Materials and Design 30 (2009) 359366

Contents lists available at ScienceDirect

Materials and Design

journal homepage: www.elsevier.com/locate/matdes

FEM prediction of welding residual stress and distortion in carbon steel considering phase transformation effects
Dean Deng *
Research Center of Computational Mechanics Inc., Department of Technical Development, Togoshi NI-Building, 1-7-1, Togoshi, Shinagawa-ku, Tokyo 142-0041, Japan

a r t i c l e

i n f o

a b s t r a c t
The objective of this study is to investigate the effects of solid-state phase transformation on welding residual stress and distortion in low carbon and medium carbon steels. In this study, based on ABAQUS code, a sequentially coupled thermal, metallurgical, mechanical 3-D nite element model is developed. In the numerical simulations, different continuous cooling transformation diagrams are used to predict the fractions of martensite for the fusion zone, the coarse-grained HAZ and the ne-grained HAZ, respectively. Effects of volume change due to austenitemartensite transformation on the nal residual stress and the welding distortion are examined. The simulation results revealed that the nal residual stress and the welding distortion in low carbon steel do not seem to be inuenced by the solid-state phase transformation. However, for the medium carbon steel, the nal residual stresses and the welding distortion seem to be signicantly affected by the martensitic transformation. 2008 Elsevier Ltd. All rights reserved.

Article history: Received 17 October 2007 Accepted 23 April 2008 Available online 1 May 2008 Keywords: Welding residual stress Welding distortion Phase transformation Finite element method Numerical Simulation Cooling rate

1. Introduction One of the major problems in welded structures is the welding residual stress and the welding distortion due to local heating. Residual stresses that develop in and around the welding zone are detrimental to the integrity and the service behavior of welded structures. The welding residual stress may promote brittle fracture, reduce the buckling strength and the fatigue life and promote stress corrosion cracking during service. Residual tensile stress also promotes cold cracking associated with hydrogen in certain steels before the welded part is put into service [1]. Welding distortion often results in problems such as dimensional inaccuracies during assembly and cost increase of the product. Several factors may contribute to the formations of residual stress and deformation. The plastic deformation produced in the base metal and weld metal is a function of design (structure), material, and fabrication parameters. The design parameters include the joint type and the thickness of plates. The material parameters reect the metallurgical condition of base metal and the weld metal. Fabrication parameters include the welding method, heat input, preheating, welding sequence and the restraint condition. In certain steel welded parts, the solid-state austenitemartensite transformation during cooling has a signicant inuence on the residual stresses and distortion [2]. The martensitic trans* Tel.: +81 3 3785 3033. E-mail address: deng@rccm.co.jp 0261-3069/$ - see front matter 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.matdes.2008.04.052

formation is a diffusionless solid-state shear deformation [3]. In steels, martensite is formed from austenite containing carbon atoms and, in view of the diffusionless nature of its formation, it ideally inherits the carbon atoms of the parent austenite. The carbon atoms are trapped in octahedral interstitial sites between iron atoms, producing a body centered tetragonal (bct) structure, and are in super-saturation relative to the body centered cubic (bcc) ferrite. In addition to the fact that the chemical composition of the austenite is directly inherited by the martensite, the martensitic shear deformation is accomplished by a plain-strain shape change parallel to a set of crystallographic planes of the parent austenite. Therefore, when the martensite is formed, the volume of metal is increased, and the transformation plasticity is also produced. During the welding process, the magnitude of the volumetric expansion in the heat-affected zone (HAZ) and the fusion zone (FZ) depends upon the volume fraction of martensite that formed. Accurate prediction and reduction of welding residual stress and deformation are critical in improving the quality of welded structures. For certain steels, to evaluate the residual stress and deformation accurately, metallurgical phase transformation must be considered. In this study, a nite element computational procedure considering solid-state phase transformation is developed based on the existing researches [16], and the effectiveness of the proposed numerical method for analyzing the residual stress and the distortion in carbon steels specic to tungsten inert gas (TIG) arc welding is demonstrated. The nite element analysis package ABAQUS [7] is used in this study.

