You are on page 1of 14

Sheet Metal Welding Conference XII Livonia, MI May 9-12, 2006

Gas Metal Arc Welding of Advanced High-Strength Steel –


Developments for Optimized Welding Control and Weld Quality
Vaidyanath Rajan and Dennis Hartman
The Lincoln Electric Company
Armando Joaquin and Adrian N.A. Elliott
Ford Motor Company
Chonghua (Cindy) Jiang
AET Integration
Chris Karas
Metro Technologies

Abstract

Gas metal arc welding (GMAW) is used in the automotive industry, for joining heavier gauge
mild and high-strength low-alloy (HSLA) steels, where it is recognized for its versatility and
speed. The only constraints typically encountered relate to fatigue performance of the joint as a
result of poor design or manufacturing fit-up. Advanced high-strength steels (AHSS), however,
do not offer the same ease of welding, and process control is significantly more critical. The
process window represents the range of acceptable process parameters, primarily heat inputs,
to achieve an acceptable weld, which is a measure of the robustness of the process. AHSS
differ from mild steels in chemical composition and thermal processing, resulting in a different
microstructure, designed with a richer chemistry to have higher strength at equivalent thickness
compared to mild steels. As a result, the sensitivity to heat input is greater and AHSS has a
narrower process window in which acceptable welds can be achieved. Optimization of the
welding process, through the understanding of the influence of the weld process parameters,
has allowed the weld window to be enlarged to encompass AHSS. Empirical data generated so
far has shown an excellent correlation between the weld geometric profile and one of the most
critical mechanical properties, fatigue strength.

Introduction

The current trend in the automotive industry toward a greater use of AHSS for both strength and
weight reasons, as a result of FMVSS safety and CAFE requirements, has necessitated in-
depth studies in to the joining of these steels. GMAW is commonly used for joining heavier
gauge mild and HSLA steels, where it is recognized for its versatility and speed. The only
constraints typically encountered relate to fatigue performance of the joint as a result of poor
design or manufacturing fit-up. AHSS (DP, TRIP, CP, and martensitic), however, do not offer
the same ease of welding and process control is significantly more critical. The process window
represents the range of acceptable parameters, directly influencing heat input [controlled
primarily by wire feed speed (WFS), head travel speed (TS), and contact tip to work distance
(CTWD)], and other parameters such as torch angle, push angle, and wire placement relative to
the joint, to achieve an acceptable weld (defined according to minimum, and sometimes
maximum, leg, throat, and penetration dimensions, with burnthrough being deemed
unacceptable). The process parameter range over which an acceptable weld is obtained, is a
measure of the robustness of the process. A wide range, typically found with mild and HSLA
steels, indicates that variations in the process can occur while maintaining weld quality;

1
Sheet Metal Welding Conference XII Livonia, MI May 9-12, 2006

whereas, a narrow range indicates that minor variations in process conditions can result in
unacceptable quality.

AHSS differ from mild steels in chemical composition and thermal processing, resulting in a
different microstructure designed with a richer chemistry to have higher strength at equivalent
thickness compared to mild steels. As a result, the sensitivity to heat input is greater, which
leads to a more significant heat-affected zone (HAZ) and, therefore, means that AHSS has a
narrower process window in which acceptable welds can be achieved. DP steel was first used
in flash welding applications,(1) where it was noted that the HAZ of the welded joint softened
below the hardness of the base material. When the joint was loaded, necking occurred in the
softer HAZ and the joint failed below the base material strength. By using forced cooling,
Ghosh et al.(2) were almost able to eliminate HAZ softening while increasing the joint strength
above that of the base material; however, this was done at the expense of weld hardness and is
not a practical alternative in the automotive industry. Optimization of the welding process,
through the understanding of the influence of the weld process parameters, will allow the weld
window to be enlarged to encompass AHSS.

The purpose of this study was to understand the influence of the GMAW process parameters
over their effective working range, on the characteristics (dimensional and strength) of a mild
steel to DP 600 welded joint and, subsequently, to determine the optimum process settings to
support a future pickup frame. Based on screening experiments, ranges for all of the significant
process parameters (WFS, robot/torch TS, CTWD, torch angle, push/drag angle, and wire
placement relative to the joint) were established. A response surface (RSM) design of
experiment (DOE) was then conducted using lapped metal plates. Post-weld analysis consisted
of static and dynamic mechanical testing (lap-shear tensile, fatigue life, and micro-hardness)
and dimensional assessment on coupons cut from the welded plates. In addition, the fracture
mode of the joints was carefully examined relative to the fatigue life. The microstructure of the
welds was also analyzed.

