Joining for Titanium Risers and Flowlines

Originally presented at a seminar - 'Titanium Risers and Flowlines' - held in Trondheim, Norway, 17 February 1999; published by SINTEF, Trondheim, Norway, 1999

Abstract

Current technologies for the exploitation of deepwater oil and gas resources have reached their limit and new materials are required for the development of future deepwater fields. Titanium alloys have been identified as offering the necessary combination of low modulus, high specific strength and exceptional corrosion resistance for future deepwater projects. Ultimately, the success of titanium risers and flowlines lies in the production of high quality joints. Fortunately, titanium is one of the easiest of metals to weld by both fusion and solid state processes. Even so, exceptional quality is required to maximise joint performance. Weld metal porosity and contamination are amongst the most significant problems which can be encountered in titanium weldments, since their presence can be very difficult to detect. Porosity is particularly insidious and its control, especially that near the surface, is key to increasing maximum allowable cyclic stresses. A greater understanding of the formation and sources of porosity in titanium welds has been gained and this information may prove crucial to the development of welding strategies that eliminate the most damaging porosity defects. Methodologies have been identified which offer hope in identifying significant porosity and contamination levels, but further work is required to develop and establish these techniques.

Those processes that offer the best combination of maturity (especially as regards welding titanium) and potential quality are TIG, MIG, keyhole plasma, reduced pressure electron beam, laser, rotary friction and radial friction welding. All of these processes are capable of producing high quality joints, but the friction welding processes and possibly keyhole plasma welding offer the particular advantage of immunity to weld metal porosity. All of the alternatives to TIG welding also offer increased productivity.

Introduction

Titanium has many attributes that are desirable for riser systems including high specific strength, exceptional corrosion resistance in most environments and low modulus. Generally, most titanium alloys, including grades 5, 23, 24 and 29 (the Ti-6Al-4V-base alloys), are quite easy to weld either by fusion or solid state processes. For example, titanium is one of the easiest of metals to weld by the TIG process: the weld pool is fluid and its combination of density and surface tension enables good control of surface profile and penetration, even when unsupported. Titanium can also be joined readily by friction processes and is mostly immune to many of the weld cracking problems that plague ferrous fabrications (fusion and solid state welded). Despite these beneficial characteristics, titanium is considered by many to be difficult to weld, mostly because it is normally only handled by specialist fabricators and due to its particular requirements with regard to gas shielding. Porosity can also be encountered which, although irrelevant in most instances, can impair performance in fatigue critical applications. These latter two negative aspects do not present insurmountable difficulties, provided that necessary preventative measures are taken. Although weldment fatigue properties are often influenced most by weld geometry, this is unlikely to be applicable for titanium risers and flowlines, for which geometry-specific stress-raisers will probably be removed. The present paper is intended to highlight those aspects of titanium which, in practice, will most influence weld quality (porosity and contamination) and identify procedures, welding processes and NDE requirements which are best able to guarantee superior joint performance. A variety of welding processes is considered, including those used currently and those that show promise for future application.

Fusion processes

Porosity

To date, only one titanium riser has been commissioned, but it is likely that the procedures adopted for the Heidrun riser will, in the short term at least, be used for any future projects. In situations where the weld reinforcement is left intact, fatigue cracking can be reliably predicted to initiate from the weld toe. In all likelihood, however, all geometry-specific stress-raisers, such as weld toes, will be removed from titanium riser systems in order to maximise the fatigue performance of the joints. Under these circumstances fatigue strength can be lowered significantly by the presence of weld porosity, especially that near the surface. The influence of pores in titanium welds has been known for some time. Lindh and Peshak (1969) ^[11] gave S-N curves for machined flush TIG welds in Ti-6Al-4V, showing expected performance according to the presence of porosity and its location. These curves are reproduced in Fig 1. The lower curve shows the influence of pores closer than two pore diameters to the surface, whilst the middle curve demonstrates the dramatically improved performance that can be gained by eliminating near surface defects. In the absence of other weld defects and geometry-specific stress-raisers, cracking always initiated from weld pores when present, regardless of pore size. Even pores as small as 50µm diameter were found to have initiated cracking, although their effect on performance was less marked than for larger defects. The conclusions of Lindh and Peshak ^[11] still apply today. Unduly restrictive maximum specified pore sizes in welding codes may not have any profound effect on fatigue performance. Indeed, the removal of such defects and subsequent weld repair may be more detrimental to fatigue performance than the original defect. Near-surface pores, however, should be eliminated and ultimately the performance of a machined-flush titanium weld may depend on the availability of NDE methods which can detect very small pores closer than two pore diameters to the surface. Until such methods have been proven, it seems unlikely that design codes can be moved above the lower curve shown in Fig 1, for welds produced using processes known to be susceptible to porosity.

