67

ISSN 2466-2232
Online ISSN 2466-2100

Correlation of Flux Ingredients with HAZ Width

in Submerged Arc Welding
Brijpal Singh*,†
*Associate Professor, Mechanical Engineering Department, Maharaja Surajmal Institute of Technology,
New Delhi, 110058, India

†Corresponding author : brijpalsingh101@gmail.com

(Received July 20, 2017 ; Revised November 13, 2017 ; Accepted January 3, 2018)

Abstract
Fluxes play an important role in deciding weld metal properties. The properties of the welds in HAZ area are
severely affected. In this study attempt has been made to evaluate the effect of flux composition on the width of the
HAZ. The study reveals that FeMn and NiO both are having significant effect on HAZ width. The combined effect
of CaF2, FeMn and CaF2, NiO is also significant. This study may help in selection of flux for minimum HAZ
width in weldments.

Key Words : SAW, BI, HAZ, Flux

silicate or sodium silicate. After bonding, the wet mix-

1. Introduction ture is pelletized and baked at a temperature lower than
that for fused fluxes. After this, the pellets are broken
In submerged arc welding, arc is produced between a up, screened to size and packaging is done. These flux-
electrode and the workpiece. Since the arc is submerged es have lower bulk density and hence less flux is melted
under a heavy coating of granular flux, hence it is called for a given amount of weld deposition. The main ad-
submerged arc welding. The welding process may be vantage of using agglomerated flux is that deoxidizers
manual or automatic. This welding process gives high and other alloying elements can be added during the dry
quality joint as double protection is obtained from at- mixing but its main limitation is the absorption of mois-
mospheric gases1).SAW is an automatic process. This ture by the flux3).
welding process is used for carbon steels as well as nickel
based alloys, stainless steels and other non-ferrous metals. 2. Heat Affected Zone
Fluxes are the chemical substances that are used as a
cleaning agent in welding. The SAW fluxes contain lime, Heat affected Zone may be defined as the area of base
silica, manganese oxide, calcium fluoride and other metal which is not melted but its properties are affected
compounds. In SAW the weld pool is protected from by the heat of welding. The properties of the HAZ are
the atmospheric contamination by being submerged un- changed due to change in the microstructure. The mi-
der a blanket of granular fusible flux. In the molten crostructure is changed due to heating and subsequent
state, the flux becomes conductive and provides a cur- cooling. The extent and magnitude of property change
rent path between the electrode and work piece. Fluxes depends on the base metal, welding process, flux com-
can be categorized depending upon the method of man- position and heat input by the welding process4). The
ufacture, the extent to which they can affect the alloy change in HAZ takes place in two stages. In first stage,
content of the weld deposit and the effect on weld de- the microstructural changes like grain growth takes
posit properties2). place and in the second stage precipitation hardening
Agglomerated fluxes: These are also known as bonded and embrittlement of the metal takes place5).
fluxes. In agglomerated fluxes, the raw materials are The literature review shows that many researchers
powdered, dry mixed and bonded with either potassium have studied the effect of welding process parameters

