Page 1
101
5.1 Introduction
The need to achieve higher productivity and stringent safety requirement have
put growing emphasis on the use of automated welding systems, submerged arc
welding is employed in semiautomatic or automatic mode in industry (Brien, 1978). In
such automated applications, a precise means of selection of the process variables and
control of weld bead shape has become essential because mechanical strength of weld
is influenced not only by the composition of the metal, but also by the weld bead shape
(Hould, 1989). The acceptable weld bead shape depends on factors such as line power
which is the heat energy supplied by an arc to the base plate per unit length of weld,
welding speed, joint preparation, etc. To do these precise relationships between the
process parameters and the bead parameters controlling the bead shape are to be
established. This may be achieved by the development of mathematical expressions,
which can be fed into a computer, relating the weld bead dimensions to the important
process control variables affecting these dimensions. Also, optimization of the process
parameters to control and obtain the required shape and quality of weld beads is
possible with these expressions. A macrophotograph of a real weld bead is shown in
Fig.5.1 (A). Cross section of an ideal weld bead showing the bead geometry is given in
Fig.5.1 (B).
CHAPTER – 5
EFFECT OF WELDING PARAMETERS ON BEAD
GEOMETRY AND FLUX CONSUMPTION
Page 2
102
(a)
(b)
Fig.5.1 (a) Photograph of a real weld bead (b) Cross-section of an ideal weld bead
Where P: height of penetration (mm); R: height of reinforcement (mm); W: width of the bead
(mm); WPSF: penetration shape factor =W/P; WRFF: reinforcement form factor = W/R.
In the present study, an attempt has been made to investigate the effect of open
circuit voltage, welding current, welding speed and basicity index on bead geometry
and shape relationships (bead width, weld penetration and height of reinforcement,
weld penetration shape factor and weld reinforcement form factor), using developed
fluxes, through experiments based on design matrix. The analysis of variance
(ANOVA) technique has been adopted to check the level and degree of the direct or
interactive effect of welding current, voltage, welding speed and flux basicity index on
features of bead geometry and shape relationship. Response surface methodology has
been applied to derive mathematical models that correspond to the welding phenomena
using developed fluxes. Predictive equations have been used to represent graphically
Page 3
103
the effects of process parameters on various responses. No work so far has been
performed which considers the four important process parameter used in this study
using fluxes developed from waste flux dust.
5.2 Operating Variables
Control of the operating variables in submerged arc welding is essential if high
production rates and the welds of good quality are to be obtained. The following are
the important variables:
(i) Welding amperage
(ii) Welding voltage
(iii) Welding speed
(iv) Electrode size
(v) Electrode work angle
(vi) Electrode stick-out
(vii) Depth of flux
(viii) Polarity
(ix) Melting rate
(x) Flux basicity index
5.2.1 Welding amperage
Welding current is the most influential parameter because it affects bead shape,
controls the rate at which electrode is melted and therefore also controls the deposition
Page 4
104
rate, heat affected zone, the depth of penetration, and the amount of base metal melted.
Penetration and reinforcement increase with the increase in welding current.
If the current is too high at a given welding speed, the depth of fusion or
penetration will also be too high so that the resulting weld may tend to melt through
the metal being joined. High current also leads to waste of electrodes in the form of
excessive reinforcement and produces digging arc and undercut. This overwelding
increases weld shrinkage and causes greater distortion. Bead width increases with
welding current until a critical value is reached and then starts decreasing if the
polarity used is DCEP. When DCEN polarity is employed bead width increases with
the increase in current for entire range (McGlone, 1982). For the same flux, heat
affected zone also increases with the increase in welding current (Kaushal and Gupta,
1988). If the current is too low, inadequate penetration or incomplete fusion may
result. Too low current also leads to unstable arc, inadequate penetration and
overlapping.
5.2.2 Welding voltage
Welding voltage varies with the length of the arc between the electrode and
molten weld metal. With the increase in arc length, the arc voltage increases because
lengthening of the arc exposes more of the arc column to the cool boundary of the arc.
Also, the arc column continuously loses the charge carriers by radial migration to the
cool boundary of the arc and therefore, imposing a greater requirement of potential for
Page 5
105
maintaining appropriate charge carriers between the electrode and weld plate
(Weiman, 1981).
The voltage principally determines the shape of the weld bead cross section and
its external appearance. Increasing the welding voltage with constant current and
welding speed produces flatter, wider, less penetrated weld beads and tends to reduce
the porosity caused by rust or scale on steel. Higher voltage also bridges an excessive
root opening when fit-up is poor. Increase in arc voltage also increases the size of
droplets and hence decreases the number of droplets. The time of the movement of
droplet transfer also increases. Further increase in voltage increases the possibility of
breaking the arc and disrupting the normal welding process. Increase in voltage also
enhances flux consumption which increases pick up or loss of the alloying elements
and therefore affects the mechanical and metallurgical properties of the weld metal
(Gupta and Gupta, 1988; Pandey and Mohan, 2003).
Excessively high voltage produces a wide bead shape that is subject to
cracking, increases undercut and creates difficulty in removing slag. Lowering the
voltage produces stiffer arc, which improves penetration in a deep weld groove and
resists arc blow. An excessively low voltage produces a narrow bead and causes
difficult slag removal along the bead edges.
5.2.3 Welding speed
Welding speed is the linear rate at which an arc is moved along the weld joint.
With any combination of welding voltage and welding current, the effect of changing
Page 6
106
the welding speed confirms to a general pattern. If the welding speed is increased,
power or heat input per unit length of weld is decreased and less filler metal is applied
per unit length of the weld, resulting in less weld reinforcement. Thus, the weld bead
becomes smaller.
Weld penetration is affected more by welding speed than any variable other
than current. This is true except for excessively slow speeds when the molten weld
pool is beneath the welding electrode. Then the penetrating force of the arc is
cushioned by the molten pool. Excessive speed any cause undercutting, porosity, arc
blow, uneven bead shape, cracking and higher slag inclusion in the weld metal. Higher
welding speed results in less heat affected zone and finer grains (Aksoy et al.1999).
Within limits, welding speed can be adjusted to control weld size and
penetration. Relatively slow welding speed provides time for gases to escape from the
molten metal, thus reducing porosity. An excessive slow speed produces a convex
bead shape which is subject to cracking and excessive arc exposure which is
uncomfortable for the operator. Too low welding speed may also result in a large
molten pool that flows around the arc, resulting in rough bead, slag inclusions and burn
through of the weld plate. Jackson and Shrubsa (1953) reported that the welding speed
did not affect the metal deposition rate significantly.