GMAW Experimental Procedure

Materials

The material combination chosen for the first DOE was DP 600 hot rolled/uncoated (3.4-mm
thickness) from U.S. Steel Corporation as the lower substrate, with a low carbon (mild) steel
(3.8-mm thickness), meeting the Ford Motor Company ESA-M1A33-C specification, as the
upper. The specified minimum mechanical properties for the DP 600 were 300 MPa (43,500
psi) yield strength and 580-MPa (84,100-psi) ultimate tensile strength and for the mild steel 221-
MPa (32,000-psi) yield strength and 324-MPa (47,000-psi) ultimate tensile strength. The DP
600 material specification sheet, which includes the chemical analysis, can be found in
Appendix A. The materials were prepared as plates measuring 300 × 150 mm (12 × 6 in.), and
were fixtured for welding with a 25-mm (1-in.) overlap, and no gap between. This combination
represented a bracket to rail combination proposed for the truck frame. The plates were
oriented so that the weld was laid transverse to the rolling direction of the steel. Subsequently,
additional welding experiments - DOE 2 with 3.4-mm DP 600/4.7-mm mild steel and DOE 3 with
3.4-mm DP 600/2.7-mm mild steel combinations - were conducted to extend the findings from
the first DOE to these other thickness combinations.

2
Sheet Metal Welding Conference XII Livonia, MI May 9-12, 2006

Welding Equipment and Consumables

The welding equipment used in this experiment was a Lincoln Electric Power Wave 455M power
source with a Tregaskiss water-cooled gun with 22-degree torch (push only), Lincoln
PowerFeed 10 wire-feeder, and integrated with a Fanuc LR mate 200iL/RJ3 robot and
controller. Although the Power Wave offers the ability to adjust the waveform via software, and
so allows manipulation of the peak current, background current, the relative timing of each of
these, and the pulse frequency. This experiment was conducted with constant voltage (CV). It
is believed that CV would allow a clearer definition of the process window, which can be
expanded using a pulse waveform when in production.

The filler wire used was 0.045-in. SuperArc L-56 which complies with AWS specification A5.18
ER70S-6. The shielding gas was a blend of 90% argon and 10% CO2.

Design of Experiment

Based on screening experiments, assessing visual weld quality and size, the ranges for WFS,
robot/torch TS, CTWD (shown in Figure 1), torch angle, push/drag angle, and wire placement
relative to the joint were selected (Figure 2). Although the extremes of some of the process
parameter ranges could be extended, the combination effect with the other factors did not allow
an acceptable weld. Voltage and current are integrated with WFS in the power system and so
interdependent; therefore, these factors are controlled solely as WFS. Combinations of WFS
and TS at the extremes of their range may have resulted in an unacceptable weld size, and
since weld size is not fully independent of WFS or TS, the square root of WFS/TS ratio (which is
directly proportional to the weld size), with specified constraints, was used in place of TS.

Nozzle

Contact Tip

Electrode
Stickout Contact Tip to
Nozzle to Work Distance
Work Distance
Arc Length

Figure 1. GMAW Electrode

3
Sheet Metal Welding Conference XII Livonia, MI May 9-12, 2006

Factors:
• Wire Feed Speed (WFS) @ 159-201 mm/s (375-475”/min)
• Travel Speed (TS) @ 13.5-24.1 mm/s (32-57”/min)
– (Actual variable SQRT WFS/TS (2.89-3.45)
• Voltage (varied with WFS)
– 23.8V @ 159; 24.5V @ 182; 25.0V @ 201 mm/s WFS
• Contact Tip to Work Distance (CTWD) @ 12.5-19.0 mm
(0.50-0.75”)
• Torch Angle (25-55º from vertical)
• Push/Drag Angle (5-25º from vertical)
• Wire Placement (0-2 wire diameters from joint)
Response:
• Weld Profile from Laser Scan (Leg Size/Bead
Convexity/Toe Angle)
• Weld Dimensions from Section (Toe Angle & Penetration)
• Load to Failure (Tensile Lap-Shear)
• Fatigue Life (Tensile Shear - 3 Loads, 3 Replicates, R=0.1)
• Micro-Hardness Traverse/Micro-Structure

Figure 2. Experiment Factors and Responses

The DOE was designed as a modified cubic D-Optimal model, involving 58 experimental runs.
To provide adequate samples for test, each run was duplicated. The DOE was directly
downloaded to the robotic welding system, so that the complete experiment could be conducted
without manual intervention and repeated set-up of each of the process conditions.

The responses measured from the experiment (also shown in Figure 2) were the weld
dimensions (horizontal and vertical leg, bead convexity, toe angle, and depth of penetration),
micro-hardness, lap-shear tensile load to failure and tensile fatigue life. Laser scans using
WISC equipment provided an initial indication of the weld profile; however, subsequent
transverse sections added more precise dimensions for the weld and also the depth of
penetration, both horizontal and vertical.

Specimen Preparation and Test Methods

The test specimens were water-jet cut from the lap-welded plates to a width of exactly 25.4 mm
(1 in.). A total of nine specimens were cut from the central area of each plate, giving a total of
18 available for test from each experimental run, labeled according to position and designated
test. Three specimens were used for tensile testing, one for micro-hardness and microstructure
analysis, and nine specimens for fatigue life. The coupons for each multiple sample test were
selected from a range of different positions, consistent for each run, in order to average any
process or material inconsistencies that existed across the plate. The test for fatigue life, on the
nine specimens from each experimental run, was conducted at three load levels with three test
replicates at each. The maximum fatigue load levels [8.9, 11.1, and 15.6 kN (2000, 2500, and
3500 lbf)] equated to approximately 25, 30, and 40% of the static tensile, load to failure, values
obtained, the R value (ratio of minimum load to maximum load) was equal to 0.1. The test
frequency was 10 Hz. The Vickers hardness test was conducted using a load level of 500 g.

4
Sheet Metal Welding Conference XII Livonia, MI May 9-12, 2006

Statistical Correlation

The test results, both dimensional and mechanical, were analyzed using Design Expert®*
statistical software to understand the variance of the responses for each factor, and
subsequently the significance of each process parameter. The aim was to develop three
correlation models; the first model correlating fatigue life to weld dimensions, the second, weld
dimensions to welding variables, and the third, fatigue life to welding variables.

Results and Discussion

Dimensional Data

Dimensional data, determined by laser scan, caliper, and scale from section photographs, was
averaged for the sample from each of the two plates, collated and compared for the different
measurement methods. The correlation between the laser scan and section measurement for
one particular feature, toe angle, is shown in Figure 3.

Toe Angle - Section Measurement vs. Laser Scan

170.0

160.0

150.0
Toe Angle - Section (º)

140.0

130.0

120.0

110.0

100.0
100.0 110.0 120.0 130.0 140.0 150.0 160.0 170.0
Toe Angle - Laser (º)

Figure 3. Toe Angle Correlation between Laser Scan and Section Measurement

Tensile Lap-Shear Data

The tensile lap-shear data was collated and the average load to failure, and corresponding
extension, of the three coupons for each experimental run determined. The average tensile
load for all samples was 36.2 kN (equivalent to the UTS of this mild steel: 360 MPa or 52,000
psi) with a range of 3.3 kN for the run averages.

Ninety seven percent of the 174 coupons tested broke in the base material of the mild steel; the
remainder broke between the weld metal and the HAZ on the mild steel side. The samples that
broke at the HAZ/weld interface exhibited under-filled vertical legs (Figure 4), and the load to
failure was lower than the experiment mean at an average of 34.0 kN.

* ®
Registered Stat-Ease Inc.

5
Sheet Metal Welding Conference XII Livonia, MI May 9-12, 2006

Toe Angle: 129.3 Degree

Tensile broke between


MS Weld & HAZ

Run 36

Figure 4. Small Vertical Leg at Mild Steel

The tensile failure at the weld to HAZ interface is most likely caused by the extreme welding
parameters. Samples were from the only runs welded with at least four of the five following
factors:
• All four runs produced the smallest weld size
• All were welded at the highest TS (22.9 or 24.1 mm/s)
• All were welded with -2 wire placement (wire center 2 diameters from upper plate)
• Three of four were welded with 25 degrees from vertical torch angle (directs arc
force/heat to bottom plate)
• Three of four were welded with 5-degree push angle (directs more of arc force/heat into
bottom plate).

A combination of small weld with high TS results in lower heat input and less penetration.
Locating the arc out two wire diameters, with a 25-degree torch angle and 5-degree push angle,
directs less heat into upper plate.

