Al-Sa’ady, M. H., et al.

: Finite Difference Simulation of Low Carbon …

THERMAL SCIENCE, Year 2011, Vol. 15, No. 1, pp. 207-214 207

FINITE DIFFERENCE SIMULATION OF LOW CARBON STEEL

MANUAL ARC WELDING

by

Moneer H. AL-SA’ADY a, Mudar A. ABDULSATTAR b*, and Laith S. AL-KHAFAGY a

a
Foundation of Technical Education,Technical College-Baghdad, Baghdad, Iraq
b
Directorate of Materials Science, Ministry of Science and Technology, Baghdad, Iraq

Original scientific paper

UDC: 621.791.053:66.011:669.15-194
DOI: 10.2298TSCI100206055S

This study discusses the evaluation and simulation of angular distortion in weld-
ing joints, and the ways of controlling and treating them, while welding plates of
(low carbon steel) type A-283-Gr-C through using shielded metal arc welding.
The value of this distortion is measured experimentally and the results are com-
pared with the suggested finite difference method computer program. Time de-
pendent temperature distributions are obtained using finite difference method.
This distribution is used to obtain the shrinkage that causes the distortions ac-
companied with structural forces that act to modify these distortions. Results are
compared with simple empirical models and experimental results. Different
thickness of plates and welding parameters is manifested to illustrate its effect on
angular distortions. Results revealed the more accurate results of finite difference
method that match experimental results in comparison with empirical formulas.
Welding parameters include number of passes, current, electrode type and geo-
metry of the welding process.
Key words: finite difference method, welding, low carbon steel

Introduction

Welding processes depends on local heating of the welded parts up to melting

temperature. Simulations of welding relays on the solution of heat transfer equations
accompanied with the appropriate mechanical equations. These solutions started at the thirties
of the twentieth century as analytical solutions [1]. After 30 years more accurate numerical
solutions started to appear [2]. The evolution of these solutions was a direct result of the
advances in computer speeds. Numerical methods include mostly two methods, finite element
method (FEM) and finite difference method (FDM) and their derivatives. Other less known
methods also exists [3]. Finite difference method is more than a mid way between solutions of
the over complicated FEM and simple analytical solutions. It is always possible to reduce the
size of the uniform mesh steps encountered in FDM to account approximately for curved
geometrical parts [4]. FDM main variables are temperature, time, geometry, and materials
properties. This method is still evolving to reduce large computer times of the 3-D case [4].

*nCorresponding author; e-mail: mudarahmed3@yahoo.com

 Al-Sa’ady, M. H., et al.: Finite Difference Simulation of Low Carbon …
208 THERMAL SCIENCE, Year 2011, Vol. 15, No. 1, pp. 207-214

Many kinds of distortions of the welded pieces occur during the welding process such as
longitudinal, transverse, buckling, and angular distortions. These distortions are due to the
heating and cooling of both welded and filler materials. In this work we shall concentrate on
angular distortions. Simplified empirical equations were suggested previously to calculate
angular distortions such as [5, 6] which we will compare with our results. Numerical
simulations of welding are not confined to arc welding but can be also found in other methods
such as laser welding [7-10]. Sophisticated numerical programs and methods are now in a
position that can yield accurate and reliable results [11-15]. In spite of being one of the first
suggested numerical methods, finite difference method is still used in many branches of
thermal and mechanical calculations [16-22] because of its speed and simplicity especially in
simple geometrical cases.

