Thermal Modelling of Laser Welding

Available online at www.sciencedirect.
com
Optics & Laser Technology 37 (2005) 99 115

www.elsevier.com/locate/optlastec
Thermal modelling of laser welding and related processes:

a literature review
A.P. Mackwooda , R.C. Craferb;
a Physics
b Abington
Centre, The University of Essex, Colchester CO4 3SQ, UK

Consultants, 78 High Street, Great Abington, Cambridge CB1 6AE, UK
Received 24 November 2003; received in revised form 23 February 2004; accepted 24 February 2004
Abstract
The main emphasis of this review is on thermal modelling and prediction of laser welding in metals. However as similar techniques
are employed to model conventional welding processes such as arc, resistance and friction, as well as related processes such as alloying,
cladding and surface hardening, part of this review is given over to the modelling of these processes where appropriate. The time frame
of the review is up to the year 2002.
? 2004 Elsevier Ltd. All rights reserved.
Keywords: Modelling; Simulation; Laser welding; Welding; Surface treatment; Literature review
1. Introduction
The source material for this review is taken from a Doctoral Thesis entitled Numerical Simulations of Thermal
Processes and Welding by one of the authors (A.P.M),
submitted to the University of Essex, UK, in January
2003. References are therefore included up to the year
2002. One principal outcome of this doctoral work is a
=exible thermal modelling code named TS4D 1 (Thermal Simulation in 4 Dimensions, there being 3 spatial
dimensions + time). It is the culmination of modelling expertise at Essex University over a period of more than a
decade by the team supervised by Professors Phiroze Kapadia and John Dowden. TS4D can handle multiple sources
in both surface and volume format, surface sinks, multiple
material workpieces, a variety of surface cooling mechanisms, and produces both steady state and time-dependent
predictions.
Although the main emphasis of this review is on prediction of laser welding in metals, similar techniques
are employed to model conventional welding processes
such as arc, resistance and friction, as well as alloying,
cladding and surface hardening. Indeed modellers in one
Corresponding author. Tel.: +44-1223-891576.

E-mail address: roger@abingtonconsultants.co.uk (R.C. Crafer).
1 Details of TS4D can be found on the Abington Consultants website
abingtonconsultants.co.uk.
0030-3992/$ - see front matter ? 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.optlastec.2004.02.017
Deld tend to publish in the other Delds as well, so part

of this review is given over to related processes where
appropriate.
Laser keyhole welding is often referred to as a high energy density or power beam technique. The fact that absorption of a laser beam increases with temperature has enabled
the use of the laser beam as a practical heat source for welding. For a CO2 laser, the absorptivity of carbon steel [1]
varies from 4% at room temperature to more than 30% at
melting temperature and reaches about 90% at vaporising
temperature.
For all laser welding irrespective of the type of laser
employed, energy is absorbed at the surface of the metal
in a layer only a few nanometres thick by a process known
as Fresnel Absorption. If the intensity is high enough,
vaporisation occurs with some of the metal electrons becoming free (ionisation). These free electrons then absorb
energy directly from the beam by a process known as inverse Bremsstrahlung. This results in higher temperatures,
increased ionisation and increased absorption leading to
vaporisation of the surface which forms a small depression
in the workpiece. As the depression deepens, a keyhole
forms and the laser light is scattered repeatedly within
it, so that Fresnel Absorption occurs at the keyhole walls
too, thus increasing the coupling of laser energy into the
workpiece. As the keyhole develops, the power of the
source can now be absorbed at greater depths, not just at
100
A.P. Mackwood, R.C. Crafer / Optics & Laser Technology 37 (2005) 99 115
the surface. Some absorption or scattering of the power

also occurs in the plasma vapour within the keyhole that
can emerge as a plume which can obstruct or defocus the
beam (especially with CO2 lasers). However, the plume
can also radiate energy back to the specimen. While the
laser energy is applied, surface tension and gravitational
forces have the eKect of closing the keyhole whereas vapour
pressure and ablation help keep it open. Laser welding is
usually realised at speeds much higher than conventional
processes.
All the references discussed in this review are based on
solutions of the heat conduction equation. The analytical solutions on which most models have been based are described
in Conduction of Heat in Solids by Carslaw and Jaeger [2].
This contains a collection of solutions for most simple geometries and shapes of heat source, both steady state and
time-dependent. Readers interested in the mathematics and
solution methods are referred to this work.
2. Basic analytic solutions
2.1. Moving point source in a medium of in=nite thickness
Rosenthal [3] published one of the earliest analytic solutions applicable to welding. He considered a point source
incident on, and moving relative to, an inDnite material. His
solution simulates surface melt runs relating to conduction
welding. The solution can be rendered Dnite in depth and
width (assuming the source moves along the length of the
material) by using the method of images. Since he did not
account for the Dnite area of a real source, it is inaccurate
in the region where the point source is incident.
2.2. Moving line source in an in=nite or semi-in=nite
material
Rosenthal also derived the solution for an inDnite line
source extending through the depth of the material, its axis
perpendicular to the top and bottom surfaces. This 2-D solution simulates full penetration welding of sheets of any thickness and can also be applied to conduction, or nonfully penetrating welding of thicker sheets provided the length scale
(the thermal diKusivity divided by the processing speed) is
large enough. For aluminium this value is 6 mm and for
steel it is 0:6 mm. A recurring problem with such point
and line source models is that they lead to inDnite temperatures at the source.
With these solutions, Rosenthal produced an approximate single formula capable of predicting the time and rate
of cooling for a wide variety of thicknesses of steel, and
for ranges of temperature and welding conditions. For both
high- and low-processing speeds, Swift-Hook and Gick [4]
approximated Rosenthals solution and predicted that the
proportion of power needed to cause melting as a function
of the incident power reaches a maximum of 48%.
On the subject of laser cutting, Bunting and CornDeld [5]

obtained a solution due to a cylindrical beam by integrating the line source solution over the area of a circle from
which they were able to evaluate the cut speed (i.e. translation speed) as a function of the incident power per unit
thickness. They found that the eNciency of the cutting process could be maximised for a certain power density and cut
speed, depending on the jet diameter. Their results, other
than in a few cases, were quite diKerent from those found
experimentally, but fared no worse than the listed results of
other authors.
A review of mathematical models of laser cutting of steels
has been presented by ONeill and Steen [6].
2.3. Continuous Gaussian surface source in an in=nite
solid
By integrating point and line source solutions over an area
it is possible to calculate the heating from top-hat, Gaussian
and cylindrical sources, etc. Lax [7] studied the steady-state
temperature distribution due to a stationary Gaussian beam
in a semi-inDnite cylindrical medium, while Nissim et al. [8]
presented a 3-D solution for a moving elliptical Gaussian
heat source. For the case of silicon and gallium arsenide they
also incorporated formulae for a varying thermal conductivity. Their model was generalised by a numerical algorithm
by Moody and Hendel [9].
Miyazaki and Giedt [10] solved the heat conduction equation for a cylindrical molten region having an elliptical
cross-section.
Davis et al. [11] considered laser transformation hardening of En8 steel under the in=uence of a Gaussian surface
source. Ignoring surface melting and also the heat of transition required to convert pearlite to austenite, they calculated
the depth at which hardening occurs and an estimation of
the power required.
2.4. Moving hypersurface line source for a medium of
semi-in=nite thickness
In order to avoid the inDnite temperatures produced by
a point source or line source, Ashby and ShercliK [12],
with reference to transformation hardening, derived a solution (as a development of the AshbyEasterling hyper
point source model [13] for high processing speeds, by
representing the heat input as a Dnite line source situated
above and parallel to the surface with its axis lying along
the width of the specimen. Since it was not Dnite along the
direction of motion, it was positioned above the surface to
avoid the inDnite temperatures generated where the beam
impinges. They also derived the beam energy required to
cause melting.
2.5. Combined moving point/line sources
The point and line source solutions of Rosenthal can
simulate simple conduction welding and keyhole welding,
respectively. However, a typical keyhole weld is usually accompanied by a distinct nail head appearance at the top.
A line source model (where the source extends throughout the entire thickness of the domain) can only lead to a
parallel-sided weld of inDnite extent (i.e. a 2-D solution)
while a point source leads to a weld of roughly hemispherical shape.
Steen et al. [14] combined the point and line source to
model more eKectively a keyhole weld. This idea enabled the
estimation of the power actually absorbed by the weld. The
position of the point source could be variedon or within
the metal. Here the line source represents absorption down
the keyhole and the point source represents the plasma radiation from the plume. It was suggested that the melt width
in the lower part of the weld is directly proportional to the
strength of the line source. As this model assumes an inDnitely thick specimen, it is inappropriate for thin workpieces.
2.6. Extensions of the point/line source models
Akhter et al. [15], extended the solution of Steen et al.
[14] by producing a point and line source model for a
medium of Dnite thickness using the method of images. A
number of point sources were combined with a line source,
the strength of which varied with depth. This enabled more
realistic-looking melt cross-sections. The model was applied
to laser lap-welding of thin sheets of zinc-coated steel. Since
absorption of the laser beam often exhibits a local maximum at the interface of the two sheets, an additional point
source was used at this boundary. However, they were still
unable to model accurately the wine glass appearance of
some welds.
Hamoudi and Ducharme [16] produced a steady-state
welding model based on the point and line source model
of Steen et al. [14] with which they studied a series of
experimental welds in steel under the in=uence of helium
shielding gas. By varying the point and line source strengths
they matched the melt cross-sections as closely as possible
to those of experiment, determining the absorbed power.
They showed that the strength of the line source increases
with processing speed, including a sharp rise when the weld
penetration ceases to be full. In some cases the point source
strength was very small, indicating the eNciency of helium
as a shielding gas.
Another point and line source model was that of Lankalapalli et al. [17]. The keyhole was represented by a Dnite line
source whose strength varies with depth. They showed that
the temperature on the bottom surface of the workpiece is a
consistent indicator of penetration depth.
Postaciouglu et al. [18] produced a solution for the temperature Deld outside the melt pool using a spheroidal coordinate system, approximating the molten region as a half
prolate spheroid.
Kaplan [19], accounting for the fact that the keyhole
is not straight but curved, integrated point sources of an
101
appropriate intensity along a curve in the interior of the

material, rather than along a straight vertical line.
2.7. Periodic moving point and line source solutions for a
medium of semi-in=nite thickness
Dowden et al. [20] constructed analytic solutions for
the temperature Deld in workpieces of inDnite area and
semi-inDnite thickness, under the in=uence of periodic point
and line sources. Their solution generalises the solutions of
Steen et al. [14] and Akhter et al. [15] for workpieces of
semi-inDnite and Dnite thickness, respectively.
2.8. The Stefan problem
Simple solutions of the heat conduction equation ignore
latent heat and utilise constant material properties. Analytic
solutions taking latent heat into account are not straightforward. The Drst published discussion related to phase change
was that by Stefan [21], regarding a study on the thickness
of polar ice. As a result, the solving of the problem with latent heat taken into account is often referred to as the Stefan
problem.
If phase change is considered to take place at a speciDc
temperature (as in the case of pure metals), this can be modelled by considering the solid and liquid as separate regions
coupled via moving boundary conditions. The material properties now take on separate values in the solid and molten
regions. Alternatively, the enthalpy formulation of the heat
conduction equation can be solved which enables an inDnite
speciDc heat (eKected as a zero diKusivity) spanning the enthalpy change between solid and liquid. This method is only
soluble (other than in one dimension) by numerical means.
However, the abrupt change in speciDc heat can hinder convergence. Fortunately, alloys exhibit a range of temperature
over which melting occurs and providing the melting range
is not too small, convergence is not usually a problem. In
this case melting is represented as an increase in the speciDc
heat (which remains Dnite) over a range of temperature.
Very few applications of heat =ow in welding with the
inclusion of latent heat eKects have been reported in an analytic manner owing to their complexity. Analytic solutions
include Neumanns solution [2] and those of Tao [22], Frederick and Greif [23] and Kasuya and Shimoda [24].
Mackenzie and Robertson [25] produced a transient
1-D Dnite-diKerence (FD) code, incorporating a novel
semi-implicit moving mesh method, to solve the enthalpy
formulation of the heat conduction equation to allow for
phase changes. The convection term of the heat conduction equation was treated explicitly with the remaining terms treated implicitly hence allowing for a larger
time-increment. They used an adaptive moving mesh which
is updated every time-step. The authors present examples
of a spot welding process (using a high electric current as
heat source) and a simulation of the freezing of water.
102
Scheerlinck et al. [26] developed a Dnite-element (FE)

