Investigation of Heat-Affected Zone Cracking of GTA Welds of Al-Mg-Si

Alloys Using the Varestraint Test

Maximum crack length has a close correlation with the temperature distribution in the HAZ

BY M . KATOH A N D H. W . KERR

ABSTRACT. There have been previous investigations of HAZ liquation cracking of welds of Al-Mg-Si alloys. However, comparison of these investigations is diffi­cult, since different test procedures were used, and the results were qualitative. In the present investigation, more quantita­tive results have been obtained using the Varestraint test to investigate liquation cracking in the HAZ of gas tungsten arc welds of Al-Mg-Si alloys. The sensitivity of HAZ liquation cracking was mainly evalu­ated by the maximum crack length in the HAZ.

HAZ liquation cracks were observed normal to the weld bead. In order to cause HAZ liquation cracking in a com­mercial 6061 alloy, critical values of both heat input and augmented strain are nec­essary. The maximum crack length increased linearly with further increments of augmented strain at a given heat input. Besides, the maximum crack length increased linearly when the heat input was increased above about 11 kj/cm (28 k)/in.) (critical value) up to about 17 kj/cm (43 k)/in.) at a given augmented strain. According to observations using a SEM, many globular structures were observed on the fracture surfaces, show­ing that the cracked region is partially melted.

According to a multiple regression analysis, it was made clear that Si and Mg, which are the main elements in the base metal, and Cu are elements promoting cracking, but that Cr, Mn and V, which are grain refiner elements, suppress cracking.

The temperature at the end of the crack decreased linearly when the aug-

M. KATOH is with the Department of Metal­lurgical Engineering, Kyushu Institute of Tech­nology, Kitakyushu, Japan. H. W. KERR is with the Department of Mechanical Engineering, University of Waterloo, Waterloo, Ontario, Canada.

mented strain was increased for the com­mercial 6061 alloy. Most of the HAZ liquation cracks were observed in the region where the temperature was between the liquidus and solidus temper­atures. At very high levels of augmented strain, however, the temperature at the end of the crack was below the solidus temperature measured using a differen­tial scanning calorimeter.

Introduction

There have been investigations of liquation cracking in the heat-affected zone (HAZ) of welds for several classes of alloys, including ferritic steels (Refs. 1-4), austenitic stainless steels (Ref. 5), and Al-Mg-Si alloys (Refs. 6, 7). The elements which give rise to cracking are, of course, different in different materials. For exam-

KEY W O R D S

GTA Al-Mg-Si Welds HAZ Liquation Cracks Varestraint Testing Augmented Strain Commercial 6061 Al Minor Element GTA Spot Weld Effect Heat Input Effects Arc Time Effect Rolling Direction

pie, in both ferritic and austenitic steels the cracking is often ascribed to the liquation of sulfides (Refs. 3-5), but also may be caused by impurities such as Cu (Ref. 8).

The effects of welding procedure on liquation cracking appear to be generally understood: the severity of cracking increases with increasing heat input, increasing penetration of the weld bead and increasing restraint (Ref. 3).

However, several mechanisms have been put forward to explain liquation. In a particular alloy, the mechanism(s) of cracking, and the elements contributing to cracking, are often not clear. For example, Pepe and Savage have explained how liquation can occur due to grain boundary segregation followed by constitutional liquation (Ref. 1). In consti­tutional liquation, the presence of segre­gation lowers the melting temperature below that predicted by the equilibrium phase diagram. In other investigations, cracking is attributed to the melting of eutectics (Ref. 2). Under some circum­stances, cracks in the HAZ may be repaired by the backfilling of liquid from the weld metal (Ref. 9).

In Al-Mg-Si alloys, at least two mecha­nisms have been proposed to explain liquation cracking. Gittos and Scott sug­gested that cracking occurs when the base metal solidus is lower than the weld metal solidus temperatures (Ref. 6). Tsuji-moto, er ai, related cracking to the

Table 1—Chemical Compositions of Materials Used (wt-%)

Cu Fe Ms Mn Zn Cr Zr

Commercial 6061 Alloy 3 (6261) Alloy 10 (6261) Alloy 17 (6061) Alloy 22 (6061) Alloy 24 (6061) Alloy 25 (6351)

0.25 0.25 0.24 0.99 0.24 0.01 0.23

0.29 0.20 0.22 0.25 0.20 0.20 0.21

0.76 0.58 0.58 1.03 1.0 1.01 0.57

0.044 0.28 0.17 0.015 0.002 0.002 0.02

0.51 0.82 0.81 0.64 0.57 0.57 0.98

0.025 0.012 0.013 0.010 0.012 0.011 0.011

0.080

— — — -— —

0.16 0.001 0.001 0.12 0.11 0.11 0.11

0.011 0.01 0.18 0.009 0.009 0.009 0.01

-0.O01 0.001 0.01 0.001 0.001 0.001

360-s | DECEMBER 1987

WELDING DIRECTION l> CRACK

FORCE

1 Ju^r:

FORCE

__MARC I Sg~-' SPECIMEN

•BENDING BLOCK

f/g. 1 —Schematic representation of the Varestraint test

melting of low temperature eutectics in the HAZ (Ref. 7), but were unable to determine if the eutectics were caused by constitutional liquation or grain boundary diffusion from the weld metal. Moreover, they observed less cracking with more melting. Comparison of these investigations is difficult, since different test procedures were used, and the results were qualitative.

In the present investigation, more quantitative results have been obtained using the Varestraint test to investigate liquation cracking in the HAZ of gas tungsten arc (GTA) welds. The effects of augmented strain, heat input, and chemi­cal compositions of the base metals on cracking were examined.

Materials Used and Experimental Procedures

Table 1 shows the chemical composi­tions of the materials used. The thickness and length of the samples were 6.3 mm (0.25 in.) and 120 mm (4.7 in.), respective­

ly. A commercial 6061 alloy, plus experi­mental alloys with varying contents of Mg2Si, Cu, Mn, Cr, V and Zr, were studied.

A Varestraint-type test was performed to evaluate the sensitivity of the HAZ liquation cracking. Figure 1 shows a sche­matic representation of the Varestraint-type test. The test resulted in HAZ liqua­tion cracks which were normal to the weld bead. The radii of the bending blocks used were 200 and 100 mm (7.9 and 3.9 in.), but it was possible to change the bending radii of the specimens by changing the stroke of the cylinder used to bend the specimens.

The sensitivity of HAZ liquation crack­ing was mainly evaluated by the maxi­mum crack length in the HAZ. The crack length was observed as-welded using a projector (20X). Gas tungsten arc weld­ing (DCEP) using argon gas was per­formed under the following conditions: welding speed (v) from 67 to 193 m m / min (2.6 to 7.6 ipm), welding current (I) from 85 to 130 A, and arc voltage (E)

from 19 to 20 V. The arc was extin­guished approximately one second after bending was initiated. When the arc was extinguished at the same time as bending, or before bending, the crack length became shorter or no crack was observed. When the arc was extin­guished more than one second after bending, almost the same crack lengths were obtained as that of one second.

Results and Discussion

The Effects of Width of Specimen and Ram Speed

Heat-affected zone liquation cracks having different orientations with respect to the weld bead can be developed (Ref. 10). The crack investigated here is one which is developed normal to the weld bead. Figure 2 shows an example of such a crack. The crack is propagating along grain boundaries in the HAZ. Another crack is observed in the weld metal. In many cases, the cracks were indepen­dent of weld metal cracking or were

WELD METAL -=»«&

HAZ

:s^--o~Se&v

Fig. 2 —An example of a feature of cracks in the HAZ

WELDING RESEARCH SUPPLEMENT 1361-s

4 .0

3 .5

3.0 X ,1 H

_ l

,. X 1— ( • )

3» UJ

_) <̂

( ) -c Lt

2.