Theory

In this work transient heat conduction equation is solved using FDM. This equation
in Cartesian co-ordinates is given by:
T    2T  2T  2T  Q
     (1)
t  c  x 2 y 2 z 2   c
In this equation, T, t, λ, ρ, and c are the temperature, time, thermal conductivity, density, and
heat capacity, respectively. Q is the volumetric density of the power input. This factor also
includes the volumetric power due to phase transformation such as melting, solidification, or
reaction. The sum of Q over all the meshes that receives electrical power of the electrode is
equal to the electrical power (IU) times the efficiency. At the boundary, heat is exchanged
with the outer environment via convection or radiation. Heat convection is described by the
following equation:
qc = hA(Tw – T) (2)
where Tw is the surface temperature, T∞ – the temperature of the surrounding fluid, A – the
surface area, and h – the convection heat transfer coefficient. Heat radiation is described by
the equation:
qr  F FG A(Tw4  T 4 ) (3)
where Fε and FG are emmissivity and geometry coefficients, respectively. σ is the Stefan-
-Boltzman constant. FDM with equal meshing in 3-D (FORTRAN language program) is used
to simulate the welding processes. Input to the simulation program includes thermal
conductivities, densities, heat capacities emmissivity coefficient, convection coefficient, and
geometry of the welded pieces and their surroundings. Current, voltage, efficiency of the
welding machine are also given as an input to the simulating program. The FDM is used to
solve heat transfer equations including conduction, convection, and radiation. The simulation
process includes two steps. The first step is concerned with the simulation of electrode
heating and melting. Not all the input welding power is converted to heating the electrodes.
Some of this power is lost in other parts of the electrical circuit. The overall efficiency is
approximately 50% in most welding processes. This step shows that radiation is the dominant
heat loss process due to the high temperature of the molten electrode and high emmisivity of
steel. Convection through air and conduction of heat are less important in this step. The
second step investigates the cooling of the molten electrodes in the joints and its effects on the
Al-Sa’ady, M. H., et al.: Finite Difference Simulation of Low Carbon …
THERMAL SCIENCE, Year 2011, Vol. 15, No. 1, pp. 207-214 209

welded plates. 3-D temperature distribution in the welded pieces are given as an output of the
program and used to understand the extent at which the material of welded pieces are affected
by the heat input of the welding processes. The angular distortions are also given as an output
depending on the welding process parameters. These angular distortions are caused by
thermal stresses, plastic deformation, and phase transformation (liquid to solid transforma-
tion) [23, 24]. Cooling rates and temperatures of the various parts is shown to interpret the
microstructure and properties of the weld and base plates. These rates can be optimized and
correlated to welding quality. Simplified empirical equations were suggested previously to
calculate angular distortions such as [5]:

  2TS tan (4)
2
where a is the thermal expansion coefficient, TS – the softening material temperature which is
600 C for steel, and θ is the groove angle. Another equation is given by [6]:
U
  0.13 (5)
vH 2
where I is the current, U – the applied voltage, v – the welding speed, and H – the plate
thickness. The above empirical equations and other related equations give rise to different
errors depending on their sophistication and range of applicability.

Experimental procedure

Two plates of 200 × 100 mm sides are used. The welding is on the 100 mm side.
Different thickness of plates is also used to explore the effect of welded pieces thickness on
temperature distribution and angular distortions. This included 4, 6, and 10 mm plates. A
welding speed of 78 mm/min. is chosen. The current and voltage with the values of 24 V and
100 to 150 A is used with E 6010 and E 7018 electrodes. The electrodes diameter is 2.5 and
3.23 mm. Square and single butt joints are investigated (shape of plates prepared space for
welding). One, two and three passes is used to fill the joints. The above dimensions and
materials cover ranges that is around the consistent ASME (American Society for Mechanical
Engineers) standards and ends slightly above the ASME standards. A welding machine
(ESAB) is used. The plates are prepared by the above mentioned dimensions, cleaned and
grooved in square or single-v butt joints. Materials properties that include thermal and
mechanical properties can be found in references such as [23].

Results and discusion

Generally, the value of angular distortions increases with the increase of input
current. The kind, geometry, and shape of the joints also affect the values of angular
distortions. Mesh size of 1 mm is used in discretizing FDM in 3-D space. The shapes and
geometries described in the experimental procedure section are used in the simulating
program. In addition to the above-mentioned geometries, the two welded pieces are 2 mm
apart (at the root part of the welded pieces) at the beginning of the welding processes. In fig. 1
an upper view of the two welded pieces shows that temperature decreases quickly when the
welding process proceeds. Since the electrode melts down and fills some of space (depending
 Al-Sa’ady, M. H., et al.: Finite Difference Simulation of Low Carbon …
210 THERMAL SCIENCE, Year 2011, Vol. 15, No. 1, pp. 207-214