Fig.1. Lindh and Peshak's (1969) ^[11] fatigue life curves for grade 5 titanium TIG welds containing pores. Some data is shown on the figure for small pores (50-150µm diameter), mostly away from the surface

Porosity in titanium fusion welds can be formed for a variety of reasons, but the most insidious is undoubtedly the profound influence of the condition of the joint surfaces. A more detailed description of weld metal porosity and its prevention in titanium fabrications is given elsewhere (Smith and Gittos, 1998a) ^[14] , but the evolution, coalescence and entrapment of hydrogen bubbles during solidification is the most likely cause. The joint surfaces should be scrupulously clean prior to welding since even a fingerprint can lead to porosity, but even this precaution is insufficient. The rutile (TiO₂ ) scale that imparts such considerable corrosion resistance is hygroscopic and adsorbs moisture from the environment. The hydrated layer provides a further source of hydrogen and steps should be taken to minimise its presence. This includes using procedures that reduce surface roughness and the hydration state (i.e. x in TiO₂ . xH ₂O) prior to welding. In principle, machining without aqueous lubricants and welding in the same 24h period is advised, although acid pickling may be used successfully on 'old' joint surfaces.

One interesting facet of porosity in titanium weld metal is its seeming dependence on the presence of a prior joint surface. Autogenous melt runs on the surface of titanium plates are notoriously difficult to produce with porosity, seemingly independent of the presence of grease, oil and other contaminants. Likely reasons for this are discussed elsewhere (Smith and Gittos, 1998a) ^[14] , but it is tempting to speculate that this behaviour could be exploited to good effect in eliminating near-surface pores in fusion welds. Much development work is necessary, but, potentially, many of the concerns regarding porosity could be countered simply by applying autogenous TIG dressing passes over the completed weld surface.

Contamination

Elements that form interstitial solid solutions such as H, C, N and O increase the strength and decrease the ductility and toughness of titanium, leading ultimately to a brittle material. At elevated temperatures, Ti absorbs and reacts with all of these elements at a rate bearing an exponential relationship with temperature. Contamination from carbon is readily avoidable by degreasing prior to welding and the methods for avoiding hydrogen in the weld pool have been discussed in the previous section. Air is more of a problem: for the thermal cycles experienced in typical welds, surface-only oxidation of titanium occurs in the range ~250-850°C. More importantly, above ~850°C, O and N dissolve directly into the bulk material, potentially causing embrittlement. It is conventional for trailing gas shields to extend coverage down to a metal temperature of 250°C, to prevent the formation of coloured oxide films. When surface colour is used to monitor the quality of the shielding gas, a silver weld is a logical aim. This is because it is not possible to discriminate between coloured welds formed under impure gas (likely to be contaminated) and those made with a 'short' trailing shield (likely to be uncontaminated). Thus, in order to apply colour acceptance criteria, gas shields usually have to be more extensive than those which would be required to prevent embrittlement with clean inert gas. This includes the weld metal, both when it is molten and after it has solidified, HAZ and the parent metal adjacent to the weld. If protection is inadequate, severe embrittlement of the weld region can occur. Although fluxes have been developed for titanium welding, these have gained little commercial acceptance. In practice, the vast majority of titanium fabrication is performed employing either gas shielded processes with inert gas purge and auxiliary shielding to protect material behind and either side of the torch, or chamber welding (including by the electron beam process in a vacuum).

Despite the precautions taken to avoid contamination it is inevitable that the shielding arrangements will occasionally fail to fully protect the weld. It is, therefore, crucial that adequate non-destructive tests are available to assess the success or otherwise of the shielding. The easiest and almost universally adopted method is the application of colour criteria. This technique, although easy to apply, can give misleading results since the most detrimental form of contamination, i.e. entrainment of air into the torch shielding gas, will still give a silver weld if the trailing shield provides good protection. Furthermore, the most commonly cited sequence of weld colours has been proved incorrect (Gittos et al., 1993) ^[7] . In particular, the interpretation applied to dark blue and light blue should be exchanged in many codes; i.e. dark blue indicates better shielding than light blue. A rule of thumb guide to interpreting weld zone colour is given in Table 1. The full sequence of colours is rather complicated, progressing through first and second order colours, separated by a dull 'silvery hiatus' which itself could readily be misinterpreted as a contamination-free weld. Clearly the interpretation of weld colour is non-trivial and insufficient to guarantee the success of the shielding arrangements. A better method of inspection is required.

