Optical and Quantum Electronics 27 (1995) 1181-1191
I N V I T E D PAPER
Laser welding with filler wire U. D I L T H E Y ,
D. F U E S T ,
W. S C H E L L E R
Institut fdr Schweil3technische Fertigungsverfahren der RheinischWesffdlischen Technischen Hochschule Aachen, Pontstraf3e 49, D-52062 Aachen, Germany Received 25 October 1994; accepted 13 March 1995 New applications such as welding of material combinations and the ability to fill opening gaps between the workpieces offer new prospects for laser beam welding processes with filler wire. To guarantee good quality, vertical distance variations between wire tip and weld pool are, above all, not permissible as this causes globular metal transfer and would accordingly result in strongly rippled, unclean welds. A process-internal signal, recorded by a sensor, helps to solve this problem. The automatic tracking of the vertical wire position is possible on-line via a controller. In this way, the running process can be optimized and a consistently good weld quality can be achieved.
1. Introduction Today's market in producer goods is characterized by short product service lives, numerous variants, and price-conscious, rather critical customers. In industrial manufacturing, the application of laser beams as a tool helps to meet the demands on thin and medium-sized components. Consistent application leads to productivity increases and, consequently, to better quality. In addition to the laser's contactless, wear-resistant and high-precision operation, fast processing times are guaranteed. Its flexibility allows application in series as well as in individual production [1], and its wide range of applications, such as welding, cutting, drilling, hardening and contact-free workpiece processing, is a great advantage [2]. The use of filler wire in laser welding offers new applications such as gap filling and welding of material combinations. Laser welding with filler wire might in future replace the prevalent techniques in some fields, because in cladding and surfacing the comparatively small molten pool results in mixing ratios that are hardly achievable by other joining operations. Rapid prototyping of smaller parts by laser welding will lead to developments in prototyping since prototypes strikingly similar in shape, function and material to the eventual workpiece can be produced in very little time. The need to meet these numerous tasks drives the development of system components such as wire feeders and sensor systems for the automation of laser welding processes using filler material.
2. Background The potential of laser welding has not yet been properly exploited by manufacturers [3], although the general applicability has been examined in the laboratory and demonstrated 0306-8919
9 1995 Chapman & Hall
1181
U. Dilthey et aL convincingly. This discrepancy between laboratory testing and industrial application probably originates in the lack of system components on the one hand, and on the other hand in the lack of updating of equipment. Research is aimed at adjusting and enhancing the laser welding process using filler material to encourage industrial applications through better and simpler equipment. The advantages of laser welding with filler wire include [4]: 9 9 9 9
Ability to bridge small gaps Filling of non-ideal groove-joint geometries, such as V-shaped joints Defined weld overfills on both sides (beam entry and beam exit) Influence of metallurgical composition of the filler wire on mechanical-technological properties in the weld and fusion zones (e.g. strength, toughness, corrosion, wear resistance) 9 The effect of wider molten pools in reducing demands on the accuracy of the weld preparation, edge preparation, edge misalignment, and beam misalignment
The application of filler wire is most important for bridging gaps, as weld defects occur when wide gaps are common and no filler material is employed. The most frequently found weld defects are weld concavity, sidewall fusion defects, root suck-up and weld undercuts, defects that are intolerable or only very exceptionally admissible according to DIN 8583. A gap width of 0.14 mm in sheets of thickness 2 mm (gap 7%) leads to a weld concavity that reduces the load-carrying weld cross-section to 86%. At a gap width of 0.25 mm, the weld cross-section is only 60% of the sheet thickness [5]. The upper limit for welding, if weld overfill is tolerated, is a gap width of no more than 10-15% of the sheet thickness, but the maximum gap width of 0.3 mm must not be exceeded [6-10]. In laser welding with filler material, less precise tolerances in edge preparation are admissible. This applies to the edge surface finish as well as to variation in component dimensions. Flexible sheets are subject to intricate clamping processes, prototypes have to be processed from varying angles of direction. When using filler material, shear- or flame-cut sheet components can be welded without any additional treatments, as gaps caused by edge irregularities can always be filled reliably. Using laser welding, very thick sheets cannot be welded with a single pass, although, depending on the beam power, plate thicknesses from 0.9mm [11, 12] up to 9 or 15mm [13-15] can be welded. Even thicker sheets can be welded with the weld pass/back weld pass arrangement, provided the reverse side remains accessible for welding. Normally, pore formation increases in the root area of both welds, as the individual layers were originally made as single welds. Should more layers be necessary, the addition of filler material is a must. In general, the filler wire is fed into the weld process from the unwelded side (dragged wire feeding) [11, 16, 17]. Figure 1 shows a diagram of the lanced/dragged wire feeding method. The advantage of dragged wire feeding lies in the constrained wire guidance, caused by the gap [16]. Moreover, better reliability can be achieved through the parallel speed vectors of component and filler wire. Lanced wire feeding, however, is used sporadically in industrial practice [12, 18, 19]. This variant has the advantage that the outflowing weld pool is larger than the one running ahead of the beam, a characteristic that gives better weld appearance (finer ripple pattern). The drawback of lanced wire feeding is the possible adhesion of the wire to the freezing puddle as soon as wire feeding accuracy declines. The application of a scanning optical sensor dictates the use of lanced wire feeding. Gap widths and variations can be identified by the sensor only when the weld is clearly in view and not impeded by the feeding unit. Other techniques of wire feeding in addition to lanced 1182
Laser welding with rifler wire
Lanced wire feeding Figure I
Dragged wire feeding
Principles of lanced and dragged wire feeding.