Figure 1. Upper view shows the distribution of temperatures (in ˚C)

between the two welded pieces after 15 seconds as welding process
proceeds; numbers on the axes represent mesh numbers of FDM
(color image see on our web site)
on welding speed) between welded pieces as a liquid, the temperature profile shape would be
like a line at the highest gradient temperature region. The value of temperature is less than
half the original welding temperature just 5 mm after the advance of welding electrode. Figure
2 also shows the same reduction of temperature but in time scale. In this figure temperature re-

Figure 2. Sectional view of temperature distribution (in ˚C) of the two welded pieces at 1, 7, and 13
seconds from the starting of welding process; numbers on the axes represent mesh numbers of FDM
(color image see on our web site)
Al-Sa’ady, M. H., et al.: Finite Difference Simulation of Low Carbon …
THERMAL SCIENCE, Year 2011, Vol. 15, No. 1, pp. 207-214 211

duces to less than half the original welding temperature in just 7 seconds. Because of rapid
solidification, the electrode melt will not have time to fill the space between the two
electrodes at the weld root. This space will still have the ambient air temperature which is
categorized from 0 to 200 °C. These two figures show the sharp temperature disintegration in
both space and time co-ordinates. Figure 2 (7, 13 s) shows that temperature disintegration
slows down as temperatures reduces to less than 1000 °C. Table 1 shows experimental and
theoretical angular distortions in welding 6 mm thickness plates in one pass using single-v butt
joints. In this table we can see that increasing the current does not always increases the angular
distortion. This is due to the fact that increasing the current results in some fluid flow of the
molten filler and base metal material which is not taken into account in the present FDM.
However, the results of FDM are far better than the general empirical equations. Using more
than one welding pass complicates the situation further. It is obvious from the results of tab. 2
that increasing the number of passes decreases the value of angular distortion but this has an
economical loss of additional filler material and efforts. Tables 3 and 4 compares the effect of
using two and three passes on the angular distortions. Three passes shows less angular
distortions than two passes. However an optimal current must be chosen to obtain the minimum
distortions. In the last table (tab. 5) a square butt joint is used. The angular distortions in this
case are less than the comparable single-v joints. This is due to the fact that a contraction of the
solidified filler molten material is equal both on the surface and inner part of the joint. However,
square joints need more filler or electrode material than single v-joints. This also confirms that
geometry has an important effect on angular distortion.

Table 1. Experimental and theoretical angular distortions in welding 6 mm thickness plates in one pass
using single-v butt joints
Finite
diference Practical
Empirical Empirical
Welding Electrode computed angular
Voltage Current Electrode eq. 4 eq. 5
speed diameter angular distortion
[V] [A] type (10–2 (10–2
[mms–1] [mm] distortion (10–2
radians) radians)
(10–2 radians)
radians
78 22 80 2.4 E7018 1.44 8.14 5.865 4.971
78 22 100 2.4 E7018 1.44 10.18 6.248 6.995
78 22 125 2.4 E7018 1.44 12.73 6.648 5.369

Table 2. Angular distortions in welding 6 mm thickness plates in two opposite pass using single-v
joints; angular distortions are in radians
Amperage of 1st run of welding Amperage of 2nd run of welding
(2.4 mm diameter of welding (3.25 mm diameter of welding Angular distortion (10–2 radians)
electrode) electrode)
80 A 100 A 2.011
100 A 125 A 2.145
125 A 150 A 1.752
 Al-Sa’ady, M. H., et al.: Finite Difference Simulation of Low Carbon …
212 THERMAL SCIENCE, Year 2011, Vol. 15, No. 1, pp. 207-214

Table 3. Angular distortions in welding 10 mm thickness plates in two opposite pass

using single-v joints
Amperage of 1st run of welding Amperage of 2nd run of welding
Angular distortion
(2.4 mm diameter of welding (3.25 mm diameter of welding
(10–2 radians)
electrode) electrode)
80 A 100 A 4.482
100 A 125 A 4.612
125 A 150 A 3.719