360

D. Deng / Materials and Design 30 (2009) 359366

2. Finite element modelling The evolution of the residual stresses and distortion is investigated by means of thermal elastic plastic nite element method. In order to accurately capture the residual stress and distortion in the welded plate, a three-dimensional nite element model is developed. Because the dimensional changes in welding are negligible and mechanical work done is insignicant compared to the thermal energy from the welding arc, the thermo-mechanical behavior of the weldment during welding is simulated using sequentially coupled formulation. The heat conduction problem is solved independently from the stress problem and phase state to obtain temperature history. However, the formulation considered the contributions of the transient temperature eld to the stress analysis through thermal expansion, as well as temperature-dependent thermo-physical and mechanical properties. The solution procedure consists of two steps. First, the temperature distribution and its history in the welding model are computed by the heat conduction analysis. The temperature history is employed as a thermal load in the subsequent mechanical elasticplastic calculation of the residual stress eld. In this step, the volume fraction of martensite is also calculated, and volume change due to phase transformation is considered through modifying the thermal expansion coefcient over the temperature range in which austenite changes into martensite. In this study, a plate model as shown in Fig. 1 is used. Because of the symmetry, one half of the model is selected as the analysis model. The FE model is shown in Fig. 2 with 4000 brick elements and 5355 nodes. It has a ne grid in the welding zone. The smallest element is 4 mm 1 mm 1.5 mm. The length, the width and the thickness of the model are 200 mm, 100 mm and 6 mm, respectively. In order to clarify the effect of phase transformation on welding residual stress and deformation, two kinds of steel, namely low carbon steel (S15C) and medium carbon steel (S45C), are selected in this study. The chemical compositions and thermal properties as functions of temperature are shown in Table 1 and Fig. 3 [8], respectively. In this study, it is assumed that the thermal properties of S15C steel are the same as those of S45C steel. In the mechanical analyses, because of the lack of material data, the variation of yield strength due to martensite transformation is neglected. 2.1. Heat source and thermal analysis Tungsten inert gas arc welding is the most frequently modelled arc welding process in which the heat source is a non-consumable electrode. In the direct current electrode negative TIG process, the weldments are joined together by the following four primary mechanisms: (1) kinetic energy of the electrons that constitute

Fig. 2. Plate model used for nite element analysis.

Table 1 Chemical composition of S15C and S45C steels, wt% Steel S15C S45C C 0.15 0.44 Si 0.22 0.22 Mn 0.41 0.66 P 0.021 0.022 S 0.024 0.029 Cr 0.06 0.15

1.4 1.2 1 0.8 0.6

(102g/mm3)

Thermal Conductivity Density Specific heat

Thermal Conductivity

C (J/g/ C)

0.4
0.2

(101J/mm/s/ C)

300

600

900
o

1200

1500

Temperature ( C)
Fig. 3. Temperature-dependent thermalphysical properties.

Fig. 1. Geometry of plate and coordinate system.

the arc current, (2) heat of condensation of the electrons penetrating the solid work surface, (3) radiation from the arc, and (4) thermal conduction from the arc plasma to work-piece. The rst two mechanisms constitute the major source of energy for the weldment [9]. According to the nature in which energy is transferred from the arc, the heat input of the TIG arc welding process to the weldment can be modelled by a point source or a line source. A more realistic approach is to consider the heat ux on the surface or the heat generation distribution in the metal, or a combination of the two. In this work, the heat from the moving welding arc is applied as a volumetric heat source with a double ellipsoidal distribution proposed by Goladk et al. [10]. This is represented by the following equations: For the front heat source p 6 3ff Q 3xvtx0 2 =a2 3y2 =b2 3z2 =c2 1e p e qx; y; z; t e 1 a1 bcp p

D. Deng / Materials and Design 30 (2009) 359366

361

For the rear heat source p 6 3fr Q 3xvtx0 2 =a2 3y2 =b2 3z2 =c2 2e p e qx; y; z; t e a2 bcp p

Voulme change

where ff and fr are parameters which give the fractions of the heat deposited in front and the rear parts, respectively. Note that ff fr 2:0: This is done because the temperature gradient in the front leading part is steeper than that in the tailing edge. Therefore, two ellipsoidal sources are combined: one for the front half and the other for the rear half. Q is the magnitude of the heat input per unit time (Q = UI); v is the welding speed; t is the welding time; and x0 is the position of the heat source in the x-direction when t is zero. The arc efciency, g, is assumed to be 70% for the TIG welding process. The parameters a1, a2, b and c are related to the characteristics of the welding arc. In the two equations, the origin of the coordinate system is located at the center of the welding arc, and the coordinate system that is xed to the weldment. In the nite element model and the Lagrangian formulation of ABAQUS, the welding arc moves in this xed coordinate system. The parameters of the heat source are adjusted to create the desired melted zone. Another way to determine the parameters of heat source is to make temperature measurements at certain positions and compare these with the computed temperature eld and history. The heat source for the moving welding arc is modeled by a user subroutine in ABAQUS, and the values of these parameters used in Eqs. (1) and (2) are summarized in Table 2. When structural analysis FEM Codes such as ABAQUS and MARC are used to simulate the temperature distribution in welding process, the uid ow and solidication of material in the weld pool cannot be directly considered because the coupled problem between solid and liquid is not involved in these software at present. However, the effect of the uid ow has signicant effects on the temperature distribution and on the shape of the weld pool. If the effect of the uid ow is neglected, the highest temperature in weld pool will be very high. Sometimes, it is over 3000 C when the Gaussian heat source with a double ellipsoidal distribution is used. This phenomenon is much different from the realistic situation. Okagaito, co-workers [11] measured the surface temperature distribution on TIG weld pool in SUS304 steel. Their research suggests that the highest temperature on the molten pool surface is approximately 1750 C. In this study, in order to consider the uid ow an articially increased thermal conductivity is used. When temperature is higher than the melting point, the value of the thermal conductivity is assumed to be approximately twice as large as that of room temperature. Because the articially increased thermal conductivity is used in the present simulation, the highest temperature of the weld pool is about 1800 C, which is much closer to the realistic situation. The thermal effects due to solidication of the weld pool are modeled by taking into account the latent heat for fusion. The value of the latent heat is 270 J/g [12]. The liquidus temperature TL and the solidus temperature TS are assumed to be 1480 C and 1430 C, respectively. The same values of the liquidus temperature and the solidus temperature are used for both S15C steel and S45C steel because their difference is very small. In this study, the mid-

dle value between the solidus temperature and the liquidus temperature is assumed to be the melting point. To account for heat losses, both the thermal radiation and heat transfer at the weld surface are assumed. A user subroutine in ABAQUS is developed to simulate the heat losses. In order to obtain a high accurate result in the mechanical analysis, the maximum temperature increment is less than 5 C in the nonlinear thermal analysis. 2.2. Phase transformation In this study, a metallurgical analysis based on phase transformation laws [13] is performed to simulate the phase transformation during welding. When a base metal is heated above A1 (cementite disappearance temperature) temperature, pearliteferrite partly transforms into austenite, and when the temperature is higher than A3 (a-ferrite disappearance temperature), pearliteferrite completely changes into austenite. For carbon steel and low alloy steel, the A1 and A3 temperatures can be calculated according to the following empirical equations [14]: A1 723 10:7Mn 16:9Ni 29Si 16:9Cr 290As 6:4W p A3 912 203 C 15:2Ni 44:7Si 104V 31:5Mo 13:1W 30Mn 11Cr 20Cu 700P 400Al 120As 400Ti 4 3

When the base metal is heated above A1 temperature, the body centered cubic structure transforms to the face centered cubic (fcc) structure, and the volume decreases. During rapid cooling, the austenite with fcc structure changes to martensite with a body centered tetragonal structure, and the volume increases. The volume change due to phase transformation during heating and cooling is schematically shown in Fig. 4. In this study, depending on the peak temperature that an integration point of an element reached during heating process and the cooling time (t8/5) from 800 C to 500 C, the decision is made whether the point underwent the martensitic transformation or not. From the continuous cooling transformation (CCT) diagram, the relationship between the cooling rate and the fraction of the martensite can be obtained. In principle, an innite number of CCT diagrams would be required to describe transformation behavior in every region of the weld metal and the HAZ. Good understanding of the weld transfor-

Austenitic Transformation

Heating

Cooling

Table 2 Parameters of the heat source Parameter a1 a2 b c ff fr Value 5.0 mm 10.0 mm 3.5 mm 5.0 mm 1.33 0.67

Martensitic Transformation

Mf
ff = 2a2/(a1 + a2) fr = 2a1/(a1 + a2)

Ms

A1 A 3

Temperature(oC)
Fig. 4. Schematic diagram of volume change due to phase transformation.

362

D. Deng / Materials and Design 30 (2009) 359366

mation behavior can usually be obtained by examining three typical CCT diagrams, namely those for the ne-grained HAZ (FGHAZ), the coarse-grained HAZ (CGHAZ) and FZ [15]. The austenization temperature for the CGHAZ is usually taken in the range 1050 1450 C, and that for FGHAZ and the FZ is A3-1050 C and 1450 C, respectively. These temperatures take grain growth of austenite into account. Then the decomposition of the austenite into different phases at different cooling rates can be obtained from the cooling curve of the CCT curve corresponding to a particular cooling rate under continuous cooling conditions. In this study, only the fraction of martensite is predicted using this method. The relationships between t8/5 and the fraction of martensite can be obtained from CCT diagrams for S15C steel and S45C steel as shown in Figs. 5 and 6. Because the CCT diagram applicable to weld metal has not been prepared by cooling from liquid state, the same CCT diagram is used to predict the fraction of martensite for the fusion zone and the CGHAZ in this work. Additionally, the inter-critical zone (ICHAZ) is neglected. The CCT diagram [8] in which the austenitizing temperature is 900 C is selected to predict the fraction of martensite for the

FGHAZ, and the CCT diagram [8] is selected with austenitization at 1050 C for the CGHAZ and the FZ. The ASTM grain size number is 8 for the FGHAZ, and 53 for the CGHAZ and the FZ. On cooling, the martensitic transformation is described using the KoistinenMarburger relationship [3,13] fm T f 1 expkMs T T 6 M s 5

where fm is the fraction of martensite at current temperature; T is the current temperature during cooling; f represents the phase fraction obtained at an innitely low temperature; and Ms and k characterize initial transformation temperature and evolution of the transformation process according to temperature, respectively. For carbon steel, the value of constant k is 0.011 [3,13]. The initial transformation temperature Ms can be calculated using the following equation [13]: M s 561 474C 33Mn 17Ni 17Cr 21Mo 6

The phase transformation temperatures (A1, A3, Ms and Mf) of S15C steel and S45C steel are summarized in Table 3. 2.3. Mechanical analysis

0.8

Martensite fraction (S15C) Austenitized at 900C,5 min ASTM grain size number: 8 Marten sitefraction (S15C) Austenitized at 1050C, 5min ASTM grain size number: 5-3

0.6

0.4

0.2

The same FE mesh used in thermal analysis is employed here, except for the element type and the boundary conditions. The analyses are conducted using the temperature history calculated by the thermal analysis as the input information. Temperature dependent material properties are taken into account, as shown in Fig. 7 [8]. In the mechanical analysis, boundary conditions are just used to prevent rigid body motion. The boundary condition is shown in Fig. 8. Because of the symmetry of the FE model, the symmetry plane is xed in Y-direction. The point P1 in Fig. 8 is constrained in the X-direction and the Z-direction; and the point P2 is constrained only in the Z-direction. P1 and P2 are the start point and the end point of the centerline in the symmetry plane, respectively. During the welding process, an additional strain is induced by microstructure evolution during solid-state phase transformation along with the thermal strain. At the same time, transformation-inTable 3 Phase transformation temperatures of S15C steel and S45C steel Steel S15C S45C A1 (C) 726 724 A3 (C) 843 787 Ms (C) 476 328 Mf (C) 280 140

Martensite fraction

0 0.1

10

100

Cooling time t8/5 (s)

Fig. 5. Martensite fraction as a function of cooling time t8/5 for S15C steel.

0.8

Martensite fraction (S45C) Austenitized at 900C,5min ASTM grain size number 8 Martensite fraction (S45C) Austenitized at 1050C,3min ASTM grain size number 5-3

350 300
Youngs Modulus (GPa) (S45C) Youngs Modulus (GPa) (S15C) Yield Stress (MPa) (S45C) Yield Stress (MPa) (S15C) Thermal Expansion Coefficient (10 )
-7

Martensite fraction

Mechanical Properties
1 10 100

250 200 150 100 50

0.6

0.4

0.2

300

600

900
o

1200

1500

Cooling time t8/5 (s)

Fig. 6. Martensite fraction as a function of cooling time t8/5 for S45C steel.

Temperature ( C)
Fig. 7. Temperature-dependent mechanicalphysical properties.

D. Deng / Materials and Design 30 (2009) 359366 Table 4 Simulation cases Case Material Welding condition Current (A) Case Case Case Case A-1 A-2 B-1 B-2 S15C S15C S45C S45C 300.0 300.0 300.0 300.0 Voltage (V) 21.0 21.0 21.0 21.0 Speed (mm/s) 10.0 10.0 10.0 10.0

363

Phase transformation

No Yes No Yes

V DV DeD c f fk exp kT M s g DT c ec

10

Fig. 8. Mechanical boundary condition.

_) duced plasticity is also produced. Therefore, the total strain rate (e can be written as the sum of the individual components of the strain rate as _e _ e _ e _ e _ e
E P T DV

where DT c is the increment of temperature during cooling. In the numerical simulations, it is assumed that when the temperature cools to Mf, the martensitic transformation is nished. A subroutine to ABAQUS code is developed to compute the fraction of the martensite f and volumetric changes due to martensitic transformation. 3. Simulated cases To clarify the inuence of phase transformation on the residual stress and the welding deformation of carbon steel, two kinds of steel, namely S15C with a low carbon content and S45C with a medium carbon content, are selected. Four cases are performed to simulate the residual stress and the welding deformation. The simulation cases are shown in Table 4. All simulations are performed by a personal computer with Pentium-4 CUP 3.2 GHz. The computational time of the thermal analysis is approximately 184 min, and the average computational time of the four mechanical analysis cases is about 600 min. 4. Results and discussion 4.1. Analysis of residual stress For the sake of convenience in the following discussion, the middle cross-section of the model is dened as shown in Fig. 9. The distributions of computed values, such as stress and displacements, will be plotted on this middle cross-section. In the analyses considering phase transformation, the solidstate phase transformation model described in Section 2 is used. Fig. 10 shows the simulated martensite distributions in the middle section of the nite element model for Cases A-2 and B-2. From this gure, it can be seen that martensitic transformation appeared both in Case A-2 and in Case B-2. Because the chemical compositions of S15C and S45C are different, the fractions of martensite are different. For steel S45C, the martensite fraction is larger than that of steel S15C because of a relatively high carbon equivalent. Fig. 11 shows the residual stresses in the welding direction along the line 1 in the middle cross-section for Cases A-1 and A2, and Fig. 12 shows the residual stresses in the welding direction

_ e

Trp

The various components in this equation represent strain rate due to elastic, plastic and thermal loading, volumetric change and transformation plasticity, respectively. In this study, the volume change is considered, but transformation-induced plasticity is not taken into account. Ignoring this component, the strain increment can be written as follows: De DeE DeP DeT DeDV
DV

The component De is caused by volume change due to phase transformation. As mentioned previously, the volume of martensite increases, and an additional stress is produced. The strain due to volume change associated with full martensitic transformation assigned V to S45C steel and S15C steel is eD 8:0 103 and c 3 DV ec 2:0 10 [8], respectively. The volume change due to ausV tenite transformation is assumed to be eD 2:88 103 [5]. h In principle, when pearliteferrite changes into austenite between A1 temperature and A3 temperature, the strain increment due to volume change can be calculated according to the lever law. For the sake of simplicity, a linear relation is assumed in this study. By using the linear approximation, the strain increment can be estimated as follows:
V DeD h

DT h eDV A3 A1 h

where DT h is the temperature increment during heating. According to the KoistinenMarburger relationship, the strain increment due to volume change in the course of the martensitic transformation process can be determined by the following equation:

Fig. 9. Middle cross-section of the analysis model.

364

D. Deng / Materials and Design 30 (2009) 359366

Fig. 10. Martensite distribution in the middle section.

300
Stress in welding direction (Case A-1)

250 200

Stress in welding direction (Case A-2)

150 100 50 0 -50

20

40

60

80

100

Y-Coordinate (mm)
Fig. 11. Residual stress in welding direction of middle section in Cases A-1 and A-2.

400
Stress in welding direction (Case B-1) Stress in welding direction (Case B-2)

is relatively small and the transformation temperature is relatively high, phase transformation has little effect on the residual stress. From Fig. 12, a clear difference is observed between the residual stresses in Cases B-1 and B-2. Generally, without lower temperature phase transformation, the residual stress level in the FZ and HAZ is almost as large as the yield stress of the welded material at room temperature. In Case B-2, because the phase transformation is neglected, large longitudinal stresses are generated in the FZ and HAZ. In Case B-2, the phase transformation is considered, so low longitudinal stresses are produced in the FZ and HAZ. The reason is that the volume dilation due to martensitic transformation is very large, and the transformation temperature is relatively low for S45C steel. In this study, experiments have not been carried out to verify the numerical model. Here, some existing literatures are cited to verify the effectiveness of the computational procedure developed in this paper. Talijiat et al. [1] measured the welding residual stress in spot welds made in an HY-100 steel disk by a neutron diffraction technique. Their experimental results show that the stresses in the FZ and HAZ are lower than those in the base metal adjacent to HAZ. Lower stresses in the FZ and HAZ resulted from the volume change of the material due to martensitic transformation in a relatively low temperature. Similarly, Cho and Kim [5] investigated the welding residual stress in plate models made in AISI1045 steel. They also simulated the welding residual stress using a 2-D plane strain FE model. Their experimental results show that a compressive longitudinal stress produces in the weld zone. In the present study, the results of Case B-2 reect the same tendency even though the welding condition and the size of the FE model are different from their model. The above two examples have indirectly veried the effectiveness of the 3-D computational procedure proposed in this study. Fig. 13 shows the histories of stress in the welding direction and temperature at point A for Cases B-1 and B-2. In the course of heating, a compressive stress is produced in the heated zone. With further heating, the compressive stress decreases because of the small yield stress at high temperature. When the temperature reaches the melting point, the compressive stress is very small. Upon cooling, a tensile stress in the welding direction is produced because of the shrinkage of the weld. Within the martensitic transformation range, a compressive stress is generated again because of a large volume change. After the completion of martensitic transformation, the longitudinal stress increases gradually again. After cooling to room temperature, a compressive stress remained nally. Under

Stress (MPa)

300 2000

Stress (MPa)

200 1500 100

Temperature ( C) Stress (MP a) S45C Case B-1 Stress (MP a) S45C Case B-2

Temperature (C)
0 20 40 60 80 100

1000

500

-100

Y-Coordinate (mm)
Fig. 12. Residual stress in welding direction of middle section in Cases B-1 and B-2.

-500 0

20

40

60

80

100

along the line 1 for Cases B-1 and B-2. From Fig. 11, it is clearly seen that there is no signicant difference between Case A-1 and Case A-2. Because the volume dilation due to phase transformation

Time (s)
Fig. 13. Temperature history and stress history of point A for Cases B-1 and B-2.