Journal of Welding and Joining, Epub ahead of print

https://doi.org/10.5781/JWJ.2018.36.1.9
68 Brijpal Singh

on HAZ microstructure and dimensions6,7). The main centrations of the additives were varied in the range
aim of the studies was to have a minimum HAZ area by 2-8%. The control parameters (additives) and their lev-
selecting optimum welding parameters. By selecting the els are shown in the coded form in the table 1. The three
flux for minimum HAZ from such studies, mechanical levels of the aforesaid additives are shown in table
properties of the weldments can be improved. (Adler et. 2.The composition of wire and base plate are given in
al., 1975), (Fisher, 1952).Lee et al. (2000)8-10) discussed Table 3. The welding parameters were made constant
the effect of welding parameters on the size of the HAZ for all the welds. These parameters are given in Table 4.
and concluded that in HAZ, the grains are of large size All the components, base constituents and additives
.So, the toughness is reduced in HAZ area. While in a were mixed in a container and Potassium silicate was
narrow HAZ, the temperature gradients are high but due used as a binder for making these fluxes. After prepara-
to finer grain size, the toughness is improved. tion the fluxes were heated in a furnace up to 400℃ for
Gunaraj and Murugan (2002)7) studied the effect of more than six hours to remove any traces of moisture.
welding process parameters and heat input on HAZ area
and other metallurgical characteristics. The study con- Table 1 Design matrix in coded form
cludes thatfor reducing HAZ proper selection of weld- No of
CaF2wt% FeMn wt% NiO wt%
ing process parameters is required. The study reveals Experiment
that HAZ is increased with increasing heat input and 1 +1 -1 -1
wire feed rate while it is reduced with increasing weld- 2 0 +1 0
ing speed. Mathematical models were developed to 3 +1 -1 +1
study the effect of process variables and heat input on 4 -1 -1 -1
HAZ width, grain refinement and other metallurgical 5 0 0 0
aspects
6 0 0 0
Heat affected zone affects the mechanical properties of
7 +1 +1 +1
the weld material as the HAZ microstructure has a
strong influence on the weld joint properties. The ulti- 8 0 0 0

mate tensile strength, h toughness and hydrogen crack- 9 0 -1 0

ing are the main problems which are associated with 10 +1 0 0
HAZ. The flux composition also has a definite effect on 11 0 0 +1
HAZ width, depth, area of penetration, area of re- 12 -1 -1 +1
inforcement and weld dilution in mild steels, so the 13 0 0 0
study and control of HAZ is very much essential. HAZ 14 0 0 0
also increases the probability of fatigue failure at the
15 +1 +1 -1
weakest zone caused by heating and cooling of the weld
16 -1 0 0
zone.
17 0 0 0
18 0 0 -1
3. Experimental Procedure
19 -1 +1 +1
The fluxes were prepared by agglomeration technique. 20 -1 +1 -1
The base constituents CaO, SiO2, and Al2O3 were mixed
in the ratio 7:10:2 based on ternary and binary phase Table 2 Showing three factors and their levels
diagrams. The additives CaF2, FeMn and NiO were se-
Additives Lower Middle High
lected as control parameters. These additives were add- Factors
% level (-1) level (0) 1evel (+1)
ed to the base constituents and their effects on elements
A CaF2 2 5 8
transferred to the welds were investigated. To inves-
B FeMn 2 5 8
tigate the effects systematically, twenty fluxes were de-
signed using response surface methodology. The con- C NiO 2 5 8

Table 3 Showing wire and plate composition.

Composition Carbon % Silicon % Manganese % Sulphur % Phosphorus % Nickel %

Base Plate 0.03 0.07 0.34 0.017 0.022 -
Wire 0.11 0.09 0.45 0.021 0.021 -
The electrode wire used having diameter 3.15 mm and its specification was AWS- A5.17 El-8

68 Journal of Welding and Joining, Epub ahead of print

Correlation of Flux Ingredients with HAZ Width in Submerged Arc Welding 69

Table 4 Welding Parameters

S.No. Voltage Current Travel speed

1 30 volts 475 ampere 20 cm/minute.

Table 5 The measured parameters

Flux CAF2 FeMn NIO HAZ width
Mn ∆Mn Ni
NO (%) (% ) (%) (mm)

1 8 2 2 3.45 0.17 -0.267 0.177

2 5 8 5 3.16 0.37 -0.069 0.702
3 8 2 8 4.38 0.23 -0.209 0.544 (a)