enthalpy method for solving nonlinear phase change heat
transfer for arbitrary 3-D geometries (e.g. freezing and thawing of food). They incorporated all nonlinearities caused by
temperature-dependent material properties in a functional
relationship between the volumetric speciDc enthalpy and
the KirchhoK function, hence ensuring that the peak of the
speciDc heat and the abrupt change in thermal conductivity
are not missed.
Nedjar [27] used an iteration scheme for the steady-state,
and a Dnite-diKerence scheme for the transient problem, in
taking account of phase change. The algorithm used is applicable for phase changes which take place either over a
temperature range or at a given temperature.
3. Numerical solutions
3.1. Standard heat transfer solutions
During the last two decades many of the existing weld
processes have been simulated by numerical methods, especially Dnite-element [28]. Other methods such as FD [29] are
simpler to use and understand, and Patankar, using a combination of FD and FE methods, produced the Dnite-volume
(FV) method [28].
Welding and heat treatment are complex processes involving the interaction of thermal, mechanical, electrical
and metallurgical phenomena. As a result most numerical
solutions neglect a number of eKects. One of the Drst numerical solutions of heat transfer for laser materials processing was due to Mazumder and Steen [30] who produced a 3-D model using FD techniques and a relaxation
procedure in solving the heat conduction equation. They
assumed a Gaussian heat source, a workpiece of inDnite
length, 100% absorption of power at temperatures in excess
of the boiling point, radiative heat losses and also convective heat losses due to the shielding gas =ow. Their model
was modiDed by Chande and Mazumder [31] who took
into account temperature-dependent material properties, radiative and convective heat losses and the latent heat of
phase change via an increased speciDc heat over a melting/solidifying temperature range.
Lim and Chan [32] used the boundary element method
(BEM) to model deep penetration welding. In the solid region, heat transfer was simulated by BEM; lubrication theory was applied to the liquid region and non-equilibrium vaporisation and gas dynamics were modelled by a 1-D model,
the solutions in each region being combined via boundary
conditions.
Dowden and Kapadia [33] produced a formula for the
question of what was the thickest workpiece that could be
reliably welded with full penetration occurring at all times.
They assumed a beam of uniform intensity and a keyhole
diameter which was three times greater on the top surface than on the bottom surface. Results from their formula
agreed well with experimental data for input powers of 1,
2, 4 and 6 kW incident on 304 stainless steel. They also

studied a curved keyhole wall and found that it made little
diKerence to their results.
Norman et al. [34] by using the same model and procedure as Hamoudi and Ducharme [16], calculated the temperature Deld, the thermal gradients and the growth rates in
a series of welds in 2090 aluminium alloy. From these data
they determined cooling rates and macroscopic grain structure along the weld centerline. They found fair agreement
between dendrite secondary arm spacings measured by experiment and those predicted using the thermal model.
Mohanty and Mazumder [35] produced a 3-D numerical
keyhole welding simulation model complete with graphical user interface and visualisation modules. The model is
based on the balancing of surface forces and energy at the
liquid/vapour and solid/liquid interfaces.
Semak et al. [36] produced a laser welding model based
on a hydrodynamic physical model of lasermaterial interaction including the eKect of evaporation recoil pressure and
convection on melt =ow. Their simulations show that convective heat transfer induced by recoil pressure, is significant at intensities corresponding to laser welding, cutting,
and drilling due to melt ejection. They also showed that motion in the molten zone outwards from the beam results in a
secondary temperature maximum. In a related publication,
Semak et al. [37] conDrmed that maximum penetration was
dependent on the speciDc location of the beam focus, the
latter being positioned above or below the surface of the
metal depending on whether there is strong or weak beam
absorption in the keyhole plasma.
3.2. Welding dissimilar metals
Joining dissimilar metals is very diNcult, especially when
the thermal conductivity of each is substantially diKerent
(e.g. copper and steel), since this aKects absorption of laser
light. Generally speaking, in order to weld two dissimilar
metals, mutual solubility is required [38]. In some cases a
third metal which is soluble with the other two is needed
to produce a successful joint, for example copper and steel
can be welded by the inclusion of a nickel alloy which is
soluble with both of them. If the thermal expansion coefDcients of dissimilar metals diKer greatly, there will be internal stresses set up in the intermetallic zone during any
temperature change of the weldment.
Chung and Wei [39] produced a fully implicit staggered
time-dependent FD method [28] to investigate electron beam
or laser welding of dissimilar metals. They obtained velocity
and temperature Delds at the maximum melt cross-section
as functions of the dimensionless surface tension coeNcient,
which compared favourably with experimental results. They
used a nonlinear mesh which was Dner under the heat source,
the discretisation equations solved by a line-by-line scheme
applied to the tri-diagonal matrix algorithm (TDMA) [28].
The continuity and momentum equations were solved by the
SIMPLE [28] procedure.
Phanikumar et al. [40] studying the laser welding of dissimilar metals, attempted modelling the welding of copper
and nickel. A pressure-based FV technique was employed
which involves phase change (using the enthalpy-porosity
technique [41,42]), convection in the melt pool, melting and
mixing. The dissimilar metal properties were handled using
appropriate mixture theories.
3.3. Multipass welding
Reed and Bhadeshia [43] constructed a partly analytic
partly numerical model to describe the thermal cycles occurring in multipass welds by using a matrix approach where
each element of the discretised workpiece is associated with
one element of a multidimensional matrix.
Lindgren et al. [44] produced a computer code to simulate
multipass butt welding. Since adding Dller wire increases
the volume of the workpiece, each pass either needs a reconstruction of the grid where new elements are added, or
it requires quiet elements which is more time-consuming.
The authors discuss the relative merits of both inactive and
quiet approaches and produce results for both methods. Although both methods gave similar results they could not
be compared with experimental results due to a number of
problems. Pulsed laser welding allows very low heat input
to the metal resulting in low distortion and enabling welding of heat-sensitive parts. Due to very rapid heating and
cooling, a continuous melt pool is not a feature and the role
of convection in pulsed laser welding is therefore minimal.
Frewin and Scott [45] produced a time-dependent 3-D
model of heat =ow during pulsed Nd:YAG laser welding,
using the FE code ANSYS. 2 Convective =ow in the melt
pool was ignored. Making use of temperature-dependent
material properties and experimentally measured beam
cross-sections, their model calculated transient temperature
cross-sections along with the dimensions of the fusion and
heat aKected zones. They found that the fusion and heat
aKected zones generated by their code were extremely close
to those produced experimentally, provided they assumed
a Gaussian energy distribution, but varied signiDcantly if a
top-hat distribution was used instead.
3.4. Melt pool models/thermocapillary >ow
The existence of =uid =ow and convection and their effects were Drst discussed in the 1950s [46]. Convection can
be regarded as the single most important factor in=uencing
the geometry of the melt pool leading to a change in pool
shape, the aspect ratio and the production of surface ripples. Furthermore it can cause various defects such as variable penetration, porosity and lack of fusion. As convection
2
ANSYS is a commercial software package that solves for the combined eKects of multiple forces, accurately modelling combined behaviours
resulting from multiphysics interactions. The software also features advanced nonlinear material simulation.
103
causes mixing, it therefore aKects the composition of the

melt pool.
Natural convection or buoyancy results from the change
in density of the molten =uid as it changes temperature.
It is more signiDcant in a large basin than in a small volume such as a typical melt pool as shown by Yokoya and
Matsunawa [47,48]. Surface tension induced convection,
also known as thermocapillary =ow or Marangoni =ow, is
caused by surface temperature gradients resulting in shear
stress and induced =uid motion [49,50]. These forces are
generally far larger than natural convection/gravity induced
forces and tend to dominate the =ow process causing intense
mixing and hence the possibility of surface alloying taking
place. Convection in the melt pool aKects a large number of
laser processes such as melt quenching, alloying and welding. Hence a knowledge of melt dynamics is important for
the understanding of laser processing. However, Matsunawa
[51] observes that in arc welding, experiments have shown
no evidence that Marangoni =ow is predominant. He also
suggests that although the surface tension gradient in laser
melted pools is greater than that in arc pools due to the higher
temperatures, surface contamination is signiDcantly less, so
that Marangoni =ow is not signiDcant in either case. In a
deep keyhole, the exposed liquid surfaces are at or close to
the vaporising temperature, so that the eKect of the surface
tension gradient is minimal. In this case =ow is likely to be
caused by a diKerence in the evaporation rates in the region
close to the beam, resulting in =uid =ow from the front to
the back of the keyhole.
In addition to convective heat transport, the temperature
distribution is aKected by boundary conditions for absorption due to surface deformation. As well as internal forces
such as temperature-dependent surface tension, buoyancy
and ablation pressure, melt =ow is also driven by external
forces such as the =ow of shielding gas and the supply of
powder or Dller wire.
Numerical methods can be used to solve for the conservation of mass, energy, momentum and vorticity in the molten
region, producing =ow velocities and in the more advanced
programs, surface waves.
With respect to experimental results displaying uniform
solute redistribution during laser surface alloying, Chande
et al. [49] suggested that this could be described by an eKective large diKusivity if molecular diKusion was assumed as
the governing mechanism. An eKective value of 106 cm2 =s
was suggested (aluminium for example has a diKusivity of
1 cm2 =s), indicating the large role played by =uid =ow in
the melt pool.
The transient 2-D FE model of Chan et al. [52] solved
the NavierStokes equations for the Marangoni driven =ow
within the melt pool. They considered a body of Dnite thickness but inDnite in area, irradiated by a rectangular source
of constant power distribution. The momentum and energy
equations were coupled through the surface boundary condition and the liquid/solid interface was a time-dependent
moving boundary. Heat losses due to radiation were
104
neglected. The model assumes the following: no heat =ow

along the z-direction (parallel to the beam axis), the surface tension depends linearly on temperature, the thermal
properties of the liquid and solid phases are considered
constant, the thermal conductivity is the same in liquid and
solid phases, latent heat of melting is neglected, and the
molten pool is considered =at. Their model calculated the
melt pool size and predicted that the cooling rate at the
edge of the pool is higher than at the bottom of the pool
below the centerline. It also predicted that the recirculating
velocity is one or two orders of magnitude higher than that
of the processing speed.
Their calculations suggest that the melt pool rotates approximately Dve times before solidifying. The model does
not take account of internal turbulence required to explain
the mixing action and has diNculty in describing the formation of waves.
The 3-D FD =ow model of Chande and Mazumder [53]
simulates laser welding by means of a point-successive overrelaxation scheme. The model incorporates keyholing with
the radiation penetrating the specimen being absorbed according to the BeerLambert law. The specimen was assumed inDnite in length but of Dnite width. They found that
increasing the thermal conductivity can have a signiDcant
eKect on the shape of the melt pool and that the decrease in
melt pool depth may additionally explain why materials of
high thermal conductivity are hard to weld.
Dowden et al. [54] considered the =ow around the keyhole and its temperature during cw laser welding by means
of four computational models for low PReclet Numbers. The
Drst model assumed a viscosity which was constant and the
second, one which depended linearly on temperature. The
third model considered the liquid region divided into a viscous cooler part and an inviscid hotter part whereas the
fourth model assumed a wholly inviscid region. Surface tension and thermocapillary =ow eKects were neglected. Each
of the models displayed a downstream displacement of the
solidliquid boundary which was in agreement with experimental observations.
In a subsequent publication, the authors followed this up
[55] with a 2-D steady-state model which calculated the temperature and =ow distribution in an inDnite workpiece outside of a circular keyhole whose boundary is at the vaporising temperature. The latent heat of phase change was accounted for and a constant viscosity assumed. They elected
for a viscosity independent of temperature using the results
from their paper [54]. Having reformulated the problem in
a form suitable for solution by the isotherm migration technique, they solved it using the method of lines 3 [56]. For
low-processing speeds, results from their model were in very
3
The method of lines (or semi-discretisation) eKectively transforms

a system of partial diKerential equations in two or more independent
variables into a system of ordinary diKerential equations in one of these
variables. The method of lines can be used to solve every type of PDE as
well as ODEs apart from those of eigenvalue type. However, the domain
of the independent variable must be a topological rectangle.
close agreement with those determined analytically from