?

. 1 .

1.

0.5

0 .0

£=3 . 2%,Q=20. 4kj"/cm

C k la V

/%.

40 50 60 70 80 90 100 110 WIDTH OF SPECIMEN, w (mm)

2. 2

2 .0 o E

_J . 1. 8

nz I— ( J

5 1.6 _ i •XL

(_) Jg 1 . 4

X -c

1.2

1.0

£=3 . 27., w=50mm, Q=14. 6kJ/cm

O ̂ ^

rT"

10 RAM SPEED,

15 S (mm/s)

20

fig. i -length

Effect of the width of the specimen on the maximum crack Fig. 4 — Relation between the maximum crack length and the ram speed

observed to be separated from cracks in the weld metal, as seen in Fig. 2.

Figure 3 shows the effect of the width of the specimen on the maximum crack length (Lmax) when the commercial 6061 alloy was welded at the heat input (Q) of 20.4 kj /cm (52 kj/in.) (augmented strain t = 3.2%, assuming Q = 60 X lE/v). The augmented strain was calculated using Equation 1:

c = t/2R (1)

where t is thickness of the specimen (6.3 mm) and R is bending radius (mm). When the width of the specimen was increased, the maximum crack length decreased gradually. This is due to the change of the temperature distribution near the fusion boundary plus the decreased ram speed possible with increased width. The width of the specimens was fixed at 50 mm (2

t- BEND IN

Fig. 5 —Schematic representation of the HAZ liquation cracking in GTA spot weld

in.) for the remainder of the investiga­tion.

The maximum crack length changes when the ram speed of the cylinder used for bending the specimen is changed. Figure 4 shows the relation between the maximum crack length and the ram speed for a fixed augmented strain of 3.2%. The ram speeds shown in Fig. 4 were measured when bending the base metal without welding. When the ram speed was approximately doubled, from about 8 to 17 mm/s (0.31 to 0.67 in./s), the maximum crack length increased approximately linearly by only 20%, from about 1.5 to 1.8 mm (0.06 to 0.07 in.). In the remainder of the investigation, the ram speed of 12.3 mm/s (0.48 in./s) was used.

Effect of Augmented Strain

In addition to traveling welds, some spot welds were made. Figure 5 shows a schematic representation of the HAZ liquation cracking in such spot welds after bending. The lengths of cracks observed in the HAZ were longest in the directions of twelve and six o'clock, and decreased towards the directions of three and nine o'clock.

The reason for the variation in crack length is the variation in augmented strain. When a specimen is bent as shown in Fig. 5, strain e in the transverse direction is given by Equation 1. Strain t', which is operating in the direction inclined at an angle 8 to the transverse direction, is given by Equation 2:

c 2 a = e COS^ (2)

e' is the strain which causes a crack at the orientation 8, as shown in Fig. 5.

Figure 6 shows the relation between crack length and the augmented strain in the commercial 6061 alloy for different spot welds. Parameter t in Fig. 6 is the arc time. When the augmented strain was less than 1%, no cracking was observed in the HAZ. As the strain was increased above 1%, the crack length increased linearly. At a given strain, the crack length became longer when the arc time was lengthened.

Figure 7 shows the relation between the maximum crack length and the aug­mented strain when bead-on-plate weld­ing was performed on the commercial 6061 alloy. The same relationship was obtained as in Fig. 6, i.e., no cracking was observed when the augmented strain was less than 1%, and the maximum crack length increased linearly as the strain increased above 1%. At a given strain, the maximum crack length became longer when heat input was increased, corresponding to the effect of arc time in Fig. 6.