4 2 2 2 2.63 0.17 -0.269 0.374

5 5 5 5 2.19 0.35 -0.089 0.388
6 5 5 5 1.91 0.31 -0.129 0.25
7 8 8 8 2.35 0.38 -0.059 0.477
8 5 5 5 1.77 0.34 -0.099 0.27
9 5 2 5 2.17 0.34 -0.099 0.474
10 8 5 5 3.56 0.42 -0.019 0.744
11 5 5 8 1.28 0.15 -0.289 1.33
12 2 2 8 3.80 0.57 0.131 0.44 (b)
13 5 5 5 1.75 0.38 -0.059 0.32 Fig. 1 (a) Photographs of bead on plate welds, (b) Showing
14 5 5 5 2.23 0.39 -0.049 0.32 width of HAZ

15 8 8 2 5.10 0.24 -0.199 0.054

plate and wire is given in table 3.Welding parameters
16 2 5 5 3.68 0.25 -0.189 0.344
such as voltage, current and travel speed were made
17 5 5 5 2.25 0.50 0.061 0.452 constant table 4. Four beads were laid one over the other
18 5 5 2 2.02 0.29 -0.149 0.366 in order to minimize the effect of dilution. In this study
10% dilution from the base plate has been considered.
19 2 8 8 2.64 0.33 -0.109 0.502
After making beads, the powder was extracted from the
20 2 8 2 6.036 0.33 -0.109 0.288
top bead with the help of a drill machine for chemical
analysis. The measure responses are given in Table 5
Before making weld the fluxes were again heated up to and the polished samples of bead and HAZ width are
100℃. CaCO3 was used in place of CaO because of its shown in Fig. 1(a) and (b).
hygroscopic nature. The transfer of manganese was cal-
culated by a ∆ Delta quantity = Analyzed composition 4. Result and Discussion
- Expected composition. The Table 5 shows that ∆Mn
is negative for most of the welds and it can be taken as As the flux composition has a definite effect on pene-
an indicator of weld oxygen content. The expected tration, element transfer, bead geometry, physical prop-
composition was calculated from the below given rela- erties of flux and chemical characteristics affect the
tion as given in equation 1: HAZ in the weld. So an attempt has been made to study
the effect of flux composition on width of the HAZ.
Expected composition The predictive equations of HAZ width have been de-
   veloped in terms of flux composition. The experimental
=
 results for HAZ width are given in Table 5 and ANOVA
–  results for HAZ width are given in Table 6.
+ (1)

5. Development of model for width of HAZ
Bead on plate welds using submerged arc welding were
made on 18 mm thick plates. The composition of base The second order quadratic model developed for width

Journal of Welding and Joining, Epub ahead of print 69

70 Brijpal Singh

Table 6 ANOVA Table for width of heat affected zone

ANOVA for Response Surface Reduced Quadratic Model

Analysis of variance table [Partial sum of squares - Type III]
Sum of Mean F p-value
Source Squares df Square Value Prob. > F
Model 27.88987 8 3.486 70.05 < 0.0001 significant
A-CaF2 0.00025 1 0.000 0.01 0.9448
B-FeMn 0.81796 1 0.818 16.44 0.0019 significant
C-NiO 2.29441 1 2.294 46.10 < 0.0001 significant
AB 0.864613 1 0.865 17.37 0.0016 significant
BC 8.507813 1 8.508 170.95 < 0.0001 significant
A^2 6.743196 1 6.743 135.49 < 0.0001 significant
B^2 1.026327 1 1.026 20.62 0.0008 significant
C^2 0.449046 1 0.449 9.02 0.0120 significant
Residual 0.547448 11 0.050
Lack of Fit 0.274114 6 0.046 0.84 0.5899 not significant
Pure Error 0.273333 5 0.055
Cor Total 28.43732 19

Std. Dev. 0.224 R-Squared 0.98107

Mean 2.92 Adj R-Squared 0.9667

C.V. % 7.65 Pred R-Squared 0.9203

PRESS 2.19 Adeq Precision 31.91

of the HAZ, using surface methodology based on ex- Normal plot of residuals
perimental results is given in the equation. The pre-
dicted equation is given in terms of actual factors is giv-
99
en in equation 2.
Normal % probability

