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

Subscribe >
Skip to content

The potential for solid state welding of titanium pipe in offshore industries

Philip L Threadgill

Paper presented at Symposium on the Right Use of Titanium, Stavanger, Norway, 4 - 5 November 1997


The advantageous properties of titanium alloys in terms of specific strength and corrosion resistance in many environments are well recognised, as are the advances in manufacturing technologies which enable a wide variety of product forms to be supplied for numerous applications. The applications for titanium alloys are constantly increasing, and a particular area for growth is in the offshore industry, for example for catenary riser pipes for deep water production. In addition to the high specific strength and corrosion resistance, the lower modulus of titanium alloys provides a degree of flexibility which is advantageous for this application. Other factors such as the uncertain price and sometimes long delivery times for titanium products are very important considerations. The choice of materials is influenced by the ease with which they can be fabricated, in particular the weldability, machinability etc must be considered when determining the overall economic and technical benefits of using any alternative material. This paper is concerned primarily with weldability aspects, particularly related to offshore applications of titanium pipes.

Although there are many titanium alloys in production, the overwhelming tonnage (>90%) is supplied as either commercial purity (CP) grades, or grades based on Ti-6Al-4V. The former are used primarily where corrosion resistance is needed, although the higher oxygen contents of some CP alloys results in reasonable strength levels. The Ti-6Al-4V is very much the workhorse of the titanium industry, and combines much higher strength with very good corrosion resistance. The latter is often improved by small alloy additions. Both CP titanium and Ti-6Al-4V are reasonably easy to weld by fusion processes, although great care is required to avoid contamination. Gas shielded processes can produce welds of excellent quality, but the productivity, particularly with TIG, is low. For some applications this is seen as a disadvantage, in particular for welding circumferential pipe joints, where a single weld joint can easily take a whole shift or even longer to complete. One shot processes are therefore potentially very attractive for joining titanium pipes, and it is the purpose of this paper to review the processes available, emphasise the advantages and disadvantages of each, and speculate on the prospects for further development of the processes in the offshore industry.

Review of solid state processes

There are a number of solid state processes which are applicable to titanium alloys, and those described below are considered to be the most suited for the offshore industry.

Rotary friction welding

Titanium alloys in general respond very well to rotary friction welding, and it is widely used for example in the aerospace industry for joining many engine components. Frictional heat is developed by rotating one axially symmetrical component against another stationary component under an applied force. Frictional heating causes the materials to soften at the interface, and after a short time the interface is sufficiently hot to allow the rotation to be stopped, and a higher forging force is then usually applied to consolidate the joint. Much of the softened material is expelled as flash, and experience has shown that almost all of the material contaminated by oxygen will be removed from the weld in this way, thus removing the need for gas shielding. Rotary friction welding can be divided into two distinct sub-processes. In the continuous drive variant, the rotating component is constantly driven by an electric or hydraulic motor, which can be braked as the forge force is applied. An alternative is the use of inertia friction welding, where the rotating component is attached to a flywheel, and the non-rotating consumable is used as a brake, thusconverting the kinetic energy of the flywheel to heat at the interface. Continuous drive friction welding is more common in Europe, and inertia friction welding is more common in the USA. A feature of inertia friction welding, which can sometimes be an advantage, is that the rate of energy transfer is high at the start of the weld, and decreases, whereas the rate of energy transfer in continuous drive friction welding is more constant. Although there may be subtle metallurgical differences in the welds made by the two processes, the end results are very similar, and either process can be used to make high quality welds.

Several derivatives of these processes have been proposed. In the first, a short length of pipe, or pup piece, is rotated between two stationary pipes. This has the possible advantage of avoiding rotating a whole pipe length, but makes two welds instead of one, requiring two flash removal operations. It is also difficult to grip the pup piece in such a way as to transmit the required forces, and the process has not been developed to a commercially viable state.

Other process variants have been proposed in which a plate is rotated between two stationary pipes. In one case, the pipe material is supposed to drill through the plate, eventually forming a weld. In a further proposition, the circular plate has teeth machined on the outer diameter to enable the plate to be rotated, and welded to the two pipe pieces. Neither of these variants has been fully developed or used for titanium, and will not be considered further here.

Radial friction welding

One potential drawback of rotary friction welding is the necessity to rotate one of the components. With small parts this is not normally a problem, but with long lengths of pipe there are obvious potential difficulties. One solution to this is to use radial friction welding, in which the pipes are held stationary, and a V sectioned ring of narrower angle than the edge preparation in the pipe is rotated between them using a continuous drive mechanism, and simultaneously radially compressed to force the insert into the joint. The equipment required for this process is more complex than that required for rotary friction welding, as it requires a radial compression device, and also an internal mandrel to resist the high radial loads. One advantage of the internal mandrel is that the internal flash is eliminated, although there is generally a small reduction of internal diameter which may need to be removed.

