A Full Factorial Analysis Of Microhardness

Variation In Bead Welds Deposited By The Process

Cold Wire Gas Metal Arc Welding (CW-GMAW)
R. A. Ribeiro, P. D’Angelo Assunção, E. M. Braga

Abstract—The microhardness in weld beads is a function of the
microstructure obtained in the welding process, and this by its time
is dependent of the input variables established at the outset of the
process. In this study the influence of angle between the plate and
the cold wire, the position in which the cold wire is introduced
and the rate in which this introduction is made are assessed as
input parameters in CW-GMAW process. This paper looks to show
that ordinary changes in the frame of CW-GMAW can improve
microhardness, which is expected to vary as the input parameters
change. To properly correlate the changes in the input parameters
to consequent changes in microhardness of the weld bead, a full
factorial design was employed. In fact, changes in the operational
parameters improved the overall microhardness of the weld bead,
which in turns can be an indication of improvement in the resistance
to abrasive wear, constituting a cheap way to augment the abrasion
wear resistance of welds used for cladding.
Keywords—Abrasion, CW-GMAW, Full Factorial Design,
Microhardness.
I. I NTRODUCTION
structure could be useful in reducing the wear rate to abrasion
when this is the unique mechanism involved in wear.
According to [3] hardfacing materials for wear resistance
applications can have high hardness in the as welded condition
but also achieved through heat treatment or mechanical
work, this paper will show that in a novel Gas Metal Arc
Welding (GMAW) process called CW-GMAW (Cold Wire
Gas Metal Arc Welding) that consists of an conventional
GMAW apparatus with an extra non-energized wire feeder,
that initially was developed to increase the deposition rate,
the hardness of the deposited metal increases as function of
the input parameters that will be: the injection angle of the
cold wire feeding, the entrance position of the cold wire along
the energized wire and the cold wire feeding rate ratio, that
is defined as percentage of the nominal feeding of energized
or hot wire; the hardness can be increased with variation
in the input parameters which associated with the deposition
of hardfacing alloy wires could improve the hardness of the
deposited metals. According [4] there is a correlation between