It should be noted that discoloration away from the weld bead does not necessarily indicate poor shielding. Indeed dark straw to blue 'tramlines' parallel to the weld bead are commonly encountered in fusion welds. Work performed at TWI has established the origin of these features, which are believed to be non-harmful and occur irrespective of the shielding integrity (Smith and Gittos, 1998b) ^[15] .

 

Table 1. A rule of thumb classification system for judging the acceptability of titanium welds

Colour of weld zone	Interpretation
Silver Light straw Dark straw Dark blue Light blue Grey blue Grey White (loose deposit)	Correct shielding, satisfactory Slight contamination, but acceptable Slight contamination, but acceptable Heavier contamination, may be suitable depending on service Heavy contamination, unlikely to be suitable Very heavy contamination, unacceptable Very heavy contamination, unacceptable Very heavy contamination, unacceptable

 

Hardness has been shown to provide a reasonable indication of contamination level in grade 2 titanium (Gittos and Scott, 1993; Harwig and Castner, 1997 ^[8] ), but is less easily interpreted for the more highly alloyed grades, such as grade 5 (Ti-6Al-4V). Work is currently in progress at TWI to develop more appropriate alternative NDE methodologies for detecting contamination in grades 2 and 5, including portable hardness and electromagnetic inspection techniques (Gittos, 1999) ^[6] . Initial results are promising and it is hoped that the work will enable the quality of titanium weldments to be assessed more reliably than has previously been possible.

TIG welding

The TIG welding process is an established technology for joining titanium and is the process with which most fabricators are familiar. The choice of TIG welding for risers and flowlines will no doubt be influenced by these factors, but the process does suffer several disadvantages for these applications. For pipe joining, in sizes appropriate in these applications (12-25 mm), the process is slow and many passes must be made to fill the joint. Furthermore, grains develop in the solidifying weld metal by epitaxial growth, giving a coarse grained structure with a grain size roughly proportional to the number of passes ( Fig 2). The coarse grains can increase the difficulty in detecting small defects using ultrasonic inspection techniques and so are generally undesirable. Despite a coarse grain size, mechanical properties compare favourably with those of the base material ( Table 2), in part due to the finer transformation products that develop in the weld metal. Porosity is a persistent problem in TIG welds which, together with the need for many weld passes, may necessitate a greater proportion of weld repairs to be performed than for the other processes discussed here.

Fig.2. A multipass TIG weld in 14mm thick Grade 23 titanium pipe

Table 2. Typical tensile properties for an as-welded multipass TIG weld in 14mm thick grade 23 titanium pipe

Zone	0.2% Proof Stress (MPa)	Tensile Strength (MPa)	Elongation (%)	Location of Failure	Maximum Hardness (Location) (HV10)
Cross-weld	-	915	-	Parent	325 (weld metal)
Weld metal	760	910	15	-	325
Parent	850	915	14	-	310

 

Productivity enhancement has been shown to be possible by the use of Ar-He torch shielding gases (Howes and Gittos, 1998) ^[9] , enabling a more rapid travel speed to be used than would be possible for a pure Ar shielding gas. Developments such as A-TIG and hot-wire TIG (Crement, 1993) ^[4] also offer scope for improved productivity.

MIG welding

The MIG process has not been applied as widely to titanium as it has been to ferrous and other non-ferrous alloys. Many of the reasons why MIG welding has not traditionally been favoured for titanium are historical and no longer apply. High currents are required for stable metal transfer and the poor wire surfaces produced originally caused rapid contact tip wear. More recently, these problems have been mitigated by power source developments and improvements in wire quality, respectively. The combination of modern inverter power sources with pulsed currents has made MIG welding feasible. Similarly, the improved surface finish of titanium wire has reduced contact tip wear such that the problem is barely apparent in development work, although it may become more of an issue in production situations. No assessments have been made of fatigue performance, but the static properties of MIG welds in titanium pipe compare favourably with those of TIG welds (Smith and Gittos, 1998c) ^[16] . An example of a typical multipass MIG weld is presented in Fig 3 and typical properties are shown in Table 3. Weld metal microstructures are very similar to those of TIG welds.