D. Deng / Materials and Design 30 (2009) 359366

365

the condition of ignoring the phase transformation effect (Case B1), a very high tensile stress existed in the weld zone as shown in Fig. 13. According to the simulation results, it can be concluded that martensitic transformation has a signicant inuence on the welding residual stress for medium carbon steel but has less inuence for low carbon steel. 4.2. Analysis of welding deformation Fig. 14 shows the vertical deformation in Cases A-1 and A-2 along the centerline (line 2 in Fig. 9) in the middle cross-section of the FE model. From this gure, it can be seen that there is a very small difference between the two cases. Fig. 15 shows the displacements in the Y-direction for Cases A-1 and A-2 along the centerline. Similar to the vertical deformation, there is only a small difference between the two cases. Therefore, it can be concluded that phase transformation does not seem to affect the welding distortion in the low carbon steel. Fig. 16 shows the vertical deformation of Cases B-1 and B-2 along the centerline (line 2 in Fig. 9) in the middle section. It is clear that there is a large difference between the two cases. The deection of Case B-2 is smaller than that of Case B-1. Because

Deflection (Case B-1) Deflection (Case B-2)

1.5

Deflection (mm)

0.5

0 0 20 40 60 80 100

Y-Coordinate (mm)
Fig. 16. Deection along line 2 for Cases B-1 and B-2.

1.5

Deflection (Case A-1) Deflection (Case A-2)

0.5

20

40

60

80

100

Y-Coordinate (mm)
Fig. 14. Deection along line 2 for Cases A-1 and A-2.

0.05
Y-Displacemen t (Case A-1) Y-Displacemen t (Case A-2)

Y-Displacement (mm)

-0.05

the volume increased in the course of martensitic transformation, a relatively small angular distortion is produced in Case B-2. Hence, the deection of Case B-2 is smaller than that of Case B-1. Bhadeshia [6,16] investigated the angular distortion in buttweld joints made in two kinds of steels using multi-pass welding process. The chemical composition, calculated transformation temperature range and measured angular distortion for two manual metal arc multi-pass weld deposits are shown in Table 5. The experimental results show that the angular distortion depends on the temperature at which the majority of the transformation is completed. The angular distortion is clearly smaller for the case where the transformation is exhausted at lower temperature. From Fig. 17, it can be seen that the angular distortion of Case A-1 is smaller than that of Case B-1. In this gure, D1 is the difference of maximum deection between Cases A-1 and B-1. The difference between the two cases is caused by the different mechanical properties such as yield strength and Youngs modulus in the two kinds of steels. Fig. 17 also shows that the angular distortion of Case A-2 is signicantly larger than that of Case B-2. Similarly, D2 is the difference of maximum deection between Cases A-2 and B-2. Without considering the effects of phase transformation, the angular distortion of Case B-1 (S45C steel) is larger than that of Case A-1 (S15C steel). On the contrary, when the effects of phase transformation are considered, the angular distortion of Case B-2 (S45C steel) is much smaller than that of Case A-2 (S15C steel). The results tell us that the temperature range of phase transformation has a signicant inuence on the welding distortion. This phenomenon is similar to Bhadeshias experimental result. Similar to vertical deection, a large difference of transverse displacement in the Y-direction is observed between Cases B-1 and B-2 as shown in Fig. 18. Because of the dilation induced by a

Deflection (mm)

-0.1

Table 5 Chemical composition, calculated transformation temperature range (DTtr), and measured distortion (b) for two manual metal arc multi-pass weld deposits [6,16] Composition, wt% C Si Mn Ni Mo Cr DTtr (C) b ()

-0.15

20

40

60

80

100
0.06 0.06 0.5 0.3 0.9 1.6 1.7 0.4 0.35 802400 422350 14.5 8.0

Y-Coordinate (mm)
Fig. 15. Y-directional displacement along line 2 for Cases A-1 and A-2.

366

D. Deng / Materials and Design 30 (2009) 359366

2
Deflection Deflection Deflection Deflection (Case A-1) (Case B-1) (mm) Case A-2 (mm) Case B- 2

DB-1

1.5

welding process and to simulate the residual stress and the welding distortion. The proposed method is applied to predict the residual stress and the welding distortion in both low carbon steel and medium carbon steel, and the following conclusions have been drawn.
DA-2

Deflection (mm)

DA-1 1=DA-1-DB-1= -0.33mm 2=DA-2-DB-2=+0.41mm

1
DB-2

0.5

0 0

20

40

60

80

100

Y-Coordinate (mm)
Fig. 17. Deection along line 2 for the four cases.