95
90
HAZ Width = 3.491 − 1.530 * CaF2 + 0.172*FeMn 80
70
+ 0.862 * NiO − 0.0360 * CaF2 * FeMn
50

- 0.145* FeMn * NiO + 0.1739CaF22 30

20
+ 0.0678 * FeMn 2 − 0.0448 * NiO 2 (2) 10
5

1
6. ANOVA Results for width of HAZ
The ANOVA results for HAZ width are given in Table -2.05 -1.18 -0.30 0.58 1.45
6. The model F- Value of 70.05 implies that the model Internally studentized residuals
is significant. There is only a 0.01% chance that a
Fig. 2 Normal plot of residuals for HAZ width
“model F- value” this large could occur due to noise. It
can be seen from the Table 6 that the factors B and C,
Interactions of BC and AC and quadratic terms A2, B2 observed from Fig. 2 that the residuals fall on a straight
and C2 all are having significant effect on HAZ width. line and the Fig. 3 illustrates that the difference between
The multi regression coefficient of the model is the predicted and actual value of the response is
0.9812. This means 98.12% of the variance in HAZ insignificant. The lack of fit is insignificant relative to
width can be explained by the independent variables. the pure error. The “pred. R-Squared” of 0.9203 and “Adj.
The Figs. 2 and 3 depict the normal plot of residuals R-squared” of 0.9667 are in close agreement. This im-
and predicted vs actual value plot for the response. It is plies that the proposed model is adequate and it can be

70 Journal of Welding and Joining, Epub ahead of print

Correlation of Flux Ingredients with HAZ Width in Submerged Arc Welding 71

7
Predicted vs. actual
6.50 6

HAZ width (mm)

5

4
5.80
3

2
5.10
1

0
4.40 0.1 0.2 0.3 0.4 0.5 0.6

Weld Mn %
3.70 Fig. 5 Variation of HAZ width with weld Mn
3.70 4.38 5.06 5.75 6.42
7
Actual
6

HAZ width (mm)

Fig. 3 Predicted vs actual values for HAZ width
5

4
used to determine the effect of various parameters on 3
the responses.
2

1
7. Effect of FeMn additive on width of HAZ
0
-0.35 -0.3 -0.25 -0.2 -0.15 -0.1 -0.05 0
The effect of FeMn content on the width of HAZ is Delta Mn quantity
depicted in the Fig. 4. This shows that after an initial
decrease, the width of HAZ increases. It can be attrib- Fig. 6 Variation of HAZ with weld Oxygen.
uted to the change of oxygen content in the weld, which
consequently affects the surface tension of the liquid in Fig. 5. The variation of HAZ width with the weld
flux and the bead geometry. A small amount weld oxy- oxygen content has been given in Fig. 6. This shows
gen act as a surface active agent which changes the di- that HAZ width is small when the oxygen content is ei-
rection of surface tension gradient from negative to pos- ther very low or very high.
itive, thus reversing Marangoni convection and the flow
of liquid takes place in downward making deeper pene- 8. Effect of NiO additive on HAZ width
tration11). So the heat may be confined to a small area. If
the oxygen may be large the liquid flow will take place The effect of NiO additive on HAZ width is given in
in the out ward direction causing to spread the heat over the Fig. 7. It can be observed from the Fig. 7 that the
a large area. Weld Mn content does not show any sig- HAZ width decreases with increase in NiO additive in
nificant effect on width of heat affected zone as shown the flux. Two possible reasons may exist for it, one may
be the weld oxygen content and the other may be the in-
3.0
creased penetration with increasing NiO. The HAZ
2.2
2.8
HAZ width (mm)

2.0
HAZ width (mm)

2.6
1.8

2.4 1.6

1.4
2.2
1.2

2.0 1.0
2 3 4 5 6 7 8 2 3 4 5 6 7 8
FeMn (%) NiO (%)