T HE cost of the damage caused by wear to the Canadian

industry is estimated as 2.5 billion a year according [1]
and according the same author in the oil sands of northern
abrasive wear and hardness. Generally, a better performance
against 3 body abrasion will be obtained by increasing the
material hardness. Thus, with the measures proposed in this
Alberta the down time cost in any of the production units due study, a low quantity of hardfacing material could saved and
to wear is estimated to be 40 million per year. These data hardness of the cladding maintained by the input parameters
show the importance of wear mitigation to the industry as all. changes.
The same frame of wear caused losses that can be found in
mining and pulp industries. The down times caused by wear
II. E XPERIMENTAL PROCEDURE
to the production of these industries impels the development
of new techniques or materials resistant to wear. A. Materials
This work aims to show that simple changes in the input
In this paper to manufacture the weld bead was used an
parameters of a new welding procedure can improve the
electronic welding source Digiplus A7 of 400 A, the electrode
hardness of claddings against wear. The wear form considered
feeder is a STA-20D IMC, and was used an additional wire
in this work is pure abrasion where an improvement in the
feeder for the cold wire ESAB MEF 30, the welding gun used
hardness can be directly related to an improvement in abrasion
as a GMAW TBi 511 automated, this gun is water cooled
resistance. According to [2] a strong diminution of the abrasive
with maximum current 400 A. The shielding gas used was a
wear rate happens when the hardness of the material exceeds
mixture of 25% CO2 and 75% Argon (Ar) with a volumetric
the hardness of the abrasive. The method described in this
flow of 15 l/min.The base metal used in this work was a
paper associated with a deposit of hardfacing alloys over a
normalized structural steel ASTM 131 grade A, normally used
in ship construction, with a low level of carbon, whose nominal
R. A. Ribeiro is a masters Graduate Student at the Faculty of Mechanical
Engineering (FEM), Federal University of Pará, Belém, PA, CEP 6607-110, chemical composition is showed in Table II. And the wires
Brazil, e-mail: ra− ribeiro@oi.com.br.
E.M. Braga is a Associate Professor at the Faculty of Mechanical TABLE I
Engineering (FEM), Federal University of Pará, Belém, PA, CEP 66075-110, N OMINAL CHEMICAL COMPOSITION OF THE BASE METAL IN % WT.
Brazil, e-mail: edbraga@ufpa.br
P. D’Angelo Assunção is a Ph.D. Student, PRODERNA, at the Faculty of C Si Mn P Cr V Fe
Chemical Engineering (FEQ), Federal University of Pará, Belém, PA, CEP 0.21 0.5 2.3 × C 0.035 0.02 0.03 Balance
66075-110, Brazil, e-mail: dangeloassuncao@bol.com.br.
used were of the AWS ER70S-6 with different diameters the calculated, the average was used as response of the full
electrode with 1,2 mm and the non-electrode with 1,0 mm. factorial experiment.
The nominal chemical composition of the wires can be seen After this, the commercial software MINITAB was used to
in Table I. perform the analyses of the data obtained in the experiments
described in this paper.
TABLE II
N OMINAL C HEMICAL COMPOSITION OF THE ER70S-6 WIRE IN % WT.
III. R ESULTS
C Si Mn P Cr V Fe
0.21 0.5 2.3 × C 0.035 0.02 0.03 Balance Table V presents the five values of microhardness for each
coupon, as said the average value was used as response, the
data show low variability as indicated in the low values of
standard deviations.
B. Full Factorial Design
TABLE V
To assess to hardness variation as function of the angle, V ICKERS H ARDNESS VALUES , AVERAGES AND S TANDARD D EVIATIONS .
position and rate of cold wire feeding, a 2k full factorial
experiment will be run, with k factors, corresponding to the Standard
Run Order I II III IV V Average
Deviation
three aforementioned input parameters, with no replicates. 1 142 151 141 148 150 146 5
Table III presents the factors used in this experiment as well 2 149 145 145 157 148 149 5
as the levels considered in it. 3 150 142 143 143 151 146 4
4 150 149 149 146 141 147 4
TABLE III 5 158 155 165 160 156 159 4
FACTOR L EVELS FOR THE E XPERIMENT. 6 145 147 143 142 158 147 7
7 156 163 145 163 158 157 8
Factors Low (-1) High (+1) 8 142 141 130 143 123 136 9
Angle (◦ ) 45 55
Position (mm) 0.0 1.2
R (%) 20 40 The hardness responses were introduced in MINITAB work
sheet to perform the analysis of the factorial experiments. The
first model, which considered angle, position and cold wire
Table IV presents in coded values the experimental
feed rate ratio, was saturated and therefore MINITAB did not
matrix used to conduct the experiment, and all the possible
show the results of F-Test and P-values. This model showed
combinations between the parameters levels.
that the position at which the cold wire is introduced has little
TABLE IV effect, -2.75, when compared with the effects produced by
E XPERIMENT M ATRIX . the other two input variables, -6.25 and 7.25 for the angle and
cold wire feed rate ratio, respectively, Table VI. Thus a second
Run Order Angle (◦ ) Position (mm) R (%) model was built considering only the effect of angle and cold
1 -1 -1 -1
2 +1 -1 -1 wire feed rate ratio, which allowed MINITAB to calculate the
3 -1 +1 -1 F and P values. Table VI shows the results of the preliminary
4 +1 +1 -1 model and Table VII shows the results of the second improved
5 -1 -1 +1
6 +1 -1 +1 model.
7 -1 +1 +1
8 +1 +1 +1 TABLE VI
M AIN E FFECTS AND I NTERACTION FOR THE P RELIMINARY M ODEL .

Term and Interactions Effect

Angle (◦ ) [A] -6.25
C. Welding Procedure Position (mm) [B] -2.75
Cold Wire Feed Rate Ratio (%) [C] 7.25
In this work the welding was flat position and an automatic [A×B] -1.75
horizontal displacement was used. The welding was made [A×C] -4.75
in the direction of wire feeding, what guaranteed the wire [B×C] 1.75
[A×B×C] 2.75
injection before the electric arc and the welding pool. The
angle of injection was varied according the values showed in
Table III.
TABLE VII
In this work, the experiments were done with a constant E STIMATED E FFECTS AND C OEFFICIENTS FOR AVERAGE
voltage of 37 V for the electrode feeding rate of 12 m/min and M ICROHARDNESS (HV).
the cold wire feeding rate ratio varied as showed in Table III.
To measure the microhardness the weld beads were cut from Term and
Effect Coef SE Coef Fisher T P-Value
Interactions
the mid length using a hacksaw and then the coupons were Constant - 148.875 1.152 129.18 0.000
submitted to the microhardness tests were made according Angle (◦ ) [A] -6.25 -3.125 1.152 -2.71 0.053
ASTM E384 [5], a load of 2 kgf was used and five indentations R (%) [C] 7.25 3.625 1.152 3.15 0.035
[A×C] 7.25 -2.375 1.152 -2.06 0.108
were made and the average and the standard deviation were
 Table VIII shows the goodness of fit parameters of the Interaction Plot for Average Microhardness (HV)
Data Means
model, showing a relatively good accuracy of the model. Angle (º)
157,5 45
55
TABLE VIII 155,0
G OODNESS OF F IT S TATISTICS PARAMETERS .
152,5