or dragged feeding are possible, but just a few special cases are mentioned in the literature [16, 20]. The angle of wire feeding, that is the angle between welding plate and wire, is another important parameter in welding with filler wire. Normally, angles of 35 ~ to 45 ~ are selected [17, 19-21], but angles from 20 ~ up to 60 ~ are possible [4, 19, 21]. For smaller angles, owing to better access to the component, the wire feed nozzle must be positioned farther from the weld point, leading to expansion of the exposed, unguided wire end (stick-out) in front of the wire feed nozzle. The inflection caused by the wire feed roll can lead to off-centred fusion of the wire tip in the laser focus. Wire feed angles greater than 60 ~ also involve the positioning of the wire feed nozzle farther from the welding point, as otherwise the feed nozzle contacts the laser beam. Moreover, a steep wire feed angle entails problems when adjusting the wire, since very small displacements cause the beam's point of contact with the wire to shift vertically so strongly that reliable welding is not guaranteed. However, access to the component can be much easier when welding with large wire feeding angles. Normally, the wire is positioned at the focal point of the focused laser beam. If the gap is distinctly smaller than the wire diameter, the intersection of the laser beam and the wire is positioned on the sheet surface [20, 22]. With dragged feeding, sheet sizes and gap widths bigger than the wire diameter require an intersection point well inside the gap [13, 16]. The beam power reflected from the wire tip leads to even fusion of the edges. The three-phase-model of the interactions between laser beam and wire describes melting behaviour at different wire feed speeds; see Fig. 2 [16]:
Phase 1: Melt-off without screening of the laser irradiation. At low wire feed speeds, the wire fusion front is positioned almost horizontally underneath the wire, at the wire tip. The melting wire does not contact the laser beam directly. At the fusion front the molten wire forms large drops, which oscillate to and thus establish contact with the laser beam. As soon as droplet surface and laser beam make contact, blue metal plasma vapour develops and escapes at an angle of 85-90 ~ to the laser point of irradiation. Phase 2: Melt-off with partial screening of the laser irradiation. At increasing wire speed, the fusion front changes from the horizontal to a vertical position. The normal of the fusion front is then positioned at an angle of 85 ~ to the laser point of irradiation. Increasing the wire feed speed enhances the wire energy requirements. Phase 3: Melt-off with total screening of the laser irradiation. A further increase of the wire 1183
U. Dilthey et aL
Figure 2 The three-phase model.
feed speed involves the laser irradiation being totally screened. Metal melt-off continues to be complete, but the underside of the wire is crossing the laser beam cone in the solid state. This part of the wire forms a groove, by means of which the molten metal escapes. The melting process is completed only after the laser beam has been crossed. In this phase there is no longer any transmission of laser energy into the area underneath the wire, that is the workpiece surface underneath the wire tip is not subject to melting. Experiments on the model show that phase 2 is best suited for laser welding with filler wire. Applications that require high levels of standardization and/or long weld lengths are typical 1184
Laser welding with filler wire
sensor
~ator
,=surface
..vrkpiece area
Figure 3 Sensor design.
for laser welding with filler wire, as, for example, in shipbuilding (panel manufacturing), in the steel processing industry and in tank construction [13, 18, 23]. Processing of more complex parts requires the application of sensor systems for seam tracking and gap-width identification so that variations at the track end can be rapidly transmitted to the CNC controller of the welding machine.
3. Sensor system for seam tracking and gap-width recognition The demands put on sensor systems for seam tracking and gap-width recognition are considerably higher in laser welding than in conventional welding processes, and for this reason new sensor systems have had to be developed. Figure 3 shows schematically an optical sensor recently designed by the Welding Institute (ISF). The sensor observes the gap-width directly in front of the welding point by means of a CCD camera positioned at right angles to the joint. A laser diode illuminates the CCD line's field of vision. The sensor, operating with reflected light, recognizes the gap as dark in comparison to the workpiece surface. The greyscale pictures taken by the camera are then digitized and interfaced to a transputer (parallel computer) where they are evaluated using 'fuzzy logic'. The transputer calculates the gapwidth and the corresponding wire feed speed. The calculated values are passed via the interface to a wire feed unit. Another interface enables synchronization with the CNC controller for seam tracking as well as for the required adaptation of laser power, weld speed, etc., to the workpiece geometry. Sequential scanning of the gap by the CCD line involves the adaptation of the sensor orientation, especially in the case of strongly curved or closed trajectories. This adjustment of the sensor orientation is necessitated by the fact that a vertical orientation of the sensor line is required to measure the gap-width.
1185
U. Dilthey et aL
Figure 4 CNC rotationalaxis.
For sensor tracking of the joint path, a CNC rotational axis, specially designed at ISF (Fig. 4) was applied. This axis enables a rapid extra twist of the sensor/wire feed unit coaxially around the laser beam. As the distance between welding point and the sensor's field of vision is very small in relation to possible radii of curvature along the trajectory, the adjustment of the measuring field orientation and the tracking of the measuring field position can be accomplished on one rotational axis [25]. Extreme demands are made of the filler wire feeding system owing to the small weld pool dimensions; these allow very restricted applications for conventional cold wire feeder units such as used in TIG welding. For easy attachment to the various handling systems, e.g. robots, or directly to the weld optical system, the wire feed unit should be of a compact but lightweight design. These prerequisites are met by the computer-controlled wire feed system manufactured by WeldAix, a system that, based on experience achieved at the ISF, was specially designed for optimum process control and all laser-specific demands. The device comprises a feed unit and feed controller (Fig. 5) [25]. Wire diameters of 0.6 mm up to 1.6 mm can be fed almost slip-free at speeds of up to 15 m min -1. The feeding unit is characterized by its high dynamics, compact size and low weight of only 800 g. It can easily be attached to various handling systems in immediate proximity to the process. The controller unit is freely programmable and a number of different feeder programs can be stored in the unit's memory. The sensor efficiency was tested and proven in butt-welding studies on mild steel sheets (St37) of a 3 mm thickness. For this, a defined stepwise-widening gap was milled onto the abutting surface. The sections, each of length 80 mm, started at a gap width of 0.1 mm and increased in intervals of 0.1 mm up to a gap width of 0.6mm. The test sheets were fixed by a pneumatic clamping device and then welded with a beam power of 5.2 kW at a speed of 2 m min -1 (filler 1186
Laser welding with filler wire
Figure 5 Computer-controlled wire feeding system. 0.4 E
0.2.
E
t,O m
i
i
:
!
i
0.0.
t/) O
~" - 0 . 2 " ---I . . . . : . . . . F--0.4
80
160
(a) 0.6
i
0.5 .............
;. . . . . .
240 320 X-position (mm)
i ........
! .........
0.4" - - I . . . . . . . . . }-. . . . . . . . 1. . . . . . . . .
v
400
i ........
1. . . . . . .
t-
-~
0.3-
o.
0.2-
t~
0.1 80
160
(b) 5
~
~ ~'" -o .c_ 3
240 320 X-position (ram)
;
400
;
;
...............................
i
~
--~ . . . . . . . . . ~. . . . . . . . . . . . . . . . .
~. . . . . . . . . ~-. . . . . .
i
i
'
;
i
. . . . L. . . . . .