[54], but deviated somewhat for higher processing speeds.
Prakash et al. [57] proposed a Dxed-grid numerical
methodology to account for phase change during welding
and took into account convection in the melt pool in two
dimensions. The governing equations were solved for the
net velocity i.e. the combined eKect of convective velocity and the material inlet velocity (equal in magnitude but
opposite in sign to the welding velocity).
Chan et al. [58] developed a 3-D numerical axisymmetric
model of =uid =ow and heat-transfer as an extension of their
1984 model [52]. Since the scanning velocity is small compared to the recirculating velocity, an asymptotic solution
was sought. By splitting the velocity Deld into a scanning velocity and a recirculating velocity, the governing equations
can be written in a form where there are two convective
termsone for the scanning velocity and one for the recirculating velocity. Non-dimensional forms of the governing
equations are derived, from which four dimensionless parameters are obtained: the Marangoni number, the Prandtl
number, the dimensionless melting temperature and the radiation factor. A Gaussian beam was assumed. A standard
alternating direction implicit (ADI) [28] iteration method is
employed to solve the governing equations. As with their
previous model [52], conduction and thermocapillary =ow
in the z-direction are neglected and only the surface tension
is considered dependent on temperature, being assumed linear. Phase change is neglected and the thermal conductivity
is considered the same for both liquid and solid. The surface of the melt pool is assumed =at with the xy plane
semi-inDnite. They found that the surface tension gradient
can set up a recirculating =ow within the melt pool, resulting
in an increase of up to 150% in the aspect ratio (width/depth)
when compared with pure conduction.
The same authors [59] developed a 3-D perturbation
model of thermocapillary =ow within the molten region
during laser surface heating. The basic solution corresponds
to the stationary axisymmetric case, and the perturbation model was based on a small scanning velocity. The
advantage of a perturbation solution is that 3-D =ow is
modelled by two sets of 2-D equations, which are easier to solve. The equations were solved by an FD ADI
method and a staggered grid. Again they assumed a linear
temperature-dependent surface tension. Graphical output
includes =ow Delds, particle trajectories, a 3-D plot of the
solidliquid interface and melt pool cross-sections obtained
for diKerent Prandtl numbers (the ratio of momentum to
thermal diKusivity), showing the melt pool becoming wider
and shallower as convection increases. They found that a
particle recirculates many times before re-solidifying, implying that the solute is well mixed. They also showed that
a positive surface tension coeNcient leads to a reverse of
the circulating =ow leading to deeper penetration.
The phenomenon of humping has been investigated by
Matsunawa [6062], Beck et al. [63] and Gratzke et al. [64].
For high-speed laser welding, Beck et al. considered =ow in
the melt pool and solved a set of diKerential equations involving the heat conduction equation, an equation describing
the incompressibility condition together with the Navier
Stokes equation (to describe the viscous motion of the
molten material). Neglecting phase change, the diKerential
equations were converted to their dimensionless form and
solved numerically with FE methods. Diagrams of streamlines, velocity contours and temperature and pressure contours are shown. The molten material produces a long thin
jet behind the keyhole resulting in high pressure at the rear
end of the melt pool. The authors of the paper suggest that it
is this pressure which is responsible for the phenomenon of
humping (opposite of undercut) during fast laser welding.
Gratzke et al. [64] investigated humping in arc and laser
welding. Their theoretical model calculated the processing
speed needed for humping to occur and rejected the in=uence
of Marangoni convection but instead applied Rayleighs theory of instability of a free liquid cylinder due to surface
tension.
Lambrakos et al. [65] calculated the time-dependent
temperature and =uid velocity Deld for deep penetration
welding. The deposition of the power from the beam is
represented by time-dependent boundary conditions on the
equations of momentum and energy transfer. The boundary conditions are speciDed at each time-step on a surface
which changes with time. The model includes the eKect of
the surface tension gradient on the surface of the =uid and
the eKect of the buoyancy force. The coupled equations
of energy, momentum transfer and continuity, combined
with the time-dependent boundary conditions representing
the keyhole and the moving boundaries of the workpiece,
are solved using a speciDc implementation of the SIMPLE
[28] algorithm. With this model they were able to gauge
the relative in=uences of the electromagnetic force, surface
tension and the buoyancy force on =uid convection in the
melt pool. Their results demonstrate that =uid convection
in the melt pool is in=uenced far more by the stirring action
due to the keyhole vapour/liquid interface than by surface
tension.
Kar and Mazumder [66] produced a mathematical model
of laser keyhole welding to predict the depth and diameter of the keyhole, the surface velocity, the temperature
distribution and the melt pool shape. They considered
the forces generated at the liquidvapour interface due
to the surface tension gradient and also the energy balance at the liquidvapour and solidliquid interfaces.
They concluded that before the keyhole forms, the velocity Deld is large in the radial and azimuthal directions but thereafter it changes to a radially and axially
dominant Deld.
Including =ow in the melt pool, Kroos et al. [67] calculated the collapse time of the keyhole when the laser power
is suddenly shut down during stationary welding. They
assumed that the recoil pressure from ablating particles balances the surface tension at the keyhole wall. The keyhole
collapse time then determines the minimum required
105
frequency for pulsed welding. They also concluded that the

keyhole can oscillate with frequencies of several hundred
Hz under typical conditions.
Mazumder et al. [68] produced a comprehensive review
of the mathematical modelling of a number of laser materials processes, discussing the transport phenomena, boundary
conditions, governing equations, solution techniques, applicability and limitations of each process.
3.5. Keyhole models
As well as being the source of energy conducted into
the metal, the keyhole also acts as an obstacle to the liquid
region upstream of it, in=uencing the ultimate molten zone
geometry. Modelling =ow in the keyhole (in addition to
modelling =ow in the melt pool) has been carried out by a
number of authors. The subject is now duly discussed.
Andrews and Atthey [69] used a perturbation expansion
to Dnd the motion and position of the evaporating boundary for a stationary workpiece undergoing deep penetration
welding. Klemens [70], considering the steady-state situation, showed that the transport of material from the front to
the back of the keyhole is mainly due to =ow of the liquid
around it rather than transported in the vapour across it.
Given that the keyhole is maintained by a balance of the
forces closing it (hydrostatic pressure and surface tension)
and those keeping it open (gas pressure), Klemens argued
that since pressure is a function of depth, the keyhole radius must also be so. Dividing the keyhole into an inner
region of constant absorption, conDned within a critical
radius, the temperature of this boundary being at the critical temperature and, assuming that heat =ows radially, he
derived a formula for this critical radius as a function of
the keyhole radius, the vaporising temperature, the power
absorbed per unit length of the keyhole and the critical
temperature.
Andrews and Atthey [71] modelled the steady-state keyhole welding process and obtained keyhole cross-sections
taking into account gravity, both with and without the
inclusion of surface tension and assuming atmospheric
pressure in the cavity. They assumed that all the incident
power was used in evaporating the surface, i.e. that no heat
was required to raise the temperature of the material to its
boiling point. They found that when the surface tension
was included, the penetration depth reduced by up to a
factor of three. They also found that the penetration depth
(in the absence of surface tension) was proportional to the
beam power divided by the beam radius for deep holes and
proportional to the beam power divided by the beam area
for shallow holes. Dowden et al. [72], whose work is an
extension of Andrews and Atthey [71] and Klemens [70],
constructed a general model of the liquid and vapour motion in the keyhole. They investigated how the area of the
keyhole varies with depth and the parameters on which it depends. Their results gave keyhole shapes plausible for open
keyholes.
106
The pressure gradient arising in the liquid region surrounding the keyhole results in =ow parallel to the axis of
the laser. Postaciouglu et al. [73] calculated this velocity
of =ow, as well as the volume =ow rate, in modelling the
bulging in the liquid region at the mouth of the keyhole.
An estimate of the elevation or depression of the surface of
the weld and the shape of the surface cross-section is also
discussed.
SchuTocker [74] produced a model of deep penetration
welding in which the shape of the keyhole was interpolated
from the calculated temperature distribution. The keyhole
is kept open by a balance between the vapour pressure inside it (estimated from the ClausiusClapeyron equation)
and the hydrostatic pressure tending to close it. Dowden et
al. [75] produced a simple time-dependent model which describes the vapour =ow parallel to the axis of the keyhole
for medium processing speeds and also the possible presence of shock waves at the end of the keyhole. The authors
concluded that as the translation speed is increased, the laser
coupling eNciency becomes progressively more dominated
by direct absorption at the keyhole walls.
Kar et al. [76] developed a 2-D axisymmetric model of
material damage caused by melting and vaporisation during pulsed laser welding by taking into account multiple
re=ections of the laser beam at the keyhole wall. Both
solidliquid and liquidvapour interfaces were modelled
by applying the Stefan condition at various positions on
the interface. The whole problem was formulated using
the energy conservation equation and solved numerically
by the RungeKutta method [69]. They concluded that as
the laser intensity increases, the depth and cylindricity of
the cavity increase, whereas the recast layer thickness decreases, with multiple re=ections exaggerating the eKects.
They also found that liquid =ow had little eKect on cavity
depth and recast layer thickness, and that the dependence of
the cavity depth on the gross laser intensity was nonlinear
when multiple re=ections are taken into account, but linear
otherwise.
Ducharme et al. [77] produced a steady-state FD code
to compute the temperature distribution in and around the
hot plasma during laser processing with argon assist gas.
The refractive index of the plasma along with the relevant
gas properties vary with temperature. Equations are derived
relating the curved paths of the light rays to the temperature
gradients in the plasma with the attenuation of the laser
light calculated along these paths. The authors showed that
variations in the electron density due to iron vapour in the
plume cause spatial variations in the refractive index, giving
rise to thermal defocusing of the incident laser light well
before the critical density is achieved.
Considering a laser beam of Gaussian distribution, Metzbower [78] calculated the size and temperature of both the
keyhole and melt pool taking into account evaporation and
assuming that a minimum laser power density is required to
generate a keyhole. To study the mechanism of heat transfer
inside the keyhole, Tix and Simon [79] studied the transport
of electrons, ions, and neutral atoms in the partially ionised