Mean
St. Deviation 3.2596
PRESS 170 150,0

R2 84.31%
R2 - pred
147,5
37.24%
R2 -adj 72.54% 145,0

20 40
Cold Wire Fedd Rate Ratio (R)
Table IX presents the ANOVA table to the experiment,
indicating the variability due to each factors as well as the
Fig. 2. Interaction Plot.
variability due to errors.
TABLE IX
A NALYSIS OF VARIANCE FOR AVERAGE M ICROHARDNESS (HV). interactions can be separately estimated, independently which
ease the analysis of the results. The same table also asserts
SOURCE DF Seq SS Adj SS Adj MS F P that independent main effects are more significant than the
Main Effects 2 183.25 183.25 9162 8.62 0.035
Angle () 1 78.12 78.12 78.12 7.35 0.053
combined interaction, because they have lower P-values than
[C] (%) 1 105.12 105.12 105.12 9.89 0.035 the interaction effect.
2-Way Interactions 1 45.12 45.12 45.12 4.25 0.108 Table VIII shows goodness to fit statistics, The R squared,
[A×C] 1 45.12 45.12 45.12 4.25 0.108
Residual Error 4 45.2 42.5 10.62
according to [6] measures the proportion of total variability
Total 7 270.87 explained by the model. The result presented in Table VIII
shows that 84.31% of the experiment variability can be
Fig. 1 and 2 show respectively the main effects, the explained by this model, a problem that can posed to the use
interaction effects of the experiment and asserts how the this statistic is that it increases as measure as more factors are
output parameter Vickers microhardness is affected by the put in the model, even if these factors play not a significant
input parameters and their mutual combination indicating that role in the model, to ameliorate this problem and give better
to low values of angle of injection and high values of cold wire understanding, the table also presents the adjusted R squared
feed rate ratio the microhardness would be improved merely that depends of the sum of squares due to error and the total
by these changes in the operational parameters. sum of squares, this statistic is said adequate to size of the
model, because it decreases if non-significant factor are added
Main Effects Plot for Average Microhardness (HV)
to the model, the of 72.54% shows a well pondered model
Data Means with a regular accuracy.
Angle (º) Cold Wire Feed Rate Ratio
153 Other statistic presented in table 9 is predicted R squared, it
152 gives a measure of how new data are explained by the model,
151 the value of 37.24% is low, and probably due to the reduced
150 number of runs, just eight, used in this study, but it wants
Mean

