High power non-vacuum EB welding...

TWI Bulletin, May - June 2005

Could it be the thick section welding process of the future?

Allan Sanderson is the Technology Chief of the Electron Beam Group. He started work at TWI (then BWRA) in 1966 after qualifying as a metallurgist in Manchester. His pioneering work in the 1970s on the generation and physics of high power electron beams formed the core of his PhD carried out in collaboration with City of London Polytechnic. Since that time he has been responsible for numerous research and development activities on in-vacuum, Reduced Pressure and non-vacuum equipment including the radio frequency excited gun innovation featured here.

Exciting new developments at TWI have revived interest in welding with an electron beam at atmospheric pressure. This process has much to offer and ongoing innovative work involving advanced pulsed beam technology is beginning to show great promise. Allan Sanderson examines some of the recent Co-operative Research Programme work carried out.

In recent years there have been some quite remarkable advances in both welding equipment and processes, largely stimulated by the need to extend single pass welding capability. These have included higher power laser beam equipment, the hybrid laser-arc process, friction stir welding and a variety of arc welding innovations. All these have much to offer, but none yet rival the deep welding capability of conventional electron beam welding (EBW) carried out inside a vacuum chamber. Unfortunately, the provision of a vacuum chamber is still seen by many as a major obstacle and certainly, for very large components, the cost of providing the chamber and associated pumping equipment is high. Also loading, evacuation and unloading operations can slow down component throughput and increase costs.

These shortcomings of high vacuum (

But what of the future; where will we be in three to ten years time? Will laser-arc or friction stir welding command the thick section welding scene or can EB welding be used at much higher pressures for thick section welding?

Non-Vacuum EBW

The concept of launching an electron beam into the atmosphere is almost as old as high vacuum EBW, but electrons are easily scattered by shielding gas atoms and dense metal vapour. This reduces the beam current density, diminishes the range over which the beam is potent and reduces the thick section penetration capability. As a consequence, industrial application of Non-Vacuum Electron Beam (NVEB) has, with few exceptions, been limited to the welding of simple thin section products. Nevertheless, in the USA, NVEB was the favoured process for high speed welding of automobile components for several decades before laser beam welding was introduced. Perhaps surprisingly, interest in NVEB welding has been rekindled by new developments in Europe, where the ability of NVEB to deal with fit-up gaps, combined with a high speed capability, was seen to be a great advantage compared with other processes.

Research and development activities on NVEB started at TWI in 1976 with the introduction of a 60kW unit and in the late 1980s a European sponsored Eureka project was undertaken to extend the NVEB process to thicker section products. Before then, NVEB typically operated at an accelerating voltage of up to 175kV but it was believed that the use of higher accelerating voltages of up to 300kV would be beneficial.

In principle, this is true. Increasing the accelerating voltage increases electron speed and extends the distance electrons travel before multiple electron-atom collisions cause significant deviation from the nominal axial flight path. In reality, although weld depths of some 50mm in steel and 40mm in copper were achieved ( Fig.1) at an accelerating voltage of 200kV, weld widths, particularly at the crown of the weld, tended to be excessive, especially for flat position welding.

It might be thought that increased penetration depth ought to be achievable by reducing welding speed, as is the case for in-vacuum EBW, however, for NVEBW this generally exacerbates the weld width problem, with disappointingly little increase in weld depth. One further problem encountered with deep NVEB welds in steel was the occurrence of fusion zone voids and cracks associated with an apparently re-melted region in the lower fusion zone, see Fig.2. More encouragingly, all partially penetrating NVEB welds made to date have exhibited a rounded tip to the fusion zone. This is of great benefit in avoiding root defects in circumferential and planetary welds. This is not the case with conventional EBW where the root defect problem when welding thick section materials is by no means fully solved.