vapour formed in the keyhole.
Ducharme et al. employed a cylindrical coordinate system
in producing integrated keyhole and melt pool models of the
cw welding of thin and thick sheets. In the Drst of these [80],
an analytic expression was formulated for the temperature
distribution outside the keyhole with the heat conduction
equation being solved by means of an FD model for the
distribution inside it. By assuming that the melt pool size
and shape are constant for all depths, a simple 2-D solution
was enabled. Predictions of the surface dimensions of the
keyhole compared favourably with those determined with a
high-speed video camera. Their model was extended to allow for the simulation of the welding of thick sheets [81]. By
considering a series of point sources of varying strength, the
generation of a 3-D temperature distribution was obtained
by superposition of much simpler solutions. Variation of the
properties of the plasma with temperature were taken into
account. Returning to the welding of thin metal sheets, the
model was extended in a diKerent way [82]. This time the
in=uence of diKerent shielding gases on diKerent materials
was modelled, again taking into account the temperature
variation of the properties of both the plasma and the degree
of ionisation, by applying the Saha equation [83]. By matching the melt pool lengths calculated by the model to those of
experiment, they were able to calculate the shape of the melt
pool and keyhole. Their results were in close agreement with
experiment.
Clucas et al. [84] produced an FV model to solve for the
=ow in the keyhole during laser welding. They assumed that
the keyhole is kept open by the balance of the ablation and
hydrodynamic pressure against the surface tension force.
The motion of gas in the keyhole was modelled by the
NavierStokes equation, and they considered three diKerent
keyhole geometries. The velocity of the ablating vapour
was calculated by applying a pressure balance relation on
the keyhole wall. They assumed a uniform temperature distribution in the keyhole along with incompressible gas =ow,
and neglected the diKusion of the shielding gas into the
keyhole.
Klein et al. [85] introduced an analytic expression for
the excess pressure in the keyhole due to gas =ow, and
also considered the oscillations in the axial and azimuthal
directions as well as the radial direction. Solana and Negro
[86] constructed an axisymmetric model to analyse keyhole
cross-sections and intensity distributions as a function of
depth for both Gaussian and top-hat distributions of the
laser beam. They took account of the eKects of inverse
Bremsstrahlung and the eKect of multiple re=ections in
the keyhole in order to model more accurately the Fresnel
absorption process. Instead of Dxing the keyhole geometry, the keyhole wall is considered to be a free boundary
whose shape changes after each iteration of the numerical process. They concluded that inverse Bremsstrahlung
radiation has the eKect of producing wider, but less deep,
keyholes.
Solana and OcaUna [87] determined the full 3-D melt pool
and keyhole geometry by setting the appropriate energy and
pressure balances. The energy balance takes into account
heat conduction, ablation losses and evaporation eKects at
the open surfaces of the keyhole as well as Fresnel absorption and inverse Bremsstrahlung, the keyhole being kept
open as a result of the balance between the ablation pressure
and the surface tension pressure. A time-dependent model
of the behaviour of the front keyhole wall during high translation speed laser welding was constructed by Matsunawa
and Semak [88]. This model was developed assuming that
(i) only the front part of the keyhole wall is exposed to the
high-intensity laser beam, (ii) recoil pressure exceeds surface tension, and (iii) propagation of the keyhole wall is
due to melt expulsion similar to that of laser drilling. Their
work shows that the keyhole wall velocity component parallel to the translation velocity can be greater than, equal to,
or less than the beam translation speed, and when higher,
the formation of humps on the keyhole are observed.
Kapadia et al. [89] produced two stochastic models to
describe the keyhole. The Drst model includes instabilities
and considers the formation of a partially penetrating keyhole under the action of a cw laser. The energy absorption
in the material is simpliDed so that it occurs exclusively in
instability events with each event moving a Dxed quantity of
energy per unit time from the incident laser beam power at
the top of the keyhole. In the second model the laser power
was absorbed in the keyhole both continuously as well as at
speciDc instability events.
Fabbro and Chouf [90,91] produced a model which studied the keyhole geometry as a function of the main operating parameters such as the processing speed, laser incident
intensity and material being welded. The model is based
on a drilling speed (i.e. the speed with which the keyhole
deepens) which when combined with the processing speed,
causes the inclination of the front keyhole wall. A stationary analytic model and a dynamic numerical model (which
takes into account a ray-tracing technique) enables the complete keyhole geometry to be determined by taking into account multiple re=ections inside the keyhole along with a
description of the closure process of the rear keyhole wall.
Pastor et al. [92] discuss the development of models in
order to study the eKect of positive and negative defocusing
on keyhole shape and weld temperature distribution during
laser keyhole welding.
3.6. Reviews
A review of analytic formulations and numerical models relating to deep penetration welding has been given by
Kapadia and Dowden [93]. Dowden and Kapadia [94] presented a further review of analytic solutions of the laser keyhole welding process by point and line sources. Lindgren
reviewed the modelling of welding by Dnite-element methods. His review is split into three parts [9597]. A comprehensive treatment by Dowden of the mathematics of thermal
107
modelling with speciDc reference to laser processing can be

found in [98].
4. Arc welding
The term arc welding comprises a group of welding processes which fuse metals by heating them with an arc, with
or without the application of pressure, and with or without
the use of Dller wire. Arc welding processes account for
about 80% or more of todays welding. Arc processes include MIG, MAG, TIG, gravity arc, plasma arc, shielded
arc, submerged arc and plasma arc. Two common processes
are TIG and MIG. TIG welding, also known as GTAW, is
a process which fuses metals by heating them with an arc
between a tungsten electrode and the workpiece. Shielding
is obtained from a gas or gas mixture. Pressure and Dller
wire may or may not be used. The positive electrode does
not melt and hence gas tungsten arc welding can be autogenous or non-autogenous if a Dller wire is employed. MIG
welding, also known as GMAW, diKers from TIG in that
the positive electrode is consumable and hence welding under MIG is non-autogenous. Again many authors have modelled the processes involved. Compared to laser conduction
welding, arc welding is aKected by an extra source of heat
ohmic heating due to the electric current.
The majority of arc welding is carried out on thick materials which require preparation in terms of cutting out a
section around the joint which is then welded with the use of
Dller wire. For the interested reader, numerical simulations
of MIG welding are described in references [99118].
In order to simulate the =ow and heat transfer associated
with autogenous arc welding processes, Eraslan et al. [119]
developed a completely transient 3-D computational model
(WELDER code) [120] Non-autogenous welding is considered in [121]. Surface tension curvature, Marangoni =ow,
the electromagnetic and buoyancy force eKects are all accounted for, with the momentum equations containing the
MHD force terms. The model can be applied to complex
shapes and inclined workpieces and has the ability to vary
the value of the gravitational force. The model allows for
actual deformation of the molten metal surfaces by considering the surface as a principal solution variable, allowing
for, amongst other things, the simulation of surface rippling
and surface gravity wave phenomena which can partially explain the experimentally observed pulsating behaviour of
melt pools. The publication contains many =ow diagrams,
both on the top surface and on xz planes, both for developing and fully penetrated melt pools, and for combinations of
the above eKects. Their results show that the surface tension
curvature has a marked eKect on melt pool shape and that
it should be considered in conjunction with the Marangoni
shear eKect.
For the interested reader, numerical simulations of TIG
welding are described in references [122146] other arc
welding publications are listed in [147156].
108
5. Other forms of welding and heating

5.1. Induction heating
Induction heating takes place when an electrically conducting object is placed in a varying magnetic Deld. Induction heating is due to hysteresis and eddy-current losses
and there is no contact between the workpiece and the heat
source. It is quicker at bringing iron alloys to their austenitising temperature than other forms of heat treatment. Some
publications contributing to the investigation of induction
heating are [157159].
5.2. Transformation/surface hardening
Transformation hardening is ideally performed by a laser
since distortion is minimal when compared with =ame and
arc processes or induction heating. Transformation hardening requires a quench which is performed eKectively by
the bulk metal in laser processes. Transformation hardening improves the wear resistance and also, under certain
conditions, the fatigue strength. Since the heat from a laser
passes into the surface and is not generated internally as
with induction heating, thermal distortion is low and depth
of hardening can be performed on very thin layers. In these
processes, no melting or keyholing is allowed and hence
power densities must be low. However, this means that the
natural re=ectivity is high and hence absorbent coatings are
often used.
Steen and Courtney [160] found that the surface hardening of certain steels is possible through a martensitic
structure produced by rapid cooling from an elevated temperature, provided the laser power, beam diameter and processing speed are kept within certain limits. They developed
an analytic expression for the depth of hardening and found
that it was closely related to the power divided by the square
root of the product of the beam diameter and processing
speed.
In the context of surface hardening, Bokota and Iskierka
[161] adopted the method of integrating Greens function over the considered region to predict the transient
temperature Deld during laser heat treatment, the method
having been originally proposed by Maier et al. [162].
They considered a rectangular element inDnite in the direction of motion of the heat source but Dnite in the other
dimensions.
Fuhrmann and Homberg [163] put forward a model
to simulate the surface hardening of steel and hence the
solidsolid phase transitions. The latent heats of phase
change were taken care of with a nonlinear energy balance equation coupled with a system of ODEs for the
volume fractions of the occurring phases. In a subsequent
paper, Fuhrmann et al. [158] incorporated Maxwells equations in order to consider the induction hardening of steel:
induction-heating heat transfer solidsolidphase
transitions. It consists of a reduced set of Maxwells equations, the heat transfer equation, and a system of ODEs for
the volume fractions of the occurring phases.
5.3. Surface/heat treatment
Singh and Mazumdar [164] described a time-dependent
3-D heat =ow model to estimate thermal Delds associated
with various heat treatment processes (e.g. heating above
critical temperatures, air cooling, water quenching operations, etc.). An appropriate coordinate system was chosen
in order to carry out numerical calculations for a wide range
of geometries (e.g. slab, parallelepiped, cylinder).
Gu et al. [165] constructed a 3-D model to calculate the
transient temperature Deld and the phase transformation
depth obtained in a Dnite workpiece during laser surface
treatment, allowing for the temperature dependence of the
material properties and taking into account convective and
radiative heat losses. Experimental results for the laser
hardening of a zirconium alloy agreed well with results
from their model. Rana et al. [166] incorporated an FD
implicit splitting method [167] in a model to estimate the
time-dependent temperature distribution in a Dnite-sized
sample during laser heat treatment. The discretised equations were solved using the tri-diagonal matrix algorithm
(TDMA) [28]. The heat source was assumed to have a Gaussian distribution and the thermal diKusivity was assumed
constant. Yilbas [168], using an explicit scheme, produced a
3-D solution to predict the Drst and second law eNciencies
and the entropy generation number [169] during laser heat
treatment.
Crafer and Mackwood [170] presented a fast 2-D
steady-state analytic model (where the temperature is independent of the lateral coordinate) for the temperature
distribution during surface treatment carried out by a moving source. The laser beam intensity is represented as a
number of parallel contiguous strips extending across the
entire width of the specimen and constituting a rectangular
source. The model is applicable to cases where the lateral
size of the workpiece is the same as the breadth of the raster
scan. The model has the ability to build tailored source
distributions and produces very quick solutions. The same
authors have since extended the approach with a 3-D model
with no restrictions on the lateral size of the beam or on the
lateral size of the specimen 4 .
Mazumder [171] gives a comprehensive review of laser
heat treatment and Tosto [172] reviews analytic and numerical laser surface treatment models.
5.4. Electron beam processes
Electron beam welding (EBW) is a high energy density
fusion welding process similar in its eKect to laser keyhole
4
This software is available as the thermal source optimiser (TSO)

package from Abington Consultants, see footnote 1.
welding. A high velocity stream of electrons is used as the

