EBW enters fourth decade

TWI Bulletin, January/February 1990

Allan Sanderson is Head of the Electron Beam Department which is concerned with design development and application of electron beam equipment for both welding and metal evaporation. He joined TWI in 1966 and in parallel with his activities at the Institute obtained his doctorate in the generation and control of electron beams. It was this pioneer work which led to the breakthrough in high power EB welding in the early 1970s, resulting in single pass welding of steel in thicknesses up to 300mm.

Since that time, he has been responsible for further innovative work on gun design, power sources, beam monitoring and magnetic trap devices and, in the last three years, the development of real-time seam tracking detectors.

He is currently the Director of the EU86 EUREKA project which involves new R&D initiatives which will lead to the design and construction of a large 15OkW non-vacuum facility at the Abington Laboratories. It is expected that the electron beams from this system will be capable of single pass welding 100mm of steel at atmospheric pressure.

Adopted initially by nuclear and aerospace industries, electron beam welding is now used worldwide in many applications. Since its inception, power and penetration levels have been increased by over an order of magnitude, but intensive R&D continues to keep abreast of ever-increasing industrial expectations. Allan Sanderson reviews equipment and process developments at TWI and elsewhere.

This article is based on a presentation at the conference 'Advances in welding and cutting processes 89', Harrogate, 31 October - 2 November 1989.

As electron beam welding (EBW) enters its fourth decade of commercial application, what has been achieved and where is the process going? Great strides have been made in power and penetration capability, in material weldability and process studies, and in equipment design. Substantial improvements have been made in the stability of electron sources and in power supplies, permitting extended welding to take place even at high power levels.

Better means of measuring, monitoring and controlling beams have been introduced which have resulted in improved weld reproducibility, and the development of real-time seam tracking has taken the process one step nearer to full automation. In parallel, solutions have been found to many problems such as residual magnetism, gun discharging and fade-out defects, although research continues in order to keep pace with ever-increasing application and material thickness demands.

Traditionally, EBW is thought of as a vacuum process, which limits component size and throughput and increases capital cost, but this can be offset by various equipment possibilities including large chamber machines, partial-vacuum and local-vacuum devices, rapid transfer systems and even non-vacuum machines. Moreover, recent advances within the EU86 EUREKA initiative are expected to result in a unique 150kW non-vacuum facility being established at Abington, capable of welding 100mm thick steel in a single pass.

Power and penetration

In the late 1950s and early 1960s, EBW was limited to a few kilowatts achieving typical maximum penetrations of 25mm in steel. In 1964 Sciaky announced the launch of a 30kW machine at an accelerating voltage of 60kV, achieving penetration levels of some 50mm in steel. Later, Hamilton Standard in the USA introduced a 25kW 150kV machine capable of 100mm penetration in steel. Figure 1 illustrates the worldwide activity stimulated by work at TWI where a 75kW, 150kV development machine resulted in the successful extension of single-pass penetration capability in steel to 200mm in 1972 and 300mm in 1975. Similarly, using a power level of 105kW, single-pass penetration was extended to 130mm in copper and over 500mm in 5083 aluminium alloy; see Figure 2.

Gun discharge problems

The pursuit of higher powers, particularly for welding volatile materials such as light alloys, caused a number of equipment problems. Ingress of metal vapour into the electron gun during welding can cause high voltage breakdown between gun electrodes, interrupting the welding process. Since the process relies on smooth translation of a vapour-filled liquid-lined keyhole along the joint seam, interruption in this way produces gross defects which can result in the component being welded having to he scrapped.

Over the years, as larger EBW chambers and higher power guns have been introduced, the process has been progressively applied to more and more expensive components. Moreover, as components have grown larger, beam-on times have moved from minutes to hours. These trends greatly increase the risk of gun discharge, which has therefore become a major problem, to whose solution most EB equipment manufacturers have devoted substantial effort.

Fundamentally, there appear to be three possible solutions:

1. A vapour-and-particle filter system in the gun column;
2. A special-purpose high-voltage supply which will sense the onset of gun discharge and rapidly reduce the accelerating voltage to arrest and suppress arc formation;
3. Improved gun design, which will restrict the flow of extraneous material through the anode aperture, combined with good high-voltage stressing to increase the gun's tolerance to metal vapour and particles.

In the early 1970s TWI introduced the magnetic trap, a means of blocking the direct line of sight of the weldpool to the gun region. This invention is now used routinely both at the Institute and in various EBW machines in the UK and abroad. It is particularly beneficial for welding material containing volatile constituents such as aluminium-zinc-magnesium and copper alloys. However, although it is very effective in preventing ingress of metal vapour and particles generated at the workpiece, it can do nothing to prevent discharging caused by miscellaneous contaminants in the gun column head itself. Similarly, although discharge frequency has been reduced by gun design changes, this does not in isolation appear to prevent discharges especially when welding volatile material at high power. For these reasons most EBW equipment manufacturers have recently concentrated their efforts on power source design.

Power sources

Design

The filter circuit, in particular the smoothing capacitor, must retain sufficient stored energy to hold the voltage level between subsequent sinusoidal peaks as indicated by the hatched regions in Figure 3c. For a full-wave rectified single-phase 50Hz supply, the time between peaks is 10ms, corresponding to a ripple frequency of 1OOHz. Under these conditions a substantial smoothing capacitor is required to minimise the voltage ripple.

a) 50Hz input
b) Rectified
c) Filtered

An attempt has been made to employ semiconductor inverter technology operating at a base frequency of 5000Hz. With this approach, the ripple frequency for a single-phase inverter feeding a full-wave rectifier is raised to 10000Hz, which reduces the time between pulses to less than 0.02ms ( Figure 4). The pulses are essentially square topped so that a low voltage ripple can be achieved while employing a small smoothing capacitor. In addition, since the ripple frequency is high, its effect on welding performance is undetectable even for high-speed welding.

a) 5kHz input
b) Rectified
c) Filtered

Conventional power sources generally contain a current overload circuit which is set to trip after a certain time if discharging persists. These tend to be crude devices which cannot discriminate between a sub-microsecond discharge and major discharges lasting over 10ms. If the device is too finely set, the supply frequently trips off, causing unnecessary weld interruption; on the other hand, 'slugging' the unit can cause the supply to remain on for longer than 20ms with disastrous results to component, gun and supply.

At the outset of a recent 100kW, 200kV power source development programme, it was therefore decided that the source should incorporate two main features:

1. Low stored energy;
   
2. Means of discriminating between minor and major discharge events.

Gun discharge control

Three modes of discharge control were implemented in the first 100kW laboratory prototype unit built. For microdischarges lasting less than a microsecond, the current-detection circuits were set to take no action since microdischarges have little or no effect on the welding process. For a series of microdischarges, or where microdischarges could potentially degenerate into an arc lasting for longer then 50 microseconds, the supply was current limited. If for any reason the gun discharge did not self-quench after a controlled period (up to 20 milliseconds), the supply switched off automatically.

More recently, it has proved possible to replace the current-limiting mode of operation by a circuit which switches off the power inverter completely within some 30 microseconds of detection of the current rise. This is followed by an off time and a voltage ramp up time, both of which can be pre-selected. By this means gun discharges can he suppressed without interrupting the welding process, and the need to resort to power supply shut down has been eliminated.

Vacuum chambers

The need to provide a vacuum in which to conduct EBW has been considered by many engineers to be a disadvantage since it has been seen as restricting component size and increasing floor-to-floor time, and the vacuum chamber and pumping system can represent a considerable proportion of total machine costs.

Nevertheless, although elimination of the vacuum chamber is feasible for common engineering materials such as steels, light alloys and copper alloys, which are not particularly prone to oxidation or nitrogen pick-up, for zirconium alloys, titanium and other materials, shielding gas requirements are far more demanding. This is particularly the case where complex parts are involved, or double-sided gas shielding is necessary, or high welding speeds are to be employed. Moreover, the cost of shielding gas could become a significant factor. For these reasons it is likely that many fabricators of refractory metal components in both the aerospace and nuclear industries will prefer to continue using EBW in a vacuum chamber where freedom from gas pick-up can be guaranteed. Of course, most welding in the UK, measured either by weight or by joint area, is carried out in steels, where a small quantity of oxygen or nitrogen pick-up is usually tolerable. Here, the vacuum chamber and pumping requirements therefore remain a key issue.