E E
2
0
y
I
l
l
I
I
80
160
240
320
400
i
i
i
~-r
X (c)
I
Figure 6 (a) Edge position, (b) gap-width and (c) wire feed speed. 1187
U. Dilthey et aL
Figure 7 Gap bridgeability (a) with and (b) without sensor control, for the example of a stepwise-widening joint gap. Material, mild steel (St37); sheet thickness, 3mm; filler wire, SG2; wire diameter, 1.0mm; laser power, 5.2 kW; welding speed 2 m min -1 . In (b) the wire feed speed was a constant 4 m min -1 .
wire, SG 2; diameter, 1.0 mm). The wire feed speed depended on the gap width and was sensorcontrolled during the welding process; 160 measurements per second were taken. Figure 6a shows an example of the trajectory and the sensor-determined Y-position values of the left edge and of the right edge. The difference of the signals gives the gap width (Fig. 6b) from which is derived the wire feed speed (Fig. 6c). To demonstrate the action of the sensor, the test welds were made with and without sensor control: Fig. 7 shows the weld cross-sections. While the test welds with sensor control show completely even weld overfills without undercuts for all gap widths, those made without sensor control but at a constant wire feed speed show distinct weld overfills for small gap widths and/ or weld sinkage for large gap widths. The upper and lower weld beads of the test welds made with sensor control show even formation over the total length of the weld; gap width variations are barely recognizable [24, 25].
4. Sensor system for the vertical positioning of the filler wire Besides seam tracking and gap-width recognition, sensor systems are also required for exact determination of the distance between the wire tip and the component surface. The vertical distance between the wire tip and the molten pool should not change during the process, as this results in droplet-shaped, strongly rippled, and dirty welds. Variations in this distance can be caused by tolerances in the workpiece design or by welding distortions. Optical sensor systems help to avoid this trouble, but their application is often prevented by their high costs. In a new ISF development that might help to solve this problem, a signal internal to the process is acquired by a sensor for on-line automatic tracking of the vertical wire position 1188
Laser welding with filler wire
LaserbeamFilewir re WorkpieV
[~ ~ Figure 8 Principle of the vertical positioning
sensor.
via a controller. This allows control over the welding process, allows optimization of it, and helps to achieve consistent highest weld quality [27]. The principle is based on the application of a voltage between the filler wire and the workpiece (Fig. 8). The signals (voltage variations) caused by short-circuit, transition, and insulation phases provide reliable assessment of the process. Thus, bridging of the vertical gap to the molten pool causes, even with a very light contact, a voltage drop in the circuit. The transition to the short-circuit represents the starting point for the formation of a stable molten metal bridge between wire and workpiece. The stable molten metal bridge between wire tip and workpiece is the most important criterion for achieving faultless welds. Welding of contoured sheets with sharp edges demonstrates the application possibilities of this newly developed positioning unit: weld surfacing tests were made on a sheet featuring two circular, 4-mm-deep milled-in cavities (Fig. 9). Using a highly dynamic linear axis for tracking the optics/wire feed unit, the work distance could be kept constant over the total length of the weld. Even welds were achieved and no changes in filler wire melting behaviour could be observed. The weld ripples were even and fine, with no seam width variations. The positioning unit, designed as a separately operating Z-axis, compensates all vertical workpiece profile variations and sets the optimum distance of the filler wire and laser focus to the weld point (Fig. 10). The maximum speed of 7mmin -1 as well as the starting time of less than 12 ms and the measured maximum acceleration of 5.2 m s -2 also allow the welding of profiles in the shape of steps. 5. S u m m a r y a n d p r o s p e c t s With the prerequisite optimized process control, laser welding delivers high and consistent manufacturing qualities. The application of sensors for seam tracking and gap recognition is unavoidable to guarantee the optimum process control. The addition of filler wire in the welding process makes it possible to reduce precision requirements in weld preparation with regard
Figure 9
Laser
welded workpiece
profile.
1189
U. Dilthey et aL
Figure 10 Positioning unit.
to edge misalignment, edge preparation, as well as bridgeable gaps, and also make it possible to metallurgically influence the molten pool. The sensor systems currently available on the market are characterized by insufficient resolution and low sampling rates, and are therefore not suited for application in laser welding owing to its high process speeds. Accordingly, a sensor system that meets the requirements of the laser welding process has been developed in which a highly dynamic wire feed unit serves the purpose of cold wire feeding. Owing to its low inertia and slip-free wire feeding, even rapidly widening gaps can be reliably filled. The coupling of sensor system and wire feed unit in addition to the coupling to a revolving axis enables welding of narrow-contour radii on complex workpieces. For perfect weld quality, it is essential to keep a constant distance between the wire tip and the molten pool. Variations in this distance, caused by tolerances during workpiece manufacturing and weld distortions, influence the formation of the molten metal bridge between the wire tip and the workpiece. The sensor designed to cope with this problem controls the formation of a stable molten metal bridge. Vertical positioning leads to good results in joining as well as in rapid prototyping. Laser welding with filler wire is therefore an attactive alternative to conventional rapid prototyping methods.