Statistical analysis for the tensile data did not show any significance for any of the process
factors.

Microhardness and Microstructure

Microhardness traverses were generated on one sample from each experimental run. The
hardness was determined at a depth of 0.5 mm below the material top surface, at 0.4 mm
intervals across the HAZ for both materials over a distance of approximately 2.5 mm, before
continuing in the base material at 2-mm intervals, and at regular intervals of approximately 1
mm diagonally across the weld (Figure 5). The mild steel exhibited an average hardness of 130
HV with negligible softening, the DP 600, however, exhibited an average hardness of 210 HV
with noticeable softening (average 180 HV) in the HAZ. The range of hardness values for the
DP 600 HAZ was small (165-188 HV), with no apparent significance based on welding
parameters. The hardness of the weld was consistently around 220 HV with the exception of
the lowest heat input runs in which values of up to 250 HV were measured (Figure 6).

6
Sheet Metal Welding Conference XII Livonia, MI May 9-12, 2006

Toe Angle: 120.4 Degree

Figure 5. Hardness Indentations Across Joint

Microhardness Traverse of Samples


with Three Different Heat Inputs
300
Microhardness (HV500g)

250
DP600

200
Mild Steel
150
Weld
100
0 2 4 6 8 10 12 14 16 18 20 22 24
Distance (mm)

A-14 Lowest Heat Input A-50 Best High Cycle Fatigue A-19 Highest Heat Input

Figure 6. Microhardness Profile for Different Runs

Microstructure analysis of the polished samples at 200× magnification showed a consistent


grain structure for the base metal, HAZ and weld metal, respectively, in all samples with the
exception of Run 19 where the heat input was excessively higher. In this sample, the grain
structure of the HAZ appeared coarser.

Fatigue Life and Fracture Location

All fatigue failures occurred between the weld and HAZ, with the location denoted as A (failure
at the DP 600 side of the weld – crack initiates at the toe and propagates through the DP 600
HAZ) or B (failure at the mild steel side of the weld – crack initiates near the root of the weld and
propagates through the mild steel HAZ or through the weld). There was an equal distribution of
each type of failure.

The number of cycles to failure averaged 35,713 (range of 5,770 to 54,515) for the highest (15.6
kN or 3,500 lbf) load condition, 117,520 cycles (7,080 to 151,426) for the 11.1 kN (2,500 lbf)
condition, and 321,922 (8,381 to 380,616) for the lowest load (8.9 kN or 2,000 lbf) condition
(Figure 7). The tests, in which the minimum numbers of cycles were achieved, at each load
level, were for those experimental runs that failed between HAZ and weld in the static tensile
tests. As would be expected, these coupons predominantly failed on the mild steel side (B).
Statistical significance existed for all factors, with the exception of CTWD, on fatigue life.

7
Sheet Metal Welding Conference XII Livonia, MI May 9-12, 2006

S-N Curve
logten(S) = 3.094 - 0.2163 logten(N)
Regression
95% CI
95% PI
200
S 0.0444567

S, Stress Range ( MPa )


R-Sq 80.3%
R-Sq(adj) 80.2%
150

100
90
80

70

60
10000 100000 1000000
N, Cycles to Failure

Figure 7. S/N Curve for Fatigue Cycles to Failure for all 58 Experimental Runs

Statistical Correlation Models

Model 1 - Effect of Weld Geometry on Fatigue Life

The statistical analysis indicated that fatigue life at the 8.9-kN (2000-lbf) load gave a Response
Surface Linear Model R-Squared value = 0.65, and life increases with increasing vertical leg
size, decreasing bottom plate (DP 600) penetration, decreasing top plate (mild steel)
penetration, and decreasing upper fusion line angle (angle formed at weld root between
horizontal lower surface of upper substrate and vertical weld leg). The failure location at this
load is more likely to be in the HAZ of the mild steel base metal with decreasing vertical leg size
and increasing horizontal leg size.

The statistical analysis indicated that fatigue life at the 11.1-kN (2500-lbf) load gave a Response
Surface Linear Model R-Squared value = 0.55, and life increases with increasing vertical leg
size and decreasing bottom plate (DP 600) penetration. The failure location at this load is more
likely to be in the mild steel base metal with decreasing vertical leg size, increasing bottom toe
angle and increasing bottom plate penetration.