Effect of Heat Input

As shown in Figs. 6 and 7, cracking increased when the arc time or heat input increased. Here, the effect of heat input on the maximum crack length was inves­tigated for different welding conditions. Figure 8 shows the relation between the maximum crack length and the heat input for the commercial 6061 alloy, using an augmented strain of 3.2%. No cracking was observed when the heat input was

362-s I DECEMBER 1987

less than about 11 kj/cm (28 kj/in.). When the heat input became larger than 11 kj /cm, the maximum crack length increased linearly. When the heat input became larger than about 17 k)/cm (43 kj/in.), the slope of the maximum crack length versus heat input was reduced. This tendency was consistent for welding speeds from 67 to 193 mm/min (2.6 to 7.6 ipm).

The effect of rolling direction was also investigated. In Fig. 8, open marks show the cases when the welding was normal to the rolling direction of the specimen, and solid marks show the cases when the welding direction was parallel to the roll­ing direction. Almost the same relation was obtained for both cases.

Figure 8 summarizes the results when the augmented strain was 3.2%. Using Fig. 7, it is possible to determine the relationship between the maximum crack length and the heat input for several levels of strain, as shown in Fig. 9. Figure 9 shows the result obtained by interpola­tion, using the result shown in Fig. 7. Over the range from about 11 to about 17 kj /cm, the maximum crack length increased linearly when the heat input increased, for all of the strains investigat­ed. With increased augmented strain, the slope of the maximum crack length ver­sus heat input curve increased.

It is clear that critical values of both heat input and augmented strain are nec­essary to cause HAZ liquation cracking in a GTA welded commercial 6061 alloy. For a fixed specimen size, the critical value of the heat input can be used as a measure to evaluate the sensitivity of an alloy to HAZ liquation cracking.

Effect of Chemical Compositions of Base Metal

Figures 10A and B show the relation­ship between the maximum crack length and the heat input for different base metals. In each case, the maximum crack length increased linearly with increased heat input above the critical heat inputs. The values of critical heat inputs depend on the chemical compositions of the base metals. The slopes of the maximum crack length versus the heat input also depend on the base metals. Alloys 24, 22 and 17 are experimental 6061 alloys with differ­ent contents of Cu: 0.01, 0.24 and 0.99%, respectively. Figure 10A shows that the maximum crack length becomes longer with increased Cu content. Figure 10B shows that the relationship between the maximum crack length and the heat input was almost the same in the cases of Alloys 10 and 25. These alloys, plus Alloy 3, all contained about 0.24% Cu and 0.58% Mg, but with various amounts of other alloy additions (Mn, Si, Cr and V). Alloy 3 had the longest crack length at a given heat input. As is clear from Fig.

0. or> 0 .0 0 .5 1.0 3.0 3 .5 4 .0 1.5 2 .0 2 .5

STRAIN, £' (%) Fig. 6 — Relation between the crack length and the strain in the case of GTA spot welding, for different times "f"

2 .0 AUGMENTED

6.0

Fig. 7-Relation between the maximum crack length and the augmented strain for different values of Q (heat input per unit distance)

10, the maximum crack length was small­est in the case of the commercial 6061 alloy. This is due to the difference of grain sizes in the HAZ, as is mentioned below.

A multiple regression analysis was per­formed by taking the maximum crack length Lmax (mm) as the dependent vari­able and heat input Q (kj/cm), and Cu, Mg, Mn, Si, Cr and V content (wt-%) as predictor variables, and by using a com­puter, obtaining Equation 3.

Lmax = -462.60 + 0.308 Q +38.31 Cu + 840.3 Mg - 3016 Mn

+ 1020 Si - 8615 Cr - 1895 V (R2 = 0.87) (3)

Since contents of Fe and Ti were approx­imately constant in all alloys, their effects could not be determined. Figure 11 shows the relationship between the esti­mated maximum crack lengths using Equation 3 and the true maximum crack lengths. A good correlation was obtained

Q.

o

> UJ Q \ I O tx < UJ <r> ui tr

Q.

o > ui

O CC < Ul t/i ui cc

aaE Q .

