You are on page 1of 24

CHAPTER - V

MODELLING PROCEDURE

5.1 INTRODUCTION

The development of high capacity computers has promoted the scope of

numerical models in simulating all the engineering processes including welding.

Appropriate numerical techniques can be effectively used to solve a group of

partial differential equations that represent the welding and associated thermal,

mechanical and micro structural changes to get a quantitative understanding of the

welding process. The principal phenomenon during the welding process is the heat

transfer. The phenomena like melting, solidification, fluid flow and stress

development are associated with heat transfer. In this study an attempt has been

made to understand the effect of constant current welding and friction stir welding

techniques on temperature distribution and residual stress of AA 5059 aluminium

alloy using the general purpose Finite Element Analysis code COMSOL.COMSOL

is an FEA software package that allows for the simulation of theoretical and

engineering applications. The software contains smaller packages, called

"modules," that contain well-known physical models to describe a particular

simulation. These physical equations can also be manually configured to describe

other models that are not built-in. The modules can be easily combined to couple

the governing equations of the respective physics used. [Comsol (2006)].

Geometries can also be modeled within COMSOL itself, along with defining mesh

type and element size. For the modelling of residual stress in welds, an advantage
will be the ability to couple physics of solid mechanics and heat transfer

relationships.

With respect to COMSOL, the modules involved in this simulation are

Structural Mechanics, along with the nonlinear add-on module. The Structural

Mechanics module allows for simulations which, couples heat transfer and solid

mechanics, while the nonlinear Structural Mechanics module is needed to add

plasticity and hardening to a material.

5.2 FINITE ELEMENT MODELLING OF GTAW

A three-dimensional transient model was proposed with the general-

purpose finite element analysis code COMSOL. The governing equation [Little et

al (1998) Komanduri et al (2000)] for heat transfer in three dimensions arrived

from the Helmholtz equation is

∂  ∂T  ∂  ∂T  ∂  ∂T 
 kx  +  ky  +  kz  + Q = 0 (5.1)
∂x  ∂x  ∂y  ∂y  ∂z  ∂z 

where, kx, ky and kz are the thermal conductivities in the x, y and z

directions and Q the internal heat generation

The continuum to which Equation 5.1 applies is represented by a mesh of

finite elements in which the temperature distribution was discretized as

M
T ( x , y , z ) = ∑ N i ( x , y , z )Ti = { N }[T ] (5.2)
i =1

94
where, Ni (x,y,z) is the interpolation function associated with nodal

temperature Ti, [N] is the row matrix of interpolation functions, and M is the

number of nodes per element.

Model Assumptions

The following assumptions are made during the development of the model:

The work piece is initially at room temperature

For weld modelling , it is considered that the heat source is moving with

fixed work piece

Convection and radiation is applied.

Only heat is transferred to the work piece, material deposition is not

considered.

Heat conduction through the clamps is ignored.

5.2.1 Thermal Efficiency of Welding Arc

Only a fraction of the power of the arc causes the heating of the work piece

and formation of the weld pool and the rest is lost as radiant heat and light. This

fraction is represented by the arc thermal efficiency defined as the ratio of net

thermal input to the plate to the net power dissipated by the welding arc. Usually

an empirical value is used for welding arc efficiency and the variety of this value

with other welding parameters like voltage, current and speed can be neglected.

For GTAW, the arc thermal efficiency is approximated as 0.71 by most of the

researches, but actually it varies from 0.41 to 0.816 [Dutta et al. (2007)]. This large

variation in the arc thermal efficiency can reduce the accuracy of thermal models.

95
The primary parameters that determine the efficiency are welding current, voltage,

arc gap and welding speed. There are many other parameters like argon gas flow

rate, electrode properties, electrode tip angle, type and thickness of the base metal

and torch position which have a little influence on the welding arc.

Based on the literatures and depending on the polarity and shielding gas a

value of 0.50 was prescribed as an arc efficiency in this study, as the results would

become unstable at higher efficiencies. Table 5.1 lists the welding parameters used

in this analysis.