Homopolar welding

Homopolar welding is a new method currently under development in the USA, where it has been developed primarily for welding steel pipes. In this process, kinetic energy stored in a flywheel is rapidly converted to a high direct current low voltage electrical pulse using a homopolar generator, and this high current pulse is passed across a closely butted weld joint, causing a resistance weld to be made. A high axial load is also applied, causing softened material to be expelled. In essence, the homopolar system acts as a huge capacitor. Like radial friction welding, neither of the components has to be rotated, and as with all the friction processes described above, no shielding gas is required, even for titanium. Although most work to date has been on steel, preliminary trials have been undertaken on titanium, apparently successfully, although no published data are available.

Friction Stir Welding

This novel process has been well developed for aluminium alloys, but progress is being made for its application to titanium alloys, although it will be some time before it can be considered a competitive process. However, it has a number of advantages already demonstrated for aluminium that may also apply to titanium. Friction stir welding involves moving a small rotating tool between close butted components. Frictional heating causes the material to soften, and the forward motion of the tool forces material from the front of the tool to the back, where it consolidates to form a solid state weld. Although the process will never be as quick as the processes described above, it will still take only minutes to complete a pipe weld, and there will be no internal flash to remove. The properties of aluminium friction stir welds have all been very good, in particular fatigue strength is most impressive compared to fusion welds. There are disadvantages to the process, in that a shielding gas is required for titanium, and there is a hole left at the end of the weld which must be plugged by either fusion welding or friction welding.

Flash welding

Flash welding is a forge welding process in which heat is generated by resistance heating, such that a large current is passed across the pipe ends being joined. During the initial flashing stage points of contact resistance heat, melt and blow out of the joint as the pipes are progressively moved together at a predetermined rate. When a critical metal displacement has been reached the pipes are forged together rapidly to consolidate the weld.

The process has been developed for steel pipes, in fact several machines have been built in the former Soviet Union, where it has been widely used, and the USA where the process has not been commercially used. There is no known application of the process to titanium, but titanium is an alloy which can be flash welded.

Explosive welding

In this process, an explosive charge is used to generate a bond between two pipe components. As a conventional butt weld is not appropriate for this joint, initial trials were made on other geometries, in particular an external sleeve or a bell and spigot arrangement. Although success has been claimed for steel pipes, there is no known application for titanium pipes. However, titanium is routinely clad onto steel by explosive bonding, and there may therefore be some possibility for this type of process.

SAG welding

In this process, originally developed in Norway for welding steel pipes, pipe ends are machined to a special profile and held in internal or external clamps. Localised heating at the interface is accomplished using high frequency resistance heating. Reduction of oxides at the interface is achieved by a hydrogen shroud, and the weld is made by a forging action when a sufficient temperature has been reached. A particular feature of the process is that the wall thickness of the pipe is reduced at the interface in such a way that the deformation during the forge stage will restore the joint to the required thickness, and it is claimed that no internal flash removal is required.

Usage of the process has been small, and there is no known application to titanium pipes. Hydrogen would not be a suitable gas, but perhaps argon could be used to prevent oxidation.

MIAB (Magnetically impelled arc butt) welding

This process is a forge welding process for tubular section material that has seen considerable application for small diameter tubes. A DC arc is struck between the pipe ends and is driven around the circumference at high speed by the influence of a radial magnetic field, which generates heat. When the pipe ends are melted, and sufficient depth of heating has been developed along the pipe, the pipe ends are forged together under a preset load, expelling all the fused material. The process has been scaled up for 60-80mm diameter pipe in Japan, and machines for welding 102 to 305mm OD pipe of up to 5mm wall thickness have also been developed. There is no known application to titanium pipes.

Diffusion bonding

Titanium is the easiest of all common engineering materials to join by diffusion bonding, due to its ability to dissolve its own oxide at bonding temperatures. Conventional diffusion bonding is a slow process, and requires careful control of temperature, and alignment of the pipe ends, which need to be machined to a very fine finish and tight tolerance on perpendicularity, diameter and wall thickness. The process also needs to be undertaken in a vacuum. Underideal conditions a bond of very high quality can be made, with no flash formation, but the process is slow, and requires considerable precision, making it unattractive for field use, although it has been widely used in the aerospace industry, in particular in conjunction with superplastic forming. A development of the process has been proposed by Sumitomo, which is designed to speed up the process by using an amorphous filler, and heating the joint by induction. This latter process has been demonstrated for steel, but there is no known use of the process for titanium.

