Subscribe to our newsletter to receive the latest news and events from TWI:

Subscribe >
Skip to content

An investigation into the effect of weld strength mismatch on the assessment of HAZ fracture toughness (September 2002)

   
H G Pisarski (1) and P L Harrison (2)

(1) TWI  (2) Corus Research Development and Technology

Paper presented at ECF 14: 14th European Conference on Fracture, 8-13 September 2002, Cracow, Poland

Abstract

The effect of weld metal strength mismatch on the fracture toughness of the weld heat affected zone (HAZ) in two ferritic steels is described. Mismatch was achieved in two different ways which ensured that the HAZs were the same, irrespective of mismatch level. It is shown that fracture toughness is more affected by the local strength of the HAZ rather than differences in strength between weld metal and parent plate. Where HAZ strength is lower than both the parent plate and weld metal, HAZ fracture toughness is reduced. On the other hand, where HAZ strength is greater than both weld metal and parent plate, differences in strength between weld metal and parent plate have little influence on HAZ fracture toughness. 

Introduction

Opinions on the effect of weld strength mismatch on heat affected zone (HAZ) fracture toughness in ferritic steels are conflicting. Some consider overmatching to be beneficial to HAZ fracture toughness and others that it is harmful. [1,2,3] In order to better understand the effect of strength mismatch on HAZ fracture toughness, a systematic study was undertaken in which mismatch was achieved by two different methods.

In the first method, mismatch was achieved by employing the same parent plate and welding conditions, and changing the welding consumable to provide two different weld strength levels. Since HAZ fracture toughness is primarily dependent on parent plate composition and weld cooling rate, the HAZs produced by the two welds are the same. A 550MPa yield strength roller quenched and tempered steel was employed and yield strength mismatch, M, (where M is defined as the ratio of weld metal to parent plate yield strength at room temperature) levels of 0.65 and 1.51 were achieved by changing welding consumables.

The second method employed the same welding consumable and welding procedure, so maintaining the same weld metal strength throughout, and same parent plate composition. Parent plate strength was changed by prior heat treatment of a roller quenched tempered steel with a yield strength of 690MPa. A reduced strength plate was obtained by normalising the plate. In this way M values of 0.95 and 2.15 were achieved, but the HAZ properties remained the same.

HAZ fracture toughness was assessed using fracture mechanics specimens in which the target microstructure was the grain coarsened HAZ close to the weld fusion boundary. The fracture mechanics parameter measured was J integral estimated from crack mouth opening displacement.

Experimental details

Fabrication of test panels

The two steels chosen for this study were 25mm thick plates, with different yield strengths, referred to in this project as QT1 and QT2. Their compositions are given in Table 1

Table 1: Plate composition

Element, wt %
PlateCSiMnPSNbMoBAlTiV
QT1 0.11 0.28 1.29 0.012 0.003 <0.005 0.18 - 0.022 - 0.05
QT2 0.13 0.41 1.42 0.015 0.002 0.034 0.003 0.0024 0.042 0.038 0.06

To facilitate testing of the HAZ, the plates were machined to provide edge preparations in which one side was vertical and the other inclined at 10° to 20° from the vertical. The welds were made with wide root gaps and backing bars to support the weld root beads. Submerged arc welds were made in the QT1 plates using Oerlikon S2 wire and OP121TT flux, to achieve undermatch (M<1), and Oerlikon Fluxochord 45 wire and OP121TT flux to achieve overmatching (M>1). These were deposited at an arc energy of 2.4kJ/mm. The average width of the weld was 17mm. Subsequently, the welds were post-weld heat treated at 580°C for one hour with the primary purpose of reducing residual stresses to low and uniform levels. Tensile properties measured after heat treatment are given in Table 2.

With the QT2 plate, two strength levels were achieved by using the plate in the as-received, QT condition (M<1), and softening the plate by normalising at 900°C (M>1); subsequently referred to as QT2N. The welds were all made at an arc energy of 1.25kJ/mm using a gas-shielded flux core arc process with a 1.2mm diameter Fluxofil 35 wire and Argonshield 20 shielding gas. Subsequently, these welds were post-weld heat treated at 560°C for one hour. Tensile properties after heat treatment are given in Table 2.

Table 2: Tensile properties after post-weld heat treatment

MaterialRp0.2, MPaRm, MPaM
QT1 plate 566 660 -
GCHAZ, Gleeble+ 451 632 0.8*
SAW, weld metal 1 372 471 0.65
SAW, weld metal 2 859 986 1.51
QT2, plate 675 753 0.95
Plate after normalising (QT2N) 297 510 2.15
GCHAZ, Gleeble+ 721 823 1.07/2.4*
FCAW, weld metal 639 709 -
+ Test conducted on GCHAZ obtained by Gleeble simulation
* M relative to parent plate strength

Fracture toughness tests

Square and rectangular section single edge notch bend specimens (SENB BxB and Bx2B, where B = 25mm) were prepared from the QT1 plate, and square section specimens (B = 23mm) from the QT2 and normalised plates. Two notch orientations were employed. Specimens were notched in the through-thickness direction (TTN) and surface notched (SN) from the original plate surface. Notching and fatigue precracking was conducted with the intention of placing the final crack tip into the grain coarsened HAZ, as close as possible to the weld fusion boundary. With the TTN specimens, the crack depth to specimen width ratio (a/W) was nominally 0.5. With the surface notched specimens a/W was in the range 0.3 and 0.5 for tests on HAZs in the QT1 plate, and 0.5 in the QT2 plate.