The statistical analysis indicated that fatigue life at the 15.6-kN (3500-lbf) load gave a Response
Surface Linear Model R-Squared value = 0.56, and increases with increasing vertical leg size,
increasing bottom toe angle and decreasing bottom plate (DP 600) penetration. The failure
location at this load is more likely to be in the mild steel base metal with decreasing vertical leg
size, increasing bottom toe angle, decreasing bottom plate penetration, and increasing
horizontal leg size.

Using the model for weld dimension optimization for highest fatigue life, with constraints as
defined in the AWS D8.8 welding specification (e.g., minimum vertical leg size at 90% of 3.4 mm
= 3.1 mm and maximum bottom plate penetration at 70% of 3.4 mm = 2.4 mm), process plots,
such as the one shown in Figure 8, can be generated. The plot indicates how threshold values
for fatigue life, such as 12,500 cycles min at 15.6-kN (3500-lbf) load, 39,000 cycles min at 11.1-
kN (2500-lbf) load, and 81,000 cycles min at 8.9-kN (2000-lbf) max load, can be applied. Other
constraints (weld dimensions or process parameters) can be applied accordingly. In this
example, maximum upper fusion line angle = 98 degrees and maximum top plate penetration =
1.1 mm were specified.

8
Sheet Metal Welding Conference XII Livonia, MI May 9-12, 2006

160.0
Toe Angle
(º) 11.1 kN (2500 lbf) 8.9 kN (2000 lbf)
Fatigue Cycle: Fatigue Cycle:
150.0
39000 min 81000 min

140.0

130.0
15.6 kN (3500 lbf)
Fatigue Cycles:
120.0 12500 min

110.0

100.0
2.00 2.50 3.00 3.50 4.00

Vertical Leg Size (mm)

Figure 8. Process Plot with Fatigue Constraints

Model 2 - Effect of Weld Variables on Weld Dimensions

Through the statistical analysis, the influence of the weld variables on each of the important
weld characteristics was determined:

• Bottom plate penetration (linear model R-Sq = 0.78) decreases with decreasing WFS,
increasing CTWD, increasing torch angle, increasing push/drag angle and decreasing
placement away from joint.
• Top plate penetration (quadratic model R-Sq = 0.64) decreases with decreasing
WFS/TS ratio, dependence being stronger at low WFS, decreasing torch angle and
increasing wire placement away from joint.
• Vertical leg size in mild steel (linear model R-Sq = 0.39) increases with increasing
WFS/TS ratio and decreasing wire placement away from joint.
• Undercut in DP 600 (2-factor interaction model R-Sq = 0.66) decreases with increasing
WFS/TS ratio and decreasing wire placement away from joint.
• Horizontal leg (2-factor interaction R-Sq = 0.79) increases with decreasing WFS,
increasing WFS/TS ratio, increasing push angles and increasing wire placement away
from joint.
• Bottom toe angle - DP 600 (quadratic model R-Sq = 0.76) increases with decreasing
WFS (Figure 9), increasing WFS/TS ratio at larger torch angles, increasing CTWD, at
intermediate push angles and increasing wire placement away from joint.
• Upper fusion line angle (linear model R-Sq = 0.32) decreases with decreasing WFS,
increasing CTWD and increasing push angle.

Model 3 - Effect of Weld Variables on Fatigue Life

The statistical model for the influence of weld variables on fatigue life showed that, at the 8.9-kN
(2000-lbf) load (Response Surface 2FI Model R-Sq = 0.71) fatigue life increases with the
WFS/TS ratio (decreasing TS) at a given WFS, and with decreasing WFS at a given WFS/TS
ratio. It also increases with increasing torch angle at high push angles only, decreasing wire
placement at low WFS/TS ratios (high TS) and increasing wire placement at high WFS/TS ratios
(lower TS) (Figure 10).

9
Sheet Metal Welding Conference XII Livonia, MI May 9-12, 2006

Toe Angle
X = A: Wire Feed Speed
Y = B: SQRT(WFS/TS)

Actual Factors
C: CTWD = 12.50 134.8
D: Torch Angle = 40
132.1
E: Push/Drag Angle = 15
F: Wire Placement = -2
129.5

Toe Angle
126.8

124.2

159.00 3.45

169.50 3.31

180.00 3.17

190.50 3.03B: SQRT(WFS/TS)


A: Wire Feed Speed
201.00 2.89

Figure 9. 3-D Plot of Toe Angle with WFS and (SQRT (WFS/TS)

