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

Subscribe >
Skip to content

Electron beam welding of crack sensitive nickel super alloy MAR-M-002 (July 2006)

T. P. Mitchell, R. Sanderson, and B. G. I. Dance


Paper presented at the THERMEC 2006 conference in Vancouver, 4-8th July 2006


There is an ongoing drive to reduce the operating costs of aero-engines and this may be achieved partially via an increase in engine efficiency. To achieve this, industry needs to utilise new materials that can withstand higher operating temperatures and stresses. Many of the nickel-based alloys suitable for these applications, eg MAR-M-002, are difficult to join using conventional welding techniques. This paper describes a technique used to successfully weld2.5 mm thick plates of cast MAR-M-002. The technique used FEA modelling to analyse stresses during welding and multiple axis electron beam deflection to alter the temperature and stress distribution around the fusion zone to prevent the initiation or propagation of cracks. FEA modelling of the process has been used to reduce the total number of practical tests, and hence, to conserve the limited material supply. This technique has made it possible to produce crack free welds in what is usually classed as an 'impossible to weld' alloy.


One important trend in aeroengine design and development in both the civil and defence sectors has been the drive towards reducing 'whole life' costs. This can be achieved by reducing operating costs through increased efficiency, improvements in design, materials, and methods of manufacture. This can yield benefits not only for the end customer but also for airline operators.

To achieve performance improvements, recent research and development has looked at materials that can withstand higher operating temperatures and stresses, typically nickel based alloys. It is unfortunate that, of the nickel based alloys suitable for these applications, eg MAR-M-002, many are difficult to join using conventional welding techniques because they are susceptible to both solidification and liquation cracking. [1,2]

A potential route to welding such alloys is to use low heat input welding techniques and electron beam (EB) welding has been the subject of investigation for this task.

In previous work, an advanced programmable multiple axis EB deflection system was developed. [3] This system allows the operator to generate multiple discrete heat sources/heating zones that are separate from the main welding beam. These heating zones in practical terms are simultaneous to the main welding beam and can be used to control the stresses and temperatures around the weld both during and after welding in an attempt to prevent the initiation or propagation of cracks.

At present the full capabilities of this approach are unknown and the number of variables is extensive making experimental investigation complex and time consuming. Using finite element analysis (FEA) modelling of the welding process in conjunction with experimental investigation we have been able to reduce the time taken to achieve a successful result.

The focus on nickel based alloy welding is of prime interest to Rolls-Royce plc (RR) and the UK MOD for military engines, however, the technique is also applicable to other difficult to weld alloys such as heat treatable aluminium alloys and intermetallics. This also makes it of interest to civil aero-engine manufacturers.

Experimental work

MAR-M-002 is a cast nickel based super alloy with a nominal chemical composition: 10.1Co, 9.9W, 8.9Cr, 5.8Al, 2.5Ta, 1.5Ti, 1.5Hf, 0.1Mo, 0.03Zr, 0.13C, Balance Ni. Supply of flat plate material was limited, as special batches had to be cast.

To reduce the number of practical trials, FEA welding simulations were produced using sequentially coupled thermal and elastic-plastic stress analyses. ABAQUS version 6.4 was used to carry out the modelling. The model simulated the joining of two plates, 25mm wide, 103mm long, and 2.5mm thick. Appropriate boundary conditions were applied in order to simulate a plane of symmetry, and since the plate was thin, the heating zones were all assumed to be volumetric.

For the welding experiments, TWI's 6kW 150kV EB machine was used. This is equipped with a TWI developed version of the Rogowski electron gun and a specialist beam deflection system. The multiple axis beam deflection system is capable of producing a wide variety of complex patterns containing over 16,000 points via a simple software toolbox.