149 not to, at this level, predict microhardness in weldments, thus,

148 this value represents little importance to the objectives of this
147 study. The PRESS presented in this table is an acronym for
146 prediction error sum of squares, a model with a small PRESS
145 generally is a good predictor, the value of 170 is hardly small
45 55 20 40
and is related to the motive presented before, relative the
predictive feature of this model.
Fig. 1. Main Effects Plot.
Table IX shows the results of the ANOVA table, to this
model of microhardness it shows that the significant values are
These figures tend to show that the microhardness values
the main effects, the interaction has a certain non-significance
could be better improved as function of the operational
when compared with the main effects, this can be saw by
parameters, as said before, to different values of the input
P-values, the interaction is greater than 0.05, the Fisher test
variables, this can be the subject of a future study.
also show this information, the values of two-way interaction
are lower than main effects, however this two-way interaction
IV. D ISCUSSIONS model was maintained because its values were higher than
The results stated that results the microhardness could those of the other interactions in the preliminary model, if
improved by simple changes in the operational parameters. a regressive study were to be developed this model would
Table VII shows the effects of the consolidated model SE depend thoroughly of the main effects, but was not the case
Coef. column shows equal values to all factors being the in this study; it was performed, as already said, to demonstrate
design orthogonal, this states that the effects of factors and that there is an influence of the operational parameters on the
values of microhardness obtained in the process CW-GMAW. ACKNOWLEDGMENT
Fig. 1 and 2 show the main effects and two-way interaction This work was supported by the Brazilian Innovation
between the factors. Fig. 1 shows clearly that the higher values Agency (FINEP), the National Counsel of Technological and
of microhardness are achieved for the lower value of angle Scientific Development (CNPq) and the Brazilian Ministry
and higher values of cold wire feed rate ratio, indicating of Science and Technology through agreement number
that the injection of cold wire at these values alters the 01.11.0024.00. The authors would like to provide special
microstructure resulting of the process, probably producing thanks to the team from the Laboratory of Materials
harder microconstituents such as acicular ferrite, due to the Characterization (LCAM) of the Federal University of Pará
metallurgical variations caused by the introduction of this cold (UFPA), where all the experimental tests were performed.
wire that could act as heat sink diminishing the weld pool
solidification gradients generating harder microstructures. R EFERENCES
Fig. 2 shows that to lower values of cold wire feeding the
[1] P. F. Mendez, N. Barnes, K. Bell, S. D. Borle, S. S. Gajapathi, S. D.
two values of angles tend to produce the same mean value of Guest, H. Izadi, A. K. Gol, and G. Wood, “Welding processes for wear
microhardness, indicating that in a future study the opposite resistant overlays,” Journal of Manufacturing Processes, vol. 16, no. 1,
direction should be pursued in order to achieve better values of pp. 4–25, 2014.
[2] J. J. Coronado, H. F. Caicedo, and A. L. Gómez, “The effects of welding
microhardness and possibly an increase of abrasion resistance. processes on abrasive wear resistance for hardfacing deposits,” Tribology
This increase in hardness would imply in an reduction of International, vol. 42, no. 5, pp. 745–749, 2009.
[3] A. Gualco, H. G. Svoboda, E. S. Surian, and L. A. d. Vedia, “Effect of
costs, once that the same hardness can be achieved using a welding procedure on wear behaviour of a modified martensitic tool steel
wires less noble and cheaper. hardfacing deposit,” Materials & Design, vol. 31, no. 9, pp. 4165–4173,
2010.
[4] M. Kirchgaßner, E. Badisch, and F. Franek, “Behaviour of iron-based
hardfacing alloys under abrasion and impact,” Wear, vol. 265, no. 5, pp.
V. C ONCLUSION 772–779, 2008.
[5] ASTM.E384/2011e1, “Standard Test Method for Knoop and Vickers
1) The model used in this paper showed the microhardness Hardness of Materials.” West Conshohocken, PA: ASTM International,
in the process CW-GMAW is function of the operational 2011, doi:10.1520/E0384-11E0. http://www.astm.org.
[6] D. C. Montgomery, “Design and analysis of experiments,” pp. 222–227,
parameters used in the manufacturing process. 2009.
2) The input parameters investigated in this paper were:
angle of injection, the position of injection of cold wire
on the energized wire and could wire feed rate ratio,
the preliminary model run in MINITAB showed that
position has little effect on the value of microhardness.
3) The second model was run in MINITAB were only
the main factors angle, cold wire feed rate ratio and
the two-way interaction between them, the same model
showed that the interaction has little effect in possible
predictive model.
4) It was showed that to values of angle and high values
of cold wire feed rate ratio there is a tendency to
an increasing in microhardness, showing that these
parameters affect the output variable.
5) The two-way interaction plot showed that the interaction
between the input variables is negligible, indicating
contrarily that a model should be based only on the two
main effects showed on 4.
6) The hardness possibly was affected by the injection
of cold wire that acts as heat sink, diminishing the
solidification rate of the weld improving the growing
of harder microconstituents.
7) This work implies the possibility of use a cheaper alloy
in cladding against abrasive wear, due to the increase
in hardness occasioned by certain, already discussed, in
the operational parameters.
8) This work has as limitations the number of runs used,
just eight, the fact that other factors that could interfere
in the microhardness were not studied such as the
shielding gas constituents proportion, its flow rate, the
plate thickness, the values of current and voltage as well
the energized wire feed rate.