The fracture toughness specimens were instrumented with a pair of clip gauges mounted on knife edges bolted to a pair of steel shims. The shims were attached to the notch mouth by short laser welds. This arrangement enabled crack mouth opening displacement (CMOD) to be estimated. CMOD was then used to calculate J (designated J CMOD) in accordance with equations developed by Kirk and Dodds. [4]

They are considered by the present authors to be more appropriate for the assessment of welds than those given in standards (e.g. BS 7448, ASTM E1290), because they are valid for a wider range at a/W (0.05 to 0.7), and J is more accurately estimated because it is derived directly from CMOD rather than from an estimate of load line displacement. Although the equations are for homogeneous material, they have been shown to be applicable to the assessment of welds and HAZs for a wide range of strength mismatch conditions. The J CMOD values have been transformed into more familiar stress intensity factor (K) values using the standard plane strain transformation.

An important feature of the testing procedure was post-test metallography conducted on the fracture toughness specimens. This ensured that results analysed only represented those from specimens in which the fatigue crack tip was located in grain coarsened HAZ (GCHAZ) or fusion boundary. In addition, post-test metallography enabled identification of the microstructure in which fracture initiation occurred.

Results and discussion

HAZ hardness and tensile properties

Hardness traverses were conducted across the welds at approximately mid-plate thickness and the results are presented in Fig.1 and 2. In the QT1 steel, HAZ softening had taken place in the overmatched weld (M=1.51), see Fig.1. In both the QT2 (M=0.95) and QT2N (M=2.15) steels, hardness in the GCHAZ close to the fusion boundary was higher than in the weld metal or parent plate, see Fig.2. Some HAZ softening is observed in the outer, intercritical HAZ of the QT2 steel.

Gleeble simulation of the GCHAZ was undertaken so that a direct estimate of the tensile strength could be made. The peak temperature achieved was 1350°C and the cooling time between 800°C and 500°C was 32s for the SAW welds (in QT1 steel) and 8s for the FCAW welds (in QT2 steel). Tensile properties, measured after post-weld heat treatment, are compared with the parent plate and weld metals in Table 2. Despite the weld metal strength overmatch in the QT1 steel, the HAZ strength undermatched the strength of the parent plate by a factor of 0.8. In contrast, the HAZs in the FCAW welds overmatched both the strength of the QT2 and QT2N steels by factors of 1.07 and 2.4, respectively.

Fig.1. Hardness traverse for welds made in QT1 steel
Fig.1. Hardness traverse for welds made in QT1 steel
Fig.2. Hardness traverse for welds made in QT2 and QT2N steels
Fig.2. Hardness traverse for welds made in QT2 and QT2N steels

Fracture toughness

The fracture toughness results from specimens where post-test metallography confirmed that the GCHAZ/fusion boundary had been tested by the fatigue crack tip, or in which initiation occurred, are given in Fig.3 to 6. Results from specimens in which the crack tip sampled other microstructures have been excluded. Despite scatter some clear trends are apparent, which confirm that strength mismatch does affect HAZ fracture toughness. To clarify this perception, the maximum likelihood (mml) procedure, developed in the European SINTAP project, [5,6] was applied to the data. The transition curves shown in the graphs represent the 50 th percentile estimates of fracture toughness based on the mml procedure.

The mml procedure assumes that fracture toughness follows a three parameter Weibull distribution with a shape parameter of 4 and shift or threshold parameter of 20MPam 0.5. Furthermore, the shape of the transition curve is uniquely defined by the Master Curve. The mml procedure is particularly suited to inhomogeneous materials and is intended to provide a realistic lower bound estimate of fracture toughness. A censoring and curve fitting process is applied to the data to: i) exclude results which do not fail under small-scale yielding conditions, ii) exclude results which represent the upper tail of the toughness distribution (these results are considered to be unrepresentative of lower toughness microstructures present). In this way, the transition curve is uniquely defined by a temperature T 0, for a median (P f=0.5) fracture toughness of 100MPam 0.5 for a reference, 25mm thick specimen. The fracture toughness (K mat) transition curve is defined by:

[1]
[1]

where:

T    = temperature, °C
B    = specimen thickness, mm
P f  = probability of the specimen failing at a given K mat. (In Fig.3 to 6, P f = 0.5)

Figures 3 and 4 show that with both SN and TTN fracture toughness specimens, weld metal strength overmatching (M=1.51) appears to result in an increase in transition temperature by 35 to 70°C compared with undermatching (M=0.65). However, the cause of this reduction in toughness is considered to be local strength undermatching in the GCHAZ. It is hypothesised that when a crack is located in the GCHAZ, strain will be concentrated into this narrow, lower strength region (e.g. QT1 steel in the overmatched condition, see Fig.1). High triaxial stresses will then develop hastening the onset of cleavage. On the other hand, when the strength of the GCHAZ is higher than in adjacent regions (such as the weld metal of the undermatched weld, M=0.65, in the QT1 steel, see Fig.1), straining is shed into weaker material, which is wider than the HAZ. Triaxial stresses therefore develop slowly in the GCHAZ and inhibit the onset of cleavage.