Fig.3. A MIG weld in 14mm thick grade 23 titanium pipe (two MIG fill passes, one TIG root pass)

Table 3. Typical tensile properties for an as-welded multipass MIG weld in 14mm thick grade 23 titanium pipe

Zone	0.2% Proof Stress (MPa)	Tensile Strength (MPa)	Elongation (%)	Location of Failure	Maximum Hardness (Location) (HV10)
Cross-weld	-	915	-	Parent	320 (weld metal)
Weld metal	720	860	9	-	320
Parent	850	915	14	-	310

 

Despite the aforementioned developments, MIG welding continues to be perceived as being lower quality than TIG and has only been utilised where superior quality was not required. MIG welds have been rumoured to be more susceptible to porosity than TIG welds, but this may be more a reflection of the general belief concerning quality in titanium MIG welds than being based on any fair comparison of the modern processes. Weld spatter can be a problem, but this can be reduced to some extent by the use of Ar-He shielding gas mixtures (Howes and Gittos, 1998) ^[9] .

Keyhole Plasma Welding

The keyhole plasma welding process offers great potential for joining titanium risers and flowlines, its main advantages being productivity and an apparently decreased susceptibility to porosity when compared with TIG welding. The process is broadly similar to TIG welding, but a copper alloy nozzle, through which flows a portion of the shielding gas (commonly referred to as the plasma gas), constricts the arc. A portion of the plasma gas is ionised forming a fully penetrating plasma jet. Single pass welds up to 20mm thick are feasible in the flat (1G) position, allowing dramatic productivity gains to be achieved (Boucher et al., 1994^[2] ; Ramsland et al., ^[13] 1997; Smith and Gittos, 1998c ^[16] ). Trials performed at TWI have shown that joint surface conditions, known to give porosity in TIG welds, show no such tendencies for keyhole plasma welds. Other advantages over TIG welding include a reduced weld zone grain size and the simple square butt joint geometry required by the process.

Keyhole plasma welds will probably require the application of a dressing pass (i.e. a TIG melt run) over the weld cap to eliminate the minor underfill that typically occurs ( Fig.4). This should pose little risk of porosity since autogenous melt runs very rarely exhibit pores. More extensive gas shielding is necessary for thicker sections since the process operates at greater heat input than TIG welding, requiring greater coverage of the joint and surrounding region. The main challenges of the process, as applied to pipe welds, are weld closure and positional welding. Gradual alterations of the process parameters are required for weld closure in order to avoid any defects, such as voids, at the stop position. By taking such precautions defect-free closure positions can be readily obtained (e.g. Fig.5).

Fig.4. A single pass keyhole plasma weld in 14mm thick grade 23 titanium pipe

Fig.5. The stop position of a single pass keyhole plasma weld in  14mm thick grade 23 titanium pipe

Although thickness up to 16mm have been welded successfully in the horizontal (2G) position, difficulties in the control of large weld pools may significantly reduce the maximum thickness that can be welded in 5G welds. Even so, predevelopment trials have demonstrated that 5G keyhole plasma welds are possible in titanium pipe (Schley, 1998) ^[17] . Microstructure and mechanical properties ( Table 4) are very similar to those of TIG welds, although those of multipass TIG welds tend to offer superior ductility.

Table 4. Typical tensile properties for an as-welded keyhole plasma weld in 14mm thick grade 23 titanium pipe

Zone	0.2% Proof Stress (MPa)	Tensile Strength (MPa)	Elongation (%)	Location of Failure	Maximum Hardness (Location) (HV10)
Cross-weld	-	915	-	Parent	335 (HAZ)
Weld metal	810	940	10	-	320
Parent	850	915	14	-	310

Reduced Pressure Electron Beam Welding

Electron beam welding is firmly established in the aerospace and other high performance sectors for which joint performance is critical. Narrow welds can be produced reproducibly with little distortion and very good control can be maintained on weld geometry. Whilst conventional EB welding seems inappropriate for use offshore and for many other requirements, reduced pressure EB (RPEB) welding processes (Belloni and Punshon, 1997) ^[1] have been developed which are suitable for welding large pipelines in-situ. Conventional EB welding is performed under a medium to high vacuum, whereas RPEB is performed in a partial pressure of helium. Thus RPEB can be performed in a chamber environment which seals to the work piece, rather than the fully enclosed chamber required for conventional EB welding. The process does not require a high vacuum capability and so mechanical vacuum pumps are more than adequate.

Fig.6. A reduced pressure electron beam weld in 14mm thick grade 23 titanium pipe (weld made in the horizontal (2G) position) 

Equipment for laying steel pipes using RPEB is under development, but satisfactory RPEB welds have been made in titanium pipe in material up to 16mm thick (Smith and Gittos, 1998c) ^[16] ; Fig.6, Table 5.

