Complex, three-dimensional laser-cut parts can be laser welded. To highlight important practical considerations, such as component preparation and presentation, the following TWI project serves as an informative case study.
Laser cutting and welding of 3-D stainless steel pressings
One example of the flexibility of lasers as manufacturing tools has been laser welding of laser-cut edges. To date, most work has been restricted to straight-line geometries; but TWI has completed a project to demonstrate the advantages (and limitations) of using 5 axis manipulation of CO2
laser beams to trim the edges of a series of convoluted, 3-D stainless steel pressings. The same beam manipulation equipment was then used to weld the trimmed sections together.
The component used in the project consisted of pairs of differently shaped and convoluted stainless steel pressings, trimmed and welded together. Twelve of these pairs of so-called 'chutings' were then welded together to form a corrugated cylinder. The currently used manufacturing process is entirely manual and subject to significant distortion and re-work. As a result, the finished component cost is undesirably high.
The project objective was to establish the feasibility of replacing manual operations with automated laser cutting and welding. TWI was supplied with pressed stainless steel chutes (formed from 0.7mm thick sheets of AISI 304) and asked to manufacture a complete assembly. No jigging or fixturing was supplied and the request was that this should be kept to an absolute minimum for the trials. A Howden Laser 3kW CO2
laser was used in conjunction with a 5-axis Robomatix beam manipulator equipped with separate cutting and welding heads.
shows the required trim line on one of the two types of as-supplied pressings. The width between the two vertical lines is about 180mm. Parameters for the (5-axis) cutting of these pressings were developed on flat sheet of the same material. 1.5kW of laser power was used, with a 130mm focal length lens in the beam focusing system. Nitrogen assist gas was used at a pressure of 10bar. The average cutting speed was 1m/min.
Although the cutting head was equipped with height sensing, using a standard nozzle design, this could not cope with the radii of curvature involved and, as a result, the trim line was programmed into the CNC control system. The height sensing function was then disabled. This procedure resulted in stand-off distances of between 0.6 and 1.0mm, at various positions on the trim line. As mentioned earlier, the jigging for support of the component during cutting was minimal and consisted of a simple wooden support frame.
Despite the above compromises, it proved possible to prepare the required 24 components with dross-free edges, of roughness Rz around 10 microns, and squareness between about 50 and 100 microns. The latter related to the ease of establishing a normal to the trim-line without CAD drawings of the components.
In keeping with the philosophy of minimal jigging, the 24 chutes were manually tack-welded together, and the complete structure was supported on two, purpose-built wooden rings placed on a large rotary table with a tilting axis. In this way, each of the joint-lines could be presented to the laser beam in a fixed position by rotation of the table. No physical clamping of the assembly was used.
Autogenous welding conditions were derived on flat sheet (which also established a tolerance window for gap and mismatch) but it quickly became apparent that in places along some of the weld lines, significant mismatch and gap occurred with the pressings. As a result, it was decided to employ wire feed in the welding process (0.8mm diameter, type 308L wire).
The laser welding head consisted of a mirror focusing system of focal length 150mm. Laser power was between 2.1 and 2.4kW and gas shielding was provided via a copper tube of 6mm internal diameter which fed into a shielding shoe which was also protected with fibre skirting to prevent air ingress during welding.
Final programming of the weld line involved defining 15 editing points along the 250 and 350mm weld lines. More consistency was found when welding from the outer (unconstrained) part of the assembly, towards the inner cylindrical part (which was supported by the wooden rings). Figure 2 shows a photograph of the completed assembly.
Inconsistency of parts and their presentation one to another was the biggest problem in this work. The cutting process indicated that it should be possible to produce edges suitable for laser welding, which when assembled together, would provide gaps <0.1mm. Stress relieving during cutting can produce in the resultant parts some deviation from the tolerances required, which can be difficult to predict as the components have varied mechanical history. This could result in some of the gaps and mismatch which were seen. Other mismatch was introduced by the tack welding process which was necessary in this work as no clamping was available. Component movement during the welding process, again attributed to lack of clamping, also caused problems.
For laser cutting, it is clear that a relatively simple jig capable of taking two of the chutes side by side, (but independently) and clamping the chute against a fixed reference, would allow the same cutting path to be used for production of two adjacent faces for welding. Jigging for welding such a complex structure as the one described above could prove prohibitively expensive but it may also be possible to develop a clamping system for two parts, which would also provide a mechanism for rotation of the welded sub-assemblies, until the original section in the jig is returned for the final weld completing the assembly. This option however, would only allow sequential welding of the structure.
The project showed that provided sufficient care is taken in the preparation of the material to be welded, and the presentation of the laser cut parts to one another for welding, it should be possible to laser weld complex 3-D structures assembled from laser trimmed or cut parts.