Results and discussion

MAR-M-002 is a modern high strength, high temperature cast nickel super alloy. The high temperature oxidation resistance gives a temperature capability for MAR-M-002 of approx. 1000°C for 1000hr life at 138 N/mm 2 [4] . Numerous alloying additions are needed to achieve this high strength and high temperature resistance and each one contributes to the overall performance of the alloy, be it in the form of the matrix, gamma prime, grain boundaries, carbide formation, or oxide scale.

Welding of MAR-M-002 is difficult as shown in earlier attempts in this study using conventional EB welding. Figure 1 shows a typical result from initial welding trials on MAR-M-002. Metallographic and radiographic inspection shows many cracks, which are mainly in, or very near to, the HAZ/fusion boundary, a typical feature of liquation cracking.

The many alloying additions to MAR-M-002 cause a wide solidification temperature range and a 'mushy zone' to occur. This 'mushy zone' appears in the partially melted areas of the HAZ as well as the fusion zone, and can lead to increased susceptibility to cracking during solidification, as stresses act on partially solidified regions with low ductility. This cracking during solidification is a problem that occurs in many other nickel based alloys and superalloys. [1,2]

The initial building and verification of the FEA model produced an accurate simulation of the welding process. The model was validated by comparing predicted temperature/residual stress values with experimentally measured values, for a melt run in a plate. The data was recorded from thermocouple traces during welding and residual stress measurements found via the hole drilling method. After small adjustments to efficiency and emissivity values used in the model there was an excellent agreement between the model and experiments.

Fig.1 W11, conventional EB weld in 5mm MAR-M-002, longitudinally sectioned through weld centre line
Fig.1 W11, conventional EB weld in 5mm MAR-M-002, longitudinally sectioned through weld centre line


Once the model was established and verified, it was used to assess different heating zone positions and sizes to find suitable conditions for reducing stress and hence reducing the likelihood of cracking in the weld. Different cases were run varying the size, position, and intensity of the heating zones. The model was used to predict the maximum residual stress in and around the weld area and the preheat level in the heating zones for each of these cases. By comparing the data from all the cases (see Fig.2), initial welding parameters could be identified.

Figure 2 shows that most of the cases reached high stresses rapidly (with respect to temperature) after the weld had passed. However, model h54 predicted that the stresses start to rise at a lower temperature and therefore the probability of initiation or propagation of cracks should be reduced in this case.

Using the FEA model in this way gives an approximate indication of the split of beam power to use in order to achieve the desired thermal contours for stress reduction in the weld. The model does not simulate multi-physics effects such as the molten metal, the keyhole and the focussing of the electron beam. It is therefore necessary to identify these parameters experimentally to ensure a good weld bead and surface quality in the preheat region.

After FEA modelling had identified suitable parameters for the remote heating zones, the data was used to create a beam deflection pattern using the software toolbox. This was then tested on sheets of 2.5mm thickness 316L stainless steel to obtain the correct pattern amplitude. Preliminary tests were performed on stainless steel to conserve the limited supply of MAR-M-002.

In the first few welds made, it was observed that the remote heating patch areas suffered slight surface melting. To overcome this, further welds were carried out with small changes in welding speed and focus. However, it was found that small changes to correct the surface melting caused other problems to appear, such as a change to a partially penetrating weld and small solidification cracks at the weld start.

Fig.2. Evolution of maximum principal stress against temperature during welding in the middle of a plate
Fig.2. Evolution of maximum principal stress against temperature during welding in the middle of a plate


A brief review found that the over simplification of through thickness heat transfer causes the model to overestimate the heat input in the remote heating zones before surface melting occurs. To counter this problem, but still stay close to the same heat input, the deflection pattern was changed to account for this. Previous work has shown that for similar energy input, surface melting was reduced when using such a pattern. [5]

The model was then re-run to find the effect on the maximum residual stresses with the new power distribution case n30 (biased heat patch). Figure 3 shows the predicted maximum residual stresses versus temperature during welding. Comparing with cases h40 and h54, case n30 shows that the stresses generated during welding and cooling are not very different from the previous 'best' case, h54, particularly during welding and in the important high temperature (800°C+) region.