The first commercial EBW machines produced were of the high-vacuum type where both the gun and work chamber were maintained at a pressure below 5 x 10 ^-1mbar and were fitted with a gun which could be mounted externally or internally.

This specification demanded large mechanical and oil-diffusion pumps and relatively long chamber pump-down periods. The introduction of differential chamber and column pumping, a gas-drift tube, and a column valve between gun column and chamber units, permitted chamber vacuum requirements to be relaxed to 5 x 10 ^-2mbar. The need for an oil-diffusion pump was obviated, thus reducing pump-down time and pump costs. One further means of achieving higher component throughput involved multiple gun arrangements. Later, for small components, Sciaky introduced a dual-feed system and Wentgate Dynaweld the so-called 'Rapid Transfer System'. In these configurations the chamber volume is minimised for a given component size and one or more chamber envelopes are pre-pumped while others are being used for welding. This permits a production throughput of up to 1000 parts per hour.

For very large components two in-vacuum solutions have been pursued: the large chamber approach and local vacuum chambers. In 1979 TWI began to design and build a 150m ³ chamber on site for prototype component manufacture. This houses a 10t capacity traverse with full CNC capability. The unit carries two external linear sliding seals, one in the flat position and one in the horizontal-vertical, providing ± 200mm and ± 600mm movement respectively. The arrangement allows greater use of chamber volume, increases the number of welding axes and permits tracking in real time for seam lengths up to 3.5m.

Local vacuum sealing is certainly a viable approach, provided that the components to be welded conform to a particular geometry. One notable example is the J-pipe laying EB installation designed and manufactured by SAF and Total, where a gun housed in a local vacuum chamber surrounding the pipe produced horizontal-vertical circumferential butt joints in steel up to 32mm thick. Similarly Sciaky, France, has developed both linear and girth sealing devices permitting welding of large area plate and large diameter vessels.

Non-vacuum systems

Use of electron beams at atmospheric pressure is not new. but commercial exploitation has been confined almost entirely to the automobile industry in the United States. In 1980 it was reported that 40 non-vacuum (NVEB) machines had already produced some 150 million components over a period of 12 years. Typical applications are catalytic converter cases, manifolds, collapsible steering columns, transmission carriers, torque converters and starter rings. But wider exploitation of NVEB has been limited by both the working distance and beam penetration capabilities.

Development of higher power NVEB equipment has been pursued in the USA, Europe and Japan. In the 1970s TWI purchased a 60kW welding head from the Westinghouse Research Laboratories in the USA and, after some modifications, succeeded in achieving the 40mm penetration in steel ( Figure 5) previously demonstrated by Schumacher in the USA. The fusion zone was rather wide but the fact that deep welds could be produced at atmospheric pressure stimulated a much larger scale R&D programme launched under the EUREKA initiative. In this project greater penetration is expected as a result of improved gun design and particularly by the use of high accelerating voltage.

In the first stage of the programme it is intended to design and build a 300kV, 150kW system which is expected to achieve up to 100mm penetration in steel at speeds of at least 500 mm/min. Work on the NVEB welding bay is nearing completion ( Figure 6), and the 300kV power source has undergone initial tests on a resistive load. The bay, whose internal dimensions are 10 x 6 x 8m high, houses an overhead crane of 10t capacity and is served by a roadway and 6 x 3m door enabling large components to be welded. Following commissioning of the welding head, power supply and control system in 1990, it is expected that the gun column will be mounted on a multi-axis cartesian robot to provide maximum flexibility ( Figure 7). One of the main aims of the R&D programme is therefore to produce a compact gun column with minimal peripheral pipework and cable attachments.

The equipment will become a facility for use by TWI Industrial Members for weldability studies, prototype component manufacture and thick section jobbing activities. This project is particularly relevant to industries involved in tube and vessel fabrication for example:

• Steam turbine components;
• Large gas and water turbines;
• Pressure vessel and chemical plant equipment;
• Pipes;
• Heat exchangers;
• Large complex assemblies;
• Shipbuilding and marine components;
• Nuclear waste encapsulation vessels;
• Offshore thick-wall tubular components;
• Coal conversion plant and equipment;
• Combinations of rolled and forged products;
• Nuclear reactor components;
• Heavy vehicle components.

NVEB will eliminate the need for a vacuum chamber, thus greatly shortening component floor-to-floor time, and the cost of major pumping equipment. There will still remain a need for X-ray shielding, which becomes more significant as power levels, and particularly accelerating voltages, are raised.

Real-time seam tracking

Seam tracking is an important requirement of modern industrial EB machines. In most machines, alignment is solely under the visual control of the operator. Although for small or simple components (where the joint line is a short linear or circular seam) this task is trivial, with large objects seams can deviate from an idealised geometry because of machining tolerances. In addition, where the total heat input relative to the component and jigging mass is large, the joint line may even distort during welding. These problems preclude use of 'teach and playback' methods of seam tracking.

Recent improvements in real-time seam tracking performance have resulted from new detector designs for collecting back-scattered electrons. For welding ferritic materials, where residual magnetism can cause deflection of the beam, the device shown in Figure 8 has been devised. The collector electrode is concentric with the magnetic shielding tube and a surrounding canopy forms the reflector. Using this type of detector in conjunction with specially-developed signal processing, statistical sorting and servo-control hardware, we have been able to carry out real-time seam tracking of several large components including steam turbine diaphragms. Figure 9 shows an example of a curvilinear testpiece successfully welded using the real-time seam tracking unit. This demonstration specimen comprised two 75mm deep, 10mm thick mild steel bars with a joint feature 0.8mm wide and 10mm deep. The specimen was successfully tracked at a beam power of 150kV x 74mA and a welding speed of 125 mm/min. The same basic hardware has already been successfully applied in the nuclear industry, where thin-section refractory metal is welded using very small joint tracking features, and more recently a similar unit has been installed in an aerospace company in the USA.

Weldability and weld integrity

The ability to perform single-pass autogenous thick-section welding offers many outstanding production advantages not only in terms of welding speed but also in the ability to respond quickly to customer demand. EBW for example offers the possibility of greatly reducing the total fabrication time from material acquisition to product realisation since, in comparison with arc welding, pre-weld machining costs and weld floor-to-floor time can be significantly reduced. Moreover in some cases it may be possible to minimise component stock holding by using the process to fabricate more complex units rapidly from simple modular stock items.

However, to produce high-integrity welds, autogenous EB welding is more dependent on parent material properties than is arc welding, where filler additions are inherently part of the process. It is therefore important to consider carefully the chemical composition, gas content and response to the rapid heating and cooling rates produced by EBW.

To avoid porosity and cracking problems when welding ferritic steel, the speed of welding must be limited to that imposed by the oxygen level, and sulphur, phosphorus and carbon content. Small quantities of minor elements such as niobium and vanadium in the presence of carbon can have a marked adverse influence on the mechanical properties of EB welds, particularly toughness, even after post-weld heat treatment.

In-vacuum applications

The earliest applications of EBW were associated with the nuclear and aerospace industries, but over the years process use has spread into the automobile component and general engineering sectors. It is, perhaps, not surprising to find a process with such a wide thickness range and material welding capability being used industrially for such diverse applications as ultra-thin-section pressure capsules, thick-section pressure vessels, micro-electronics, high-power electric motors, nuclear fuel element cans, nuclear waste disposal vessels and a host of others.

The EBW machines at TWI were developed primarily for thick-section welding: however, great care has been exercised in the design of the gun-column electron optics ( Figure 10) to provide well-controlled beams over the full power range. This enables the machines to be used for material thicknesses from less than a millimetre to more than 500mm in aluminium alloys. Two application examples are briefly described below.

The 5154A forged aluminium-magnesium alloy nuclear reactor core box shown in Figure 11 required four longitudinal seam welds in 83mm material. For this material, the magnetic trap device was found to be essential to avoid gun discharging. Two core boxes were welded and successfully passed full ultrasonic and X-ray examination to stringent code requirements. One of these vessels is already in service in the Studvik plant in Sweden.

EBW of thick-wall steel tubulars is an attractive possibility for producing various offshore and pressure vessel components. Two longitudinal joints of a 2m long demonstration tubular with 100mm wall thickness ( Figure 12), were successfully welded for the Whessoe Company in a single chamber pump-down in a floor-to-floor time of less than three hours. The welding speed was 200mm/min at a power of 120kV x 240mA.