Properties of welds

The data available on the properties of one-shot pipe welds in titanium alloys are very limited. Some of the available data is included here, but it is noted that at present information is limited to tensile and hardness data, with no full scale fatigue data, fracture toughness or corrosion data.

Continuous drive rotary friction welding

Fig.1 shows a continuous drive rotary friction weld made in a Ti-6Al-4V-0.5Pd pipe, of 246mm diameter and 14mm wall thickness. No shielding gas was used. It is emphasised that the procedures used to make this weld were not necessarily those which might be considered as optimum.

Fig.1. Rotary friction weld in Ti-6Al 4V-0.5Pd pipe
Fig.1. Rotary friction weld in Ti-6Al 4V-0.5Pd pipe

A macro section through the weld is shown in Fig.2. The area of the weld at and close to the bond line is seen to be completely recrystallised, and to consist of fine equiaxed grains showing the typical Widmanstätten transformation products, as shown in Fig.3. There is clear evidence for beta grain growth just outside this area, together with indications of the intense plastic deformation that accompanies this and all other friction welding processes.

Fig.2. Macro section through rotary friction weld in Ti-6Al-4V-0.5Pd pipe
Fig.2. Macro section through rotary friction weld in Ti-6Al-4V-0.5Pd pipe
Fig.3. Microstructures in rotary friction weld in Ti-6Al-4V-0.5Pd alloy Fig.3a) weld centre
Fig.3. Microstructures in rotary friction weld in Ti-6Al-4V-0.5Pd alloy Fig.3a) weld centre
Fig.3b) HAZ
Fig.3b) HAZ

A number of mechanical tests were performed on the welded sample. An initial bend test showed no evidence of failure at the weld, and indeed was bent through over 45 degrees before failure started to occur at a location away from the weld. A number of tensile tests were performed, and the data is given in the attached Table. It is evident from these data that there was no reduction in strength as a result of welding, and in fact all failures occurred away from the weld zone. Thus, the mechanical properties equal or exceed those of the parent material. A hardness traverse ( Fig.4) shows a slight increase at the weld line, and this is consistent with the tensile data.

Fig.4. Hardness traverse across rotary friction weld in Ti-6Al-4V-0.5Pd alloy
Fig.4. Hardness traverse across rotary friction weld in Ti-6Al-4V-0.5Pd alloy

Table 1: Mechanical Properties of Continuous Drive Rotary Friction Welds (Ref [1] )

Sample location0.2% PS MPaUTS MPaElongation %R of A %Failure location pipe79993015.037  pipe79393115.534  pipe79893715.534  cross-weld79693313.534pipe cross-weld78793213.534pipe cross-weld79093412.034pipe 

Inertia friction welding

Nessler et al [2] have described a fairly detailed series of inertia welding tests on rings of Ti-6Al-4V, with additional work on some other alloys. They reported good results, with little dependence of tensile strength on welding parameters, but elongation was slightly reduced at low weld upset values. The rings were 584mm diameter, 4.75mm wall thickness. Typical conditions used were reported to be 1500rev/min rotation speed, with a force of 222kN, giving an upset of about4mm. The average mechanical properties from many welds was reported as follows:

Microstructural features looked similar to those reported for the continuous drive weld, but these were not reported in detail.

Radial friction welding

Hutt [3] has made reference to preliminary radial friction welds on 100mm diameter CP titanium pipe, but has not published details of the welds.

Friction stir welding

Preliminary data only is available from work on CP titanium. Details of the welding procedure remain proprietary, but a typical macro section is shown in Fig.5. The dark etching areas are regions where the titanium has been heated above the beta transus, and has therefore transformed to a characteristic Widmannstätten microstructure on cooling. The remainder of the weldregion consists of very fine equiaxed alpha grains, with a grain size substantially less than measured in the parent material. The partial transformation in the weld zone gives a good indication of the maximum temperature reached during this process, as the alpha transus would normally occur at about 900°C. Initial hardness tests indicate that there is no loss of mechanical properties across the joint, but tensile data to confirm this are not available. Data are presented in Fig.6.

Fig.5. Macro section across friction stir weld in CP titanium
Fig.5. Macro section across friction stir weld in CP titanium
Fig.6. Hardness traverse across friction stir weld in CP titanium
Fig.6. Hardness traverse across friction stir weld in CP titanium


Although the data are sparse, there appear to be reasonable prospects of producing high quality welds in titanium alloy pipes by a number of processes. From the data available, there is no real problem in obtaining welds of sufficient strength, and achieving or exceeding parent material strength in the weld is not a problem for materials such as Ti-6Al-4V. The choice of process then must consider other aspects, such as maturity, availability of technology, process robustness, flash removal, pipe rotation etc., and there is no single solid state process which shows all the desirable attributes.