Fig. 4 Effect of FeMn on HAZwidth Fig. 7 Effect of NiO on HAZ width

Journal of Welding and Joining, Epub ahead of print 71

72 Brijpal Singh

7 2) After decreasing to a minimum, the HAZ width in-

6 creases with increasing FeMn additive.
HAZ width (mm)

5 3) The HAZ width is reduced with increasing NiO

4 content in the flux but weld Mn content does not show
3 any effect of HAZ width.However, it is reduced with in-
2 creasing Ni addition.
1 4) The width of the HAZ has a reducing trend with in-
0 creasing Ni content.
0 0.2 0.4 0.6 0.8 1 1.2 1.4 5) Quadratic modelling can be used to predict the ef-
Weld nickel % fect of flux on HAZ width.
Fig. 8 Correlation of Ni with HAZ width
ORCID: Brijpal Singh: http://orcid.org/0000-0002-5418-033X
Table 7 Predicted and Experimental values
Weld
References
Experimental
FeMn NiO HAZ Error
S.No. CaF2% value of 1. Hould Croft PT., Submerged arc welding, Second Ed.
% % width %
HAZ width London, Abington publishing, Cambridge, (1998)
predicted
1 2 2 5 3.34 3.50 4.7 2. Kumar V., Devolopment and characterization of fluxes
for submerged arc welding, Ph.D Theis, Punjabi University
2 5 2 8 6.74 6.3 6.5
Punjab, India, (2011)
3. Maheshwari S., Study of Element Transfer behaviour
width may depend upon the quantity of Nickel and oxy- during submerged arc welding, Ph.D. Thesis, Indian
gen transferred to the weld. NiO is not very stable oxide Institute of Technology Delhi, India, (1998)
4. Weman Klas, Welding process hand book, First Ed.
and at high temperature it decomposes into Ni and
New York, CRC press Boston, (2003)
oxygen. Nickel transfer reduces the HAZ width as shown
5. Lancaster JF, The metallurgy of welding Brazing and
in the Fig. 8. The HAZ width initially increases with in- Soldering, 6th Ed.London, George Allen Unvin, (1999)
creasing weld oxygen content but it is reduced slightly 6. Linert G E. Welding Metallurgy, Fundamentals, 4th Ed.
with further increase of oxygen content as shown in New York, American Welding Society, (1995)
Fig. 6. With addition of NiO both oxygen and Ni con- 7. Gunaraj V and Murugan N., Rediction of heat affected
tents increases and the basicity index also increases zone characteristics in submerged arc welding of structural
simultaneously. The correlation of Ni proportion trans- steel pipes. Welding Research, 81(2002), 94-98
fer with the width of the HAZ shows a decrease in the 8. Adler YP, Markov EV and Granovsky YV., The design
HAZ width with increasing Nickel proportion transfer of experiment to find optimal condition. USSR, MIR
Publisher, Moscow, (1975)
to the weld.
9. Fisher RA., Statistical methods for research workers,
12th Edition, Scotland, Oliver and Boyd Dinburg, (1952)
9. Validation of the model 10. Lee CS, Chandel RS and Seow HP., Effect of welding
parameters on the size of heat affected zone of submer-
The validation of the model was done by selecting two ged arc welding, Materials and manufacturing process,
fluxes within the given range .The results are tabulated 15(5) (2000), 649-666
in Table 7 given below and the error lies within the per- https://doi.org/10.1080/10426910008913011
missible limit. It shows that the model can be used for 11. Gunaraj V and Murugan N., Rediction of heat affected
predicting the values of HAZ width for the given flux zone characteristics in submerged arc welding of struc-
composition. tural steel pipes. Welding Research, 81(2002), 94-98
12. Heiple CR and Roper JR., Mechanism for minor ele-
ment effect on GTA fusion zone geometry 1982, Welding
10. Conclusions
research supplement, 97-102
1) FeMn and NiO both additives as individuals are
having synergistic effect on HAZ width but as a mix-
ture, these are having anti synergistic effect on HAZ
width.

72 Journal of Welding and Joining, Epub ahead of print