DESIGN-EXPERT Plot
0.00
8.9 kN (2000 lbf) Fatigue Cycles
8.9 kN (2000 lbf) Fatigue Cycles
X = B: SQRT(WFS/TS) Ratio
Y = F: Wire placement

Actual Factors
A: WFS = 180 -0.50
C: CTWD = 15.87
F: Wire placement

D: Torch Angle = 40.00


E: Push/Drag Angle = 15.00
135000 150000
-1.00

165000

120000

-1.50

180000

105000
-2.00
2.89 3.03 3.17 3.31 3.45

B: SQRT(WFS/TS) Ratio

Figure 10. Fatigue Life with SQRT (WFS/TS) and Wire Placement

Fatigue life at the 11.1-kN (2500-lbf) load (Response Surface 2FI Model R-Sq = 0.80) increases
with decreasing WFS, increasing WFS/TS ratio, increasing CTWD, increasing torch angle at
high push angles and higher wire placements, increasing push angle at higher torch angles,
decreasing wire placement at lower torch angles, and increasing wire placement at higher torch
angles.

Fatigue life at the 15.6-kN (3500-lbf) load (Response Surface 2FI Model R-Sq = 0.73) increases
with decreasing WFS (more so at larger wire placements), increasing WFS/TS speeds (more so
at larger wire placements), increasing CTWD, increasing torch angle, increasing push angle at
larger wire placements, decreasing wire placement at high WFS/TS ratios, and low WFS, and
increasing wire placement at larger push angles.

Transfer functions were developed to determine the optimum settings for fatigue life at each
load level.

10
Sheet Metal Welding Conference XII Livonia, MI May 9-12, 2006

Summary of Results from Experiments 2 and 3

Graphical optimization of the first DOE model, which relates weld variables to weld dimensions,
using the constraints of 120 degrees for the minimum toe angle, 3.1 mm for the minimum
vertical leg length, 0.63-in. CTWD, 40-degree torch angle, 15-degree push angle, wire
placement of -1 (one wire diameter out of the joint), and penetration in the 20-70% range,
yielded the regime in weld parameter space shown in Figure 11. This regime was used to
select the five points for the next set of experiments for the 3.4-mm DP 600 to 4.7-mm mild steel
lap welds. Table 1 shows the five points that were selected for these welds.

Figure 11. Graphical Optimization Regime in WFS and WFS/TS Parameter Space
Showing the Five Points Selected for Experiment 2

Table 1. Five Points Selected for Runs 1-5

Run
1 2 3 4 5
WFS (mm/s) 180 159 159 201 201
TS (mm/s) 18 14 19 17 23

11
Sheet Metal Welding Conference XII Livonia, MI May 9-12, 2006

Results indicate that for Welds 1, 2, and 5, the vertical leg size (on the mild steel side) was
greater than 3.0 mm, with bottom (DP 600) plate penetration of less than 50%, and bottom toe
angles in the range of 112-136 degrees. Results are listed in Table 2.

Table 2. Results for Welds 1, 2, and 5

Maximum Maximum
Penetration Penetration
Toe Vertical Horizontal Vertical Horizontal
Angle Leg Leg Leg Leg
Run (degrees) (mm) (mm) (mm) (mm)
1 136 4.6 5.5 1.49 0.69
2 115 3.4 4.2 1.79 1.24
5 112 3.2 3.6 1.82 1.26

Tensile failure during lap-shear testing was mostly in the mild steel base metal and occurred at
averaged loads of 42-45 kN. Tensile failure only occurred in the DP 600 HAZ with the smallest
weld (Run 5), where the toe angle was the lowest (112 degrees).

Fatigue lives were similar for Runs 1 and 2, but less for Run 5, where TS and WFS were the
highest, following the direction of DOE 1 results. These results are shown in Figure 12. The
fatigue location was predominantly in the DP 600 HAZ area, indicating the dominant effect of
the bottom toe angle on the fatigue failure process in these joints.

DO E #2 : DP6 0 0 3 .4 mm/M1 A3 3 4 .7 mm
200
Variab le
Ru n 1
Ru n 2
Ru n 5
Stress Range (MPa)

100

80
10000 100000 1000000
N, Cy cles t o Failure

Figure 12. S/N Curve for Fatigue Cycles to Failure for Three Runs of Experiment 2

For Experiment 3 involving lap joints between 3.4-mm DP 600 and 2.7-mm mild steel, the runs
were chosen from an extrapolation of the first DOE model for smaller weld sizes, achieved by
increasing the TS to aim for vertical leg sizes of 2.4 or greater, which represents 90% of the
thickness of the thinnest member (mild steel) in this combination. The rest of the variables were
similar to that in Experiment 2. This resulted in the five points shown in Table 3.