o — I ui >

o cc < Ul tf) Ul cr

a. O _ i ui >

X o cr < ui tr>

a. o _! Ul

> Ul Q \ I O OC

< Ul

Ul cc

WELDING RESEARCH SUPPLEMENT 1363-s

4. 0

3 . 5

'E E

" 3 . 0 X • E

_ J

. 2 . 5

C D

5 2.0 _J

:̂ L J

£ 1.5 L J

i 1.0 X •< s:

0.5

0.0

£=3.2%

— O v=67mm/

A v=94mm/

• v=124mm

V v=134mm

() v= 166mm

OPEf

SOL

1 MARK:

D MARK

i i i n

'm in

'min

'min

*. D.J-R.

:W. D.//

D.

*. D.

mf

/v

4

<̂ 6 ^^

•

O

4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20.0 22.0 HEAT INPUT, Q (k j /cm)

Fig. 8 —Relation between the maximum crack length and the heat input

4.0

3.5 'E

E 3. 0 x a

E _ l

* 2-5 I E

LD

l5 2. 0 _ l

:̂ C_)

K 1-5 L J

i 1.0 x <; 2 :

0. 5

1 e=4.5X

/S=3

/g=3

/

X"8

/ « = Z

_^-«=i

. 5*

. 0%

.5%

. OZ

.5X

0.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20.0 22.0

HEAT INPUT, Q (kJ/cm) Fig. 9 — Relation between the maximum crack length and the heat input at several different augmented strains

between them, though some scattering was observed in the data. Equation 3 shows that Si, Mg and Cu are elements promoting cracking, but Cr, Mn and V are elements suppressing cracking. That is, Si and Mg, the main elements in the base metals, are detrimental elements for HAZ liquation cracking. It is effective to add Cr, Mn and/or V in order to sup­press HAZ liquation cracking. These are grain refiner elements. To understand these results better, the fracture surfaces were examined.

SEM Observations

Figures 12 and 13 show fractographs of the commercial 6061 alloy when the welding direction is normal and parallel to the rolling direction of the specimen, respectively. In each series, fractograph A is from the weld metal, B is near the

4.0

3.5 'E

« 3 - ° a E

_! . 2.5

IE

5 2.0 b : l_J

£ 1.5 LJ

••€.

5 1.0 X

< 0.5

fusion boundary, C is near halfway along the crack, and D is near the end of the crack. Many dendrite tips are clearly observed in Fig. 12A. In Fig. 12B, lamellar patterns parallel to the rolling plane are observed, besides dendrite tips. The lamellar structures become more pro­nounced in Fig. 12C, but some rounded or globular structures are also observed, which show that this area was partially melted. To expose the crack surface, the material was subsequently fractured at room temperature. In Fig. 12D, the boundary between the HAZ liquation cracking and the subsequent ductile frac­ture (where many dimple patterns typical of ductile fracture are observed) is clearly recognized.

Almost the same fractograph features are observed in Fig. 13 as in Fig. 12, although the lamellar structures are short­er, since cracking is normal rather than

parallel to the rolling direction. Figure 8 shows that there was little difference between crack lengths for the two cases.

Figure 14 shows fractographs of Alloy 17 (experimental 6061 alloy containing 0.99% Cu). In Fig. 14, some differences from those already shown in Figs. 12 and 13 are observed. The grain sizes in the HAZ of the experimental alloys were greater than in the commercial alloy, producing larger, flatter regions in these fractographs instead of the lamellar struc­tures. These structures are clearly observed in Figs. 14B and C. In Fig. 14B, many small white globular structures are observed which contain considerable Cu —about 20 to 30% according to the EDX results. The number of the globular structures is greater near the fusion boundary. These structures clearly show that there was some liquid when the

.0.0 A «•

£=3.2% _ j _ A

O

o/ X-, / / /

/

/

/ f

JO / / /

°/ / / A

'

rfc.