Pulsed NVEB

It has been known for some time that penetration depth can be increased and weld width reduced by pulsing a power beam. For both conventional in-vacuum EB and laser beam welding, pulsing has a definite beneficial effect. The application of high energy pulses drives the heat source deep into the material being welded and during the beam off period, dispersal of metal vapour and plasma is allowed to occur. On the other hand, pulsing a power beam effectively reduces the average power level available for penetration. This can be a limiting factor for commercially available laser beam equipment because of the problems of achieving very high peak power in combination with high average power levels. Fortunately, electron beam equipment can be more readily engineered to deliver very high total power (eg >100kW) and even higher peak power levels (eg >150kW). With these thoughts in mind it was decided to explore the possibility of extending the capability of NVEB by beam pulsing and a Co-operative Research Programme was launched in 1998 to develop a suitable pulsed beam source for NVEB performance studies.

Previous laser beam welding research had indicated that fluctuations in laser plasma plumes appeared to occur in the 100msec regime, although plasma features evolved in slower timescales of the order of 1µsec. It was concluded from this work that an EB pulsing device should be designed to be capable of producing near square wave pulses at a pulse frequency of the order of 10kHz.

Although the main thrust of the work was to improve the performance of NVEB, it was expected that there could well be many spin-off benefits for Reduced Pressure and high vacuum EBW. For example:

• Beam pulsing can be used to improve beam penetration by the application of high current pulse levels for a given average beam current.
• Pulsing can be used to control heat input, and the thermal cycle of welds for microstructural improvement purposes.
• There are also advantages for back-scattered electron imaging since the image resolution can be greatly enhanced by reducing the beam current and hence the beam spot diameter during scanning.
• Beam monitoring using probing devices is made far easier if the total beam power can be reduced during the time period between monitoring periods.

Previously, TWI's NVEB and Reduced Pressure equipment incorporated an indirectly heated diode gun, using a single core high voltage cable which precluded pulsing. One purpose of this work was therefore to find a means of modifying and extending the diode gun technology in order to explore the potential benefits of beam pulsing whilst preserving, as far as possible, the high beam quality of the diode gun.

Pulsing challenges

Generally, high frequency beam pulsing can only be carried out using a triode gun, where the potential on the grid electrode surrounding the cathode is biased negative with respect to the cathode. The grid voltage is usually supplied from an auxiliary transformer-rectifier housed in the high voltage transformer tank and the auxiliary supply transformer has to be capable of standing off the full high voltage potential ( eg 200kV). In addition, with a high voltage cable-fed grid supply it is not usually possible to achieve a very high grid voltage slew rate for pulsing purposes, since the capacitance of the multi-core cable degrades the waveform of the remotely sited grid supply.

Attempts have been made to produce locally coupled auxiliary supplies by others, in which fairly conventional arrangements of transformer-rectifier elements were encapsulated in an epoxy resin and mounted directly onto the gun column. This approach would eliminate the high voltage cable capacitance, but would also increase the bulk of the gun column assembly and could suffer from high voltage breakdown problems if extended to high power operation.

In November 1973, TWI patented a novel method of generating and controlling cathode heating and the grid voltage by means of locally coupled resonant supplies. The grid voltage was produced by a transformer housed within the gun housing. An operating frequency of 80kHz was originally used and rectification of the secondary winding AC was achieved by means of two thermionic diodes making use of the filament in the gun. Beam current control was effected by merely modulating the supply carrier wave as indicated in Fig.3. This unit was built and successfully demonstrated albeit on a 30kV, low power gun using a conventional ribbon filament, powered from a remote supply housed in the main high voltage tank. Although only capable of changing the grid voltage and hence beam current at a modest rate, because of the relatively low carrier wave frequency, the principle was demonstrated. Increasing the carrier wave frequency into the tens of megahertz range offered a means of achieving very high frequency response without the risk of auxiliary power supply or cable damage, whilst retaining a compact gun column configuration.

Programme objectives

The main objectives of the CRP work programme were:

• To develop a high modulation frequency, locally coupled RF (radio frequency) pulsed grid supply.
• To design and build control electronics to adjust pulse width, amplitude and modulation frequency.
• To apply the hardware developed to an NVEB/Reduced Pressure welding head.
• To demonstrate the effect of pulsing on welding performance.