Table 4. Angular distortions in welding 10 mm thickness plates in three opposite pass

using single-v joints
Amperage of 1st run of welding Amperage of 2nd and 3rd run of
Angular distortion
(2.4 mm diameter of welding elec- welding (3.25 mm diameter of
(10–2 radians)
trode) welding electrode)
80 A 100 A 3.422
100 A 125 A 1.611
125 A 150 A 2.032

Table 5. Angular distortions in welding 4 mm thickness plates in three opposite pass

using square butt joints
Amperage of 1st run of welding Amperage of 2nd and 3rd run of
Angular distortion
(2.4 mm diameter of welding elec- welding (3.25 mm diameter of
(10–2) radians
trode) welding electrode)
80 A 100 A 0.549
100 A 125 A 1.707
125 A 150 A 0.601

Conclusions

The formation of angular distortions is greatly affected by a number of factors.

Among these are the input current, geometry of the joint, the thickness of welded plates and
the kind, thermal and mechanical properties of the welded plates and filler material. Generally
speaking we can observe the following points.
 Angular distortions increase with increased current. This is true as long as that the melt-
ing of base plates does not reach the other side of the plate. If this happens some molten
metal flows to the other side of the plates causing the distortion to decrease. A more so-
phisticated program is needed to account for the fluid flow of molten filler electrode
which is not taken into account in the present FDM program.
 Angular distortions decrease with the increased number of opposite passes. The opposite
side welding tends to eliminate the effect of each other. Shrinkages at opposite sides of
the welding tend to eliminate each other effects. This is opposed by increasing welding
expenses and efforts.
 Angular distortions decrease with the increased thickness of plates. This is due to the fact
that each plate act by its own weight torque on the joint causing it to decrease.
Al-Sa’ady, M. H., et al.: Finite Difference Simulation of Low Carbon …
THERMAL SCIENCE, Year 2011, Vol. 15, No. 1, pp. 207-214 213

 Square butt joints have fewer distortions than single-v joints. This is due to the fact that a
contraction of the solidified filler molten material is equal both on the surface and inner
part of the joint.
 FDM results are better than simple empirical results since it takes into account more
thermal and mechanical effects on the formation of the joints.

Nomenclature

A – area, [m2] Greek letters

c – specific heat, [Jkg–1C–1]
α – the thermal expansion coefficient
FG – geometry coefficient
β – angular distorsion, [radians]
Fε – emmisivity coefficient
l – thermal conductivity, [Wm–1K–1]
H – plate thicknes, [mm]
 – density, [kgm–3]
h – heat transfer coefficient, [Wm–2C–1]
σ – Stefan-Boltzman constant, [Wm–2k–4]
I – current, [A]
θ – groove angle, [radians]
Q – volumetric density of the power input,
– [Wm–3] Subscripts
qc – heat transfer by convection, [Wm–2]
 – surrounding fluid
qr – heat transfer by radiation, [Wm–2]
T – temperature, [C] Acronyms
TS – softening material temperature, [C]
Tw – surface temperature, [C] FDM – finite difference method
T – temperature of the surrounding fluid, [C] FEM – finite element method
U – Voltage, [V]
v – welding speed, [mms–1]