5.2.4 Electrode size
Electrode size affects the weld bead shape and the depth of penetration at fixed
current. Electrode size also influences the deposition rate. At any given current, a small
diameter electrode will have a higher current density and a higher deposition rate than
Page 7
107
a larger electrode. However, a larger diameter electrode can carry more current than a
smaller electrode, and produce a higher deposition rate at higher amperage. For the
same values of current, arc voltage and welding speed, an increase in electrode
diameter results in a slight increase in the spread of the bead (Cornu, 1988).
5.2.5 Electrode work angle
The electrode may be held perpendicular to the workpiece or, tilted forward or
backward with respect to the weld pool. As the arc stream tends to align itself along
the axis of the electrode, the weld pool shape is different in each case, and so is the
shape of the weld bead. It is observed that in forehand welding, molten metal flows
under the arc, the depth of penetration and reinforcement are reduced while the width
of the weld increases, whereas in backhand welding the pressure of the arc scoops the
molten metal from beneath the arc, the depth of penetration and height of
reinforcement increases while the width of the weld is reduced (Nadkarni, 1988). The
electrode in perpendicular position results in bead geometry in between those obtained
in the above two cases.
5.2.6 Electrode stick-out and melting rate
The distance between the current pick-up tip and the arc root, called electrode
stick out, has a considerable effect on the weld bead geometry. Normally the distance
between the contact tip and the work is 25-40 mm.
Page 8
108
The increase in melting rate of the electrode as a result of increase in electrode stick-
out is proportionate to the product of current density and stick-out. The electrode
melting rate in kg/min is given by the relationship,
Electrode melting rate =1
1000 0.35 +
d2
645+ 2.08 × 10−7 ×
IL × 25.4
d2
1.22
(5.1)
Where d, L and I are the diameter of the electrode, electrode stick-out in mm and
current density respectively.
Chandel et al. (1997) reported that the melting rate of the electrode increased
with the increase in the stick out. This effect is particularly more significant with
smaller diameter electrode since electrode heating is caused by the electrode electric
resistance, which increases with the decrease in the electrode diameter. The depth of
penetration decreases with the increase in electrode stick-out. This factor needs to be
given due consideration where deeper penetration is required. Gunaraj and Murugan
(1999) reported that heat affected zone decreased with the increase in stick- out.
Janez (2000) reported that a mutual influence of the arcs was quite strong and
consequently melting rate was high in twin-wire welding. He further reported that arc
energy melted more filler material per wire in twin-wire welding than in single-wire
welding and with the same welding parameters, this required higher wire feed speed in
twin-wire welding.
Page 9
109
5.2.7 Depth of flux
The depth of the layer of the granular flux influences the appearance and
soundness of the finished weld as well as welding action. If the granular flux layer is
too shallow, the arc will not be entirely submerged in flux. Flashing and spattering
will occur. Apart from injurious to the eyes of the operator, this may lead to poor
appearance of weld and it may also be porous. If the flux layer is too thick, the arc will
be too confined and a rough ropelike appearing weld will result and the weld bead may
be narrow and humped. The gases generated during welding may not be able to escape,
and the surface of the molten weld metal becomes irregularly distorted. Optimum
depth of flux can be established by slowly increasing the flow of flux until the welding
arc is submerged and flashing no longer occurs. The gases will then puff up quietly
around the electrode, sometimes igniting.
5.2.8 Polarity
The amount of heat generated at the electrode and work piece, deposition rate,
bead geometry and mechanical properties are affected by polarity. The change in
polarity from DCEP to DCEN changes the amount of heat generated at electrode and
the work piece and, hence the metal depositing rate, weld bead geometry and
mechanical properties of the weld metal (Robinson, 1983). Little (1976) observed that
the two third of the total heat was generated at the positive electrode and the one third
of the total heat was generated at the negative electrode.
Page 10
110
It has been reported by Renwick et al. (1976) that DCEN polarity produced higher
deposition rate and reinforcement than with DCEP polarity in submerged arc welding.
Ghosh et al. (1991) observed high yield strength, ultimate tensile strength and hardness
of the weld metal with DCEN polarity as compared to DCEP polarity.
5.2.10 Flux basicity index
Flux basicity index also influences the penetration (Gupta and Gupta, 1988). In
general higher penetration is obtained with the use of low basicity index fluxes due to
high viscosity which enhances the tendency of heat concentration in the narrow zone.
Patchett and Dancy (1980) reported that the penetration increased with the increase in
slag viscosity and surface tension. They also observed that an increase in viscosity, arc
stability and surface tension resulted in deeper penetration.
5.3 Weld Bead Shape
The weld bead shape is an indication of bead geometry which affects the load
carrying capacity of the weldments (Baach et al., 1981., Samiti, 1986) and number of
passes needed to fill the groove of a joint. The bead geometry is specified by bead
width, reinforcement, penetration, penetration shape factor and reinforcement form
factor.
5.3.1 Weld bead width
The weld bead width is the maximum width of the weld metal deposited. It
influences the flux consumption rate and chemistry of the weld metal. Weld bead
Page 11
111
width is directly proportional to arc current, welding voltage and electrode diameter
and indirectly proportional to the welding speed. The bead width increases with an
increase in electrode diameter (McGlone, 1982). Gupta and Arora (1991) observed that
bead width increased with an increase in current until it reaches a critical value and
then it decreases with an increase in welding current. Yang et al. (1992) investigated
that the bead width was not affected significantly by the types of power source
(constant voltage or constant current) when an acidic fused flux was used. However,
using a basic fused flux with constant current operation showed somewhat larger bead
width than with welds laid using acidic fused flux.
5.3.2 Penetration
Weld bead penetration is the maximum distance between the base plate top
surface and depth to which the fusion has taken place. The more the penetration, the
less is the number of welding passes required to fill the weld joint which consequently
results in higher production rate. It is observed that the penetration is influenced by
welding current, polarity, arc travel speed, electrode stick-out, basicity index and
physical properties of the flux. McGlone (1982) observed that penetration was directly
proportional to welding current. He also observed that the deepest penetration was
achieved when DCEP polarity was used and the least with DCEN polarity. He further
investigated that the penetration was indirectly proportional to welding speed and
electrode diameter. Penetration decreases with the increase in welding speed because
the time during which the arc force is allowed to penetrate into the material‟s surface
Page 12
112
decreases. The penetration decreases with the increase in electrode diameter due to
decrease in current density (Cornu, 1988). Chandel et al. (1987) reported that the
penetration increased with the decrease in electrode extension and included angle of
the joint. Caddle (1967) reported that the penetration increased with a decrease in
thermal conductivity of the weld metal.