References 1. M. WECK, Werkzeugmaschinen I, 3rd edn (VDI-Verlag, Diisseldorf, 1988). 2. D. RADAJ, R. KOLLER, U. DILTHEY and O. BUXBAUM, Beitriige zu innovativen Fertigungsverfuhren 116 (1994) 103. 3. G. HEIN, Die Kosten im Griff, Beitage zu Werkstatt und Betrieb, Laserpraxis (June 1991) 10. 4. A RIEF, Untersuchungen zur Verfahrensfolge Laserschneiden und schwei~en in der Rohkarosseriefertigung (Carl Hanser Verlag, Miinchen Wien, 1992). 5. C.J. DAWES, ICALEO '85, 4th International Congress on Applications of Lasers and Electro-Optics, San Francisco, 1985, p. 73.
1190
Laser welding with filler wire 6. H. SASAKI, N. NISHIYAMA and A. KAMADA, CISFFEL '83, 3rd International Colloquium on Welding and Melting by Electrons and Laser Beam, Lyon, 1983, p. 569. 7. J. ALBRECHT, A. FRINGS and W. PRANGE, Fachberiehte far Metallbearbeitung 64(2) (1987) 91. 8. L. HANICKE, Application of Laser Processing in Automobile Fabrications, Cambridge, 1987, p. 68. 9. D. BAKOWSKY, Opto-Electron. Mag. 4(3) (1988) 276. 10. F. EICHHORN, J. SCHNEEGANS and M. HENDRICKS, 'Seminar Lasermaterialbearbeitung' far Unternehmen der EBM-Industrie und Stahlformung, Aachen, 1988, p. 48. 11. M. GEIGER, P. HOFFMAN and G. DEINZER, ECLAT '90, 3rd European Conference on Laser Treatment of Materials, Erlangen, 1990, p. 689. 12. U. DILTHEY et aL, 2nd International Conference on Power Beam Technology, Stratford-upon-Avon, 1990, p. 69. 13. D. R. MATYR, Welding Review (May 1987) 106. 14. A. SHINMI, U. UTSUMI and N. ANDOH, 1CALEO '85, 4th International Congress on Applications of Lasers and Electro-Optics, San Francisco, 1985, p. 65. 15. M. PANTEN, Laser Beam Welding with Filler Wire, IIW Doc. IV-545-90. 16. J. SCHNEEGANS, Untersuchungen zum Laserstrahlschweiflen mit Zusatzdrahtzufarung an un- und niedriglegierten Sttihlen (Dissertation, RWTH, Aachen). Aachener Berichte Fiigetechnik, 2/93 (Verlag-Shaker, Aachen, 1993). 17. 1. J. STARES, R. L. APPS, J. H. MEGAW and J. SPURRIER, Power Beam Technology, Brighton, 1986, p. 223. 18. K. HAKANSSON, JOM1, Joining of Metals, Helsingor, Denmark, 1981, p. 36. 19. M. N. WATSON, Laser welding of structural steel with wire feed, Welding Institute Research Report, 264/1985. 20. J. H. MEGAW, M HILL and R. JOHNSON, Joining of Metals - Practice and Performance, Warwick, 1981, p. A1. 21. I. M. NORRIS, International Conference on Power Beam Processing, San Diego, 1988, p. 165. 22. U. DILTHEY, D. FUEST and M. HENDRICKS, Laser Applications in the Automotive Industries, 26th ISATA Conference, Aachen, 1993, suppl, p. 8. 23. K. W. CARLSON and V. G. GREGSON, ICALEO "86, 5th International Congress on Applications of Lasers and Electro-Optics, Arlington, 1986, p. 222. 24. M. HILL et al., 2nd International Conference on Power Beam Technology, Stratford-upon-Avon, 1990, p. 108. 25. U. DILTHEY, D. FUEST and O. KROPLA, Laser und Optoelektronik 26(1) (1994), p. 44. 26. A. HUWER, Sensorsystem zur Erfassung variabler Fiigespaltweiten beim Laserstrahlschweiflen im Stumpfstofl (Dissertation, RWTH, Aachen). Aachener Berichte Fiigetechnik, 3/94 (Verlag-Shaker, Aachen, 1993). 27. U. DILTHEY, D. FUEST and O. KROPLA, 5th European Conference on Laser Treatment of Materials, ECLAT '94, Bremen-Vegesack, 1994, p. 396. 28. U. DILTHEY, D. FUEST and J. SCHNEEGANS, 5th European Conference on Laser Treatment of Materials, ECLAT '94, Bremen-Vegesack, 1994.
1191