V

/

n/n ___„

/ \ \ ALLOY 17(0. 99ZCu)

^ALLOY 24(0. DIZCu)

-̂COMM :RCIAL 6061

0 6.0 10.0 12.0 14.0 16.0 18.0 20.0 22.0 24.0 HEAT INPUT, Q (kJ/cm)

10.0 12.0 14.0 16.0 18.0 20.0 22.0 24.0 HEAT INPUT, • (kJ/cm)

Fig. 10 —Relation between the maximum crack length and the heat input for different base metals. A—6061 alloy; B — 6261 and 6351 alloys

364-s | DECEMBER 1987

crack was developed, and that the amount of the liquid decreased farther from the fusion boundary.

Mechanism of HAZ Liquation Cracking

As already mentioned, both the strain and the heat input must be greater than critical values in order to cause HAZ liquation cracking. When the heat input is changed, the distribution of the maxi­mum temperature near the fusion boundary also changes. Figure 15 shows the distribution of the maximum temper­ature near the fusion boundary of com­mercial 6061 alloy, welded using two conditions. The temperature was mea­sured using a temperature-sensitive lac­quer. The maximum temperature de­creases almost linearly very near the fusion boundary. When the heat input was increased from 13.6 to 15.6 k)/cm (35 to 40 kj/in.), the slope of the maxi­mum temperature versus the distance from the fusion boundary into the HAZ decreased.

Using both Figs. 7 and 15, it is possible to obtain the relationship between the augmented strain and the peak tempera­ture at a given location in the HAZ. Figure 16 shows the relationship obtained. The hatched region in Fig. 16 represents the cracked region. Almost the same curve

4.0

3 . 5

3 . 0

2 .5

2 .0

1.5

1.0

0 . 5

0. 0

o

XC SL a A/ A A

^M&k •T3A/$7 S

ozL° i 1 J? • kr T

y o 0

A^

o

O ALLOY 3

A ALLOY 10

D ALLOY 17

V ALLOY 22

0 ALLOY 24

O ALLOY 25

|

A

o

Fig. 11 — Comparison of the estimated maximum crack length with the true maximum crack length

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3. TRUE MAXIMUM CRACK LENGTH (mm)

4. 0

was obtained for the heat inputs of both 13.6 and 15.6 kj /cm. The peak tempera­ture at the end of the crack, i.e., the lowest temperature of crack growth for each augmented strain, decreased linear­ly when the augmented strain increased. It is clear that the maximum crack length has a close correlation with the tempera­ture distribution in the HAZ.

The observation that the maximum crack length at a given augmented strain becomes longer when the heat input increases, as shown in Fig. 8, is due to the phenomenon that the temperature becomes higher over a wider region in the HAZ. The temperature measured at the fusion boundary was 649°C (1200°F), very close to the liquidus temperature

Fig. 12 — Examples of fractographs of a commercial 6061 alloy (welding direction being normal to the rolling direction). A - Weld metal; B - near the fusion boundary; C — near halfway along the crack; D — near the end of the crack

WELDING RESEARCH SUPPLEMENT I 365-s

Fig. 13 - Examples of fractographs of a commercial 6061 alloy (welding direction being parallel to the rolling direction). A — Weld metal; B - near the fusion boundary; C — near halfway along the crack; D — near the end of the crack

Fig. 14 — Examples of fractographs of Alloy 17. A — Near the fusion

boundary; B — near halfway along the crack; C — near the end of the

crack

366-s I DECEMBER 1987

2 4 6 8 10 12 DISTANCE FROM FUSION BOUNDARY, :

14 (mm)

Fig. 15 —Distribution of the maximum temperature near the fusion boundary

5.0

4.0

3.0

£ 2 . 0

1.0

0. 0 660 650 640 630 620 610 600 590 580 570

TEMPERATURE IN HAZ, T C O Fig. 16 — Relation between the augmented strain and the temperature in the HAZ

V, V, V y

1 ^ I/I y

CRACKED REGION ,

lA

A . 1

1 Ik iA

, \H> lAn X

O 0-13. 6kJ/cm

A 0=15. 6kJ/cm

(648°C/1198°F) of the commercial 6061 alloy obtained using a differential scan­ning calorimeter (DSC), whose accuracy is about ±1°C(±1 .8°F) .