5.3.3 Reinforcement
Reinforcement is the maximum distance between the base metal level and the
top point of the deposited metal. Reinforcement is the crown height of the weld bead
from the base plate. It affects the strength of the weld joint and welding wire
consumption rate. It increases with the increase in welding wire feed rate irrespective
of the welding current and the type of polarity employed (Gunaraj and Murugan 1999).
It is indirectly proportional to welding voltage, welding speed and electrode diameter.
The reinforcement is more with DCEN polarity and less with DCEP polarity. Increase
of reinforcement with an increase of welding filler wire feed rate is mainly due to the
larger amount of metal deposited per unit length. The decrease of reinforcement with
the increase in voltage is due to increase in weld bead width.
5.3.4 Weld penetration shape factor (WPSF) and weld reinforcement form
factor (WRFF)
WPSF and WRFF are also called as coefficients of internal shape and external
shape respectively. The ratio of bead width to penetration and bead width to
Page 13
113
reinforcement are termed as Weld Penetration Shape Factor (WPSF) and Weld
Reinforcement Form Factor (WRFF) respectively. The smoothness of the weld
increases with the increase in WRFF (Cornu, 1988). Mandotov (1969) and Srihari
(1992) reported that WPSF and WRFF increased with an increase in voltage.
5.4 Flux Consumption
Flux consumption influences the economic aspects of welding and chemical
composition of the weld metal. Flux consumption depends upon the welding
parameters such as welding current, arc voltage, welding speed, polarity and type of
flux. Flux consumption increases with the increase in arc voltage and decrease in
current. The electrode extension has no significant effect on flux consumption (Gupta
and Gupta, 1988). Agglomerated fluxes have low flux consumption as compared to
fused fluxes (Vishvanath, 1982).
5.5 Experimental Procedure
The machine employed for experimentation was Ador TORNADO-800. The
composition of the welding wire (4mm diameter) and base plate are shown in Table 5.1.
The mild steel plates were cleaned chemically and mechanically to remove the oxide
layer and any other source of hydrogen. Bead on plate welds were laid on the plates of
200x75x12 mm size, using the developed fluxes. The ranges of the parameters and their
level have already been reported in Table 3.2 and the parameters were varied as per the
design matrix shown in Table 3.3. The basis for selection of the range and the level of
Page 14
114
the parameters has already been discussed in Chapter-3. The experiments were
performed in random manner to avoid any systematic error.
After welding, transverse sections of the weld beads were cut from the middle
portions of the plates as specimens. These specimens were prepared by standard
metallurgical polishing methods. The properly polished specimens were etched with a
2% Nital solution for about 30 seconds, which was followed by investigation and
analysis. For each of the bead-on-plate specimens, the important dimensions of the
weld bead geometry were measured. The average response parameters (bead width,
penetration, reinforcement, weld penetration shape factor and weld reinforcement form
factor) and flux consumption were recorded by conducting experiments as per design
matrix (Table 4.2) are shown in Table-5.2. With the help of these observed responses,
models were developed.
Table 5.1 Chemical composition of base plate and electrode wire
Element
(%) C Mn Si S P Ni Cr
Base
Plate 0.23 0.42 0.127 0.039 0.056 0.065 0.113
Electrode
Wire 0.069 1.86 0.1 0.028 0.023 Nil Nil
Page 15
115
Table 5.2 Observed values of bead parameters and flux consumption
Expt.
Run
No.
Response factors
W
Bead Width
(mm)
P
Penetration
(mm)
R
Reinforcement
(mm)
WPSF
WRFF F
Flux
Consumption
(gms.)
n
(gms)
1 17.76 6.545 3.201 2.71352 5.54827 50
2 17.24 7.349 4.043 2.3459 4.26416 37.02
3 17.81 6.735 3.345 2.64439 5.32436 45.77
4 15.192 6.671 2.955 2.27732 5.14112 40.29
5 16.79 10.455 6.025 1.60593 2.78672 21.88
6 17.48 6.389 2.382 2.73595 7.33837 48.34
7 17.2 6.17 3.245 2.78768 5.30046 43.42
8 17.215 8.56 3.55 2.0111 4.8493 49.53
9 17.54 6.66 2.94 2.63363 5.96599 50.83
10 16.345 7.115 4.58 2.29726 3.56878 40.39
11 16.792 6.4 3.59 2.62375 4.67744 49.08
12 19.475 7.53 2.544 2.58632 7.65527 60.24
13 15.812 8.75 4.392 1.80709 3.60018 32.44
14 16.99 6.525 3.01 2.60383 5.64452 44
15 17.515 11.323 4.191 1.54685 4.17919 26.09
Page 16
116
Table 5.2 Observed values of bead parameters and flux consumption (Continued)
Expt.
Run
No.
Response factors
W
Bead Width
(mm)
P
Penetration
(mm)
R
Reinforcement
(mm)
WPSF
WRFF F
Flux
Consumption
(gms.)
n
(gms)
16 18.855 7.355 2.899 2.56356 6.50397 52
17 18.985 9.285 3.22 2.0447 5.89596 52.38
18 17.865 7.895 3.135 2.26282 5.69856 39.46
19 16.2 6.6 3.165 2.45455 5.11848 38.28
20 15.295 7.935 4.875 1.92754 3.13744 31.25
21 16.873 6.34 3.192 2.66136 5.28603 54.23
22 19.355 7.835 3.45 2.47033 5.61014 50.93
23 15.18 7.255 4.045 2.09235 3.75278 30.26
24 20.44 6.99 3.565 2.92418 5.73352 65
25 15.56 7.025 3.49 2.21495 4.45845 32.86
26 17.905 11.07 4.906 1.61743 3.64961 37.69
27 17.042 6.7 3.205 2.54358 5.31732 49.79
28 17.365 6.32 3.475 2.74763 4.99712 54.61
29 16.63 8.272 4.005 2.0104 4.15231 35.89
Page 17
117
5.6 Development of Model
Response surface methodology‟s Box-Bohnken design consisting of twenty
nine experiments was conducted to develop model showing the relationships between
the response Y (bead width, penetration, reinforcement, weld penetration shape factor
and weld reinforcement form factor) and the welding parameters (open circuit voltage
A, welding current B, welding speed C and flux basicity index D) for coded values of -
1 to +1 for each of the welding parameters.