thermal source which can join materials of up to 300 mm in
one pass without leading to signiDcant distortion. Thermally
sensitive materials can also be joined. On impacting with
the material, the energy of the electrons is converted to thermal energy with an eNciency of close to 100%. Most EBW
is performed in a vacuum otherwise most of the electrons
would fail to reach their target. In addition, the vacuum ensures good beam propagation and focusing. When focused
to areas of about 0:25 mm in diameter, power densities as
high as 107 W=cm2 can be achieved. At this intensity, part
of the molten material is vaporised leading to a portion
of the liquid being pushed sideways. The electron beam
drills a small hole that penetrates deep into the specimen,
creating a keyhole. The most noticeable feature of EBW is
its ability to produce a single pass, deep penetration weld
while maintaining a very low overall heat input. However,
the process is susceptible to beam wander (due to the electric and magnetic Delds) and produces hazardous X-rays
requiring the vacuum chamber to be walled with thick steel
or lead.
Electron beam melt cross-sections resemble those created by laser keyhole welding, although exhibiting less of a
wine-glass appearance, and can be modelled by an array of
point and line sources in a similar manner to laser-produced
welds.
A few authors have constructed models of the electron
beam process. A 2-D transient model was developed by
Dufrene et al. [173] to predict temperature, metallurgical
properties, stresses and strains. Semak et al. [174] simulated the dominant dynamical processes in modelling the
transient behaviour of the front keyhole wall, taking into
account the latent heat of phase change and using a set of
values for the density, speciDc heat and thermal conductivity in the solid phase and another set in the liquid phase.
They assumed that the keyhole propagation was dominated
by evaporation-recoil-driven melt expulsion from the beam
interaction zone. Their results were consistent with experimental observations of metal welding, metal cutting and
ice welding. Hemmer et al. [175] developed a 3-D FE code
capable of predicting penetration depths and microstructure
evolution of the HAZ, and validated their model against thermocouple data and optical microscope measurements. They
applied their model to the welding of tubular joints.
5.5. Solid-state welding processes
A well-known form of solid-state welding is that of friction welding. During friction welding no melting is involved.
The energy required to join the metals comes from direct
conversion of mechanical energy to thermal energy at the
interface of the metal surfaces without the need for additional energy or heat from other sources. It is essentially a
solid phase forging process with any molten ejecta being a
by-product.
109
Friction stir welding is another solid-state process. Conceived in 1991, a cylindrical shouldered tool with a proDled
probe is rotated and pushed into the two pieces of metal being joined. The friction created by the rotating tool softens,
but does not melt the workpieces. The softened material is
transferred from the leading edge to the trailing edge of the
tool which is then forged by the contact of the tool shoulder
and pin proDle. This has been heralded by some as the most
remarkable joining process invented in the 1990s.
Metals can also be welded without melting by a further
solid-phase welding technique known as cold pressure welding [176].
Benzsak et al. [177] and Fu and Duan [178] have constructed models of the friction welding process and Chao and
Qi [179] and Grong and Midling [180] produced a model
relating to friction stir welding.
5.6. Resistance welding
Resistance welding is a process in which the application
of pressure along with resistance heating of the workpieces
produces coalescence of metals at the interface between the
two parts to be joined. Spot, seam and butt welding can all
be carried out using this process. In all resistance welding
processes the metal is heated locally until molten and then
allowed to cool, resulting in a weld nugget. The electric
current is produced by electrodes made of copper, due to
its high thermal conductivity, which ensures that the heat
is generated in the workpieces rather than in the electrodes.
Resistance welding is commonly used in the automobile
industry for its high eNciency and excellent quality control.
The interested reader is referred to references [181188].
5.7. Surface alloying/cladding
Surface alloying with a laser is similar to laser surface
melting with the exception that another material is incorporated into the melt pool. Laser surface alloying, studied
by Chande and Mazumder [189], also has similarities with
surface cladding, which if performed with excess power,
can result in surface alloying. (In cladding operations, one
material in liquid form is placed on top of another to form a
sound interfacial bond without diluting the cladding metal
with substrate material). Surface alloying and cladding
when performed by a laser, require minimal post-process
re-machining of the surface.
5.8. Microstructure models
The following publications relate to microstructure models [190198].
5.9. Residual stresses and deformations in welds
Thermal stress arises from restrained thermal expansion.
When welding, very large temperatures are produced in and
around the molten region which reduce quickly with
110
distance. This diKerential heating, followed by cooling to

ambient temperature, induces stresses in the structure which
remain after the material has cooled. These are known as
residual stresses. In the absence of restraint the material
will instead distort and exhibit a reduced stress. Distortion
and residual stresses can cause signiDcant problems in the
welding of large structures and lead to cracking and fracture
problems. Preheating of joints can be eKective in reducing the cooling rate of the weld, hence reducing distortion
and residual stresses. Post-welding heat treatment can also
be used, which relieves internal stresses and controls the
microstructure of the weld.
Finite-element is the dominant numerical method for the
simulating heat treatment and predicting distortion and residual stresses. Often commercial FE programs are used where
routines for phase transformation are implemented.
A number of publications, [199218] describe models
used to determine residual stress and deformation, and Rohde and Jeppsson [218] presented a literature review of heat
treatment simulations tackling phase transformation, residual stress and distortion.
In order to model residual stress and deformation, the
domain representing the metal specimen must be able
to change its shape and volume rather than be represented as a non-deformable weld surface. In addition,
temperature-dependent material properties such as plasticity
are required.
codes used in research work on welding, crystal growth and

=ow of viscoelastic liquids.
Finally, Viswanathan and Gandy [227] have reviewed
literature on weld repair of CrMo steel piping.
5.10. Miscellaneous processes
References
Anthony and Cline [219] studied surface rippling induced by surface tension gradients during laser surface
melting and alloying and concluded that if the laser beam
sweep velocity exceeds a critical velocity, the liquid does
not have suNcient time to form ripples. Wang et al. [220]
produced a 2-D FE model to consider quenching problems involving non-isothermal phase changes, allowing for
temperature-dependent material properties. They applied
their model to the quenching of steel cylinders and an aluminium connector. Lind et al. [221,222] calculated the heat
transfer during cross-=ow quenching of a cylinder.
Garwood et al. [223] used a =uid dynamical computer
code to predict the =uid =ow distribution in an agitated
quench tank during heat treatment of superalloy forgings.
Olafsson et al. [224] used the thermodynamical computer
program ThermoCalc to predict thermal resistivities in heat
treated alloys. With the use of Matthiessens rule [225], they
determined the solubility of elements in commercial alloys
from which the resistivity was predicted.
Amberg et al. [226] developed a toolbox for a commercial maths package which can be used to generate complete
FE codes in 1, 2 or 3 dimensions from a symbolic speciDcation of the mathematical problem which enables the user
to model phase change, =uid =ow and heat transfer in material processes. The authors have used their toolbox to create
5.11. List of welding publications

Mackerle [228] has compiled a list of over 800 published
papers dealing with FE methods related to general and speciDc modelling of welding processes, the in=uence of geometrical parameters, heat transfer and =uid =ow, residual
stresses and deformations, fracture mechanics, fatigue, destructive and nondestructive evaluation of weldments and
cracks, and the welding of tubular joints, pipes, pressure
vessels, plates and other structures/components.
Acknowledgements
The authors would like to acknowledge the Dnancial support of Technology Group 04 (Materials and Structures)
of the MOD Corporate Research Programme in the United
Kingdom.
Grateful thanks are also due to Drs. J.W. Brooks of QinetiQ and S.J. Williams of BAe Systems for advice and guidance, and to Professors P.D. Kapadia, J.M. Dowden and S.P.
Smith of The University of Essex for academic support and
supervision.
[1] SchuTocker D, editor. Handbook of the Eurolaser Academy. London:

Chapman & Hall; 1998.
[2] Carslaw HC, Jaeger JC. Conduction of heat in solids. Oxford:
Oxford Science Publications; 1997.
[3] Rosenthal D. The theory of moving sources of heat and its
application to metal treatments. Trans Amer Soc Mech Eng
1946;68:84966.
[4] Swift-Hook DT, Gick AEF. Penetration welding with lasers. Weld
Res Suppl 1973;492s8s.
[5] Bunting KA, CornDeld G. Toward a general theory of cutting: a
relationship between the incident power density and the cut speed.
Trans ASME J Heat Transfer 1975;11622.
[6] ONeill W, Steen W. Review of mathematical models of laser
cutting of steel. Lasers Eng 1994;3:28199.
[7] Lax M. Temperature rise induced by a laser beam. J Appl Phys
1977;48:391924.
[8] Nissim YI, Lietoila A, Gold RB, Gibbons JF. Temperature
distributions produced in semiconductors by a scanning elliptical
or circular CW laser beam. J Appl Phys 1980;51:2749.
[9] Moody JE, Hendel RH. Temperature proDles induced by a scanning
cw laser-beam. J Appl Phys 1982;53:436471.
[10] Miyazaki T, Giedt WH. Heat-transfer from an elliptical cylinder
moving through an inDnite-plate applied to electron-beam welding.
Int J Heat Mass Transfer 1982;25:80714.
[11] Davis M, Kapadia P, Dowden J, Steen WM, Courtney CHG. Heat
hardening of metal-surfaces with a scanning laser-beam. J Phys D
1986;19:198197.
[12] ShercliK HR, Ashby MF. Master plots for predicting the case-depth
in laser surface treatments. Cambridge University engineering
publication CUed/C-Mat./TR134, 1986.
[13] Ashby MF, Easterling KE. Transformation hardening of steel
surfaces by laser-beams.1. Hypo-eutectoid steels. Acta Metall
1984;32:193548.
[14] Steen WM, Dowden J, Davis M, Kapadia P. A point and line source
model of laser keyhole welding. J Phys D 1988;21:125560.
[15] Akhter R, Davis M, Dowden J, Kapadia P, Ley M, Steen WM.
A method for calculating the fused zone proDle of laser keyhole
welds. J Phys D 1989;21:238.
[16] Hamoudi W, Ducharme R. Keyhole welding of C-Mn steel using
a 10kW CO2 laser. Int J Join Mater 1996;8:306.
[17] Lankalapalli KN, Tu JF, Leong KH, Gartner M. Laser weld
penetration estimation using temperature measurements. J Manuf
Sci Eng-Trans ASME 1999;121:17988.
[18] Postaciouglu N, Kapadia P, Dowden J. A mathematical-model of
heat-conduction in a prolate spheroidal coordinate system with
applications to the theory of welding. J Phys D 1993;26:56373.
[19] Kaplan A. Heat transfer during laser material processing. In:
Kaplan A, SchuTocker D, editors. Proceeding of the 12th Meeting
on Mathematical Modelling of Materials Processing with Lasers
(M4PL12), Section 2, Paper 5. 1997.
[20] Dowden JM, Ducharme R, Kapadia PD. Time dependent line and
point sources: a simple model for time-dependent welding processes.
Lasers Eng 1998;7:21528.
[21] Stefan J. Ann Phys Chem 1891;42:26986.
[22] Tao LN. The exact-solutions of some Stefan-problems with
prescribed heat-=ux. J Appl Mech 1981;48:7326.
[23] Frederick D, Greif J. A method for the solution of heat transfer
problems with a change of phase. J Heat Transfer 1985;107:5206.
[24] Kasuya T, Shimoda N. Stefan problem for a Dnite liquid phase
and its application to laser or electron beam welding. J Appl Phys
1997;82:36728.
[25] Mackenzie JA, Robertson ML. The numerical solution of
one-dimensional phase change problems using an adaptive moving
mesh method. J Comput Phys 2000;161:53757.
[26] Scheerlinck N, Verboven P, Fikiin KA, De Baerdemaeker J, Nicolai
BM. Finite element computation of unsteady phase change heat
transfer during freezing or thawing of food using a combined
enthalpy and KirchhoK transform method. Trans ASAE 2001;44:
42938.
[27] Nedja B. An enthalpy-based Dnite element method for nonlinear
heat problems involving phase change. Comput Struct 2002;80:
921.
[28] Patankar SV. Numerical heat transfer and =uid =ow, 1st ed. New
York; Hemisphere: 1980. ISBN 0-89116-522-3.
[29] Forsythe GE, Wasow WR. Finite-diKerence methods for partial
diKerential equations. New York: Wiley; 1960.
[30] Mazumder J, Steen WM. Heat transfer model for CW laser material
processing. J Appl Phys 1980;51:9417.
[31] Chande T, Mazumder J. Heat =ow during CW laser materials
processing. In: Mukherjee K, Mazumder J, editors. Proceedings of
Lasers in Metallurgy, TMS-AIME, 1981. p. 16577.
[32] Lim JW, Chan CL. Modelling deep penetration laser welding
using a {BEM} sensitivity scheme. Eng Anal Boundary Elements
1995;16:938.
[33] Dowden J, Kapadia P. A mathematical investigation of the
penetration depth in keyhole welding with continuous CO2 lasers.
J Phys D 1995;28:225261.
[34] Norman AF, Ducharme R, Mackwood A, Kapadia P, Prangnel
PB. Application of thermal modelling to laser beam welding of
aluminium alloys. Sci Technol Weld Join 1998;3:2606.
[35] Mohanty PS, Mazumder J. Workbench for keyhole laser welding.
Sci Technol Weld Join 1997;2:1338.
[36] Semak VV, Damkroger B, Kempka S. Temporal evolution of the
temperature Deld in the beam interaction zone during laser material
processing. J Phys D 1999;32:181925.
[37] Semak VV, Steele RJ, Fuerschbach PW, Damkroger BK. Role
of beam absorption in plasma during laser welding. J Phys D
2000;33:117985.
111
[38] Cary HB. Modern welding technology. Englewood CliKs, NJ:

Prentice Hall; 1989. p. 5559.
[39] Chung FK, Wei PS. Mass, momentum, and energy transport in a
molten pool when welding dissimilar metals. J Heat Transfer
Trans ASME 1999;121:45161.
[40] Phanikuma G, Chattopadhyay K, Dutta P. Modelling of transport
phenomena in laser welding of dissimilar metals. Int J Numer
Methods Heat Fluid Flow 2001;H11:15671.
[41] Voller VR, Prakash C. A Dxed grid numerical modelling
methodology for convection diKusion mushy region phase-change
problems. Int J Heat Mass Transfer 1987;30:170919.
[42] Bren AD, Voller VR, Reid KJ. Enthalpy-porosity technique for
modelling convection-diKusion phase-changeapplication to the
melting of a pure metal. Numer Heat Transfer 1988;13:297318.
[43] Reed RC, Bhadeshia HKDH. A simple-model for multipass steel
welds. Acta Metall Mater 1994;42:366378.
[44] Lindgren LE, Runnemalm H, Nasstrom MO. Simulation of
multipass welding of a thick plate. Int J Numer Method Eng
1999;44:130116.
[45] Frewin MR, Scott DA. Finite element model of pulsed laser welding.
Weld J 1999;78:15s22s.
[46] Houldcroft PT. Dilution and uniformity in aluminium alloy weld
beads. Br Weld J 1954;1:46872.
[47] Yokoya S, Matsunawa A. Surface tension driven =ow in
semi-cylindrical basin. Trans Jpn Weld Soc 1983;16:518.
[48] Matsunawa A. What must be resolved to construct a better
mathematical model of fusion welding? In: Zacharia T, editor.
Proceedings of the International Conference on Modelling and
Control of Joining Process. Florida: American Welding Orlando;
1993. p. 18.
[49] Chande T, Mazumder J. Mass-transport in laser surface alloying
ironnickel system. Appl Phys Lett 1982;41:423.
[50] Heiple CR, Roper JR. Mechanism for minor element eKect on GTA
fusion zone geometry. Weld J 1982;61:97s102s.
[51] Matsunawa A. Possible motive forces of liquid motion in laser
weld pool. IIW Doc.IV-7720-2000, 2000.
[52] Chan C, Mazumder J, Chen MM. A 2-dimensional transient model
for convection in laser melted pool. Metall Trans A-Phys Metall
Mater 1984;15:217584.
[53] Chande T, Mazumder J. Estimating eKects of processing conditions
and variable properties upon pool shape, cooling rates, and
absorption-coeNcient in laser-welding. J Appl Phys 1984;56:
19816.
[54] Dowden J, Davis M, Kapadia P. Some aspects of the =uid-dynamics
of laser-welding. J Fluid Mech 1983;126:12346.
[55] Davis M, Kapadia P, Dowden J. Solution of a Stefan problem in
the theory of laser-welding by the method of lines. J Comput Phys
1985;60:5348.
[56] Meyer GH. The method of lines and invariant embedding for
elliptic and parabolic free boundary problems. SIAM J Numer Anal
1981;18:15064.
[57] Prakash C, Samonds M, Singha AK. A Dxed grid numerical
methodology for phase-change problems involving a moving heat
source. Int J Heat Mass Transfer 1987;30:26904.
[58] Chan CL, Mazumder J, Chen MM. 3-dimensional axisymmetrical
model for convection in laser-melted pools. Mater Sci Technol
1987;3:30611.
[59] Chan CL, Mazumder J, Chen MM. EKect of surface-tension gradient
driven convection in a laser melt pool3-dimensional perturbation
model. J Appl Phys 1998;64:616674.
[60] Matsunawa A, Ohji T. Role of surface tension in fusion weldingpart 1. Trans JWRI 1982;11(2):14554.
112
[63] Beck M, Berger P, Dausinger F, HTugel H. Aspects of keyhole/melt

interaction in high speed laser welding. In: Orza JM, Domingo C,
editors. Proceedings of the Eighth International Symposium on Gas
Flow and Chemical Lasers (GCL), vol. 1379. Bellingham, Wa:
SPIE; 1991. p. s769s74.
[64] Gratzke U, Kapadia PD, Dowden J, Kroos J, Simon G. Theoretical
approach to the humping phenomenon in welding processes. J Phys
D 1992;25:16407.
[65] Lambrakos SG, Metzbower EA, Moore PG, Dunn JH, Monis A. A
numerical-model for deep penetration welding processes. J Mater
Eng Perform 1993;2:81938.
[66] Kar A, Mazumder J. Mathematical-modelling of key-hole
laser-welding. J Appl Phys 1995;78:635360.
[67] Kroos J, Gratzke U, Vicanek M, Simon G. Dynamic behaviour of
the keyhole in laser-welding. J Phys D 1993;26:4816.
[68] Mazumder J, Mohanty PS, Kar A. Mathematical modelling of
laser materials processing. Int J Mater Product Technol 1996;11:
193252.
[69] Andrews JG, Atthey DR. On the motion of an intensely heated
evaporating boundary. J Inst Maths Appl 1975;15:5972.
[70] Klemens PG. Heat balance and =ow conditions for electron beam
and laser welding. J Appl Phys 1976;47:216574.
[71] Andrews JG, Atthey DR. Hydrodynamic limit to penetration of a
material by a high-power beam. J Phys D 1976;9:218194.
[72] Dowden J, Postaciouglu N, Davis M, Kapadia P. A keyhole model
in penetration welding with a laser. J Phys D 1987;20:3644.
[73] Postaciouglu N, Kapadia P, Davis M, Dowden JM. Upwelling in
the liquid region surrounding the keyhole in penetration welding
with a laser. J Phys D 1987;20:3405.
[74] SchuTocker D. Modelling of the laser welding process. In:
Proceedings of ECLAT90; vol 1. Erlangen: Sprechsaal Publishing
Group; 1990. p. 5564.
[75] Dowden J, Wu SC, Kapadia P, Strange C. Dynamics of the vapour
=ow in the keyhole in penetration welding with a laser at medium
welding speeds. J Phys D 1991;24:51932.
[76] Kar A, Rockstroh T, Mazumder J. 2-dimensional model for
laser-induced materials damageeKects of assist gas and multiple
re=ections inside the cavity. J Appl Phys 1992;71:25609.
[77] Ducharme R, Kapadia P, Dowden J. A mathematical model of
the defocussing of laser light above a workpiece in laser material
processing. In: Farson D, Steen W, Miyamoto I, editors. Proceedings
of ICALEO92, vol. 75. LIA, Orlando: Laser Institution of America;
1993. p. 18797.
[78] Metzbower EA. Keyhole formation. Metall Trans BProcess
Metall 1993;24:87580.
[79] Tix C, Simon G. A transport theoretical-model of the keyhole
plasma in penetration laser-welding. J Phys D 1993;26:206674.
[80] Ducharme R, Kapadia P, Dowden J, Williams K, Steen WM. An
integrated mathematical model of the keyhole and weldpool in
the laser welding of thin metal sheets. In: Farson D, Steen W,
Miyamoto I, editors. Proceedings of ICALEO92, vol. 75. LIA,
Orlando: Laser Institution of America; 1993. p. 17686.
[81] Ducharme R, Kapadia PD, Dowden JM, Williams K, Steen WM.
An integrated mathematical model for the welding of thick sheets
of metal with a continuous CO2 laser. In: Denny P, Miyamoto I,
Mordike BL, editors. Proceedings of ICALEO93, vol. 77. LIA,
[82] Ducharme R, Kapadia P, Glowacki M, Dowden J, Williams K, Steen
WM. The welding of thin sheets of various metals with a continuous
CO2 laser and a variety of shielding gases investigated by means
of an integrated mathematical model. In: Denny P, Miyamoto I,
Mordike BL, editors. Proceedings of ICALEO93, vol. 77. LIA,
[83] Cobine JD. Gaseous conductors. New York: McGraw-Hill; 1949.
[84] Clucas DAV, Steen WM, Ducharme R, Kapadia PD, Dowden JM.
A mathematical model of the =ow within the keyhole during laser
welding. In: Mazumder J, Matsunawa A, Magnusson C, editors.
[85]
[86]
[87]
[88]
[89]
[90]
[91]
[92]
[93]
[94]
[95]
[96]
[97]
[98]
[99]
[100]
[101]
[102]
[103]
[104]
[105]
[106]
[107]
[108]
[109]
Proceedings of ICALEO95, vol. 80. LIA, Orlando: Laser Institution

of America; 1996. p. 98998.
Klein T, Vicanek M, Kroos J, Decker I, Simon G. Oscillations of
the keyhole in penetration laser-beam welding. J Phys D 1994;27:
202330.
Solana P, Negro G. A study of the eKect of multiple re=ections
on the shape of the keyhole in the laser processing of materials.
J Phys D 1997;30:321622.
Solana P, Ocana JL. A mathematical model for penetration laser
welding as a free-boundary problem. J Phys D 1997;30:130013.
Matsunawa A, Semak V. The simulation of front keyhole wall
dynamics during laser welding. J Phys D 1997;30:798809.
Kapadia P, Solana P, Dowden J. Stochastic model of the deep
penetration laser welding of metals. J Laser Appl 1998;10:1703.
Fabbro R, Chouf K. Keyhole modelling during laser welding.
J Appl Phys 2000;87:407583.
Fabbro R, Chouf K. Dynamical description of the keyhole in deep
penetration laser welding. J Laser Appl 2000;12:1428.
Pasto M, Sagaro R, Cabrera R. Modelling of the welded proDle and
the keyhole of a laser welded Al 5182 alloy. Rev Metal 2001;37:
64352.
Kapadia PD, Dowden JM. Review of mathematical models of deep
penetration laser welding. Lasers Eng 1994;3:187280.
Dowden JM, Kapadia P. Point and line sources: models of power
absorption in the theory of welding. In: Beyer E, Xiangli Chen,
Miyamoto I, editor. Proceedings of ICALEO98, vol. 85. LIA,
Orlando: Laser Institution of America; 1999. p. G16170.
Lindgren LE. Finite element modelling and simulation of welding.
Part 1: increased complexity. J Therm Stress 2001;24:14192.
Part 2: improved material modelling. J Therm Stress 2001;24:
195213.
Part 3: eNciency and integration. J Therm Stress 2001;24:30534.
Dowden JM. The mathematics of thermal modelling. London, Boca
Raton, FL: Chapman & Hall/CRC; 2001. ISBN 1-58488-230-1.
Goldak J, Chakravarti A, Bibby M. A new Dnite-element model
for welding heat-sources. Metall Trans BProcess Metall 1984;15:
299305.
Kumar S, Bhaduri SC. 3-dimensional Dnite-element modelling of
gas metal-arc welding. Metall Mater Trans BProcess Metall Mater
Process Sci 1994;25:43541.
Kumar S, Bhaduri SC. Theoretical investigation of penetration
characteristics in gas metal-arc welding using Dnite-element
method. Metall Mater Trans BProcess Metall Mater Process Sci
1995;26:61124.
Haidar J, Lowke JJ. Predictions of metal droplet formation in arc
welding. J Phys D 1996;29:295160.
Wahab MA, Painter MJ. Numerical models of gas metal arc
welds using experimentally determined weld pool shapes as the
representation of the welding heat source. Int J Pressure Vessels
Piping 1997;73:1539.
Jones LA, Eagar TW, Lang JH. A dynamic model of drops detaching
from a gas metal arc welding electrode. J Phys D 1998;31:10723.
Wahab MA, Painter MJ, Davies MH. The prediction of the
temperature distribution and weld pool geometry in the gas metal
arc welding process. J Mater Process Technol 1998;77:2339.
Wu CS, Sun JS. Modelling the arc heat =ux distribution in GMA
welding. Comput Mater Sci 1998;9:397402.
Wu CS, Sun JS. Determining the distribution of the heat content
of Dller metal droplets transferred into gas-metal arc weld pools.
Proc Inst Mech Eng BJ. Eng Manufact 1998;212:52531.
Park H, Rhee S. Analysis of weld geometry considering the
transferring droplets in gas metal arc welding. JSME Int J
2001;44:85862.
Yang Z, DebRoy T. Modelling macro-and microstructures of
gas-metal-arc welded HSLA-100 steel. Metall Mater Trans B
Process Metall Mater Process Sci 1999;30:48393.
[110] Bachorski A, Painter MJ, Smailes AJ, Wahab MA. Finite-element
prediction of distortion during gas metal arc welding using the
shrinkage volume approach. J Mater Process Technol 1999;93:
4059.
[111] Kim JW, Na SJ. A study on the 3-dimensional analysis of heat
and =uid-=ow in gas metal arc-welding using boundary-Dtted
coordinates. J Eng Ind-Trans ASME 1994;116:7885.
[112] Ushio M, Wu CS. Mathematical modelling of three-dimensional
heat and =uid =ow in a moving gas metal arc weld pool.
Metall Mater Trans BProcess Metall Mater Process Sci 1997;28:
50916.
[113] Little GH, Kamtekar AG. The eKect of thermal properties and weld
eNciency on transient temperatures during welding. Comput Struct
1998;68:15765.
[114] Wan Y, Shi Q, Tsai HL. Modelling of the eKects of surface-active
elements on =ow patterns and weld penetration. Metall Mater Trans
BProc Metall Mater Process Sci 2001;32:14561.
[115] Ohring S, Lugt HJ. Numerical simulation of a time-dependent 3-D
GMA weld pool due to a moving arc. Weld J 1999;78:416S24S.
[116] Kim IS, Basu A. A mathematical model of heat transfer and =uid
=ow in the gas metal arc welding process. J Mater Process Technol
1998;77:1724.
[117] Wang Y, Tsai HL. EKects of surface active elements on weld
pool =uid =ow and weld penetration in gas metal arc welding.
Metall Mater Trans BProcess Metall Mater Proccess Sci 2001;32:
50115.
[118] Wang Y, Tsai HL. Impingement of Dller droplets and weld pool
dynamics during gas metal arc welding process. Int J Heat Mass
Transfer 2001;44:206780.
[119] Zacharia T, Eraslan AH, Aidun DK. Modelling of autogenous
welding. Weld J 1988;67:S5362.
[120] Eraslan AH, Zacharia T, Aidun DK. Computer code for
simulating fast-transient three-phase =ow, temperature and material
composition conditions during welding. Report No. MIE-142,
Department of Mechanical Engineering, Clarkson University,
Potsdam, NY, 1986.
[121] Zacharia T, Eraslan AH, Aidun DK. Modelling of non-autogenous
welding. Weld J 1988;67:S1827.
[122] Ule RL, Joshi Y, Sedy EB. A new technique for 3-dimensional
transient heat-transfer computations of autogenous arc-welding.
Metall Trans BProcess Metall 1990;21:103347.
[123] Zacharia T, David SA, Vitek JM. EKect of evaporation
and temperature-dependent material properties on weld pool
development. Metall Trans BProcess Metall 1991;22:23341.
[124] Abouarab TW, Saleh KM, Ramadan I. A study of a
multidimensional transient heat-conduction process with a moving
heat-source-2: numerical and experimental. Mater High Temp
1991;9:418.
[125] Lho TJ, Na SJ. A study on 3-dimensional transient heat-=ow in
circumferential GTA welding of pipes using periodicity conditions.
Proc. Inst Mech Eng BJ Eng Manufact 1991;205:2718.
[126] Fan HG, Shi YW, Na SJ. Numerical analysis of the arc in pulsed
current gas tungsten arc welding using a boundary-Dtted coordinate.
J Mater Process Technol 1997;72:43745.
[127] Kim WH, Na SJ. Heat and =uid =ow in pulsed current GTA weld
pool. Int J Heat Mass Transfer 1998;41:321327.
[128] Wu CS, Zheng W, Wu L. Modelling the transient behaviour of
pulsed current tungsten-inert-gas weldpools. Modell Simul Mater
Sci Eng 1999;7:1523.
[129] Yang YP, Dong P, Zhang J, Tian X. A hot-cracking mitigation
technique for welding high-strength aluminum alloy. Weld J
2000;79:9s17s.
[130] Reddy AA, Guha B, Achar DRG. Finite element modelling of
three-dimensional transient heat transfer in stainless steel (304)
pulsed GTA weldments. Numer Heat Transfer AAppl 2002;41:
416.
[131] Tian XT, Shi QY. Preventing welding hot cracking by welding with
an intensive trailing cooler. J Mater Process Technol 2000;97:304.
113
[132] Oreper GM, Szekely J. Heat-=ow and =uid-=ow phenomena in

weld pools. J Fluid Mech 1984;147:539.
[133] Kou S, Wang YH. Weld pool convection and its eKect. Weld
J 1986;65:63s70s.
[134] Choo RTC, Szekely J, WesthoK RC. On the calculation of the
free-surface temperature of gas-tungsten-arc weld pools from 1st
principles 1. Modelling the welding arc. Metall Trans BProcess
Metall 1992;23:35769.
[135] Choo RTC, Szekely J, David SA. On the calculation of the
free-surface temperature of gas-tungsten-arc weld pools from 1st
principles 2. Modelling the weld pool and comparison with
experiments. Metall Trans BProcess Metall 1992;23:37184.
[136] Kim SD, Na SJ. EKect of weld pool deformation on weld penetration
in stationary gas tungsten arc-welding. Weld J 1992;71:S179S93.
[137] Wu CS, Dorn L. Prediction of surface depression of a tungsten
inert-gas weld pool in the full-penetration condition. Proc Inst Mech
Eng BJ Eng Manufact 1995;209:2216.
[138] Zacharia T, David SA, Vitek JM, Kraus HG. Surface-temperature
distribution of GTA weld pools on thin-plate 304-stainless-steel.
Weld J 1995;74:S353S62.
[139] Dutta P, Joshi Y, Janaswamy R. Thermal modelling of gas tungsten
arc-welding process with nonaxisymmetric boundary-conditions.
Numer Heat Transfer AAppl 1995;27:499518.
[140] Ducharme R, Kapadia P, Dowden J, Thornton M, Richardson I.
A mathematical-model of the arc in electric-arc welding including
shielding gas-=ow and cathode spot location. J Phys D 1995;28:
184050.
[141] Fan HG, Shi YW. Numerical simulation of the arc pressure in gas
tungsten arc welding. J Mater Process Technol 1996;61:3028.
[142] Kim WH, Fan HG, Na SJ. A mathematical model of gas tungsten
arc welding considering the cathode and the free surface of the
weld pool. Metall Mater Trans BProcess Metall Mater Process
Sci 1997;28:67986.
[143] Fan HG, Na SJ, Shi YW. Numerical simulation of current density
in gas tungsten arc welding including the in=uence of the cathode.
Proc Inst Mech Eng BJ Eng Manufact 1997;211:3217.
[144] Cao ZN, Zhang YM, Kovacevic R. Numerical dynamic analysis
of moving GTA weld pool. J Manufact Sci EngTrans ASME
1998;120:1738.
[145] Goodarzi M, Choo R, Takasu T, Toguri JM. The eKect of the
cathode tip angle on the gas tungsten arc welding arc and weld
pool: II. The mathematical model for the weld pool. J Phys D
1998;31:56983.
[146] Zacharia T, Chen Y, Dress W. Modelling of fundamental
phenomena in gas tungsten arc welds. Int J Mater Product Technol
1998;13:7788.
[147] Apps RL, Milner DR. Br Weld J 1995;12:475.
[148] Oreper GM, Eagar TW, Szekely J. Convection in arc weld pools.
Weld J 1983;62:307s12s.
[149] Zacharia T, Eraslan AH, Aidun DK, David SA. 3-dimensional
transient model for arc-welding process. Metall Trans BProcess
Metall 1989;20:64559.
[150] Ducharme R, Kapadia P, Dowden J, Decker I. A mathematical
model of gas =ow and heat transfer in the welding arc. Int J Join
Mater 1993;5:1316.
[151] Postaciouglu N, Trayner C, Kapadia PD, Ducharme R, Dowden JM,
Richardson I. A mathematical model of the electric arc interacting
with the anode in electric arc welding. Int J Join Mater 1993;5:
13742.
[152] Ravichandran G, Raghupathy VP, Ganesan N. Analysis of
temperature distribution during circumferential welding of
cylindrical and spherical components using the Dnite element
method. Comput Struct 1996;59:22555.
[153] Bonifaz EA. Finite element analysis of heat =ow in single-pass arc
welds. Weld J 2000;79:121S5S.
[154] Komanduri R, Hou ZB. Thermal analysis of the arc welding process:
part I. General solutions. Metall Mater Trans BProc Metall Mater
Process Sci 2000;31:135370.
114
[155] Komanduri R, Hou ZB. Thermal analysis of the arc welding

process: part II. EKect of variation of thermophysical properties
with temperature. Metall Mater Trans BProcess Metall Mater
Process Sci 2001;32:48399.
[156] Wen SW, Hilton P, Farrugia DCJ. Finite element modelling
of a submerged arc welding process. J Mater Process Technol
2001;119:2039.
[157] Wang KF, Chandrasekar S, Yang HTY. Finite-element simulation
of moving induction heat-treatment. J Mater Eng Perform 1995;4:
46073.
[158] Fuhrmann J, Homberg D, Uhle M. Numerical simulation of
induction hardening of steel. COMPELThe Int J Comput Math
Elect Electron Eng 1999;18.
[159] Tsuboi H, Ikeda K, Kurata M, Kainuma K, Nakamura K.
Eddy current analysis for the pipe welding. IEEE Trans Magn
1998;34:12346.
[160] Steen WM, Courtney GHC. Surface Heat Treatment of En8
steel using a 2kW continuous-wave CO2 laser. Metal Technol 6
1979;12:45662.
[161] Bokota A, Iskierka S. Numerical prediction of the hardened zone
in laser treatment of carbon steel. Acta Mater 1996;44:44550.
[162] Maier C, Schaaf P, Gonser U. Calculation of the temperature proDle
for laser treatment of metallic samples. Meter Sci Eng A Struct
Mater 1992;150:27180.
[163] Fuhrmann J, Homberg D. Numerical simulation of the surface
hardening of steel. Int J Numer Method Heat Fluid Flow 1999;9:
70524.
[164] Singh AK, Mazumdar D. Mathematical-modelling of thermal Delds
in some heat-treatment operations. Steel Res 1992;63:194200.
[165] Gu BW, Ma TC, Brown SK, Mannik L. 3-dimensional
numerical-model for laser transformation hardening of metals. Mater
Sci Technol 1994;10:42530.
[166] Rana J, Mishra PK, Prasad BVSSS. A transient three-dimensional
thermal model for laser heat treatment of Dnite size workpieces.
J Mater Process Manufact Sci 1997;5:28393.
[167] Fertziger JH. Numerical methods for engineering applications. New
York: Wiley; 1981.
[168] Yilbas BS. Three-dimensional laser heating model and entropy
generation consideration. J Energy Resourc TechnolTrans ASME
1999;121:21724.
[169] Bejan A. Advanced engineering thermodynamics. New York:
Wiley; 1997.
[170] Crafer RC, Mackwood AP. Flexible 2D CW laser surface treatment
model. Lasers Eng 1999;8:2419.
[171] Mazumder J. Laser heat treatment: the state of the art. J Min Metal
Mater Soc 1983;35:1826.
[172] Tosto S. Laser surface treatments: a review of models. Lasers Eng
1994;3:15786.
[173] Dufrene L, Truc J, Martin R. Modelling of thermal stresses during
an electron-beam welding process. J Therm Stresses 1996;19:
21136.
[174] Semak VV, Bragg WD, Damkroger B, Kempka S. Transient model
for the keyhole during laser welding. J Phys D 1999;32:L61L4.
[175] Hemmer H, Grong O, Klokkehaug S. A process model for the
heat-aKected zone microstructure evolution in duplex stainless steel
weldments: part II. Application to electron beam welding. Metall
Mater Trans APhys Metall Mater Sci 2000;31:103548.
[176] Houldcroft PT. Which process?. Cambridge: Woodhead Publishing
Ltd; 1990.
[177] Bendzsak GJ, North TH, Li Z. Numerical model for steady-state
=ow in friction welding. Acta Mater 1997;45:173545.
[178] Fu L, Duan L. The coupled deformation and heat =ow analysis
by Dnite element method during friction welding. Weld J 1998;77:
202S7S.
[179] Chao YJ, Qi XH. Thermal and thermo-mechanical modelling of
friction stir welding of Aluminum alloy 6061-T6. J Mater Process
Manufact Sci 1999;7:21533.
[180] Frigaard O, Grong O, Midling OT. A process model for friction

stir welding of age hardening aluminum alloys. Metall Mater Trans
APhys Metall Mater Sci 2001;32:1189200.
[181] Browne DJ, Chandler HW, Evans JT, Wen J. Computer-simulation
of resistance spotwelding in aluminum 1. Weld J 1995;74:
339s44s.
[182] Browne DJ, Chandler HW, Evans JT, James PS, Wen J, Newto
CJ. Computer-simulation of resistance spot-welding in aluminum
2. Weld J 1996;74:417s22s.
[183] Gupta OP, De A. An improved numerical modelling for resistance
spot welding process and its experimental veriDcation. J Manufact
Sci EngTrans ASME 1998;120:24651.
[184] Chang BH, Li MV, Zhou Y. Comparative study of small scale
and large scale resistance spot welding. Sci Technol Weld Join
2001;6:27380.
[185] Ageorges C, Ye L, Mai YW, Hou M. Characteristics of resistance
welding of lap shear coupons. Part I: heat transfer. Compos A
Appl Sci Manufact 1998;29:899909.
[186] Khan JA, Xu L, Chao YJ. Prediction of nugget development during
resistance spot welding using coupled thermal-electrical-mechanical
model. Sci Technol Weld Join 1999;4:2017.
[187] Xu L, Khan JA. Nugget growth model for Aluminum alloys during
resistance spot welding. Weld J 1999;78:367s72s.
[188] Khan JA, Broach K, Kabir AASA. Numerical thermal model of
resistance spot welding in aluminum. J Thermophys Heat Transfer
2000;14:8895.
[189] Chande, Mazumder J. Two-dimensional, transient model for
mass-transport in laser surface alloying. J Appl Phys 1985; 57:
222632.
[190] Carolan RA, Faulkner RG. Grain-boundary precipitation of M23C6
in an austenitic steel. Acta Metall 1998;36:25766.
[191] Faulkner RG, Jiang H. Combined model of grain-boundary
precipitation and segregation of high-strength Aluminum-alloys.
Mater Sci Technol 1993;9:66571.
[192] Faulkner RG. Nonequilibrium grain-boundary segregation in
austenitic alloys. J Mater Sci 1981;16:37383.
[193] Jiang H, Faulkner RG. Modelling of grain boundary segregation,
precipitation and precipitate-free zones of high strength aluminium
alloys 1. The model. Acta Mater 1996;44:185764.
[194] Rappaz M, Gandin CA. Probabilistic modelling of microstructure
formation in solidiDcation processes. Acta Metall Mater
1993;41:34560.
[195] Koseki T, Matsumiya T, Yamada W, Ogawa T. Numerical
modelling of solidiDcation and subsequent transformation of Fe
CrNi alloys. Metall Mater Trans APhys Metall Mater Sci
1994;25:130921.
[196] Katayama S, Mizutani M, Matsunawa A. Modelling of melting
and solidiDcation behaviour during laser spot welding. Sci Technol
Weld Join 1997;2:19.
[197] Yang Z, Sista S, Elmer JW, DebRoy T. Three dimensional Monte
Carlo simulation of grain growth during GTA welding of titanium.
Acta Mater 2000;48:481325.
[198] Sista S, Yang Z, DebRoy T. Three-dimensional Monte Carlo
simulation of grain growth in the heat-aKected zone of a 2.25Cr-1Mo
steel weld. Metall Mater Trans BProcess Metall Mater Process
Sci 2000;31:52936.
[199] Leblond JB, Devaux J. A new kinetic-model for anisothermal
metallurgical transformations in steels including eKect of austenite
grain-size. Acta Metall 1984;32:13746.
[200] Leblond JB, Devaux J, Devaux JC. Mathematical-modelling of
transformation plasticity in steels 1. Case of ideal-plastic phases.
Int J Plast 1989;5:55172.
[201] Yang YS, Na SJ. A study on the thermal and residual-stress by
welding and laser surface hardening using a new 2-dimensional
Dnite-element model. Proc Inst Mech Eng BJ Eng Manufact
1990;204:16773.
[202] Chen Y, Sheng IC. Residual-stress in weldment. J Therm Stresses
1992;15:5369.
[203] Sheng IC, Chen Y. Modelling welding by surface heating. J Eng
Mater TechnolTrans ASME 1992;114:43949.
[204] Brown S, Song H. Finite-element simulation of welding of large
structures. J Eng IndTrans ASME 1992;114:44151.
[205] Chidiac SE, Mirza FA. Thermal-stress analysis due to welding
processes by the Dnite-element method. Comput Struct 1993;46:
40712.
[206] Postaciouglu N, Kapadia P, Dowden J. Stresses and strains in
laser welding of thin metal sheets. Transport Phenomena in Mater
Process Manufact ASME 1994;280:10712.
[207] Prasad NS, Narayanan TKS. Finite element analysis of temperature
distribution during arc welding using adaptive grid technique. Weld
J 1996;75:123s8s.
[208] Michaleris P, Sun X. Finite element analysis of thermal tensioning
techniques mitigating weld buckling distortion. Weld J 1997;76:
451s7s.
[209] Myhr OR, Klokkehaug S, Grong O, Fjaer HG, Kluken AO.
Modelling of microstructure evolution, residual stresses and
distortions in 6082-T6 aluminum weldments. Weld J 1998;77:
286s92s.
[210] Brickstad B, Josefson BL. A parametric study of residual stresses in
multi-pass butt-welded stainless steel pipes. Int J Pressure Vessels
Piping 1998;75:1125.
[211] Jahanian S, Mosleh M. The mathematical modelling of phase
transformation of steel during quenching. J Mater Eng Perform
1999;1:7582.
[212] Seok CS, Suh MW, Park JH. Investigation of welding residual
stress of high tensile steel by Dnite element method and experiment.
KSME Int J 1999;13:87985.
[213] Carmignani C, Mares R, Toselli G. Transient Dnite element
analysis of deep penetration laser welding process in a singlepass
butt-welded thick steel plate. Comput Method Appl Mech Eng
1999;179:197214.
[214] Kiselev AS. Computer-assisted simulation of thermal, structural,
and deformational processes in heat-treatment technology. Ind Lab
1999;65:65966.
[215] Matos CG, Dodds RH. Modelling the eKects of residual stresses
on defects in welds of steel frame connections. Eng Struct
2000;22:110320.
115
[216] Sarkani S, Tritchkov V, Michaelov G. An eNcient approach for

computing residual stresses in welded joints. Finite Elements Anal
Design 2000;35:24768.
[217] Prime MB, Prantil VC, Rangaswamy P, Garcia FP. Residual stress
measurement and prediction in a hardened steel ring. Mater Sci
Forum 2000;347:2238.
[218] Rohde U, Jeppsson A. Literature review of heat treatment
simulations with respect to phase transformation, residual stresses
and distortion. Scand J Metall 2000;29:4762.
[219] Anthony TR, Cline HE. Surface rippling induced by surface tension
gradients during laser surface melting and alloying. J Appl Phys
1977;48:388894.
[220] Wang KF, Chandrasekar S, Yang HTY. An eNcient 2D
Dnite-element procedure for the quenching analysis with
phase-change. J Eng IndTrans ASME 1993;115:12438.
[221] Lind M, Lio N, Alavyoon F, Bark F. Flow eKects and modelling
in gas-cooled quenching. In: Proceedings of the 11th International
Conference on Heat Transfer, Kyongju, Korea, 1998.
[222] Thuvander A, Melander A, Lind M, Lior N, Bark F. Prediction
of heat transfer coeNcients and examination of their eKects on
distortion of tubes quenched by gas cooling. In: Proceedings of
the 4th ASM Heat Treatment and Surface Engineering Conference
Florence, Italy, 1998.
[223] Garwood DR, Lucas JD, Wallis RA, Ward J. Modelling of the =ow
distribution in an oil quench tank. J Mater Eng Perform 1992;1:
7818.
[224] Olafsson P, Sandstrom R, Karlsson A. Electrical conductivity
of aluminium alloys. Aluminium Alloys: Phys Mech Prop
1996;217(1-3):9816.
[225] Matthiessen A. Rep Br Assoc 1862;32:114.
[226] Amberg G, Tonhardt R, Winkler C. Finite element simulations
using symbolic computing. Math Comput Simul 1999;49:
25774.
[227] Viswanathan R, Gandy D. Weld repair of aged CrMo steel
pipingA review of literature. J Pressure Vessel Technol ASME
2000;122:7685.
[228] Mackerle J. Finite element analysis and simulation of welding: a
bibliography (19761996). Model Simul Mater Sci Eng 1996;4:
50133.

Thermal Modelling of Laser Welding

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Thermal Modelling of Laser Welding

Uploaded by

Copyright:

Available Formats

Available online at www.sciencedirect.

Optics & Laser Technology 37 (2005) 99 115

Thermal modelling of laser welding and related processes:

Centre, The University of Essex, Colchester CO4 3SQ, UK

Corresponding author. Tel.: +44-1223-891576.

Deld tend to publish in the other Delds as well, so part

the surface. Some absorption or scattering of the power

On the subject of laser cutting, Bunting and CornDeld [5]

appropriate intensity along a curve in the interior of the

Scheerlinck et al. [26] developed a Dnite-element (FE)

2, 4 and 6 kW incident on 304 stainless steel. They also

causes mixing, it therefore aKects the composition of the

neglected. The model assumes the following: no heat =ow

The method of lines (or semi-discretisation) eKectively transforms

close agreement with those determined analytically from

frequency for pulsed welding. They also concluded that the

of electrons, ions, and neutral atoms in the partially ionised

modelling with speciDc reference to laser processing can be

5. Other forms of welding and heating

This software is available as the thermal source optimiser (TSO)

welding. A high velocity stream of electrons is used as the

distance. This diKerential heating, followed by cooling to

codes used in research work on welding, crystal growth and

5.10. Miscellaneous processes

5.11. List of welding publications

[1] SchuTocker D, editor. Handbook of the Eurolaser Academy. London:

[38] Cary HB. Modern welding technology. Englewood CliKs, NJ:

[63] Beck M, Berger P, Dausinger F, HTugel H. Aspects of keyhole/melt

Proceedings of ICALEO95, vol. 80. LIA, Orlando: Laser Institution

[132] Oreper GM, Szekely J. Heat-=ow and =uid-=ow phenomena in

[155] Komanduri R, Hou ZB. Thermal analysis of the arc welding

[180] Frigaard O, Grong O, Midling OT. A process model for friction

[216] Sarkani S, Tritchkov V, Michaelov G. An eNcient approach for

You might also like