For augmented strains less than 1%, no crack was observed in the HAZ. For augmented strains of 2.0, 3.0 and 4.5%, cracking was observed in the region high­er than about 630°, 612° and 583°C (1166°, 1134° and 1081°F), respective­ly.

It is clear from the fractographs shown in Figs. 12 and 13 that the HAZ liquation cracked region is partially melted. Several mechanisms have been proposed by which the HAZ liquation occurs: constitu­tional liquation (Ref. 1), eutectic reaction

(Ref. 2), and invasion (or diffusion) from the weld metal of elements composing the low melting temperature compounds (Ref. 7). Figure 17 shows a microstructure near the fusion boundary of the commer­cial 6061 alloy, using the heat input of 14.6 k]/cm. The partially molten zone is observed near the fusion boundary, hav­ing a width of about 0.5 to 0.6 mm (0.019 to 0.023 in.). When the augmented strain was 3.5%, the maximum crack length was about 2 mm (0.08 in.) at the heat input of 14.6 kj /cm (37 kj/in.), and the peak temperature at the end of the crack was about 600°C (1112°F). This crack length is much longer than the observed width of the partially melted zone, but small

amounts of liquid may be difficult to detect after welding. According to the result of the DSC, the solidus tempera­ture of the commercial 6061 alloy was 598°C (1108°F). Hence, up to 3.5% aug­mented strain, the observed temperature at the end of the crack is above the solidus temperature, so that melting can be expected at least for equilibrium con­ditions. However, when the augmented strain was 4.5%, HAZ liquation cracking was observed in the HAZ as low as about 580°C (1076°F), i.e., considerably lower than the solidus temperature obtained using the DSC.

The degree of cracking is therefore increased by increased strain and is not

WELD METAL PARTIALLY MOLTEN ZONE =+£ =H

Fig. 17 — An example of microstructures near the fusion boundary (heat input: 14.6 kj/cm)

WELDING RESEARCH SUPPLEMENT 1367-s

limited by the solidus temperature as measured by DSC. Some of the cracking beyond the measured solidus tempera­ture may be due to lowering of the solidus temperature, due to rapid freez­ing of the liquid layer. Alternatively, con­stitutional liquation could also lower the solidus temperature. At lesser strains, the cracking is within the measured freezing range.

In the case of GMA welding, HAZ cracking which is developed normal to the weld bead can be increased by pen­etration of molten metal from the weld metal into the HAZ (Ref. 10). Hence, penetration is also expected to occur in the present case, but cannot be distin­guished from melting in place.

Summary and Conclusions

The effects of augmented strain, heat input and the chemical compositions of the base metals on HAZ liquation crack­ing were studied quantitatively for GTA weld metals of Al-Mg-Si alloys using the Varestraint test. The conclusions ob­tained are as follows:

1) HAZ liquation cracks were ob­served normal to the weld bead.

2) An augmented strain greater than a critical value (about 1%) was necessary to cause HAZ liquation cracking in a com­mercial 6061 alloy. The maximum crack length increased linearly with further increments of augmented strain at a giv­en heat input.

3) A heat input greater than a critical value was necessary to cause HAZ liqua­tion cracking in the commercial 6061 alloy. At a given augmented strain, fur­ther increases in heat input beyond the critical value resulted in the maximum crack length increasing linearly with heat

input up to a second critical value. 4) For a given plate thickness and weld

geometry, the critical value of the heat input can be used as a measure to evalu­ate the sensitivity of HAZ liquation crack­ing for a given degree of augmented strain.