12
Sheet Metal Welding Conference XII Livonia, MI May 9-12, 2006

Table 3. Five Points for Runs 11-15

Run
11 12 13 14 15
WFS (mm/s) 180 159 159 201 201
TS (mm/s) 30 26 31 29 34

Results indicate that for the welds in Runs 11, 13, and 14, the vertical leg size (on the mild steel
side) was in the range of 2.1-2.3 mm, with bottom (DP 600) plate penetration of less than 50%,
and bottom toe angles in the range of 88-117 degrees. Results are listed in Table 4.

Table 4. Results for Runs 11, 13, and 14

Maximum Maximum
Penetration Penetration
Toe Vertical Horizontal Vertical Horizontal
Angle Leg Leg Leg Leg
Run (degrees) (mm) (mm) (mm) (mm)
11 88 2.2 4 1.2 0.14
13 117 2.1 3.8 1.0 0.08
14 88 2.3 4.1 1.5 0.30

Tensile failure during lap-shear testing of the three weld runs tested was in the mild steel base
metal, and occurred at an average load of about 27 kN.

Fatigue lives were similar for the three runs (Figure 13), and fatigue failure location, with the
exception of one sample, was in the DP 600 HAZ area, indicating the dominant effect of the
bottom toe angle on the fatigue failure of these joints.

DOE #3: DP600 3.4 mm/M1A35 2.7 mm


200
Variable
Run 11
Run 13
Run 14
Stress Range (MPa)

100

80
10000 100000 1000000
N, Cycles to Failure

Figure 13. S/N Curve for Fatigue Cycles to Failure for Three Runs of Experiment 3

13
Sheet Metal Welding Conference XII Livonia, MI May 9-12, 2006

Conclusions

The experiments and, subsequent, statistical analysis conducted enabled a comprehensive


understanding of the influence of weld process variables on weld dimensions and, in turn, on
the mechanical properties.

The following conclusions can be drawn from the study:

• The work revealed that the GMAW process window when welding 3.8-mm-thick mild
steel to 3.4-mm-thick DP 600 using the CV mode is significantly large to classify as a
robust process. The additional benefit of pulsed waveforms will further expand this
window. Predictions from this work could be extended to choose welding parameters for
other thickness combinations, such as 3.4-mm DP 600 to either 4.7- or 2.7-mm mild
steel, and the results were in general agreement with trends from the first model.
• Optimum process conditions cannot be determined by static tensile strength testing of
the welded joint, which resulted in equivalent values over most of the process range
applied.
• Microhardness assessment indicated that all samples in this study, no matter what the
level of heat input, softened in the HAZ by approximately 15%. None of the process
factors was statistically significant relative to microhardness.
• Fatigue life differed significantly over the range of the process parameters and with the
resultant weld characteristics. Fatigue life of the DP 600 and mild steel lap joints
increased with increasing WFS/TS ratios (decreasing TS), decreasing WFS, increasing
CTWD, increasing torch angle, increasing push angle, and increasing wire placement
away from the joint, within the range of the defined parameters. The fatigue life was also
shown to increase with increasing vertical leg size and decreasing bottom plate
penetration at all load levels applied. The weld toe angle became significant, larger
being better, for the fatigue life at higher loads.
• The weld failure in fatigue is more likely to be at the root of the weld on the mild steel
side when the vertical weld leg decreases, the horizontal leg increases or the toe angle
increases.

References

(1) Kunishige, K., Yamauchi, N., Taka, T. and Nagao, N., "Softening in Weld Heat-affected
Zone of Dual Phase Steel Sheet for Automotive Wheel Rim," SAE Technical Paper Series
830632, SAE Int. Congress & Exposition, Detroit, MI (Feb. 28-Mar. 4, 1983).

(2) Ghosh, P. K., Gupta, P. C., Pal, O. M., Avtar, R., Jha, B. K., and Sagar Dwivedi, V.,
"Influence of Weld Thermal Cycle on Properties of Flash Butt Welded Mn-Cr-Mo Dual Phase
Steel," ISIJ Int., 33(7):807-815 (1993).

14

You might also like