Where the GCHAZ yield strength is higher than that of the weld metal and parent plate, as in the QT2 and QT2N steels (see Fig.2), the opportunity for high triaxiality in the GCHAZ is diminished, and higher fracture toughness is measured. Strength mismatch between the weld metal and parent plate has a smaller effect on HAZ toughness. This appears to be confirmed by the results from the QT2 and QT2N steels shown in Fig.5 and 6. However, overmatching (M=2.15) appears to be marginally beneficial compared with undermatching (M=0.65) because straining occurs preferentially in the weaker parent plate. This is illustrated by Fig.7 which shows a section taken through the fracture initiation point in a specimen from an overmatched weld tested at -40°C.

Fig.3. Transition curve from surface notched HAZ specimens (BxB) in QT1 steel
Fig.3. Transition curve from surface notched HAZ specimens (BxB) in QT1 steel
Fig.4. Transition curve from through thickness notched HAZ specimens (Bx2B) in QT1 steel
Fig.4. Transition curve from through thickness notched HAZ specimens (Bx2B) in QT1 steel
Fig.5. Transition curve from through thickness notched HAZ specimens (BxB) in QT2N (M=2.15) and QT2 (M=0.95) steels
Fig.5. Transition curve from through thickness notched HAZ specimens (BxB) in QT2N (M=2.15) and QT2 (M=0.95) steels

A shear step links the fatigue crack tip, located on the fusion boundary, to the fracture initiation in the outer regions of the HAZ towards the parent plate. However, because in this series of tests the yield strength always overmatches the strength of both weld metal and parent steels, the effect of strength mismatch between parent steel and weld metal on HAZ fracture toughness was small. Weld strength mismatch of 0.95 and 2.15 resulted in HAZ transition temperatures which did not differ by more than 15°C.

Fig.6. Transition curve from surface notched HAZ specimens (BxB) in QT2N (M=2.15) and QT2 (M=0.95) steels
Fig.6. Transition curve from surface notched HAZ specimens (BxB) in QT2N (M=2.15) and QT2 (M=0.95) steels
Fig.7. Fatigue crack tip in fusion boundary of surface notched specimen with cleavage initiation at edge of transformed HAZ in QT2N steel, M=2.15, (K J=280MPam 0.5 at -40°C)
Fig.7. Fatigue crack tip in fusion boundary of surface notched specimen with cleavage initiation at edge of transformed HAZ in QT2N steel, M=2.15, (K J=280MPam 0.5 at -40°C)

Conclusions

Fracture mechanics tests conducted on ferritic steel welds with yield strength mismatch between weld metal and parent plate ranging from 0.65 to 2.15 have shown that GCHAZ fracture toughness is more affected by the local strength of the GCHAZ rather than differences in strength between weld metal and parent plate. Where GCHAZ strength undermatched the strength of both the weld metal and parent plate, GCHAZ fracture toughness was reduced. For the QT1 steel, the HAZ fracture toughness transition temperature was increased by 35 to 70°C compared with a condition where GCHAZ strength was higher than that of the weld metal. On the other hand, where GCHAZ yield strength overmatched the strengths of both the weld metal and parent plate (as in the QT2 and QT2N steels), strength mismatch between the weld metal and parent plate of 0.95 and 2.15 had little influence on GCHAZ fracture toughness. In this case, the transition temperatures did not differ by more than 15°C.

Acknowledgement

The authors are pleased to acknowledge that this work is derived from projects undertaken by them for a Group Sponsored Project 'Development of fracture assessment and testing procedures for HAZ cracks' managed by EWI, USA, and the Brite-Euram project on Structural Integrity Assessment Procedures for European Industry managed by British Steel (now Corus).

References

  1. Thaulow, C. and Toyoda, C. (1997). Mismatching of interfacing and welds - Ed K-H Schwalbe and M Koçak, GKSS Research Center Publications, Geesthacht, Germany.
  2. Ohata, M., Minami, F. and Toyoda, M. (1997). Ibid.
  3. Koçak, M et al (1989). Proc. 8th Int. Conf. OMAE-ASME. The Hague, the Netherlands, pp.623-633.
  4. Kirk, M.T. and Dodds, R.H. (1993). Journal of Testing and Evaluation. JTEVA, Vol.21, No.4, pp.228-238.
  5. Structural Integrity Assessment Procedures for European Industry - SINTAP - Final Procedure European Union Brite-Euram Programme BE95-1426-BRPR-CT95-0024 (1999).
  6. Pisarski, H.G. and Wallin, K. (2000). Engineering Fracture Mechanics, 67, pp.613-624.

For more information please email:


contactus@twi.co.uk