 H. Onodera, H. Gotoh and 1. Tamura, Proc.Ist JIM Intemational Symposium on New Aspects of Martensitic ‘Transformation, 1976, pp. 327-332 LP. Patel and Mi, Cohen, Acta Metal 1, 1953, pp. 531-538 ‘SV. Radcliffe and M. Schatz, Acta Meta. 10, 1962, pp. 201-207. Inoue, K. Arimoto and D.Y Ju, Proc. first Intemational Conference on Quenching and Control of Distorsion, Chicago, September, 1992, pp 205-212 T. Inoue and K. Arimoto, J. Materials Engineering and Performance, ASM International, Vol6, No.1 (1997-1) pp51-60. F, Ikuta and T. Inoue, Proc. 17th Heat Treating Society Conference and Exposition and thelst Intemational Induction Heat Treating Symposium, Indianapolis, (1998-3) pp.541-550 M. Mukai and D.¥. Ju, Proc. 2" International Conference ‘on Thermal Process Modelling and Computer Simulation, in Naney, 2003 (to appear) Y. Nagasaka, , Metallurgical Transactions A, 24A (1993), pp-795-808 K. Okamura and H.Kawashima, Proc. 32nd Japan ‘Congress on Material Reserch (1989), pp 323-329 H. Porzner, Pam User Conference in Asia, PUCA MM, 2 (2000), pp.579-61333, 35, 56, 31, 38, 59, 60, 61 a. 6, Macro-, meso- and micro-seopic metallo-thermo-mechanies K_ Arimoto, G Li, A. Arvind and W-T. Wa, Proc. 18th ASM Heat Treatment Society Conference and Exposition, pp.23-30 K. Arimoto, D. Lambert, G Li, A. Arvind and W.T. Wu, Proc, 18th Heat Treating Conference, ASM International (998), pp.639-644 BLL. Ferguson, A.M, Freborg and GJ. Petrus, Software simulates quenching, heat treating progress, ASMI (2000), pp.H31-H36 BLL Ferguson, AM. Freborg and GJ. Pertrus, Proc. 21st ASM HTS Conference, Indianapolis (2001) ‘1H, Biffle, Proc, 21st ASM HTS Conference, Indianapolis 2001) T. Inoue and K. Okamura, Proc. S* Intemational ‘Symposium on Quenching and Distortion Control, ASM, Si Louis, October, 2000, pp.753-760. T, Inoue, Proc. Sth International Conference on Mechanical Behaviour of Materials, Victoria, Canada, (1999) pp-458-468 T. Inoue, Materials Science Research Intemational, 3-4(1997), pp.193-203 T. Inoue, Proc. 1* International Conference of Business and Technology ‘Transfer, ICBTT2002, Kyoto, 2002-10 p.133-138 ZG. Wang and T. Inoue, Material Technology, Vol. 1 (1985-10) pp. 899-903 ZG. Wang and T. Inoue, Proe. International Symposium ‘on Calculation of intemal Stresses in Heat Treatment of ‘Metallic Materials, (1984-5) pp.474-486, Science and u 61. 68, 69, 0. 7 T. Inoue and DY. Ju, Advances in Continuum Mechanics {9 Papers from Intemational Experts Dedicated to Horst Lippmann), ‘ed. 0. Briller, V. Mannl and. J. Naja, Springer-Verlag, 1991, pp.389-406 DY. Ju,T. Inoue and Hi Matsui, Advances in Engineering Pasicity and its Application, (ed. W.B. Lee), Elsevier, 1993, pps21-s28 T. Uchara, T. Tsjino and Tinoue, Proc. Mechanical Engineering Congress, in Japanese) Tokushima, 2003 pp, 337-338 T. Uchara, M. Kawahara and T. Inoue, Proc. 7th Asian Foundry Congress (AFC-1) (2001), pp.S03-812. T. Uchara and M. Kawahara, T. Inoue, Proc. Fifth World Congress on Computational Mechanics (WCCM V), Eas HLA. Mang, FG Ranmersorfer, J. Ebethardtiner, (2002), Vienna Univ. of Technology, Austria, htp/wesm, tuwien,acat (Published onthe website. ‘T. Ushara and RF. Sekerka, Advances in Comp. Eng. Sci. (Proc. 2002 Tn. Cont. Comp. Eng. Sei), Edited by Satya No Atl, Darell W. Pepper, Tech Science Pres, Encino, (2002), Chap 14. Published by CD-R) TT. Uchara and T. Inoue, Materials Science Rescarch Iteration, 41(1998) pp.45-52. TT. Uchara and Inoue, Proc. Progress in Mechanial Behaviour of Materials (ICM), VoL, (1999), pp. 7L1T] T. Uchara, T. Masago and T. Inoue, Preprints of SO ‘Anniversary Conference of the Society of Materials Science (in Japanese), Osaka, 2001, pp. 283-284