Table 5.1 Welding parameters used for the analysis

Welding Parameter Parameter value

Current 110 A

Voltage 14v

Welding speed 1.3 mm/S

Weld efficiency 0.5

Number of passes Single

Electrode diameter 3 mm

Shielding gas Argon

Polarity Alternating current

5.2.2 Heat Source Model

The ellipsoidal heat source model revealed that the temperature gradient in

front of the heat source was not as steep as expected and the gentler gradient at the

trailing edge of the molten pool was steeper than experimental measurements [John

Goldak et al (1986); Afzaal Malik (2008); Bate et al (2009); Shan et al (2009)].

96
Hence, two ellipsoidal sources were combined as shown in fig 5.1. The front half

of the source is the quadrant of one ellipsoidal source, and the rear half is the

quadrant of another ellipsoid. The estimation of the heat input was made based on

the equations 5.3 and 5.4 [Lindgren (2006), Afzaal Malik (2008)].

6 3Qff   x 2 y2 z 2  
qf = exp  − 3 w + 2 + 2   (5.3)
π a f bc  af b c 

6 3Qf r   x 2 y2 z2  
qr = exp − 3 w + 2 + 2   (5.4)
π a r bc  ar b c 

where, ff and fr are the frontal and rear fraction of the heat flux. Based on the

literature [John Goldak et al(1986)] af is 5mm and ar is 20 mm. af means forward

portion of the area of heat application. ar means rear portion of the area of heat

application. b and c are the parametric values obtained from the metallographic

data and from weld pool surface ripple markings as depicted in Fig. 5.1. Q is the

calculated heat input. Fig.5.2 shows the solid model of the welded specimen with

heat source.

Fig. 5.1 Heat source model

97
Fig.5.2 Solid Model of the Welded Specimen with Heat Source

During the early stages of welding heat transfer research, it was assumed

that all the boundaries are insulated and there is no heat loss from boundaries, later

radiation boundary conditions were incorporated by researchers as the radiation

heat transfer is prominent over convection


convection heat transfer at high temperatures. In the

present study,, in the case of modelling


mode ling of plates, all the sides are assumed to be

exposed to ambient atmosphere and radiative heat transfer boundary conditions are

applied to all boundaries expect the plane of symmetry. Fig.5.3 shows the

schematic representation of boundary conditions applied in the present study.


study

Fig. 5.3 Schematic Representation of Boundary Conditions


onditions

98
The heat loss is incorporated into the numerical analysis as a Cauchy type

boundary condition. In order to avoid non-linearities associated with the radiation

term, a combined convective and radiative heat transfer coefficient is used to

accommodate the convective and radiative heat transfer, in which the surface heat

transfer coefficient is approximated as the function of total emissivity. Kim

proposed an empirical relation for the combined convection and radiation heat

transfer coefficient during welding and it can be represented by equation [Kim and

Basu (1998)].

− 4 1.16 2
h = 24.1x10 ε T W/m K . (5.5)

5.2.3 Mathematical Formulation and Boundary Conditions

The Gaussian heat input is applied to the top, and all the other sides are

exposed to convective and radiative heat loss. Modelling requires an equilibrium

balance, a constitutive relation relating stress and strain, and a kinematic relation

displacement to strain. Newton‘s second law serves as the equilibrium equation,

which in tensor form is:

∇. + = , (5.6)

where, σ is the Cauchy stress tensor, Fv is the body force per unit volume, ρ is the

density and u is the acceleration. The constitutive equation relating the stress tensor

σ to strain ɛ is the generalised Hooke’s law = :ε

where, C is the fourth order elasticity tensor and : denotes double dot tensor

product. In COMSOL, this relation is expanded to

99
σ − σ 0 = C : (ε − ε 0 − ε inel ) (5.7)

For this application, initial stress σ0, initial strain ε0, and inelastic strain ɛ inel

are all zero. For isotropic material, the elasticity tensor reduces to the 6× 6

elasticity matrix which includes Lame constants ( λ and µ)

υ
λ= (5.8)
υ υ

µ= (5.9)
υ

where, E is the elasticity modulus and υ is the Poisson’s ratio.

Initial conditions, T(x,y,z)= Tin (5.10)

In COMSOL, h value is expressed by the following equation

− n.(k∇T ) = h.(Tamp − T ) (5.11)

h is the heat transfer coefficient, n is the normal vector of the boundary, q is the

conductive heat flux vector where

q = -K∇T (5.12)

For thermal boundary terms, all upper and lower boundaries of the weld

were set with heat loss terms from convection and radiative heat transfer.

Convection is described by Newton's law of Cooling and Radiative heat transfer is

described by Stefan-Boltzmann's law.

100
The convective and radiative heat loss is represented by equations

=ℎ − (5.13)

" "
=! − ) (5.14)

where, qc and qrare the heat loss by convection and radiation respectively, hc is the

convective heat transfer coefficient, ε is the emissivity of the material, and σ is the

Stefan Boltzmann constant. T and Tambi are local and ambient temperatures

respectively.

5.2.4 Material Properties

In this study, the following important material properties were needed to be

defined:

• Thermal conductivity

• Heat capacity at constant pressure

• Elastic modulus

• Density

• Poisson's ratio

• Surface emissivity

• Initial yield stress

• Isotropic tangent modulus

• Coefficient of thermal expansion

101
5.2.5 Heat Input Calculation

For the calculation of the heat input (Q), the relationship used for constant

current process is Q= [(V×I)/ S]η where V is the voltage, I is the current, S is the

welding speed and η is the efficiency of utilization of the heat generated

[Gonsalves et al (2006),Kim et al (1998)]. The amount of heat input was found as

the product of arc efficiency, voltage and current which were taken equal to 0.5,

14v and 110A respectively in this analysis. In this analysis, the amount of heat

input was calculated as 592.32 J/mm.

5.3 NUMERICAL PROCEDURE FOR GTAW

5.3.1 Temperature Analysis

A mathematical model is generated for the governing equation and boundary

conditions as described previously. A custom developed pre-processor is used for

discretization of the domain and initialization of the nodal values. The dimensions

of the domain and welding parameters like welding current, voltage and speed are

the major inputs to the program. The temperature dependent material properties

like thermal conductivity, enthalpy and specific heat are calculated from empirical

relations explained. The levels of discretization and convergence parameters are

given as input to the program. Different elements were used for finite element

analysis. In order to reduce the computational time and achieving accuracy in the

result, coarser mesh was carried out away from the tool and fine mesh was carried

out close to the tool. Figure 5.4 depicts the mesh pattern of the present study. The

meshed model composed of 13100 tetrahedral elements, 915 triangular

elements,415 edge elements and 20 vertex elements. Meshed size varied from

102
0.225 to 5.25. The typical thermo-mechanical properties of AA 5059 are as

tabulated in the Table 5.2.

Table 5. 2 Typical Thermo-mechanical Properties of AA 5059

Name Value Units

Heat capacity at constant pressure 904 [J/(kg*K)]

Thermal conductivity 117 [W/(m*K)]

Density 2640 [kg/m^3]

Coefficient of thermal expansion 23e-6 [1/K]

Young's modulus 70e9 [Pa]

Poisson's ratio 0.33

Fig. 5.4. Schematic Representation of Moving Mesh Technique

103
5.3.2 Solution Techniques

The program flow can be represented by the flow chart shown in Fig. 5.5.

The developed mathematical model is solved numerically by Gauss Seidel iterative

method with two decimal point accuracy in temperature values. The major steps in

the solution process can be listed as below.

1 Initialization

2 Calculation of material properties/ Boundary conditions

3 Iterations

4 Convergence

5 Storing of results

Fig. 5.5 Flow chart showing the solution procedure

104
After completion of discretization, all nodal points are assumed to be at

ambient temperature and initialized with ambient temperature for calculating the

material properties before starting of the first iteration. After completion of every

iteration, the nodal values are checked for convergence criteria. If the values do not

satisfy the convergence criteria, the iteration process is repeated till it converges.

After achieving required convergence the values are stored in a file and the time is

incremented. The torch position, material properties, were updated. These

procedures were repeated till the completion of the welding and cooling time.

For the welding process, a simulation requires description for the heat

transfer, as well the elastic, plastic, and thermal strains that contribute to residual

stress. Geometries can also be modeled within COMSOL itself, along with

defining mesh type and element size. For the modelling of residual stress in welds,

an advantage will be the ability to couple physics from solid mechanics and heat

transfer relationships.

Therefore, using small deformation theory for an isotropic material, a

thermo-elastic-plastic law is used to describe the constitutive relationship between

stress and strain. With respect to COMSOL, the modules involved in this

simulation are Structural Mechanics, along with the Non-Linear add-on module.

The Structural Mechanics module allows for simulations which coupled heat

transfer and solid mechanics, while the Non-Linear Structural Mechanics module

is needed to add plasticity and hardening to a material. For calculating residual

stresses on the plate, governing differential equations for heat transfer and elastic-

plastic structural behavior is used. The temperature distribution is defined as

105
thermal load for the Stress-Strain analysis. Materials are defined as elastic-plastic,

so that when the whole system is cooled back at room temperature, a state of

residual stress is obtained.

5.3.3. Residual Stress Analysis

In fusion welding, a weldment is locally heated by the welding heat source.

Due to the non-uniform temperature distribution during the thermal cycle,

incompatible strains lead to thermal stresses. These are compatible strains due to

dimensional changes associated with solidification of the weld metal (WM),

metallurgical transformations, and plastic deformation, are the sources of residual

stresses and distortion. During the welding process, the total strain rate can be

decomposed into three components as follows:

ε$%$&' = ε( + ε) + ε$* (5. 15)

εtotal is the total strain produced, εe is the elastic strain, εp is the plastic strain

and εth is the thermal strain. The elastic strain is modeled using the isotropic

Hooke’s law with temperature dependant Young’s modulus and Poisson’s ratio.

The thermal strain is computed using the temperature dependent coefficient of

thermal expansion. For the plastic strain, a rate-independent plastic model is

employed with the following features: the Von Mises yield surface, temperature-

dependent mechanical properties, and linear kinematic hardening model.

Kinematic hardening is taken into account as an important feature because material

points typically undergo both loading and unloading in the course of the welding

process. In this mechanical analysis, the temperature histories obtained from the

106
thermal analysis were given as input as thermal loadings into the structural model.

The thermal strains and stresses were calculated at each time increment. The final

states of residual stresses were obtained by totaling the thermal and mechanical

strains and stresses.

{σ } = [D ] {ε e } (5. 16)

{σ } = [D ] ({ε } − {α } (T − T0 )) (5. 17)

For elastic range: The stress strain relations for thermo-elastic analysis. The

materials were assumed to be isotropic and the linear strain hardening behavior

was used. The equilibrium equations were defined as in equation

σ ij + ρb i = 0 and σ ij = σ ji (5.18)

σij was the stress tensor. The following thermal elastic plastic constitutive

equations were used in the simulations.

[dσ] = [Dep][dε]-[Cth]dT (5.19)

and

[Dep]=[De]+[Dp] (5.20)

where, [De] is the elastic stiffness matrix, [Dp] is the plastic stiffness matrix, [Cth]

is the thermal stiffness matrix, [dσ] is the stress increment and [dε] is the strain

increment. The residual stresses mainly depend on the thermal histories

experienced by the work piece.

107
5.4 FINITE ELEMENT MODELLING OF FSW

Thermal aspects play a significant role in the modelling and understanding

of FSW .While the real process is thermo-mechanical in the sense that the thermal

and mechanical aspects of the process are coupled. This study considers three-

dimensional nonlinear thermal and thermo-mechanical analyses using the finite

element welding simulation code COMSOL.

Thermal models differ greatly in complexity, but all are based on the heat

conduction equation

+ ∇, = ∇. -∇ . + / (5.21)

in the case of an Eulerian formulation with a convective term:

+ ∇, = ∇. -∇ . + − 0 (5.22)

Eulerian is advantageous since the steady state condition that exists with

respect to the moving heat source.

In equations. 5.21 and 5.22, denotes the material density [kg/m3 ], C the

heat capacity[J/kg º K], T is the temperature [K], k the thermal conductivity [W/m

º K], q, the volume heat input [W/m3] and u the welding velocity vector[m/s]. The

solution procedure and its complexity depends very much on the assumptions

made regarding welding set up, geometry, boundary conditions and the type of

heat source.