 

Table 5. Typical tensile properties for an as-welded RPEB weld in 14mm thick grade 23 titanium pipe

Zone	0.2% Proof Stress (MPa)	Tensile Strength (MPa)	Elongation (%)	Location of Failure	Maximum Hardness (Location) (HV10)
Cross-weld	-	920	-	Parent	340 (HAZ) 330 (weld metal)
Parent	850	915	14	-	310

 

The advantages of the RPEB process are similar to those of keyhole plasma welding, although many of the difficulties, such as weld closure, have been comprehensively solved for EB welding. A possible disadvantage lies in the rapid cooling rates experienced which can produce martensitic microstructures. Whilst the martensite formed in Ti-6Al-4V is not as hard as that formed in more heavily β-stabilised titanium alloys, fracture toughness can be lowered compared with the base material (Boyer et al., 1994) ^[3] . However, preliminary toughness investigations suggest that RPEB weld metal performance is similar to that of the base material (Smith and Gittos, 1998c) ^[16] . Electron beam welds are susceptible to porosity and all precautions necessary for its avoidance in TIG welds apply equally to RPEB welding.

Laser Welding

Laser welding has been successfully applied to titanium alloys ( Fig.7, Table 6) (e.g. Dabezies et al., 1993) ^[5] and machines of sufficient power are now available to weld the thicknesses of interest. Maximum penetration is particularly good for titanium, due its low thermal conductivity and high absorption rate for infrared light. The process offers similar advantages to EB welding, i.e. minimal distortion with good joint reproducibility, but can be operated in air, albeit with local shielding. Laser welding will also produce similar microstructures; i.e. alpha prime martensite within prior β grains. Although laser welding is not a common process in the offshore industries, a steel pipe laying system that utilises laser welding is under evaluation. If suitably modified to incorporate auxiliary shielding, this could presumably be adapted for titanium. Laser welds are susceptible to porosity, but can also be prone to the entrapment of the helium shielding gas. Adequate joint preparation strategies will minimise the former, whilst careful selection of welding parameters should eliminate the latter.

Fig.7. A laser weld between 11mm thick grade 5 and grade 2 titanium plate

Table 6. Typical tensile properties for a laser weld in 20mm thick Grade 23 plate, after Dabezies et al. (1993) ^[5]

Zone	0.2% Proof Stress (MPa)	Tensile Strength (MPa)	Elongation (%)	Location of Failure	Maximum Hardness (Location) (HV0.5)
Cross-weld	-	1070	-	Parent	385 (weld metal)
Parent	981	1003	16	-	330

Solid state processes

Contamination

For the solid state processes considered here, any material that has been contaminated by the atmosphere will typically be expelled into the 'flash'. Since the flash is removed after welding, no embrittled material should remain in the joint, eliminating the need for gas-shielding. The correct flash geometry should be obtained, however, to ensure that the material is adequately protected. The only possible exception is radial friction welding for which no internal flash is produced on the inside of the pipe. In this instance, however, the material is in intimate contact with the internal mandrel (i.e. only a limited amount of air will be available and so contamination should be limited) and no detrimental effects have been reported.

Flash removal

Whilst the removal of weld flash should be quite straight forward on the outside of the pipe, the removal of internal flash, such as will be necessary for rotary friction welds, will be more difficult. Shearing systems have been developed for flash removal in pipes for other industrial sectors, whereby the flash is sheared away from the pipe immediately after welding, while still hot. Similar techniques could be adopted for friction welds in flowline or riser pipe systems, after suitable adaptation to cope with long pipe lengths.

Rotary Friction Welding

Inertia and continuous drive rotary friction welding are well established for joining titanium and are processes which have already found application, albeit for other materials, in the offshore sector. For pipe butt welding, frictional heat is developed by rotating one pipe against another stationary pipe. After a few seconds, when the interface has reached an elevated temperature, rotation is stopped whilst simultaneously applying a greater forging force. There are two derivatives of rotary friction welding. In the continuous drive variant, the rotating component is conventionally driven directly by a motor, which can be braked as the forging force is applied. The other variant, which has found particular favour in the US, is inertia friction welding. Here, the rotating pipe is attached to a flywheel and the non-rotating pipe acts as a brake. An example of a rotary friction weld in titanium pipe is shown in Fig.8 and typical properties are shown in Table 7 (Threadgill, 1997) ^[18] .

