Electron beam welded butt joints - a fatigue study

TWI Bulletin, September - October 1995

Siak Manteghi is a Principal Project Engineer in TWI's Structural Integrity Department. After gaining a BSc (Hons) in Civil Engineering from UMIST he continued his postgraduate studies there, working on the stress analysis and fatigue properties of welded joints. He was awarded a PhD for this in 1991. In 1988 he joined the then Fatigue Department at TWI, and has since maintained his interest in the structural integrity of welded and bonded joints.

Electron beam (EB) welding is now used widely in industry, often to produce engineering components which are subjected to fluctuating service loads. Siak Manteghi reports the results of a recent investigation showing that EB butt welds can perform at least as well as similar joints produced by the conventional arc welding processes.

Use of EB welding to produce butt joints in engineering components is increasing. These parts are often subjected to cyclic service loading which has made it vitally important for the designer to understand the factors affecting fatigue performance of such joints. Notable examples of EB welding applications are the variety of transmission components in aircraft and automotive industries.

Present UK fatigue design rules included in BS 7608:1993 ^[1] do not generally make a distinction between joints produced by different welding processes. However, the code recognises that the shape of overfill profile in transversely loaded butt welds is an important factor influencing their fatigue strength. Consequently, some factors affecting overfill profile are taken into account in assigning the relevant design classification for such joints. These include the welding process (manual metal arc welding v. submerged-arc welding), welding position (flat v. positional), and site welding (as opposed to shop welding).

Given that the guidance in BS 7608 is based on experimental data obtained from specimens welded by conventional arc welding processes, EB welds are, strictly speaking, outside the scope of the code. Furthermore, EB welds are by their nature single sided welds, which are not normally classified unless they are made onto backing bars. Therefore, design data based on suitable standard fatigue specimens, and incorporating EB butt welds with or without backing bars, are needed to fill the gap in the existing design rules.

A project completed recently ^[2] has built on previous TWI work on the fatigue strength of EB butt welds in C-Mn steels ^{[3], [4]} and Ni-Cr-Mo low alloy steels ^{[5], [6]} . These had concentrated on transverse butt welds without backing bars, mainly in the as-welded condition. The present work was conducted to generate more design data from that detail, and also to provide data from transverse butt welds made onto backing bars, which were not included in any of the previous studies. Additionally, the effects of a cosmetic pass and various grinding procedures, to improve the fatigue strength of EB butt welds, were studied.

Experimental

Material

The test specimens were fabricated from two C-Mn structural steel plates, both 12.5mm in thickness and conforming to BS 4360:1986 Grade 50D.

Welding procedure

All the specimens were produced at TWI using a 75kW, 150m ³ vacuum chamber EB welding machine. The machine was fitted with a directly heated DH Mk14 triode gun, mounted horizontally. Butt welds were made between 500 x 370mm plates, to produce panels measuring 740 x 500mm with welds parallel to the shorter edge. All welds were perpendicular to the plate rolling direction. Welds were made in the horizontal-vertical position as square butt joints with a deliberate total offset of 0.5° introduced to counteract distortion. Where used, backing bars were attached by TIG tack welds before EB welding. All evidence of tack fillet welds was removed on preparation of the final test specimens.

The welds can be grouped into three types:

• butt welds without backing bars,
• butt welds made onto backing bars
• and butt welds (without backing bars) which had received a second cosmetic welding pass, to remelt the face bead, after allowing the first pass to cool to ambient temperature.

For welds receiving a cosmetic pass, a low power defocused electron beam was used locally to melt and blend the face bead resulting from the first pass, see Fig.1. The intention was to extend the joint fatigue strength by achieving a smoother weld profile compared with untreated joints.

Specimen fabrication

Waisted specimens incorporating transverse butt welds with or without backing bars were produced, as illustrated in Fig. 2 and 3. Each welded panel was first sawn into parallel-sided specimens. The specimens without backing bars were then flame cut, using a template, to produce the required specimen dimensions. The specimens with backing bars were cut to the required size in a computerised numerically controlled (CNC) mill.

Grinding

In addition to the three types of as-welded specimen (series A-C), four other types (series D-G) were produced by grinding, as summarised in the Table above.

Uncontrolled grinding, aimed simply at flush-grinding the weld bead, was performed using a disc grinder. Care was taken to ensure that all grinding marks were perpendicular to the weld.

A number of butt welds made without backing bars were treated in this way (series D).

In some cases, the weld toes were first treated using a pencil grinder to ensure removal of any discontinuities at the toe of the face bead or at the edge of the root bead (series E). This treatment, referred to in the report as full grinding, left grooves typically 0.5mm deep at the original weld toes, see Fig. 4.

Some of the specimens with backing bars received uncontrolled grinding with the disc grinder, involving removal of the backing bar and flush grinding the root side (series F), or in some cases both sides (series G).

Fatigue testing

All fatigue tests were performed in air at ambient laboratory temperature under constant amplitude sinusoidal axial tension loading. Tests were run in load control using various hydraulic and servo-hydraulic fatigue testing machines, at frequencies in the range 3-17Hz. The machines were equipped with wedge jaws which enabled the plate specimens to be gripped. The stress ratio, R, ie the ratio of the minimum to maximum cyclic applied stress, was kept at approximately 0.1 for all tests.

In total 47 specimens were fatigue tested, covering seven different conditions (three as-welded and four incorporating various degrees of grinding as described above). The main characteristics of each condition, and the number of corresponding fatigue tests, are shown below.

Metallography

A length of 150mm was removed from two of the EB welds in the as-welded condition, one with backing bar and one without. These were sectioned to investigate the presence of welding related flaws, particularly near the toes of the face and root beads.

From each sample 10 sections were taken, transverse to the weld length, at approximately equal spacing. The sections were prepared for metallographic examination. Size, location, and nature of any flaws within the weld metal or on the fusion boundary were recorded.

Results and discussion

Untreated specimens

The results from all as-welded specimens with or without backing bars (but excluding those having received a cosmetic pass) are plotted, together with data obtained at TWI previously ^[3,4] in Fig. 5 and 6. Figure 5 represents failure from weld toes on the face side, while Fig. 6 shows data obtained from specimens which failed from the root.

Out of the 13 specimens without backing bars, 10 failed at the joint, and of these, six had fatigue cracks initiating at the edge of the root bead. This is in line with previous experience at TWI ^[3-6] which had also highlighted the critical role of the root condition in determining the joint fatigue strength.

None of the specimens with backing bars failed at the root. Of the 12 specimens tested in this series, 10 failed in the plate from the toe of the face bead, the other two remained unbroken after 10 ⁷ cycles. Interestingly, in many applications of EB welding the beam penetrates into solid material producing a blind weld similar to the detail tested here.

The fact that no root failure was observed in this series may well reflect the existence of beneficial, ie compressive, residual stresses in that location in the particular specimens tested here. The residual stresses were not measured in the present project, and in any case the residual stress fields will not necessarily be the same at the root of all EB welds. More data, preferably obtained from stress relieved EB joints are therefore needed to assess the likelihood of root failure in joints made on backing bars.

The results shown in Fig.5 and 6 suggest BS 7608 Class E as a reasonable design class for transverse EB butt welds, both for failure from the toe of the face side and (in the case of welds made without backing bars) for failure from the root.

Effect of cosmetic pass

The test results from the present and previous TWI studies ^[3] on the effect of cosmetic pass are shown in Fig. 7. Although the dataset is too small to form a basis for firm design recommendations, it is encouraging to note that all the data lie well above the BS 7608 Class D design line.

Effect of grinding

The test results from the present and previous TWI studies ^[3] on ground EB butt welds are plotted together in Fig. 8. They show clearly that uncontrolled flush grinding, ie removing the weld overfill so that the joint area is flush with the surrounding parent plate, will not always lead to significant improvements in fatigue life. This is because the treatment cannot guarantee the removal of all discontinuities associated with the original weld face/root beads. In that case, such a discontinuity will act as the site for fatigue crack initiation. In contrast, the three series of specimens representing full grinding ( ie removing discontinuities at the weld toes using a burr grinder, before flush grinding) had all resulted in significant improvements in fatigue strength, to Class D at least.

Conclusions

• Single sided transverse EB butt welds in C-Mn steel, produced without backing bars, can be assigned to Class E in the current UK fatigue design rules. This would cover the avoidance of fatigue failure in the plate, either from the toe of the face bead or from the edge of the root bead.
  
  
• Transverse EB butt welded joints in C-Mn steel made on backing bars always failed from the toe of the face bead. For this potential failure mode, the detail can be included in Class E.
  
  
• It is tentatively suggested that EB butt welds receiving a cosmetic second pass may be included in Class D. More work is needed to confirm this.
  
  
• Uncontrolled flush grinding will not necessarily result in significant increases in fatigue life.
  
  
• Burr grinding to remove all weld discontinuities, followed by disc grinding to remove the weld overfill, is likely to upgrade the joint fatigue strength to at least Class D.
  

Recommendations

The results from EB welds are encouraging, suggesting that in all the conditions studied EB welds are at least as good as, and in some cases significantly better than, similar joints produced by the conventional arc welding processes. Data will be made available to the relevant code writing bodies. In particular, it is hoped that the data will be taken into account during any future revisions of BS 7608.

References