108
5.4.1 Mesh

Different elements were used for finite element analysis. In order to reduce

the computational time and achieving accuracy in the result, coarser mesh was

carried out away from the tool and fine mesh was carried out close to tool. Fig. 5.6

depicts the mesh pattern of the present study. The meshed model composed of 851

quadrilateral elements, 915 hexahedral elements, 274 edge elements and 32 vertex

elements. Mesh size varied from 2.5 to 20. The input parameters used for FEA was

tabulated in the Table 5.3.

Fig. 5.6 Meshed model of AA 5059 Aluminum Alloy

109
Table 5.3 Input Parameters used for FEA

Name Expression Description

T0 303[K] Ambient temperature

T_melt (plate 1) 861[K] Work piece melting temperature


aluminium

T_melt (plate 2) 861[K] Work piece melting temperature


aluminium

h_upside(plate 1) 30[W/(m^2*K)] Heat transfer coefficient, upside


aluminium

h_upside(plate 2) 30[W/(m^2*K)] Heat transfer coefficient, upside


aluminium

h_downside(plate 1) 250[W/(m^2*K)] Heat transfer coefficient,


downside aluminium

h_downside(plate 2) 250[W/(m^2*K)] Heat transfer coefficient,


downside aluminium

epsilon(plate 1) 0. 3[1] Surface emissivity aluminium

epsilon (plate 2) 0. 3[1] Surface emissivity aluminium

u_weld 25[mm/min] Welding speed

mu (plate 1) 0.3 [1] Friction coefficient of aluminium

mu (plate 2) 0.3[1] Friction coefficient of aluminium

n 950[1/min] Rotation speed (RPM)

omega 2*pi[rad]*n Angular velocity (rad/s)

F_n 3.4[kN] Normal force

r_pin 2[mm] Pin radius

r_shoulder 6[mm] Shoulder radius

A_s pi*(r_shoulder^2 - Shoulder surface area


r_pin^2)

110
5.4.2 Model Assumptions

A number of assumptions have been made in developing the finite element,

thermal model which include:

• The downward force applied to the work piece from the tool creates the

uniform pressure between the tool and the work piece.

• Work piece material is isotropic and homogeneous.

• No melting occurs during the welding process

• Heat transfer from the work piece to the clamp is negligible

5.4.3 Boundary Conditions

The various boundary conditions were obtained based on the assumptions

made from data available in various published Literature [L.E.Linddgren et al

(1988), Mahoney et al(1998)]. Convective and radiative heat losses to the ambient

occurs across all free surfaces of the work piece and conduction losses occur from

the work piece bottom surface to the backing plate.

1. Mechanical Boundary Condition

The specified mechanical boundary conditions are those just sufficient to

prevent rigid body motion of the model. Zero displacement conditions were used

for constraining the butt joint which resembled the complete fixed fixturing.

2. Thermal Boundary Condition

For FSW process, the samples are clamped at the backing plate. Thus the

heat generated during welding was dissipated by heat conduction. Heat loss from

111
the entire surface other than the backing plate are assumed as heat loss due to

natural convection and surface-to-ambient radiation at the free surface. In order to

account for the conductive heat loss through the bottom surface of weld plates, a

high overall heat transfer coefficient has been assumed. This assumption is based

on the previous studies.

An effect that has a large impact on the global temperature fields is the

heat loss from the work piece to the backing plate governed by the equation

QR=h(T-Tbp) (5.23)

where, h is the heat transfer coefficient and Tbp the temperature of the backing

plate. The heat transfer coefficient is non uniform across the area of contact and is

a function of contact pressure, backing plate material, surface roughness and other

factors, thus making it hard to determine appropriate values. The material data for

the temperature dependent yield stress is shown in Fig. 5.7.


Yield stress (MPa)

Temperature in kelvn
Fig. 5.7 Yield Stress Function of Temperature

112
The yield strength is an expression of the stress essential to create

significant plastic deformation in a material under uniaxial tensile or compressive

load. As the temperature increases and the material softens, yield strength

decreases. Above the solidus temperature, the material acts as a fluid. As an

outcome, the material close to the tool/material interface will reduce its heat

generation to negligible values if exceeding the solidus temperature. Subsequently

reducing the temperature level and allowing the material to recover its strength.

Temperature dependent material properties of AA 5059 is given in Table 5.4. In

this work, tool made by H13 material was used. The property of tool is listed in the

Table 5.5.