Fig.8. A continuous drive rotary friction weld in 14mm thick grade 24 titanium pipe

Table 7. Typical tensile properties for a continuous drive rotary friction weld in 14mm thick grade 24 titanium pipe

Zone	Maximum hardness (Location) (HV10)	0.2% Proof Stress (MPa)	Tensile Strength (MPa)	Elongation (%)	Location of Failure
Cross-weld	350 (HAZ) 340 (weld zone)	-	930	-	Parent
Parent	320	800	930	15	-

 

Both rotary friction welding processes are robust and can tolerate quite large deviations from optimum welding parameters, producing high quality joints with superior reliability. The repeatability of the processes offers the potential of assessing weld quality by monitoring the welding conditions. Although long lengths of pipe will need to be rotated, pipe rotation speeds are not high and decrease with increasing diameter of pipe; e.g. 150mm-diameter pipe would require <250rev/min and <60rev/min would be necessary for 600mm diameter pipe.

Radial Friction Welding

Radial friction welding overcomes the requirement to rotate long lengths of pipe by using a consumable ring and completing two welds simultaneously (one between the consumable and each pipe). Both pipes are held stationary and the consumable ring, which has a 'V' cross-section, is rotated and compressed radially ( Fig.9). Once the rotation achieves sufficient frictional heat across the interfaces, rotation is stopped and the compressive forces are increased to consolidate the joint. Collapse is prevented by an internal mandrel that supports both the consumable and the pipe. The feasibility of radial friction welding for the offshore sector has been demonstrated (Hutt, 1995) ^[10] . Weld properties established by Torster et al. (1998) ^[19] are summarised in Table 8, for comparison with the other processes.

 

Table 8. Tensile properties for a radial friction weld in 14mm thick grade 29 titanium pipe, after Torster et al. (1998) ^[19] .

Zone	0.2% Proof Stress (MPa)	Tensile Strength (MPa)	Elongation (%)
Cross-weld	-	900	-
Consumable ring (as-welded)	925	1055	9
Parent	840	910	14
Consumable ring (as-received)	795	885	11

 

 

Fig.9. Schematic diagram of the radial friction welding process

The main advantage of radial friction welding over a rotary friction welding process is the obviation of the need to rotate long lengths of pipe. Internal flash is also eliminated, although a small change in internal profile, where the pipes are compressed against the mandrel, will typically be present. Its disadvantages over rotary friction welding include the increased complexity, and therefore cost, of the welding machine and the added difficulties in monitoring and controlling weld quality, since two welds are produced rather than one. Even so, the process is fully mechanised and quality control will probably be easier than for fusion welding processes.

 

Discussion

Titanium is remarkably easy to weld by most processes, decreasing the likelihood of most weld defects. Thus, in most instances, defects such as lack of fusion should not be encountered. Those aspects of joint quality that apply most to titanium risers and flowlines are weld porosity and contamination. Adopting stringent joint preparation procedures can generally prevent porosity, but the occasional incidence of pores must be an accepted risk for fusion welds. Since porosity near the surface is particularly detrimental to fatigue, reliance is placed on effective near surface NDE solutions. Modern electromagnetic surface inspection techniques are available which offer promise for detecting such features, but work is needed to quantify the minimum detectable pore size. A greater understanding of porosity in titanium has been gained and preventative methodologies have been suggested. Development work in this area is advised to allow increased minimum performance levels to be achieved. Keyhole plasma welding has proven more resistant to porosity, but further work is necessary to determine the limits of its apparent immunity to this type of defect. The friction welding processes offer a clear advantage in this respect, since pore formation is impossible.

Weld contamination due to poor shielding is too complicated to interpret using colour criteria alone. Whilst historically no other methodology for detecting likely contamination has been available, the latest developments offer hope for the near future. The chances of contamination are much reduced in friction welds so, again, friction welding processes offer a clear advantage in reliability. Even friction welds, however, could benefit from a reliable method of detecting contamination to provide confidence that sufficient flash has been produced to remove all atmospheric contaminants from the weld zone.

All of the processes outlined here are capable of producing high quality joints in titanium flowlines and risers. The TIG welding process should be considered as the benchmark for all comparisons, since this has seen the most frequent application and was used for the Heidrun project, for example. The disadvantages of the process are clear: productivity is low and the persistent problem of porosity must be acknowledged in the maximum allowable cyclic stresses that can be stipulated in design codes. The main attributes of the processes highlighted in the present paper are summarised in Table 9. Although MIG welding offers increased productivity, possible risks, although unsubstantiated, remain in achieving the desired weld quality. Those processes which offer the prospect of greater weld quality include keyhole plasma, laser beam, RPEB, rotary friction and radial friction welding. All of these have particular combinations of advantages and disadvantages, with no one process offering the best overall solution. For example rotary friction welding is faster than keyhole plasma welding and offers total immunity to weld metal porosity, but carries far greater capital and development costs.

 

Table 9. Attributes of the Welding Processes

Process	Position	Shielding Gas	Development Effort Required	Weld Time	Capital Costs	Susceptibility to Weld Porosity
TIG	Any	Yes	None 1	Slow	Low	High
MIG	1G, 2G	Yes	Low	Slow	Low	High
Keyhole Plasma	1G, 2G	Yes	Moderate	Fast	Low	Low
RPEB	2G	Yes	High 2	Fast	High	High
Laser	1G, 2G	Yes	High 2	Fast	High	High
Rotary Friction	Any	No	High 3	Fast	High	Immune
Radial Friction	Any	No	High 2	Fast	High	Immune

Notes:
¹ With the exception of development work for minimising and detecting porosity
² Some development work has already been performed on titanium and/or pipe systems
³ Development work may be minimised since this is an established process for titanium welding

 

The weldment mechanical properties given in the preceding sections are all for as-welded material and, with the exception of those obtained from the multipass TIG weld, which showed greater weld metal elongation, are remarkably similar. It should be noted that a single pass TIG weld would show mechanical properties more similar to the other weldments. Fatigue crack propagation rates are typically lower for weld metals than for the parent metal of Ti-6Al-4V alloys, for weldments produced at a variety of heat inputs ranging from those typical for TIG to those of EB welding (Murthy et al., 1997) ^[12] . Although there are variations in grain size, the most extreme being in the comparison of friction welds with, say, TIG welds, the microstructures tend to be reasonably similar. Acicular products are always observed, with aspect ratio and the proportions of martensitic α' and lamellar α/ β colonies, dependant on cooling rate. Ductility is promoted by lower aspect ratios and smaller β grain sizes. Higher 'aggregate heat input' processes (e.g. TIG) have the disadvantage of giving larger prior- β grains, but promote lower aspect ratios. Here, the term 'aggregate heat input' is used to describe the summation of heat inputs for each weld pass required for joint completion. Lower 'aggregate heat input' processes (e.g. laser, RPEB) produce smaller prior- β grains, but aspect ratios and martensitic α' levels are typically greater. Thus, no process has a particular advantage with regard to weldment tensile properties and fatigue crack growth rates. Weld metal toughness can be effected detrimentally by martensitic α', but V-notch Charpy toughness tests performed by Smith and Gittos (1998c) ^[16] show little difference between TIG, MIG, keyhole plasma and rotary friction welds in 14mm thick grade 23 titanium pipe: All gave greater values than the β-processed parent. Postweld heat-treatments (typically performed at temperatures >600°C) have little, if any, beneficial effect on tensile properties or fatigue crack growth rates: Heat treatments at temperatures significantly greater than 700°C are required to promote significant microstructural alteration and increase ductility (Murthy et al., 1997) ^[12] .

The above discussion pertains to microstructural effects of heat treatment. It is also important to note that heat treatment can also change the high levels of residual stress which are normally assumed to be present in as-welded structures. Such stresses have the same effect as high applied tensile stresses in maximising the detrimental effect of the stress range and will make compressive loads more damaging. Thus, any relief of residual stress is likely to be beneficial in all fatigue loading scenarios, with the exception of all tensile stress cycles. However, typical thermal stress relieving treatments for titanium alloys are likely to produce only partial relief of residual stresses.

In the short term, joint quality is likely to play the most important role in welding process and procedure selection for titanium fabrication in the off shore sector. Productivity will probably only become of utmost concern if pipe systems are to be produced on lay barges. However, all of the high quality processes identified here also offer increased productivity over TIG welding (see Table 10). Although of less importance at this stage in the development of titanium flowlines and risers, this may come to dominate process selection in the future.

 

Table 10. Approximate joint completion times for 250mm outside diameter, 14mm thick titanium pipe, including setting up.

Process	Joint Completion Time (hours)
TIG	10
MIG	6
Keyhole Plasma	1
RPEB	1
Laser	1
Rotary Friction	1
Radial Friction	1

 

Summary

The key issues concerning quality and productivity in titanium flowlines and risers have been identified. Those processes that offer the best combination of maturity (especially as regards welding titanium) and quality have been identified and their potential advantages and disadvantages discussed. None of the higher quality processes considered here (RPEB, keyhole plasma, laser, rotary friction and radial friction welding) has a clear advantage over all others, and the choice of process will depend largely on specific requirements. Weld metal porosity and contamination are potentially the most significant defects which can be encountered in titanium weldments, since their presence can be very difficult to detect. Porosity is particularly insidious and its control, especially that near the surface, is key to increasing maximum allowable cyclic stresses. A greater understanding of the formation and sources of porosity in titanium welds has been gained and this information may prove crucial to the development of welding strategies that eliminate the most damaging porosity defects. Methodologies have been identified which offer hope in identifying dangerous porosity and contamination levels, but further work is required to develop and establish these techniques.

 

References

1	Belloni, A. and Punshon, C. (1997).	Reduced Pressure EB Welding for Offshore Pipelines. IIW Doc. IV/680/97.
2	Boucher, C., Messager, F., Heuze, J.-L. and Gaillard, F. (1994).	Open-air Arc Welding of Medium Thickness Titanium Alloy. Proc. Conf. Eurojoin 2 Second Conference on Joining Technology, publ. Italian Institute of Welding, Florence, May 1994, 61-71.
3	Boyer, R., Welsch, G. and Collings, E.W. (1994).	In Materials Properties Handbook: Titanium Alloys, ASM International, 483.
4	Crement, D.J. (1993).	Narrow Groove Welding of Titanium using the Hot-Wire Gas Tungsten Arc Process. Welding Journal, 72, (4), 71-76.
5	Dabezies, B., Menager, B. and Diani, J.M. (1993).	Laser Welding TA6V Titanium Alloy of Thickness 20mm. Proc. 5 th CISSFEL Int. Conf. Welding and Melting by Electron and Laser Beams, Publ. Commissariat a l'Energie Atomique, La Baule, France, June 1993, 1, 251-258.
6	Gittos M.F. (1999).	A New Approach to Monitoring the Quality of Titanium Welds. TWI Group Sponsored Project 5702. (TWI Proposal GP/MAT/1112, 1997).
7	Gittos, M.F. and Scott, M.H. (1993).	Effect of Contamination by Air on the TIG welding of Commercially Pure Titanium Sheet. Proc. Titanium '92: Science and Technology, Publ. TMS, March 1993, 1, 601.
8	Harwig, D.D. and Castner, H. (1997).	Oxygen Equivalent Effects on the Mechanical Properties of Titanium Welds. Proc. Int. Conf. Advances in Welding Technology: The Joining of High Performance Materials, Nov 1996, Publ. EWI, 159-179.
9	Howse, D.S. and Gittos, M.F. (1998).	Using Shielding Gases for Improved Productivity Arc Welding of Titanium. TWI Bulletin, March/April, 30-32.
10	Hutt, G. (1995).	Titanium Dynamic Riser Systems. Titanium World 2 (4) 25-27.
11	Lindh, D.V. and Peshak, G.M. (1969).	The Influence of Weld Defects on Performance. Welding Journal 48, (2), 45s-46s.
12	Murthy, K.K., Potruli, N.B. and Sundaresan, S. (1997).	Fusion Zone Microstructure and Fatigue Crack Growth Behaviour in Ti-6Al-4V Alloy Weldments. Materials Science and Technology, 13, 503-510.
13	Ramsland, A.R., Ødegård, J. and Hauge, M. (1997).	Welding and Mechanical Properties of High Strength Titanium Alloys for Offshore Applications. Proc. IBC Int. Conf. Joining and Welding for the Oil and Gas Industry, October 1997, Publ. TWI.
14	Smith, L.S. and Gittos, M.F. (1998a).	A Review of Porosity and Hydride Cracking in Titanium and its Alloys. TWI Members' Report 658/1998.
15	Smith, L.S. and Gittos, M.F. (1998b).	Formation of 'Tramlines' on Arc Welds and Back Face Contamination of Keyhole Plasma Welds in Titanium and its Alloys. TWI Members' Report 661/1998.
16	Smith, L.S. and Gittos, M.F., (1998c).	High Productivity Pipe Welding of Ti-6Al-4V Alloys. TWI Members' Report 660/1998.
17	Schley, J. (1998).	Private communication.
18	Threadgill, P.L. (1997).	The Potential for Solid State Welding of Titanium Pipe in Offshore Industries. Proc. Seminar Right Use of Titanium III, November 1997, Stavanger, Norway.
19	Torster, F., dos Santos, J.F., Hutt, G. and Koçak, M. (1998).	Metallurgical and Mechanical Properties of Radial Friction Welded Ti-6Al-4V-0.1Ru Risers. Proc. 17 th Int. Conf Offshore Mechanics and Arctic Engineering, Lisbon, Portugal, July 1998, Paper 2208.