References

[1] Rosenthal, D., Schmerber, R., Thermal Study of Arc Welding, Experimental Verification of Theoretical
Formulas, Am. Weld. Journ, 17 (1938), 4, pp. 2s-8s
[2] Westby, O., Temperature Distribution in the Work-Piece by Welding, Ph. D. thesis, The Technical
University of Norway, Trondheim, Norway, 1968
[3] Rappaz, M., Modelling of Microstructure Formation in Solidification Processes, International Materials
Review, 34 (1989), 3, pp. 93-123
[4] Igarashi, K., Method for Automatic Optimization of Finite Difference Grids in Simulator, United States
Patent no. 5991526, 1999
[5] Okerblom, N. O., Technological and Structural Design of Welded Structures (in Russian),
Mashinostroenie, Moscow, 1964
[6] Okerblom, N. O., The Calculations of Deformation of Welded Metal Structures (in Russian), Mashgiz,
Moscow, 1955
[7] Shibib, K. S., Minshid, M. A., Tahir, M. M., Finite Element Analysis of Spot Laser of Steel Welding
Temperature History, Thermal Science, 13 (2009), 4, pp. 143-150
[8] Maiti, A., et al., Finite Element Simulation of Laser Spot Welding, Science and Technology of Welding
and Joining, 8 (2003), 5, pp. 377-384
[9] Jiang, W., Yahiaoui, K., Hall, F. R., Finite Element Predictions of Temperature Distributions in a
Multipass Welded Piping Branch Junction, J. Pressure Vessel Technol., 127 (2005), 1, pp. 7-13
[10] Frewin, M. R., Scott, D. A., Finite Element Model of Pulsed Laser Welding, Welding Research
Supplement (1999), pp. 17s-22s
[11] Zhu, X. K., Chao, Y. J., Numerical Simulation of Transient Temperature and Residual Stresses in
Friction stir Welding of 304L Stainless Steel, Journal of Materials Processing Technology, 146 (2004),
2, pp. 263-272
[12] Abida, M., Siddique, M., Numerical Simulation to Study the Effect of Tack Welds and Root Gap on
Welding Deformations and Residual Stresses of a Pipe-Flange Joint, International Journal of Pressure
Vessels and Piping, 82 (2005), 11, pp. 860-871
 Al-Sa’ady, M. H., et al.: Finite Difference Simulation of Low Carbon …
214 THERMAL SCIENCE, Year 2011, Vol. 15, No. 1, pp. 207-214

[13] Yaghia, A., et al., Residual Stress Simulation in Thin and Thick-Walled Stainless Steel Pipe Welds
Including Pipe Diameter Effects, International Journal of Pressure Vessels and Piping, 83 (2006),
11-12, pp. 864-874
[14] Mackerle, J., Finite Element Analysis and Simulation of Welding: a Bibliography (1976-1996),
Modelling Simul. Mater. Sci. Eng., 4 (1996), 5, pp. 501-533
[15] Koseki ,T., et al., Numerical Simulation of Equiaxed Grain Formation in Weld Solidification, Science
and Technology of Advanced, 4 (2003), 2, pp. 183-195
[16] Mezrhab, A., Bouzidi, M., Lallemand, P., Hybrid Lattice-Boltzmann Finite-Difference Simulation of
Convective Flows, Computers & Fluids, 33 (2004), 4, pp. 623-641
[17] Costa, M., Buddhi, D., Olivia, A., Numerical Simulation of a Latent Heat Thermal Energy Storage
System with Enhanced Heat Conduction, Energy Conversion and Management, 39 (1998), 3-4, pp. 319-
330
[18] Andrés, E., et al., A Hybrid Spectral/Finite-Difference Large-Eddy Simulator of Turbulent Processes in
the Upper Ocean, Ocean Modelling, 30 (2009), 2-3, pp. 115-142
[19] Tamura, A., Tsutahara, M., Kataoka, T., Numerical Simulation of Two-Dimensional Blade-Vortex
Interactions Using Finite Difference Lattice Boltzmann Method, AIAA Journal, 46 (2008), 9, pp. 2235-
2247
[20] Grzesik, W., Bartoszuk, M., Prediction of Temperature Distribution in the Cutting Zone using Finite
Difference Approach, International Journal of Machining and Machinability of Materials, 6 (2009),
1-2, pp. 43-53
[21] Bakier, A. Y., Mansour, M. A., Combined of Magnetic Field and Thermophoresis Particle Deposition in
Free Convection Boundary Layer from a Vertical Flat Plate Embedded in a Porous Medium, Thermal
Science, 11 (2007), 1, pp. 65-74
[22] Mohammed, H. A., Salman, Y. K., Numerical Study of Combined Convection Heat Transfer for
Thermally Developing upward Flow in a Vertical Cylinder, Thermal Science, 12 (2008), 2, pp. 89-102
[23] Callister, W. D., Materials Science and Engineering: An Introduction, John Wiley & Sons, Inc., New
York, USA, 2000
[24] Hughes, W. F., Gaylord, E. W., Schaum’s Outline Series in Basic Equations of Engineering Science,
McGraw-Hill, USA, 1964

Paper submitted: February 6, 2010

Paper revised: March 17, 2010
Paper accepted: May 22, 2010