5) Both lamellar structures (due to roll­ing) and globular structures (from solidifi­cation) are observed in the HAZ liquation cracked region, showing that this region is partially melted.

6) According to a multiple regression analysis, it was made clear that Si and Mg, which are the main elements in the base metal, and Cu, are elements promoting cracking, but Cr, Mn and V, which are grain refiner elements, suppress crack­ing.

7) The peak temperature at the end of the crack decreased linearly when the augmented strain was increased for the commercial 6061 alloy. Almost the same curve of limiting temperature was obtained for heat inputs of 13.6 and 15.6 kj/cm (35 and 40 kj/in.).

8) Most of the HAZ liquation cracks were observed in the region where the temperature was between the liquidus and equilibrium solidus temperatures of the commercial 6061 alloy.

9) At very high levels of augmented strain, the temperature at the end of the crack was below the solidus temperature measured using a differential scanning calorimeter.

A ckno wledgments

This work was supported by a grant from the Natural Sciences and Engineer­ing Research Council of Canada, and by the University of Waterloo. Alcan gra­ciously supplied the base metals, and also

carried out the chemical analyses and differential scanning calorimetry. Messrs. N. Wilhelm, A. Degroot and H. Kamler provided important technical support at the University.

References

1. Pepe, ). )., and Savage, VV. F. 1967. Effects of constitutional liquation in 18-Ni mar­aging steel weldments. Welding Journal 46(9):411-s to422-s.

2. Tamura, H., and Watanabe, T. 1973. Mechanisms of liquation cracking in weld heat-affected zone of high-strength steel. Trans. Japan Weld. S. 4(2): 135-142.

3. Phillips, R. H., and lordan, M. F. 1976. Liquation cracking in the weld heat-affected zone of high-strength ferritic steels. Metals Technology (12):571-579.

4. Phillips, R. H. 1980. Fractography and mechanisms of high-temperature cracking in ferritic steel weldments. Metals Forum 3(3):158-166.

5. Honeycombe,)., and Gooch, T. G. 1970. Microcracking in fully austenitic stainless steel weld metal. Metal Const. Brit. Weld. J. (9):375-380.

6. Gittos, N. F., and Scott, M. H. 1981. Heat-affected zone cracking of Al-Mg-Si alloys. Welding lournal 60(6):95-s to 103-s.

7. Tsujimoto, K., Sakaguchi, A., Kinoshita, T., Tanaka, K., and Sasabe, S. 1983. HAZ cracking of Al-Mg-Si alloys. IIW Doc. IX-1273, pp. 1-13.

8. Savage, W. F., Nippes, E. F., and Mushala, M. C. 1978. Liquid-metal embrittlement of the heat-affected zone by copper contamination. Welding Journal 57(8):237-s to 245-s.

9. Savage, W. F., and Dickinson, D. W. 1972. Electron microanalysis of backfilled hot cracks in Inconel 600. Welding Journal 51(11):555-s to 562-s.

10. Katoh, M., and Kerr, H. W. 1987. Investi­gation of heat-affected zone cracking of GMA welds of Al-Mg-Si alloys using the Varestraint test. Welding Journal 66(9):251-s to 259-s.

WRC Bulletin 323 May 1987

Monograph on Narrow-Gap Welding Technology By V. Malin

This monograph is intended to familiarize the U.S. and the international welding communi ty with narrow-gap welding. Over 180 articles published by researchers around the world were reviewed to make a comprehensive analysis of the subject.

Publication of this WRC Bulletin was sponsored by the Subcommit tee on Minimum-Volume (Reduced-Gap) Welding of the Weldability Commit tee and the Interpretive Reports Commit tee of the Welding Research Council. The price of WRC Bulletin 323 is $20.00 per copy, plus $5.00 for postage and handling. Orders should be sent with payment to the Welding Research Council, Suite L301, 345 E. 47th St., New York, NY 10017.

368-s | DECEMBER 1987