Fig.3. Evolution of maximum principal stress against temperature during welding in the middle of a plate. New pattern n30 based on a biased heating patch
Fig.3. Evolution of maximum principal stress against temperature during welding in the middle of a plate. New pattern n30 based on a biased heating patch


Initial specimens welded with the new pattern still showed some surface melting and some incidence of solidification cracking. To combat surface melting, further changes were made to the remote heating zone pattern using the deflection system; this adaptation expanded the previous point-raster pattern to include added circular movement at each point in the pattern. This increased the complexity of the pattern and increased the number of programmed points to over 4500.

Further welds were performed on plates of MAR-M-002 and these welds showed no sign of cracking upon examination Fig.4. To test the welding further, two welds were laid next to each other on the same plate. This simulates a heat treatment of the first weld and indicates its susceptibility to re-heat cracking, which is sometimes problematic in nickel super alloys.

a) Cross weld
a) Cross weld
b) mid thickness planar sections
b) mid thickness planar sections

Fig.4. Results from metallographic examination: a) Cross weld; and b) mid thickness planar sections (W85-86)



In spite of the simplifications of the model, the combined use of FEA and the novel EB welding technique has been shown to be capable of producing crack free welds in an 'impossible-to-weld' cast nickel superalloy. The elimination of cracking has been attributed to the redistributed stress and temperature around the weld as predicted by FEA modelling. The use of FEA has greatly decreased the number of tests needed to produce crack free welds by providing valuable data and insight into the best location, shape, and intensity of heating zones.

The implementation/commercialisation of the technique for either military or civil applications using MAR-M-002 is not without its problems. Most current EB machines are not equipped with the required deflection systems to enable the generation of the types of patterns used here, although suitable systems could be retrofitted to existing machines. EB machines from different manufacturers generate beams with different geometric and intensity characteristics, which give rise to differences in welding performance. Therefore, welding parameters (including beam deflection patterns) identified in this work are likely to require adjustment to achieve similar results on other machines.

From the work reported here the main conclusions are:

  • A suitable and accurate FEA model has been built that accurately predicted stresses in the weld region.
  • Crack free welding of MAR-M-002 has been achieved using a novel EB welding technique that combines welding and heating in a simultaneous process.



This project was one section from a larger programme of work titled 'Design, Manufacture and Integrity of High Performance Welded Metallic Structures' (DEWMIPS) which was funded by the UK Ministry of Defence/Department of Trade and Industry (UK MOD/DTI) research competition 'Structural Joints and Joining Technology'. This work was conducted in collaboration with RR, who supplied the material for the experiments, and who also supplied the physical material data ofMAR-M-002 for the FEA modelling for which we are grateful. [6]


  1. M. B. Henderson, D. Arrell, R. Larsson, M. Heobel, and G. Marchant: 'Nickel based super alloy welding practices for industrial gas turbine applications'. Science and Technology of Welding and Joining, 2004, Vol. 9, No.1, p13-21
  2. K. Shinozaki: 'Welding and joining of Fe and Ni-base superalloys'. Welding International, 2001, Vol. 8, No.15, p593-610.
  3. O. Nello, B. G. I. Dance: 'Scanning the field - electron beam deflection technology reviewed'. TWI Bulletin, 2001, 42 (5), p79-82 (Available to TWI Industrial Members only).
  4. L. La Braca: 'The use of minor metals in cast nickel base superalloys and market considerations'. Speaker & Metal-pages & MMTA, Proc. Minor Metals 2003.
  5. B. G. I. Dance: 'Welding the unweldable: electron beam welding of crack-sensitive alloys'. TWI Connect, 2000, (106), p7
  6. S. Walloe: 'DEWMIPS A5: material data for MARM002'. Letter from S. Walloe to R. Sanderson: April 2003

For more information please email: