Andrew Woloszyn*, Dave Howse*
*TWI, Granta Park, Great Abington, Cambridge, CB1 6AL, UK
Paper presented at 8th NOLAMP Conference, Copenhagen, 13-15 August 2001.
Abstract
A stiffened panel demonstrator component typical of that used in the shipbuilding industry was conceived to demonstrate the advantages of three dimensional laser processing using robot manipulated 4kW Nd:YAG lasers. BS EN 10025:1991 Grade S275JR (formally BS 4360: Grade 43D) material in the 6-8mm-thickness range was used. Initial parameter development preceded the fabrication of a 1mx1m test panel and the eventual production of a 4.8mx1.9m full size panel section for destructive testing. Equipment set-up, joint design, effect of fit-up, processing techniques and distortion are described. The completed panel was used to compare the level of distortion induced between laser and conventional arc welding processes, before being used to conduct a similar buckling test comparison.
1. Introduction
CO2 laser welding is already established within the shipbuilding industry [1,2]. However, difficulties with beam manipulation associated with this type of laser tend to limit their application to linear joints with good access.
One of the main advantages of Nd:YAG laser sources is that the wavelength of the laser light (1.064µm) allows the beam to be delivered via an optical fibre with relatively small energy losses. This allows flexible delivery of the laser beam at the welding head. Consequently, Nd:YAG lasers can be manipulated using multi-axis robots, making them ideal for joining relatively complex three-dimensional components.
A stiffened panel demonstrator component typical of that used in the shipbuilding industry was conceived to demonstrate the advantages of three dimensional high power Nd:YAG laser processing. The component was manufactured using a 4kW Nd:YAG laser. The principle objectives of the work were:
- To demonstrate the feasibility of introducing Nd:YAG laser welding in shipbuilding construction by fabricating a stiffened panel.
- To illustrate high automation robot manipulation for subsequent use in an industrial production environment.
- To compare the level of distortion induced between laser and conventional arc welding processes.
- To assess the mechanical properties of the completed panel by a buckling test.
This paper describes the work undertaken delivering the first three objectives, performed as part of a larger Group Sponsored Project managed by TWI. The larger project, part funded by the DTI, aimed to demonstrate the potential for using high power Nd:YAG lasers in a range of industry sectors, one of which was the shipbuilding industry.
2. Experimental details
2.1 Materials
The material used during process development, and for the eventual construction of the stiffened panel conformed to BS EN 10025:1993 Grade 275JR. The composition ranges (wt.%) were as follows: C: 0.10-0.14, Si: 0.10-0.22, Mn: 0.63-1.03, P: 0.08-0.015, S: 0.006-0.014. Three thicknesses of steel were used; the base plate (7.5mm thick) and rolled bulb flats for the longitudinal and transverse stiffener components (6 and 8mm thick).
2.2 Equipment
A GSI-Lumonics Ltd CW 4kW Nd:YAG laser was used for all of the welding. A step index type silica fibre with a 0.6mm core diameter was used to transmit the laser beam to a process head. This housed collimating optics with a magnification ratio of 1:1, and a 200mm focal length; giving a 0.6mm diameter focused laser spot. Gas shielding was delivered using a co-axial nozzle attached to the process head and both argon and helium were used during procedure development trials.
During welding, the process head was manipulated by one of two floor mounted 6-axis Kawasaki Heavy Industries JS-30 robots located in the processing cell. The robot was used to manipulate the process head around the components be welded, whilst the workpiece remained stationary at all times. The sequence of welding was controlled by pendant teach and playback. Some offline programming techniques were used to design the angled fixture for the process head for optimum access and also to define the maximum reach of the robot arm and hence the welding sequence. Figure 1 shows one of the robots, and the process head.
Fig. 1. Robot mounted laser process head used for conducting welding trials, and construction of stiffened panel demonstration component (left) and close-up of head in operation (right)
Clamping and fixturing were basic, consisting primarily of plates, bars, and G-clamps to hold down the components during welding. A pneumatically operated bridging clamp was also used to ensure close fit-up of the stiffeners to the base plate during manufacture of the full size panel.
2.3 Experimental procedure
The stiffened panel design consisted of a 4.8m x 1.9m x 7.5mm butt-welded base plate, onto one side of which longitudinal and transverse stiffening members were attached. Four 80mmx6mm sets of longitudinal members ran the length of the panel, slotting through four sets of 160mmx8mm transverse members running across the width of the panel.
The production of the full size demonstration component was approached in three stages:
- Initial parameter development using small T-joint test pieces
- Production of 1m2 panels (equivalent to one 'cell' in the full size panel) for demonstration
- Production of the full size 4.8mx1.9m panel
Figure 2 shows the test pieces and panels and their dimensions for each stage. For the first two stages, the welding trials centred on the development of suitable parameters for producing a fully fused T-butt weld between the stiffener-to-base plate and the through corner stiffener-stiffener joints. All of the laser welds were autogenous throughout the work programme except for instances where shims were required.
Fig. 2. Testpieces used during procedure development (Stages 1 and 2) prior to manufacture of full sized panel (Stage 3)
During the first two stages, process factors which included laser welding parameters, robot manipulation and jigging; were all developed with the base plate laying flat; as it would throughout the assembly sequence of the full size panel. Weld quality was assessed by visual inspection, and macro examination of weld cross-sections. In addition, ultrasonic non-destructive testing was used to inspect the stiffener-to-base plate welds made on one of the 1m2 panels.
The full size panel was constructed in two 2.4x1.9m sections (one of which is illustrated in Fig.2). In the final stage of assembly, the two half panel section base plates were butt-welded together, after which the final four sections of the longitudinal stiffeners were positioned and welded into place. This was done to demonstrate that a panel to panel joint could also be fabricated using Nd:YAG laser processing.
3. Results
3.1 Initial parameter development using small T-joint test pieces
Figure 2 illustrates the welding direction and weld position for each of the joints. To achieve full penetration, one weld was made from each side of the respective T-joints. The joints were located using G-clamps. A low approach angle was used to maximise weld penetration; 20° to the horizontal for the stiffener-base plate joints and 20° from the plane parallel to the transverse stiffener for the vertical joint. This required a reduced angle co-axial nozzle to be made. A laser power of 3.5kW at the workpiece was used with travel speeds in the 0.7-1.0 m/min range, using both helium and argon shielding.
The optimum travel speed was 0.7 m/min; giving adequate penetration in the 6-8mm thickness range. Visual inspection of the welds indicated that acceptable quality welds, with consistent profiles could be achieved with both shielding gases for each of the T-joint test pieces. The factors that had the greatest influence on procedure development and resulting joint quality were joint access and joint fit-up.
3.2 1m2 panel for demonstration
The second test piece was constructed using a 1m
2 section of the 7.5mm base plate, onto which stiffeners were attached (
Fig.2). Initially, the vertical corner welds between the four stiffeners were made to produce a box section. This section was then tacked onto the 1m
2 base plate. The materials were located using simple G-clamps, then fixed into position using 40mm length laser tack welds prior to the full weld runs. The parameters developed in the previous trial were used as starting conditions.
The production of a larger test-piece further highlighted the potential problems associated with poor joint fit-up. Difficulties arose as a result of variable tolerances on the parent materials which included bowing of the base plate, some waviness of the bottom surface of the stiffeners and loose tolerance of the vertical slots. All of these factors resulted in gaps up to 1mm in size between the mating faces on parts of the assembly. This gave poor quality welds in corresponding positions of the panel, where undercut was noticed in certain areas of the relatively narrow welds. Figure 3 shows a section taken through vertical weld made between the two stiffeners on one corner of a T-joint in a 1m2 panel.
Fig. 3. Section taken through stiffener-to-stiffener vertical-up weld. Welding conditions: 3.5kW power at workpiece, 0.7m/min travel speed, 0mm focus position, argon shielding gas.
Note misalignment of parent materials and resulting undercut
The welding parameters were altered to reduce the undercut on weld bead profile and improve tolerance to variations in fit-up in the stiffener-to-stiffener joints. This comprised a reduction in the travel speed to 0.3m/min coupled with defocusing the beam +3mm above the material surface. Helium shielding gas was also used. The resulting welds showed a marked improvement in visual appearance profile. Figure 4 shows a comparison of a weld made with the original 0mm focus condition (4(a)) and at the defocused condition (4(b)).
Fig.4. Surface profile of welds between stiffeners on the 1m2 testpiece 3.1kW laser power at workpiece, 0.3m/min travel speed, +3mm focus position
Ultrasonic testing of one of the panels showed a minimal number of indications (0-4) along the length of three of the four stiffener-to-base plate welds. One of the welds (6mm longitudinal material) showed multiple indications (13) along its length. The majority of indications were spot type defects, except those in the corners of the panel, where the robot was required to track round 90°.
Another issue identified during production of the panels was the programming time associated with robot manipulation of the laser beam. This was a significant factor in the total process time taken for the production of the test panel when compared with the actual welding time.
3.3. Full Sized Stiffened Panel
Based on the results of the preceding phases of the project, several steps were taken to improve both the ease of manufacture of the panel assembly and the resulting quality of the laser welds.
The factors responsible for causing variation in fit-up where addressed in turn. To solve the problem of waviness in the under-surface of the stiffeners, the bottom of the stiffeners were milled flat. This was performed after all of the stiffener-to-stiffener welds were completed, prior to welding of the stiffener sub-assembly onto the base plate. Any gaps between the vertical joints in adjacent stiffeners where bridged using shims. The corresponding base-plate surface was prevented from bowing by ensuring an adequate level of clamping was applied. Close fit-up between the stiffeners and the base plate was maintained during tack welding using the pneumatic bridging clamps. The adjoining faces of the base plate to be butt-welded were also milled square to ensure good fit-up.
Both the tacking and main welding procedures were designed to minimise the level of distortion by careful control of the sequence to spread the heat input and resulting distortion evenly around the area of each panel half-section.
Each half of the panel followed an identical assembly route using the optimum parameters developed during the preceding trials. The stiffener sub-assembly was welded using a vertical up procedure and the curved bulb top sections of the longitudinal stiffeners were arc welded. The base plate and pre-welded/milled stiffener sub-assembly were clamped to a work bed. The stiffeners were tacked into position using laser welds ~40mm in length. Once all of the tack welds were complete, all of the stiffener-base plate joints were welded. Each internal cell was welded in one continuous movement of the robot with some overlap at the stop/start position. The two halves of the panel were then positioned and butt-welded in the PA position giving full penetration. Finally, the four sections of longitudinal stiffener were tacked then fully welded into place. On completion of welding, the panel showed minimal distortion. Figure 5 shows the completed panel.
Fig. 5. Completed stiffened panel
4. Discussion
The stiffened panel successfully demonstrated that a 4kW Nd:YAG automated system is capable of producing low distortion welds, of an acceptable visual appearance in stiffened panel components. The difficulties highlighted by the work relate to the need to maintain consistent beam-to-joint alignment; which is a consequence of the small energy spot size associated with the laser welding process. This had a major influence on factors such as acceptable joint fit-up and the choice of suitable welding parameters.
The alteration of the welding parameters to minimise undercut (0.6m/min at 0 focus, to 0.3m/min at +3mm focus) resulted in an increase in the total production time of the component. Although increased productivity benefits through high travel speeds are one of the main advantages of laser processing, one significant other advantage was the observation that no significant increase in the distortion level was witnessed following laser welding with the increased heat input.
The whole fabrication sequence progressed with the base plate remaining in a fixed (flat), as would be applied in practice within the industry. This necessitated the use of a robotic manipulation system to allow accurate laser beam placement. Whilst the application of a CO2 laser system would theoretically be possible, the complexity of such a system would negate its use for such a component in the shipbuilding industry. This demonstration component highlighted one of the major advantages of Nd:YAG laser systems; arising from the use of flexible fibre optic beam delivery.
The work also illustrated the effect teach and playback type programming of robot systems can have on the overall production time for stiffened panel components. For the large-scale production of components with good fit-up and where a single repeatable welding sequence is required, the programming time is of little consequence. The cellular design of stiffened panel structures suggests that only a limited number of such welding sequences need be programmed into a robot control system, which should not impose significant time requirements. In reality, when applying the Nd:YAG laser welding process, minor variations in parent material tolerances and fit-up meant that in welding each cell, the process head required accurate positioning (and therefore re-programming) of the robot movement to ensure acceptable quality welds could be produced. This was very time consuming and highlighted the importance of being able to achieve repeatable accuracy in joint positioning.
Maintaining accurate and repeatable position and orientation of the joints relative to the programmed movement of the robot-welding head is one option to ensure repeatable performance. However, an alternative is the use of sensor technology to ensure the process head has consistent alignment to the joint line by adaptive control. Various methods of seam tracking such as tactile, eddy current and vision-based sensors; are available. The linear joints on the stiffened panel, together with the high resolution and signal update rate required for laser welding would suggest that vision-based systems would be best applied. However, the system would need to be capable of controlling the rotation of the process head and change in welding direction at each corner of a given cell and as such would probably require independent movement of both the tracking system and the processing head at these points. The results of the ultrasonic inspection of the stiffener-to-base plate joints showed that this position would need careful control of the laser positioning if sound welds are to be produced.
An alternative option is to alter the process to be more tolerant to variations in fit-up. This was achieved to a certain extent by defocusing the laser. The use of a larger spot can relax the beam-to-joint alignment and fit-up criteria, although this is offset by a reduction in travel speed and/or penetration associated with reduced power density. The process could also be made more tolerant by augmenting the laser with an arc process such e.g. MIG/MAG welding. Hybrid CO2 laser/MAG welding processes have already been successfully applied in the ship building industry for unidirectional welding of stiffened panels [3] .
It is recognised that some of the actions taken to ensure the production of acceptable quality welds would not be practical in industry should Nd:YAG lasers be used for this application (e.g. insertion of shims in areas of poor fit-up). However, two of the primary objectives in producing the demonstration component were to illustrate both fully automated robotic manipulation and reduced distortion associated with high power Nd:YAG laser processing for industrial production environments. To this end, the project was successful.
5. Conclusions
The three-dimensional welding capability of Nd:YAG lasers was demonstrated on a stiffened panel component, typical of the type employed by the shipbuilding industry. The work highlighted several issues related to the need for accurate and repeatable fit-up and/or the implementation of steps to compensate for a poor fit-up e.g. seam tracking, defocusing of the laser spot or arc-augmented welding.
6. Acknowledgements
The authors would like to acknowledge the contribution of Mr Arnaud Ales who co-ordinated and planned much of the practical work and Mr Andrew Hassey and Mr Frank Nolan who carried out the laser welding trials. The authors also wish to thank the each of the Sponsors for funding the work, their support throughout the project and for their permission to publish this paper.
7. References
- Manzon L: 'Welding in the shipbuilding industry: the experience at Fincatieri'. Welding Review International 1998 17 (2) 6-8.
- Russell J D: 'High power CO2 laser welding for high quality fabrication of steel structures' Proc conf '37th Laser Material Processing Conference', Nagoya, Japan Laser Processing Society, 1996, 19-34.
- Roland F and Lembeck H: 'Laser Beam Welding in Shipbuilding - Experience and Perspectives at Meyer Shipyard'. Proc Conf 'High Productivity Joining Processes', 7th International Aachen Welding Conference, Aachen, Germany, 2001, 463-476.