Pulsing equipment

The electron gun design was based on the diode gun which forms the electron source for all TWI's high power Reduced Pressure and in-chamber equipment. For high power and high voltage RPEB machines (100kW; 175-220kV) the gun is conveniently housed in a 300mm cube vacuum enclosure with demountable or hinged doors, Fig.4. In this figure an 84MHz cathode supply primary winding is attached to the hinged gun access door. The grid supply consists of a similar RF transformer associated with the high voltage cable insulator. This produces an AC voltage of in excess of 3000V. The main gun cathode component also provides a source of electrons for a thermionic diode to rectify the RF waveform in much the same way as described above. Since the RF carrier voltage can be modulated at the supply amplifier, with rise and fall times of the order of 10µsec, grid voltage and beam current pulse frequencies can be very high.

Beam characteristics for extraction

Currently the best means of extracting a high power electron beam from the gun region, held at typically 5 x 10 ^-5 mbar, into atmospheric gas, is to use a gun column consisting of cavities separated by fine bore nozzles. These cavities are pumped by a series of vacuum pumps to provide a gradient of pressure along the beam path. In the final stage, helium gas is injected to create an outward flow of low atomic number gas to reduce electron scattering and prevent ingress of weld pool vapour.

Minimising the gas flow into the gun region necessitates small bore nozzles, but of course, the electron beam has to pass through these nozzles without appreciable power absorption. The current gun design is based on the well-established RF excited diode technology that produces a high quality beam, Fig.5.

To date non-vacuum beams of up to 100mA have been transmitted into the atmosphere at an accelerating voltage of 150kV (15kW) and 170mA at 175kV (30kW). Peak pulsed beam current of up to 270mA at 150kV (40.5kW) and 300mA at 175kV (52.5kW) have also been successfully generated and transmitted.

Preliminary welding results

Continuous and pulsed beam welding has been carried out on ferritic and austenitic steels and the results compared on the basis of weld penetration, fusion zone top bead width and mid-depth fusion zone width. Welding trials have been conducted over a wide range of pulse frequencies, power levels, welding speeds and pulse mark-space ratios (duty cycles). Pulsing has been shown to increase penetration, for example by 46% and reduce top bead width by 22% at certain pulse frequencies.

Figure 6 shows a comparison between pulsed and non-pulsed NVEB welds made in the flat position in low alloy steel. The welds were made at an average power of 15kW with a welding speed of 170mm/min using an accelerating voltage of 150kV. The pulsed weld shows a distinct increase in penetration depth (20.4 compared with 14.0mm) and a corresponding reduction in weld width, particularly at the weld top (10.4 compared with 13.4mm).

At higher accelerating voltages and peak power levels of 175kV and 52.5kW, near parallel sided fusion zones were achieved as illustrated in Fig.7. In this case the welding speed was 480mm/min. Fusion zone width was only 4-5mm and the penetration depth 22mm.

Work is now in progress to extend the performance of the equipment to higher voltage and power levels. In order to achieve this, improvements have been made to high voltage components and a new inverter driver has been introduced for the 150kW, 300kV high voltage tank.

Conclusion

It is unlikely that NVEBW will match the ultimate versatility of high vacuum and Reduced Pressure EBW, particularly in terms of working distance. However, being free of the vacuum chamber constraint is a tremendous advantage and there are many large-scale, planar and circular components where the process could score highly. Further, it is anticipated that pulsed NVEB beams, at average power levels approaching 150kW will, in the not too distant future, be able to weld section thickness in excess of 75mm in steel. This would open up a large number of applications in power generation, nuclear, marine, chemical, aerospace and nuclear waste disposal fields.

In order to carry out work on large thick section prototype components, a special purpose laboratory has been built at TWI. This consists of an X-ray enclosure with internal dimensions of 10m x 6m x 8m high, served by a 10 tonne crane. Presently, the enclosure houses the experimental electron beam generator mounted on a temporary frame, but it is anticipated that, with continuing advances in welding performance, a cartesian robot will be installed to carry out welding of large-scale components as depicted in Fig.8.

To find out more about the development of NVEBW at TWI please contact Allan Sanderson.