To test the goodness of the fit and validation of the developed models,
adequacy was determined by the analysis of variance technique (ANOVA). The
analysis of variance test was performed to evaluate the statistical significance of the
fitted quadratic models and factors involved therein for response factors W, P, R,
WPSF,WRFF and flux consumption (F). In addition to this, the goodness of fit of the
fitted quadratic model was also evaluated through „lack of fit test‟. The "Prob > F" for
all these tests was found in excess of 0.05, implying that the lack of fit is insignificant.
The results obtained are summarized in Tables-5.3 to 5.8.
All the fitted models are found to be significant, since for all the responses, the
Prob. > F are observed to be less than 0.0001. In other words, there is only a 0.01%
chance that "Model F-Value" larger than those reported in Tables-5.3 to 5.8 could
occur due to noise. The values of "Prob > F" less than 0.05 observed for some factors
involved in model equations, indicate that the contribution of these terms to the model
is significant. On the other hand, the value of "Prob > F" greater than 0.10 indicates
that the impact of model terms are not significant.
Page 18
118
Table 5.3 ANOVA results for bead width (W)
Source
Sum of
Squares DF
Mean
Square
F
Value Prob. > F Remarks
Model 45.46 14 3.25 29.75 <0.0001 significant
A 12.79 1 12.79 117.19 < 0.0001 significant
B 0.73 1 0.73 6.69 0.0216 significant
C 9.39 1 9.39 86.03 < 0.0001 significant
D 0.18 1 0.18 1.64 0.2213 not
significant
A2 0.25 1 0.25 2.33 0.1493
not
significant
B2 1.86 1 1.86 17.03 0.0010 significant
C2 0.56 1 0.56 5.09 0.0405 significant
D2 3.57 1 3.57 32.66 < 0.0001
signific
ant
AB 0.46 1 0.46 4.24 0.0587 not
significant
AC 1.44 1 1.44 13.21 0.0027 significant
AD 0.27 1 0.27 2.43 0.1410 not
significant
BC 0.34 1 0.34 3.07 0.1016 not
significant
BD 0.065 1 0.065 0.60 0.4522 not
significant
CD 0.019 1 0.019 00.17 0.6851 not
significant
Residual 1.53 14 0.11
Lack of
Fit 0.88 10 0.088 0.54 0.8037
not
significant
Pure
Error 0.65 4 0.16
Cor
Total 46.99 28
Page 19
119
Table 5.4 ANOVA results for penetration (P)
Source
Sum of
Squares DF
Mean
Square
F
Value Prob. > F Remarks
Model 54.66 14 3.90 48.99 < 0.0001 significant
A 0.76 1 0.76 9.57 0.0079 significant
B 7.51 1 7.51 94.26 < 0.0001 significant
C 4.10 1 4.10 51.47 < 0.0001 significant
D 39.50 1 39.50 495.75 < 0.0001 significant
A2 0.68 1 0.68 8.49 0.0113 significant
B2 0.53 1 0.53 6.70 0.0215 significant
C2 0.21 1 0.21 2.62 0.1276
not
significant
D2 5.10 1 5.10 64.02 < 0.0001 significant
AB 2.5E003 1 2.5E003 0.031 0.8619 not
significant
AC 0.16 1 0.16 1.96 0.1835 not
significant
AD 0.054 1 0.05 0.68 0.4244 not
significant
BC 1.34 1 1.34 16.76 0.0011 significant
BD 1.64 1 1.64 20.54 0.0005 significant
CD 1.35 1 1.351243 16.96 0.0010 significant
Residual 1.12 14 0.08
Lack of
Fit 0.88 10 0.088 1.50 0.3692
not
significant
Pure
Error 0.23 4 0.059
Cor
Total 55.77 28
Page 20
120
Table 5.5 ANOVA results for reinforcement (R)
Source
Sum of
Squares DF
Mean
Square
F
Value Prob. > F Remarks
Model 16.67 14 1.19 27.64 < 0.0001 significant
A 6.95 1 6.95 161.30 < 0.0001 significant
B 0.3 1 0.3 7.05 0.0188 significant
C 0.47 1 0.47 10.92 0.0052 significant
D 4.49 1 4.49 104.22 < 0.0001 significant
A2 1.03 1 1.03 23.90 0.0002 significant
B2 0.14 1 0.14 3.21 0.0947
not
significant
C2 0.33 1 0.33 7.57 0.0156 significant
D2 3.17 1 3.17 73.52 < 0.0001 significant
AB 0.71 1 0.71 16.52 0.0012 significant
AC 6.56E003 1 6.56E003 0.15 0.7022 not
significant
AD 1.19 1 1.2 27.71 0.0001 significant
BC 4.556E003 1 4.556E003 0.11 0.7498 not
significant
BD 1.16E004 1 1.16E004 2.693E003 0.9593 not
significant
CD 1.020E003 1 1.020E003 0.024 0.8799 not
significant
Residual 0.60 14 0.043
Lack of
Fit 0.38 10 0.038 0.69 0.7128
not
significant
Pure
Error 0.22 4 0.055
Cor
Total 17.27 28
Page 21
121
Table 5.6 ANOVA results for weld penetration shape factor (WPSF)
Source
Sum of
Squares DF
Mean
Square
F
Value Prob. > F Remarks
Model 4.02 14 0.29 29.30 < 0.0001
significant
A 0.42 1 0.42 42.72 < 0.0001
significant
B 0.31 1 0.31 31.64 < 0.0001
significant
C
6.806E003 1 6.806E003 0.69 0.4185
not
significant
D 2.40 1 2.40 244.91 < 0.0001
significant
A2
0.062 1 0.062 6.35 0.0245 significant
B2
0.17 1 0.17 17.34 0.0010 significant
C2
0.06 1 0.064 6.56 0.0226 significant
D2
0.55 1 0.55 55.77 < 0.0001 significant
AB 4.727E003 1 4.727E003 0.48 0.4988
significant
AC
8.384E004 1 8.384E004 0.086 0.7741
not
significant
AD
0.021 1 0.02 2.18 0.1617
not
significant
BC 0.21 1 0.21 21.31 0.0004
significant
BD
0.027 1 0.027 2.74 0.1200
not
significant
CD 0.051 1 0.05 5.22 0.0384
significant
Residual 0.14 14 9.794E003
Lack of
Fit 0.11 10 0.01 1.36 0.4123
not
significant
Pure
Error 0.031 4 7.803E003
Cor
Total 4.16 28
Page 22
122
Table 5.7 ANOVA results for weld reinforcement form factor (WRFF)
Source
Sum of
Squares DF
Mean
Square
F
Value Prob. > F Remarks
Model 34.29 14 2.45 11.98 < 0.0001
significant
A 16.10 1 16.10 78.72 < 0.0001
significant
B
0.39 1 0.39 1.91 0.1886
not
significant
C
2.737E005 1 2.737E005 1.338E004 0.9909
not
significant
D 4.51 1 4.51 22.07 0.0003
significant
A2
0.49 1 0.49 2.40 0.1436
not
significant
B2
0.01 1 0.01 0.054 0.8193
not
significant
C2
1.49 1 1.49 7.31 0.0171 significant
D2
9.61 1 9.61 46.98 < 0.0001 significant
AB 1.30 1 1.30 6.34 0.0246
significant
AC
0.10 1 0.10 0.49 0.4964
not
significant
AD
0.74 1 0.74 3.64 0.0772
not
significant
BC
3.918E003 1 3.918E003 0.019 0.8919
not
significant
BD
0.037 1 0.037 0.18 0.6754
not
significant
CD
2.482E003 1 2.482E003 0.012 0.9138
not
significant
Residual 2.86 14 0.2
Lack of
Fit 2.03 10 0.2 0.98 0.5583
not
significant
Pure
Error 0.83 4 0.21
Cor
Total 37.15 28
Page 23
123
Table 5.8 ANOVA results for flux consumption (F)
Source
Sum of
Squares DF
Mean
Square
F
Value Prob. > F Remarks
Model 2842.67 14 203.05 27.12 < 0.0001
significant
A 1352.9 1 1352.9 180.7 < 0.0001
significant
B 88.34 1 88.34 11.80 0.0040
significant
C 137.08 1 137.08 18.31 0.0008
significant
D 600.1 1 600.1 80.17 < 0.0001
significant
A2
2.84 1 2.84 0.38 0.5481
not
significant
B2
118.56 1 118.56 15.84 0.0014 significant
C2
2.39 1 2.39 0.32 0.5810
not
significant
D2
175.59 1 175.59 23.46 0.0003 significant
AB
3.26 1 3.26 0.44 0.5201
not
significant
AC
0.10 1 0.10 0.014 0.9084
not
significant
AD
28.50 1 28.50 3.81 0.0714
not
significant
BC
2.36 1 2.36 0.31 0.5836
not
significant
BD 35.62 1 35.62 4.76 0.0467
significant
CD
20.17 1 20.17 2.69 0.1230
not
significant
Residual 104.80 14 7.49
Lack of
Fit 66.72 10 6.67 0.70 0.7054
not
significant
Pure
Error 38.08 4 9.52
Cor
Total 2947.47 28
Page 24
124
The ANOVA results for bead width (Table5.3) show that A, B, C, B2, C2, D2, AC are
significant model terms. The ANOVA results for penetration that (Table 5.4) reveals
that A, B, C, D, A2, B2, D2, BC, BD, CD are significant model terms. The ANOVA
results for reinforcement (Table 5.5) shows that A, B, C, D, A2, C2, D2, AB, AD are
significant model terms. The ANOVA results for weld penetration shape factor (Table
5.6) shows that A, B, D, A2, B2, C2, D2, BC, CD are significant model terms. The
ANOVA results for weld reinforcement from factor (Table 5.7) reveals that A, D, C2,
D2, AB are significant model terms. The ANOVA results for flux consumption (Table
5.7) reveals that A, B, C, D, B2, D2, BD are significant model terms.
Tables-5.9 to 5.14 show the model summary statistics for all responses. The
coefficients of correlation (R2) for all the models are observed in excess of 0.92 which
inspire confidence in the developed models. The predicted and adjusted R2
values for
all the response models were in reasonable agreement which again validates the fitness
of developed models. The coefficient of variation (C.V.) defined as (S.D./Mean x 100)
of model is measurement of error. The low value of C.V. obtained for all the models
indicates improved precision and reliability of the experiments performed. The
adequate precision values, defined as signal to noise ratio for the fitted value, are
significantly higher than 4 indicating the suitability of models for future prediction.
Page 25
125
Table 5.9 Model summary statistics for bead width
Std. Dev. 0.33 (R2) 0.967
Mean 17.27 Adjusted (R2) 0.935
C.V.(%) 1.91 Predicted (R2) 0.845
PRESS 7.28 Adequate Precision (AP) 22.143
Table 5.10 Model summary statistics for penetration
Std. Dev. 0.28 (R2) 0.98
Mean 7.59 Adjusted (R2) 0.96
C.V.(%) 3.72 Predicted (R2) 0.9087
PRESS 5.09 Adequate Precision (AP) 25.087
Table 5.11 Model summary statistics for reinforcement
Std. Dev. 0.21 (R2) 0.965
Mean 3.61 Adjusted (R2) 0.930
C.V.(%) 5.75 Predicted (R2) 0.841
PRESS 2.75 Adequate Precision (AP) 24.08
Table 5.12 Model summary statistics for weld penetration shape factor
Std. Dev. 0.099 (R2) 0.967
Mean 2.34 Adjusted (R2) 0.934
C.V.(%) 4.24 Predicted (R2) 0.829
PRESS 0.71 Adequate Precision (AP) 18.614
Page 26
126
Table 5.13 Model summary statistics for weld reinforcement form factor
Std. Dev. 0.45 (R2) 0.9229
Mean 5.01 Adjusted (R2) 0.8459
C.V.(%) 9.03 Predicted (R2) 0.6748
PRESS 12.08 Adequate Precision (AP) 15.638
Table 5.14 Model summary statistics for flux consumption
Std. Dev. 2.74 (R2) 0.9644
Mean 43.58 Adjusted (R2) 0.9289
C.V.(%) 6.28 Predicted (R2) 0.8135
PRESS 549.81 Adequate Precision (AP) 22.272
To test the accuracy of the models in actual applications, conformity test runs
were conducted by assigning different values for process variables within their
working limits. Specimens were cut from the conformity test plates and their bead
profiles were traced. All bead dimensions were measured. The percentage of errors,
which give the deviation of predicted results of responses from the actual measured
values, were also calculated and presented in Table-5.15. It is found from the table
that the average error for all models is less than 3%.
5.7 Results and Discussion
The developed mathematical models can be employed to predict the geometry
of weld bead and shape relationships for the range of parameters used in the
investigation by substituting their respective values in coded form. The predicted
Page 27
127
values of response factors W, P, R, WPSF, WRFF and Flux consumption F from
regression equations (5.2) to (5.7) corresponding to different combination of welding
variables reported in Table-3.2 are compared with the corresponding experimental
values. A nice agreement is observed between these values, as evident from Figs.5.2-
5.7.
Table 5.15 Comparison of actual and predicted values of weld bead parameters
% Error = Act. Value − Pred. Value ÷ Pred. Value X100
Trial No.
Predicted values of
bead parameters
Actual values of bead
parameters
% Error
W P R W P R W P R
1 17.83 6.6.48 3.24 17.29 6.54 3.3 -3.02 1 1.85
2 17.28 6.53 3.27 17.63 6.51 3.19
2.02 -0.3 -2.44
3 16.17 7.30 4.66 16.63 7.12 4.62
2.87 -2.46 -3
4 18.89 7.37 2.79 18.87 7.39 2.82
-0.1 0.27 1.07
5 17.45 6.33 2.6 17.57 6.51 2.67
0.68 2.84 2.69
6 18.03 8.23 3.24 18.12 7.99 3.06
0.49 -2.91 3
Av.error(%)
0.49 -0.26 0.52
Page 28
128
Fig. 5.2 Comparison between measured and predicted value of bead width (W)
Fig. 5.3 Comparison between measured and predicted value of penetration (P)
Page 29
129
Fig. 5.4 Comparison between measured and predicted value of reinforcement (R)
Fig. 5.5 Comparison between measured and predicted value of WPSF
Page 30
130
Fig. 5.6 Comparison between measured and predicted value of WRFF
Fig. 5.7 Comparison between measured and predicted value of flux consumption (F)
Page 31
131
Based on these models, the main and the interaction effects of the process parameters
on the bead geometry were computed and plotted as depicted in Figs.5.8-5.40. The
results show the general trends between the cause and effect. The possible causes for
the effects of different welding variables on bead geometry and shape relationships
were analyzed and discussed below:
5.7.1 Effect of process parameters on bead width
Direct effect
The regression equations obtained for bead width by using multiple regressions
are given below:
Bead Width W = 18.28 + 1.23 ∗ A + 0.29 ∗ B1.06 ∗ C0.10 ∗ D + 0.20 ∗
A2 0.54 ∗ B20.29 ∗ C20.90 ∗ D2
+ 0.34 ∗ A ∗ B0.60 ∗ A ∗ C + 0.19A ∗ D +
0.29 ∗ B ∗ C 0.097 ∗ B ∗ D + 0.052 ∗ C ∗ D (5.2)
Figs.5.8-5.10 show the effect of process parameters on bead width. It is
apparent that bead width increases with the increase in open circuit voltage. As shown
in Fig.5.8, bead width (W) increases from 17.24 to 19.70 mm with the increase in open
circuit voltage from 32 to 38 volts. It can be attributed to the increase in arc length with
the increase in open circuit voltage, which in turn results in spreading of the arc cone at
its base which further results in more melting of work piece instead of penetrating the
plate. This extension in bead width causes corresponding reduction in penetration and
reinforcement. In fact excessive increase in voltage can result in nearly flat bead.
Page 32
132
Fig. 5.8 Effect of voltage on bead width
Bead width increases from 17.44 to 18.03 mm with increase in welding current from
375 to 475 amperes, as shown in Fig. 5.9. This effect is due to increase in heat input
and the weight of the weld metal deposited (Gunaraj and Murgun, 1999). These factors
contribute to increase in weld pool size and consequently increase the bead width. As
shown in Fig.5.10, weld bead width decreases steadily with the increase in welding
speed. The bead width decreases from 19.04 to 16.92 with increase in welding speed
from 24 to 30 m/hr. This negative effect of speed on W is due to the fact that when
speed increases, the thermal energy transmitted to the base plate from the arc or line
power per unit length of the weld bead decreases and less filler metal is deposited per
unit length of weld bead, resulting in thinner and narrower weld bead. Hence, at lower
travel speeds, the weld bead is larger in mass, whereas at higher travel speeds, it is
lesser in mass. If speed decreases, the bead becomes wider, flatter and smoother
(Olson et al., 1990). It can be explained on the basis of decrease in metal deposition
Page 33
133
rate and heat input with the increase in welding speed. The effect of basicity index on
bead width is not significant.
Fig. 5.9 Effect of current on bead width
Fig. 5.10 Effect of welding speed on bead width
Page 34
134
Interaction effect
It is apparent from Figs.5.11 and 5.12, showing interaction of open circuit
voltage and welding speed on bead width (W) that the increase in voltage increases W
for all values of speed. The bead width increases from 17.40 to 21.07 mm and from
16.49 to 17.75 mm with the increase in voltage from 32 to 38 volts, at the welding
speed 24 and 30 m/hr respectively. It shows that the increasing trend of bead width
with the increase in open circuit voltage decreases with the increase in welding speed.
It is due to the fact that open circuit voltage has a positive effect whereas welding
speed has a negative effect on bead width. Therefore, the combined effect of these
parameters causes the decrease in increasing trend of bead width with the increase in
Open circuit voltage.
Fig. 5.11 Interaction effect of voltage and speed on bead width
C = 24m/ hr
C = 30m/hr
Page 35
135
Fig. 5.12 Response surface due to interaction of voltage and speed on bead width
5.7.2 Effect of process parameters on penetration
Direct effect
The developed model for the penetration is shown below:
Penetration (P) = 6.990.30 * A + 0.95 * B0.70 * C1.53* D+0.32 * A2+ 0.29 *
B2+0.18 * C2+1.07* D20.025* A* B0.20 * A * C + 0.088 * A * D-0.58 * B *
C0.48 * B * D + 0.44 * C * D (5.3)
As shown in Fig.5.13, the penetration (P) increases from 6.33 to 8.22 mm with
the increase in welding current from 375 to 475 amperes. Increase in current gives rise
to enhanced line power per unit length of the weld bead and higher current density,
causing larger volume of the base material to melt and hence, deeper penetration. As
current increases the temperature, the heat content of the droplets also increases, which
results in more heat being transferred to the base material. Increase in current also
Page 36
136
increases momentum of the droplets, which on striking the weld pool causes a deeper
penetration. An increase in welding current, with other variables remaining constant,
results in increased depth of penetration, increased deposition rate and increased weld
bead size and shape at a given cross-section. It is also attributed to the increase in
digging power of the arc with the increase in welding current. As the current increases,
the intensity of the arc and hence the digging power of the arc and penetration
increases. This is also consistent with the study of MacGlone (1982).
Fig. 5.13 Effect of current on penetration
As depicted in Fig.5.14, the penetration decreases from 7.86 to 6.47 mm with
the increase in welding speed from 24 to 30 m/hr. This could obviously be due to the
reduced line power per unit length of weld bead as speed increases. Also, at higher
welding speeds, the electrode travels faster and covers more distance per unit time. The
combined effects of lesser line power and faster electrode travel speed result in
Page 37
137
decreased metal deposition rate per unit length of weld bead (Box et al., 1976). It is
also attributed to decrease in heat input, metal deposition rate and digging power of the
arc with the increase in welding speed resulting in decrease in weld metal penetration.
Fig. 5.14 Effect of welding speed on penetration
From Fig. 5.15, it is observed that P decreases from 7.61 to 7.01 when open
circuit voltage increases from 32 to 38 volts. This is obviously due to the fact that the
increase in voltage results in increased arc length and spreading of arc cone at its base
which results in more melting of work piece surface instead of penetrating the plate.
This is consistent with the study conducted by Murugan and Gunaraj (2005). In fact,
excessive increase in voltage can result in nearly flat bead. Flux basicity index also
influences the penetration. It is observed from Fig.5.16, the higher value of
penetration i.e 9.59 mm is obtained with using low basicity index flux (0.6), because
Page 38
138
low basicity index fluxes have high viscosity which enhances the tendency of heat
concentration in the narrow zone and hence high penetration. This is consistent with
the study conducted by Gupta and Gupta (1988).
Fig. 5.15 Effect of voltage on penetration
Fig. 5.16 Effect of basicity index on penetration
Page 39
139
Interaction effect
From the Fig.5.17, it is evident that P increases with the increase in welding
current for all values of welding speed. It shows that the weld metal penetration
increases from 6.63 to 9.67 mm and from 6.38 to 7.12, with the increase in current, at
the welding speed of 24 to 30 m/hr respectively. The rate of increase in P with the
increase in current decreases gradually as speed increases. These effects on P are due
to the reasons that current has positive effect but speed has a negative effect on P as
discussed already in the direct effects of current and speed on P. It is found that at
lower values of speed, the positive effect of current on P is stronger but at higher
values of speed, the negative effect of speed on P is stronger. These effects are further
explained with the help of a response surface plot as shown in Fig.5.18. From the
contour surface, it is noted that P is maximum (about 9.67 mm) when current and
speed are at their maximum (+1) and minimum (−1) limits, respectively, and the
lowest value of P (about 6.39 mm) is obtained when current and speed are at their
minimum and maximum limits, respectively.
From Figs. 5.19 and 5.20, it is observed that penetration increases from 8.45 to
11.31 mm and from 6.38 to 7.28 mm, with increase in current, at the basicity index of
0.6 and 1.2 respectively. It is evident form Figs.5.21 and 5.22 that penetration
decreases form 10.91 to 8.63 and from 6.97 to 6.45 with increase in welding speed
from low basicity index to higher value of basicity index. These results can be
explained with the help of effects of welding variables such as welding speed and
basicity index on penetration.
Page 40
140
Fig.5.17 Interaction effect of current and speed on penetration
Fig. 5.18 Response surface due to interaction of current and speed on
penetration
C = 24m/ hr
C = 30m/hr
Page 41
141
Fig. 5.19 Interaction effect of current and basicity index on penetration
Fig. 5.20 Response surface due to interaction of current and basicity index on
penetration
D = 1.2
D = 0.6
Page 42
142
Fig. 5.21 Interaction effect of welding speed and basicity index on penetration
Fig. 5.22 Response surface due to interaction of welding speed and basicity index on
penetration
D = 1.2
D = 0.6
Page 43
143
5.7.3 Effect of process parameters on reinforcement (R)
Direct effects
The developed model for the reinforcement is shown below:
Reinforcement (R) = 2.94-0.91* A+ 0.19 * B-0.24 * C-0.52* D + 0.40 * A20.15 *
B2+0.22 * C2 + 0.85 * D2 0.42*A*B + 0.041*A*C + 0.41* A* D + 0.034* B * C +
4.071E003 * B * D+0.012*C * D (5.4)
From Figs.5.235.25, it is observed that the reinforcement (R) decreases with
the increase in open circuit voltage and welding speed, it increases with the increase in
welding current. Reinforcement decreases with increase in basicity index due to
similar reasons as described for penetration. It is seen from these graphs that
reinforcement decreases from 4.24 to 2.42 mm with change of voltage from 32 to 38
volts, and decreases from 3.39 to 2.92 mm when welding speed increases from 24 to
30 m/hr. When current changes from 375 to 475 amperes, it changes from 2.59 to 2.97
mm. As evident from Fig.5.26, its value increases from 4.29 to 3.26 mm with increase
in basicity index from 0.6 to 1.2. The reasons for these changes are due to same
reasons as described in preceding section for penetration.
Interaction effects
It is observed from Figs.5.27 and 5.28 that reinforcement decreases with the
increase in voltage, when the current changes from 375 to 475 amperes and it also
decreases with voltage from low basicity index to higher value of basicity index, as
shown in Figs.5.29 and 5.30. These interaction effects can be explained on the basis of
effect of voltage, current and basicity index on reinforcement.
Page 44
144
Fig. 5.23 Effect of voltage on reinforcement
Fig. 5.24 Effect of welding speed on reinforcement
Page 45
145
Fig. 5.25 Effect of current on reinforcement
Fig. 5.26 Effect of basicity index on reinforcement
Page 46
146
Fig. 5.27 Interaction effect of voltage and current on reinforcement
Fig. 5.28 Response surface due to interaction of voltage and current on
reinforcement
B = 375 amp
B =475 amp
Page 47
147
Fig. 5.29 Interaction effect of voltage and basicity index on reinforcement
Fig. 5.30 Response surface due to interaction of voltage and basicity index on
reinforcement
D = 0.6
D = 1.2
Page 48
148
5.7.4 Effect of process parameters on weld penetration shape factor (WPSF)
and weld reinforcement form factor (WRFF)
Direct effects
The developed model for the weld penetration shape factor and weld
reinforcement form factor are shown below:
WPSF = 2.62 + 0.22*A0.19 * B+0.028 * C+0.38* D0.098 * A20.16* B20.100*
C20.35* D2+ 0.034 * A* B0.014 * A * C+0.055 * A * D + 0.23* B * C + 0.062 * B
* D0.086* C* D (5.5)
WRFF = 6.27 + 1.38 * A0.22 * B + 1.805E-003 * C + 0.52* D0.28 * A2 + 0.041*
B20.48 * C21.47 * D2 + 0.57 * A * B0.16 * A * C0.33 * A * D + 0.031* B * C
+ 0.073 * B * D + 0.019* C * D (5.6)
Figs.5.31 and 5.32 show the effect of open circuit voltage on both weld
penetration shape factor (WPSF) and weld reinforcement form factor (WRFF). WPSF
increases from 2.29 to 2.74 and with the increase in open circuit voltage from 32 to 38
volts and WRFF increases from 4.61 to 7.38 with increase of open circuit voltage from
32 to 38 volts. This is also in consistent with the studies of Mandotov (1969) and
Srihari (1992).The positive effect of voltage on both the factors is due to the reasons
that bead width (W) increases almost steadily, but penetration (P) and reinforcement
(R) decrease little as voltage increases from −1 to +1 limit as discussed already. Hence,
they increase steadily as voltage increases.
Page 49
149
WPSF decreases when current increases, but it remains nearly constant from 375 to
400 amperes and it decreases from 2.65 to 2.26 (Fig.5.33) when current changes from
400 to 475 amperes. This could be due to reason that WPSF which is the ratio of W/P,
decreases when current changes from 400 ampere to 475 amperes, because rate of
increase of P is more than that of W with in this range of current. It is also observed in
Fig.5.34 that WPSF increases 1.89 to 2.64 with increase of basicity index from 0.6 to
1.2. It is seen from Fig.5.34 that WRFF decreases from 4.28 to 5.31 with increase of
basicity index from 0.6 to 1.2. These results can be explained with the help of effects
of welding variables on bead width, penetration and reinforcement respectively.
Fig. 5.31 Effect of voltage on WPSF
Page 50
150
Fig. 5.32 Effect of voltage on WRFF
Fig. 5.33 Effect of current on WPSF
Page 51
151
Fig. 5.34 Effect of basicity index on WPSF
Fig. 5.35 Effect of basicity index on WRFF
Page 52
152
Interaction effect
Figs.5.36 and 5.37 show the interaction effects of voltage and current on
WRFF. It is evident from these graphs that WRFF increases for all values of current
when voltage increases from 32 to 38 volts. But the increasing trend of WRFF is more
at higher value of current i.e. at 475 amperes than that at 375 amperes. This is due to
fact that WRFF= W/R, W increases with increase of voltage and nearly remains
constant with change of current whereas R decreases with increasing voltage and
increases with the increase of current. Thus, voltage has a positive effect on WRFF
whereas current has a negative effect on WRFF. This increasing trend of WRFF at
higher current is due to more positive effect of voltage on WRFF. This can also be
explained on the basis of effects of voltage and current on bead width and
reinforcement.
The interaction effect of current and speed on WPSF is shown in Fig.5.38.
Fig.5.39 shows the effect of welding speed on WPSF with change in basicity index.
These can be explained on the basis of effects of welding variables on the bead width
and penetration respectively.
Page 53
153
Fig. 5.36 Interaction effect of voltage and current on WRFF
Fig. 5.37 Response surface due to interaction of voltage and current on WRFF
B: Current
B = 375 amp
B =475 amp
Page 54
154
Fig. 5.38 Response surface due to interaction of current and speed on WPSF
Fig. 5.39 Response Surface due to interaction of welding speed and basicity
index on WPSF
Page 55
155
5.7.5 Effect of process parameters on flux consumption
Direct effects
The developed model for the flux consumption is shown below:
Flux Consmption (F) = 48.11 + 12.69 * A3.24 * B4.04 * C + 5.97 * D 0.66 *
A24.28 * B2 + 0.61* C26.30 * D2 + 0.90 * A * B0.16 * A * C2.02 * A * D+0.77
* B * C + 2.26 * B *D1.70*C*D (5.7)
As shown in Fig. 5.40, flux consumption increases from 34.757 to 60.139 gms
with the increase in open circuit voltage from 32 to 38 volts. This is due to the
positive effect of voltage on bead width. This is also consistent with the findings of
Gunaraj and Murgun (1999).
In Fig. 5.41, the effect of change in welding speed on flux consumption can be
explained on the basis of effect of speed on bead width. At higher welding speed there
is less spread of bead, and hence less consumption of flux compared to that at lower
speed.
Fig. 5.42 shows the effect of basicity index on consumption of flux. Flux
consumption increases with the increase in basicity index. This is due to the fact that
because low basicity index fluxes have high viscosity which enhances the tendency of
heat concentration in the narrow zone and hence high penetration and lower bead
width. This is consistent with the study conducted by Gupta (1988). The Flux
consumption decreases with the increases in current, as shown in Fig.5.43. This can be
Page 56
156
explained on the basis of individual effect of welding current on bead width,
penetration and reinforcement.
Interaction effect
Response surface due to interaction of current and basicity index on flux
consumption is shown in Fig.5.44. These effects can be explained on the basis of
individual effects of current and basicity index on bead width and consumption of flux.
Fig.5.40 Effect of voltage on flux consumption
Page 57
157
Fig.5.41 Effect of welding speed on flux consumption
Fig.5.42 Effect of basicity index on flux consumption
Page 58
158
Fig.5.43 Effect of current on flux consumption
Fig. 5.44 Response surface due to interaction of current and basicity index on flux
consumption