Just how much does it matter?

TWI Bulletin, November - December 2003

Inadvertent welding procedure changes - investigating their influence on weldment properties...

Jack Still worked in the steel industry as a metallurgist, which involved advising customers on welding of clad steels. He transferred to the fabrication industry and worked for Motherwell Bridge, Babcock and Wilcox and eventually Redpath Dorman Long. During this period he was involved with the fabrication of pressure vessels, tanks, boilers, offshore structures and modules. In 1978 he joined the offshore oil and gas industry and during the last 24 years worked for Britoil/BP, Shell Expro and eventually Amerada Hess on projects involving jacket and production modules, construction of an FPSO and refrubishment of mature structures.

John Weston's early industrial experience involved the manufacture of a wide range of engine components for the OEM and replacement markets in New Zealand. He returned to TWI in 1975 where the emphasis has been on project and consultancy work related to robotics and mechanised welding, manufacturing systems and general welding engineering support.

David Abson gained his first degree at Cambridge University, and then a Masters degree and a PhD at the University of Sheffield. After working as a post-doctoral research fellow at McGill University Montreal and the University of Wales, and a period of work for Westinghouse in the USA, he joined TWI's Materials Department.

Welding procedures, qualified in 1999 in accordance with the requirements of BS 4515, were submitted by a jacket fabricator, as being suitable for welding riser girth welds in a new offshore installation. It was planned to install the full complement of risers during initial construction, but to use only some of them for the initial field development. The remainder were not to come into service until a second field had been developed. As Jack Still, John Weston and David Abson report all the risers were hydro-tested after fabrication at a pressure, which satisfied the design requirements for the initial development.

When the second field was developed the operating pressures were found to be considerably higher because of a higher wellhead shut in pressure. It was decided to use an Engineering Critical Assessment (ECA) approach to verify the acceptability of the riser girth welds for use at the higher test pressure. To perform such an assessment, mechanical property data of the weldment was required.

Since no Crack Tip Opening Displacement (CTOD) tests had been specified for the original Weld Procedures, which involved Gas Tungsten Arc Weld (GTAW) root and hot pass and Shielded Metal Arc Weld (SMAW) fill and cap, it was necessary to establish a programme to generate the required data. Surplus material from the original riser fabrication programme was available. To produce the required welds, it was decided to subcontract the work, which was to be carried out in accordance with the original Weld Procedure Operating Instructions (WPOI).

Unfortunately, the subcontractor involved (Subcontractor A) failed to produce the trial welds in accordance with the original WPOI, and the resulting Heat Affected Zone (HAZ) CTOD properties were lower than expected. The lowest HAZ CTOD values recorded for welds designated A and A1 were approximately 0.02 mm at -10°C and 0.12 mm at +4°C respectively. It was decided therefore to repeat the trial welds using a different subcontractor (subcontractor B), with the work being carried out under the supervision of the Operator's Project Engineering Group. The lowest HAZ CTOD value recorded for this joint, weld B, was 0.34mm at -10°C and 0.46mm at +4°C. This paper reports on the welder performance differences that resulted in the variation in the properties of welds produced in trial and production welding. See Table 1.

Table 1

Pipe material

The pipe material specified for the risers and used for the trial welds was to API 5L X65. The seamless pipe to API 5L X65 of 15.5mm wall thickness and 376mm internal diameter was used for both procedure development and the trials reported. The material used in the trials was analysed and was found to be within specification requirements, Table 2.

Table 2 Chemical analysis of the riser pipe material 

Welding consumables

The welding consumables, which were used for the trial welds were nominally the same as those used in the production welds. They were to the same specification and from the same manufacturer, and consisted of the following:
For the root and hot passes the GTAW process was used with filler wire Filarc PZ6512 (AWS A5.18:ER70S-G [nearest] ).
For the SMAW fill passes the electrode Oerlikon Tenacito 38R (AWS A5.5 E7018-G) was used.

Trial weld procedures

The welding programme involved two independent fabrication subcontractors performing trial welds. Each contractor was provided with material, details of the weld preparation and a copy of the appropriate WPOI.

Weld Procedure Qualification Record (WPQR)

The SMAW Weld Procedure Qualification Record (WPQR), shown Figure 1, the basis for a welding procedure operating instruction (WPOI) that was used in the original riser girth welds. The WPQR had been produced in accordance with the requirements of BS 4515. The weld procedure consisted of a single V butt weld (60° included angle) between pipe (15mm W/T, 406mm OD) and flange material. Figure 1 summarises the welding parameter data contained in the WPQR.

Weld Procedure Operating Instruction (WPOI)

The subcontractors were issued with WPOI, and this was passed to the welders involved in welding the riser girth trial welds. The essential details of WPOI are listed in Table 1. These welding parameters were those which were expected to be used in the trial welds.

Production of trial welds

The trial joints produced consisted of:

• Weld A - GTAW root and SMAW fill and cap, produced by subcontractor A. This joint was made approximately in accordance with the WPOI; but specified a 50° included angle single bevel butt joint ( ie half-K type preparation).
• Weld A1 - GTAW root and SMAW fill and cap produced by subcontractor A. This was a repeat test, according to WPOI; and again specified a 50° included angle single bevel butt joint.
• Weld B - GTAW root and SMAW fill and cap produced by subcontractor B. This joint was produced approximately according to the WPOI; and again the specification was for a 50° included angle single bevel butt joint, ie half-K type preparation.

Metallurgical Investigation

A macrosection was taken from each weld, and prepared for metallographic examination. The etchant used was Nital. The exact location of initiation of unstable fracture was identified on the low toughness CTOD specimens by examination in a scanning electron microscope. A metallographic section was then taken through the initiating region so that the microstructure could be examined.

Trial welds

The SMAW trial welds were produced with the parameters listed on the WPOI, as set out in Table 3.

Table 3 Summary of welding variables

Each of the trial welds were witnessed by an inspector appointed by the subcontractor A. The responsibilities of the inspector involved establishing that the bevel angle was correct and recording both the welding parameters and run sequence.

A review of the weld records and the subsequent macro examination indicated that the subcontractor had not precisely followed the WPOI for trials weld A and weld A1. In particular it was noted that the bevel angle was not 50° as specified but varied between 40° and 45°. Also, the welders appeared to have deposited the first run of each layer on the bevel face, which resulted in an excessive gap between the deposited weld and the vertical face of the weld preparation, as, illustrated, particularly for layer 4, in Fig.1. This figure also illustrates the welding sequence adopted by the subcontractor.

Subsequently weld B was produced by the second subcontractor B, and supervision during welding was carried out by a member of the Operators Project Engineering Group.

Following welding, studies were made of the welding procedures, the weld macro and micro sections. Also the joint mechanical properties were determined.

Results

Macro and Micro Examination

Macro sections of trial welds A, A1 and B, were examined; see Fig.2. Sketches of the bead placement taken from these sections are shown in Fig.3, which gives run numbers and the general run sequence.

Metallographic examination of weld A revealed a heat affected zone (HAZ) having regions of inter-critically heated grain-coarsened HAZ (ICGCHAZ) at the point of fracture initiation. These regions commonly act as local brittle zones (LBZs), and are the result of the reheating of grain-coarsened material (GCHAZ) by subsequent passes. Detailed metallographic examination revealed that these regions contained grain boundary decorations of an unresolved phase, MA (martensite-austenite), It was apparent that these regions were responsible for the low fracture toughness values. An example of the microstructure illustrating the ICGCHAZ is presented in Fig.4.

Welding parameters

The welding parameters for the three test welds and the WPQR, were examined and compared. The collated information is presented in the following: Table 4 Comparison of Amperage, Voltage and Heat Input between WPQR, A, A1 and B.

Table 4 Comparison of amperage, voltage and heat input between WPQR, A A1 and B

In examining the WPQR and the developed WPOI, it was noted that the welding current range used for the GTAW root run and hot pass had been reduced in the latter from 112-150 amperes to 90-140 amperes, see Table 1. Furthermore, the upper limit of voltage specified on the WPOI is slightly higher than that actually recorded on the WPQR (13V compared with 11.6V). For the fill and cap runs, the currents and voltages were more closely aligned. The arc energies specified and recorded were also closely aligned.

When comparisons are made between the test welds A, A1, and B, table 4, the following similarities were noted:

• Welding currents for the GTAW root runs were higher than specified in the WPOI. However, these currents were similar to those used in generating the WPQR.
• The heat input levels used in the root runs tend to be higher than those used for the WPQR, although they were within the range specified (see Table 4 and Figure 5).
• Welding currents for the SMAW fill and cap runs are more closely aligned with both the WPOI and the WPQR.

Discussion

An examination of the chemical compositions of both the available parent material and the consumables that were used to make welds A, A1 and B did not reveal any significant variations that might account for the differences in joint properties. The levels of vanadium in the parent pipe (0.06%) were not considered to be detrimental to the fracture toughness in the HAZ. Furthermore, the consumable composition, from reputable suppliers, was expected to be consistent.

Therefore, with this consistency, and in the absence of post-weld heat treatment, the changes in mechanical properties were considered to be influenced by re-heating effects, which would be caused by the layer thickness and heat input. In general, the thinner the layer deposited, the more runs will be made, the lower the heat input, the greater the degree of re-heating and refinement of the weld metal and HAZ, and the higher would be the expected toughness. This was approximately in agreement with the observed results. For example, Weld B, with the lowest heat input and greater number of layers, was found to have superior properties to the A welds.

It was considered that the weld layer thickness was more dependent upon the process variables than upon the choice of joint preparation ( eg single V butt joint compared with half-K butt joint). Decreases in the welding current and increases in the travel speed would be expected to result in decreased layer thickness. In this context, it is of note that options for such parameter variation exists within the weld procedures. It is also of note that the arc energies and the welding parameters for the bulk of the welds, ie the fill regions, were found to be within the ranges allowed in the WPOI, see Figure 6.

Another factor of importance was the sequence of weld build-up and weld bead placement. The poor placement of beads was aggravated, in the case of the A and A1 welds by the narrower weld preparation (40° - 45°) compared with the specified 50° and the generally higher deposition rates used.

Having considered all of these factors, it was concluded that if the production welds had been made in accordance with the WPOI, then the parameters found in the test welds that were welded currently would be representative of those found in production. Therefore, the range of mechanical properties observed in these welds would also be representative of those found in the production welds, and were deemed to be valid for the application.

This study reinforced the importance of fully documented weld procedures, and of having good control in their implementation when intending to replicate welding and mechanical test data.

Conclusions

The weld procedures and macro specimens from the test welds relating to riser girth welds were examined in some detail. These procedures were compared with those believed to have been used in the deposition of production girth welds in the riser. These studies led to the following conclusions:

• The arc energies and welding parameters of the bulk part of the welds, ie the fill regions, were found to be within the ranges specified in the welding procedure WPOI. On this basis it was considered that the range of mechanical properties observed in the test welds would be representative of the properties expected from the production welds and were deemed to be adequate for the application.
  
  
• The level of vanadium (0.06%) present in the pipe material was not considered to be detrimental to the fracture toughness in the weld HAZ.
  
  
• In weld A1 the presence of regions of inter-critically reheated grain-coarsened HAZ that contained grain boundary regions of an unresolved phase were responsible for the observed low fracture toughness results.
  
  
• Precise following of the WPOI is vital in replicating mechanical property values.