Table 5.4 Temperature Dependent Material Properties of AA 5059

Temperature, °C Thermal conductivity, W/mK Heat capacity, J/Kg K

37.8 162 945

93.3 177 978

148.9 184 1004

204.4 192 1028

260 201 1052

315.6 207 1078

371.1 217 1104

426.7 223 1133

113
Table 5.5 Thermo mechanical properties of Tool H13

Name Value Units


Heat capacity at constant pressure 460 [J/(kg K)]

Thermal conductivity 24.3 [W/(mK]


Density 7870 [kg/m^3]

5.4.4 Heat Input Calculation

In this process, heat input is calculated by using the following formula


μ Pω Rs (5.24)
3s

mm
where, S is the welding speed in , P is the axial force, ω is the rotational speed in rps,
s

Rs is the shoulder radius and E is the coefficient of friction

[Lakshminarayanan et al., (2009)]

F
Heat input= × 0.4 × 3.4 × 1.85 × 6
G×I." JJ

[Note- 950 rpm converted as 1.85 rps]

Heat input value was estimated as 75.848 J/mm.

5.5 NUMERICAL PROCEDURE FOR FSW

5.5.1 Temperature Analysis

The heat generated is intense locally and propagates rapidly into

subsequent regions of the plates by heat conduction according to (equation 5.1) as

well as convection and radiation through the boundary. Three-dimensional heat

transfer model for the work piece assumes the tool shoulder and pin to have a

constant heat flux. Heat generation in shoulder-work piece interface involves the

114
interactive effect of major elements such as vertical pressure, area subjected to

friction and angular velocity. It can be mathematically expressed as

qs = (μ ×F/As) × (R× ω) (5.25)

where, F represents the normal force, As is the shoulder’s surface area, μ is

the friction coefficient, R is the distance from the center axis of the tool and ω is

the angular velocity of the tool (rad/s).

The model simulates the heat generated in the interface between the tool’s

pin and the work piece as a surface heat source and given as

qp = µ/sqrt(3× (1 + μ2)) × (rp× ω) ×Ybar (T) (5.26)

where, μ is the friction coefficient, rp denotes the pin radius, ω refers to the

pin’s angular velocity (rad/s), and Ybar(T) is the average shear stress of the

material. As indicated, the average approximated shear stress is a function of the

temperature; for this model, this function with an interpolation function is

determined from experimental data.

5.5.2 Residual Stress Analysis

Calculation of the residual stress in FSW due to thermal gradient induced

strains, strains caused by micro structural changes and strains due to

mechanical loading requires thermo elasto-plastic formulations. In the stress

analysis, the temperature history obtained from the thermal analysis is taken as

input for determining thermal strains and stresses. Thermal strains and stresses

are calculated in addition to the mechanical strains and stresses due to the

axial load of the tool and the cumulative stresses are found at each load

115
step. Before the next temperature increment, these stresses are then added to

those at nodal points to update the behavior of the model.

Thermal effects are accounted for in the stress analysis by including

thermal strains and temperature dependent material properties. Also the thermal

state affects the physical yield criterion. The constitutive laws relating

stresses, strains, and temperature are nonlinear in the theory of plastic

deformation of solids. Two basic sets of equations relating to the elasto-plastic

model, the equilibrium equations and the constitutive equations are considered

[Teng et al (1998)]. One is equation of equilibrium (in Cartesian co-ordinates)

another one is constitutive equations for an elasto-plastic material. For an

isotropic material, there are only two independent elastic constants K and G,

known as Lame’s constants and the generalized Hook’s Law gives the

stress –strain relations.

5.6 SUMMARY

The modelling procedures were evolved and the various heat transfer

equations were incorporated for the two processes namely, the gas tungsten arc

welding and friction stir welding. In the preprocessing stage, the three

dimensional model was generated by defining the nodes later joining the nodes to

form the element, the boundary conditions were defined. Then the loads were

applied. The next step was to solve the set of equations which was done in the

solution stage by using solve command. The final stage is the post processing

stage during which the various results have been analyzed. The procedure for

developing a thermal and thermo mechanical model to predict the thermal cycles

and residual stresses using finite element code COMSOL has been dealt with.

116

You might also like