0.05
Y-Displacement (Case B-1) Y-Displacement (Case B-2)

1. For low carbon steel, phase transformation has an insignicant effect on the welding residual stress and the distertion because of a small dilation due to martensitic transformation and a relatively high transformation temperature range. 2. For medium carbon steel, phase transformation has a signicant effect on the welding residual stress and the distortion because of a relatively large dilation and a low transformation temperature. In this study, it is observed that a compressive stress in the welding direction is produced in the FZ. 3. According to the present simulation results, when the residual stresses and the welding deformation of medium carbon steel are predicted using numerical simulation, it is necessary to consider solid-state phase transformation. On the other hand, the solid-state phase transformation can be neglected for low carbon steel. Neglecting the phase transformation in low carbon steel weld structure will make the numerical model simple and reduce the computation time.

Y-Displacement (mm)

References
[1] Talijiat B, Radhakrishnan B, Zacharia T. Numerical analysis of GTA welding process with emphasis on post-solidication phase transformation effects on residual stress. Mater Sci Eng A 1998;246:4554. [2] Ravi Vishnu P. Solid-state transformations in weldments. ASM handbook, vol. 6; 1994, p. 7087. [3] Krauss George. Deformation and fracture in martensitic carbon steels tempered at low temperatures. Metall Mater Trans B 2001;32B:20521. [4] Leblond J-B, Mottet G, Devaux J, Devaux J-C. Matermatical models of anisothermal phase transformations in steels, and predicted plastic behaviour. Mater Sci Technol 1985;1:81522. [5] Cho SH, Kim JW. Analysis of residual stress in carbon steel weldment incorporating phase transformations. Sci Technol Weld Join 2002;4:2126. [6] Bhadeshia HKDH. Developments in martensitic and bainitic steels: role of the shape deformation. Mater Sci Eng A 2004;378(12):349. [7] Hibbitt Karlsson Sorensen Inc., ABAQUS/Standard Users Manual, vols. 13; Version 6.3. [8] METEQ Ver1.2 CD-ROM, The Society of Material Science, Japan; 2002. [9] Glickstein SS, Friedman E. Characterization and modeling of the heat source. ASM handbook, vol. 6; 1994. p. 11416. [10] Goladk J, Chakravariti A, Bibby M. A new nite element model for welding heat sources. Metall Trans B 1984;15:299305. [11] Okagaito T, Ohji T, Miyasaka F. UV radiation thermometry of TIG weld pool development of UV radiation thermometry (Report 1). Quart J Jpn Weld Soc 2004;22(1):216. [12] Zhang W, Elmer JW, DebRoy T. Modeling and real time mapping of phase during GTA welding of 1005 steel. Marter Sci Eng A 2002;333:3205. [13] Kosistinen DP, Marburger RE. A general equation prescribing extent of austenitemartensite transformation in pure FeC alloys and plain carbon steel. Acta Metall 1959;7:5060. [14] Watt DF, Coon L, Bibby M, Goldak J, Henwood C. An algorithm for modeling microstructural development in weld heat-affected zones, (part A) reaction kinetics. Acta Metall 1988;36(11):302935. [15] Adil GK, Bhole SD. HAZ hardness and microstructure predictions of arc welded steel-I review of predictive models. Can Metall Quart 1992;31(2):1517. [16] Bhadeshia HKDH. Material factors, handbook of residual stress and deformation of steel. ASM Int 2002:310.

-0.05

-0.1

-0.15 0 20 40 60 80 100

Y-Coordinate (mm)
Fig. 18. Y-directional displacement along line 2 for Cases A-1 and A-2.

low temperature phase transformation, which partly cancels transverse shrinkage in Case B-2, the displacement in the Y-direction is much smaller than that of Case B-1. According to the results, it can be concluded that phase transformation has a signicant inuence on welding deformation in the medium carbon steel. Comparing Cases A-2 and B-2, one can nd that the transverse shrinkage of Case B-2 is smaller than that of Case A-2. 5. Conclusions A three-dimensional nite element model considering phase transformation is developed to analyze the tungsten inert gas arc