Both rotary friction processes are very well established in other industries, and with other materials, and have an excellent record for reliability and consistency, but both require one of the pipe sections to be rotated, andremoval of internal and external flash. Rotary friction welding is already widely used in the OCTG industry for welding drill pipe. The rotary processes are regarded as robust, in that there is usually a reasonable generous range of parameters which will give a satisfactory weld. This range is less in radial friction welding, and unknown for homopolar welding. Current indications for friction stir welding of titanium suggest that close control of parameters willbe required.

Although pipe rotation is not an immediately attractive prospect with long pipe lengths, the rotation speeds are not high, and decrease with pipe diameter. For example, a 150mm diameter pipe would need no more than 250rev/min, and a600mm pipe would only need about 60rev/min. Pipes must be reasonably straight before welding, although support systems can cope with minor deviations. Processes such as radial friction welding and homopolar welding do not require rotation, but these are emerging technologies rather than established technologies, and although there have been attempts to apply both to welding titanium, further development and industrialisation of the processes are required, and this is believed to be underway.

Internal flash removal, and dressing of the weld will be required for all the processes except friction stir, although the volume of material to be removed will be a lot less in radial friction welding. Irrespective of the process, it is understood that the inner surface will have to be dressed to a high quality finish after welding.

Comparisons of welding time are perhaps misleading, as production rates will depend on turn around time between welds. With the exception of friction stir welding, actual welding time will be measured in seconds, and it should be remembered that traditional welding methods such as TIG and MIG may take hours. Power requirement in Table 2 is comparative. Inertia friction welding and homopolar welding rely on release of stored energy, and therefore will require less power than continuous drive welding. Radial friction welding will have the highest power requirement, as energy is required to compress the consumable ring, and in addition there are two weld interfaces to be joined.

Table 2: Mechanical Properties of Inertia Friction Welds

Upset range
0.2% PS
R of A
Failure location
1.5-2.7 966 1028 14 38 pipe
3.2-4.4 959 1034 13 39 pipe
4.6-5.6 966 1028 15 36 pipe

Concluding remarks

There is a long and successful history of using solid state processes to join titanium alloys in the aerospace and other industries, and it should therefore be of little surprise to find encouraging data on pipe welds, even though the data appear to be rather sparse. Obtaining good tensile properties is unlikely to be an issue, but fatigue data suitable for offshore design purposes are not available. The choice of process or processes to be developed will depend on many other factors, but it would appear that all processes described in Table 3 are potentially acceptable, as no single process has a clear advantage over its competitors. Established processes based on rotary friction welding already have a long pedigree in offshore and other industries, and the lower development costs and shorter development times may make them the most economically attractive solution, particularly in the short term.

Table 3: Comparison of welding processes for solid state welding of titanium pipes

 Cont. drive FWPup piece FWInertia FWRadial FWHomo-polarFlash Welding
Process maturity very good low very good low low very good
Process robustness good not known good medium not known good
Technology availability short term medium term short term short term medium term medium term
Pipe rotation required not required required not required not required not required
Interior flash removal required required required limited required required
Weld time very fast very fast very fast very fast very fast fast
Power requirement high very high moderate very high moderate very high
Consumables none pup piece none Ti alloy ring none none
Tensile properties very good good data expected very good good data expected very good not known
Fatigue properties not known not known not known not known not known not known

 MIABSAGExplosiveDiffusion BondingFriction Stir
Process maturity low low good low very low for Ti
Process robustness medium not known good good poor
Technology availability medium term medium term short term medium term long term
Pipe rotation not required not required not required not required not required
Interior flash removal required limited required not required not required
Weld time very fast fast very fast slow moderate
Power requirement moderate moderate n/a low low
Consumables shielding gas shielding gas sleeve* interlayer* shielding gas
Tensile properties not known not known not known good data expected good data expected
Fatigue properties not known not known not known not known not known

* some process variants only


The author is indebted to David Peacock of Timet for provision of the material for rotary friction welding pipes, and for providing the unpublished tensile data in Table 1. Colleagues inside and outside TWI are also thanked for useful discussions and comments, in particular P D Sketchley, R E Andrews and E D Nicholas.


  1. DA Peacock: Private communication to P L Threadgill, October 1997
  2. NesslerC G, Rutz D A, Eng R D and Vozzella P A: "Friction welding of Titanium alloys". Welding Journal Research Supplement, 379s-385s, September 1971
  3. HuttG A: 'Titanium dynamic riser systems'. Titanium World 2 (4) 25-27, December 1995

For more information please email: