- Email: [email protected]
Hot cracking in Al–Mg–Si alloy laser welding – operating parameters and their effects
Materials Science and Engineering A 395 (2005) 1–9
Hot cracking in Al–Mg–Si alloy laser welding – operating parameters and their effects E. Cical˘aa , G. Duffetb,∗ , H. Andrzejewskib , D. Greveyb , S. Ignata,b b
a Polytechnic University, Mechanical Faculty, 1 blv. Mihai Viteazu, 300222 Timisoara, Romania Universit´e de Bourgogne, Laboratoire Laser et Traitements des Materiaux (LTm), 12 rue de la Fonderie, 71200 Le Creusot, France
Received 1 June 2004; received in revised form 17 November 2004; accepted 18 November 2004
Abstract Hot cracking is a phenomenon that frequently occurs in the laser welding of some “special” alloys, such as the aluminium–magnesium–silicon type. Each occurrence of this phenomenon needs to be studied in itself, taking into account not only the individual, but also the interactive, influences of the various parameters. The advantage of using laser beams in welding processes lies in the speeds that can be reached. The disadvantage, however, is that, owing to the high cooling rates characteristic of the interaction between the laser beam and the material, the welding speed itself becomes a cause of hot cracking. The aim of this paper is to see how this disadvantage may be eliminated. We consider what the most important parameters may be, relating to tensile strength and the quantity of cracks produced, that might influence the presence or absence of hot cracking. The most influential factors in avoiding hot cracking are the welding speed and wire parameters. Also important is welding stability, as instability generates cracks. We can then determine a technological window, useful for industrial applications, which takes into account the values of these influential factors and stability. © 2004 Elsevier B.V. All rights reserved. Keywords: Laser welding; Aluminium welding; Hot cracking
1. Introduction Owing to their low density and good mechanical properties, aluminium alloys are increasingly employed in many important manufacturing areas, such as the automobile industry, aeronautics and the military. Using aluminium alloys implies the development of assembly processes, especially in welding, where conventional techniques have shown their limitations. Laser welding, therefore, has progressively attracted the attention of scientists during the last decade [1,2]. In the laser welding of some aluminium alloys, many types of defects have been revealed, such as porosity, cavities and hot cracking [3,4]. Hot cracking is a defect manifesting itself as a surface crack during the solidification of a metallic alloy. Metallic alloys do not have a good deforming capacity during the so∗
Corresponding author. Tel.: +33 3 85 42 43 17; fax: +33 3 85 42 43 29. E-mail address: [email protected] (G. Duffet).
0921-5093/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2004.11.026
lidification stage of a calorific process, and hot cracking takes place during the final stage of solidification [5,6], when the alloy is semi-solid. This cracking phenomenon occurs at a high percentage rate (85–95%). In the case of aluminium alloys, the cracks that appear during welding are produced by the direct interaction of many factors [7,8], such as: solidification shrinking and thermal tensions, which generate tensions and deformations; wide range of solidification; temperature and time-cycle of solidification speed; chemical composition of the alloy (a hot cracking domain) (Fig. 1) [9]); fastening system of the welding components, which can limit contraction. Current techniques for reducing hot cracking in the laser welding of aluminium alloys usually relate to the above factors [10–15]. In a recent review, Eskin et al. [16] presented results from the last 50 years on hot tearing of aluminium alloys. They conclude that a generic quantitative criterion that will predict hot cracking under varying conditions is still not available.
2
E. Cical˘a et al. / Materials Science and Engineering A 395 (2005) 1–9
Fig. 1. Hot cracking sensitivity of aluminium alloys dependent on the Si and Mg content.
So, knowledge of the influence of the operating parameters on hot cracking during laser welding is an important step in establishing a criterion for prediction. The method most often employed to prevent the phenomenon of hot cracking consists in chemical modifications to the molten pool. Some alloys have a chemical composition within the range that presents a high risk of cracking [17]. To reduce this risk, wires having adequate alloying elements are added during the welding process. Nevertheless, welding may still be made difficult, or even impossible, unless a good correlation is achieved between the welding parameters [18,19]. The optimum correlation is hard to find, as the optimum zone, that is to say, the one in which no cracks are present, is very narrow [6]. Initially, butt-welding of plates was carried out, so that the operating parameters that influence hot cracking could be placed in order of importance. This process is fully described in this paper. Later, T-joint welding was carried out using two laser heads. The results of this process will be presented in a future paper.
2. Experimental strategy To clarify the phenomenon under review, any general approach should be designed with the aim of establishing from the outset the appropriate response functions and contributory factors. In our case, the “product”, that is, the final result of welding is the weld, which is the direct result of the solidification of the molten pool (Fig. 2). The characteristics of the
Fig. 2. Functional design of laser welding processes.
weld can be appreciated from different points of view: technological, aesthetic, economic, etc. As for laser welding, the factors influencing the response functions that characterise the weld can be put into two categories, namely, phenomenological factors and operating factors (Fig. 2). The phenomenological factors relating to the molten pool are: size, solidification speed, chemical composition, external tensions, gas shielding system, dynamics. The operating factors are: welding speed, laser power, nature of the shielding gas, focal point position, gap, gas flow. The operating factors are the welding setup parameters. Their values and interactions induce the phenomenological factors, which can be considered, in this context, as intermediate response functions (Fig. 2). From that, the phenomenological factors give the final result, namely, the weld. In order to control any phenomenon relating to laser welding and, in particular, hot cracking of aluminium alloys, the influence of the phenomenological factors needs to be understood. Among the operating factors influencing the shape and properties of the molten pool, and, consequently, the properties of the weld, we can identify absolute factors and relative factors. The absolute factors relate to the following: – The laser beam: its wavelength, power, power distribution, operating mode, spot size. – The welded materials: their chemical composition, properties, microstructure, geometry. – The shielding gas: its composition, flow rate, flow configuration. – The added material: its chemical composition, properties, geometry, shape. The relative factors are the following: – The position of the laser beam relative to the welded materials, to the shielding gas stream and to the added material. – The angle and distance relative to the welded materials. – The movement between the laser beam and weld material (welding speed). – The movement between the added material and the laser beam (feed rate of added material). We must emphasize here that, as in all phenomena relating to laser welding, and in particular the phenomenon of hot cracking, there are many parameters that also present internal interactions. When the parameters do not have precise levels, which are generally numerical, that will be a general example of the problem under review. When they have precise values and, consequently, an experiment takes place that will be an example of a particular case of the general problem. For every experiment, a value from the response area is obtained, which is specific to the studied phenomenon. In other words, the general case can be solved using experimental methods and data processing, usually based on statistical methods [20]. Therefore, using as parameters the influences on the phenomenon of hot cracking in the laser welding of aluminium alloys, a
E. Cical˘a et al. / Materials Science and Engineering A 395 (2005) 1–9
by the industrial world as an indication of internal and external weld quality, and, secondly, the number of open cracks, which can be considered proportional to the number of all cracks in the welded zone. The use of these types of response functions makes it possible, first, to establish and characterize the presence of cracks and, secondly, to investigate the effect of many diverse factors. The outcome will have practical consequences for manufacturing industries.
Fig. 3. Welding system design. Table 1 Chemical composition of plates to weld Alloy
Al 6056
3
Element (wt.%) Si
Fe
Cu
Mn
Mg
Zn
Ti
0.82
0.07
0.55
0.57
0.69
0.17
0.02
State
4. Design and results for the preliminary factorial experiment
T4
4.1. Factors that may influence hot cracking
factorial design method seems to be an appropriate solution [21,22]. The aim of the study presented in this paper therefore is to reveal, using a factorial design method, the influence of operating parameters on the response functions that are considered representative for hot cracking during Al–Mg–Si laser welding.
3. Problem definition The experimental conditions for studying the problem were as follows (Fig. 3): AA 6056 T4 aluminium alloy (Table 1) plates, 1.6 mm thick; butt-joint configuration; compression fastening system; full penetration welding; 3 kW cw Nd:YAG type laser beam; 150 mm focusing lens; 0.45 mm diameter focalized beam. For this case, the response functions that will give a quantitative picture of the dimensions and number of cracks in the welded zone are, first, tensile strength, which is accepted
Previous experiments run in our laboratory allowed us to choose the operating factors for the preliminary experiments and to study their influence on the response functions. These operating factors are: the type and feed rate of the added material (wire); the power of the beam; the nature and flow rate of the shielding gas; the gap between the materials to be welded; the welding speed; the position of the shielding gas and the wire relative to the laser beam and to the materials to be welded; the position of the focal point of the laser beam. To be able to place the analysed factors in order of importance, the principal response function employed was the tensile strength obtained with 30 mm wide samples. At the same time, the number of open cracks was revealed by penetrative inspection and by measuring the molten pool temperature with a pyrometer. 4.2. Structure of the factorial experiments Two Taguchi [23] type factorial experiments were designed to study the effects of the operating factors. The levels are given in Tables 2 and 3. The results given
Table 2 Influential factors for the first factorial experiment Factor level
Wire (–)
1 2 3 4
No 4047 – –
Shielding gas Position (–)
Angle
Ahead Behind – –
39 49 – –
(◦ )
Type (–)
Nature (–)
Feed rate (l/min)
Direct Direct + indirect – –
He N2 – –
20 30 – –
Laser power (kW)
Welding speed (m/min)
Focal point position (mm)
2.4 2.6 2.8 3.0
3.0 4.0 5.0 6.0
+0.5 0 −0.5 −1.0
Table 3 Influential factors for the second factorial experiment Factor level
Wire (–)
Gap (mm)
Wire position (mm)
Focal point position (mm)
Shielding gas nature (–)
Shielding gas flow (direct) (l/min)
Laser power (kW)
Welding speed (m/min)
Wire feed rate (mm3 /min)a
1 2 3 4
4043 5356 – –
0 0.2 – –
0.5 1.5 – –
0 −0.5 – –
He N2 – –
20 30 – –
2.4 2.6 2.8 3.0
3.0 4.0 5.0 6.0
0.785 1.570 2.355 3.140
a
For 4043 the wire diameter is 1 and 0.8 mm for 5356.
4
E. Cical˘a et al. / Materials Science and Engineering A 395 (2005) 1–9
Table 4 Chemical composition of added material Alloy type
4043 4047 5356
Wire diameter (mm)
1 1 0.8
Element (wt.%) Si
Fe
Cu
Mn
Mg
Cr
Zn
4.5–4.6 11–13 0.25
0.8 0.8 0.4
0.3 0.3 0.1
0.05 0.15 0.05–0.2
0.05 0.1 4.5–5.5
– – 0.05–0.2
0.1 0.2 0.1
Fig. 4. Temperature instabilities.
Table 4 relate to the chemical composition of the added material.
Fig. 6. Cracks.
4.3. Analysis of results In order to gain a precise value for tensile strength, three samples obtained in the same conditions were analysed for each value of tensile strength. The data processing for these preliminary experiments led us to the following conclusions. The operating parameters under investigation influence the shape and properties of the weld zone. The importance of their influence depends on the particular operating parameters and their variation range. The most regular welds were obtained with wireless welding and high energy. The filler wire produces perturbation in the welding process. The range that corresponds to regular welds is reduced by the presence of added wire, the level of reduction being dependent on the wire’s parameters. The variation in the temperature of the molten pool gives a picture of the stability of the welding process and of the uniformity of the weld (Figs. 4 and 5). The instabilities are often associated with defects such as cracks and porosity (Figs. 5 and 6). For the welds obtained without wire, or with a small quantity of filler wire, the solidification is different and leads more
Fig. 5. Weld instabilities.
Fig. 7. Wireless weld.
rapidly to cracks (Figs. 7 and 8) than when there is a large amount of filler wire (Fig. 9). The presence of wire mitigates the tendency towards hot cracking. Figs. 10 and 11 show the effects of the relevant factors. These may be summarized as follows. The presence of wire,
Fig. 8. Wire weld (0.196 mm3 /mm).
E. Cical˘a et al. / Materials Science and Engineering A 395 (2005) 1–9
5
Table 5 Parameters of the best weld derived from the preliminary experiments Wire (–) Wire feed rate (m/min) Wire position (mm) Gap (mm) Laser power (kW) Focal point position (mm) Welding speed (m/min) Gas nature (–) Gas position (–) Gas angle (◦ ) Gas flow – direct (l/min) Wire diameter (mm)
4047 3 1.5 0 2.8 −1 3 N2 Behind 39 30 1
Fig. 9. Wire weld (0.628 mm3 /mm).
After analysing this first group of results, we were able to establish a starting-point for the next series of experiments, shown as Table 5, which presents the parameters giving the best weld.
5. Experimental modeling 5.1. Exploration of the optimal zone
Fig. 10. Influential factors – experiment 1.
and its position and feed rate, have the greatest influence on tensile strength. Welding speed is also an important influence on tensile strength. The parameters of the shielding gas, its nature, position, protection type, flow rate, angle, have the lowest importance, of the studied parameters, for tensile strength. From a mechanical angle, and for producing more regular welds without open cracks, the best results were obtained using a low welding speed (3 m/min), high laser power (3 kW) and a focal point placed either on or just under the surface with zero gap.
Fig. 11. Influential factors – experiment 2.
The preliminary experiments, referred to in the first part of this paper, enabled us to place the factors relevant to hot cracking in order of importance. The experiments also established the basis for supplementary factorial experiments. These were designed to study the influence on the response functions of the following factors: welding speed, wire feed rate, laser power, focal point position, wire position and fastening system. The experiments were conducted in the following conditions: 4047 type filler wire, 1 mm diameter, placed in front of the molten pool, at 22◦ to the horizontal; N2 shielding gas, direct – 30 l/min, and indirect (under the weld) – 15 l/min, placed behind the molten pool, at 39◦ to the horizontal; zero gap. Table 6 summarizes the different levels of operating factors. Tests for tensile strength were conducted, as before, on three identical welds to obtain more accurate values. After data processing, the following conclusions can be formulated: (1) The fastening system has a major influence on the weld’s stability and mechanical properties. The best results are obtained with a uniform compression system (two fastening points) – see Figs. 12 and 13. (2) Having the focal point at the surface gives the best results – see Fig. 14. (3) There is an optimal position for the wire relative to the laser beam and welding plates. In the configurations of this experiment, that position is at 1.5 mm in front of the laser spot and at 0.3 mm above the welded plates. This position guarantees maximum stability and the best mechanical properties for the weld (Fig. 15).
6
E. Cical˘a et al. / Materials Science and Engineering A 395 (2005) 1–9
Table 6 Influential factors for the second factorial experiment Factor level
Welding speed (m/min)
Laser power (kW)
Wire feed rate (m/min)
Focal point position (mm)
Wire position (mm)
Fastening points (–)
1 2 3 4
2.5 3.0 3.5 –
2.7 2.8 2.9 3.0
2.5 3.0 3.5 –
0 −0.5 −1.0 –
0.5 1.0 1.5 –
1 2 – –
Fig. 12. Influence of fastening system on tensile strength.
Fig. 14. Influence of focal point position.
(4) When the wire feed rate is greater than about 3 m/min, tensile strength is attenuated, leading to a substantial dispersion of values. The cause of this is instability in the wire’s movements (Fig. 16). (5) Welding speed greatly affects the weld’s tensile strength. High speeds, of over 5 m/min, lead to fewer cracks and are not detrimental to mechanical operation. The effects induced by the welding speed, wire speed and laser power, as well as the interactions between them, can be analysed from Fig. 17. The most important, in this experimental field, are the filler wire and the welding speed together with the interaction between the laser power and the wire feed rate.
Fig. 15. Influence of wire position.
Fig. 13. Welds obtained with the same operating parameters: using two fastening points (a) and one fastening point (b), respectively.
E. Cical˘a et al. / Materials Science and Engineering A 395 (2005) 1–9
Fig. 16. Wire feed rate influence on tensile strength.
Fig. 17. Factors, and their interactions, affecting tensile strength.
5.2. Experimental model for the estimation of tensile strength The results obtained allow the influence of the two most important factors to be studied, that is to say, that the response functions that characterize hot cracking in laser welding are the welding speed and the wire feed rate.
7
Fig. 18. Variations in quantity of open cracks relative to welding speed and wire feed rate.
The particular problem is defined by all the conditions mentioned above. Supplementary conditions, which guarantee the best welds, are: a two-point compression fastening system; a focal point positioned on the surface of the welding parts; an optimum wire position of 1.5 mm; laser power of 3 kW. The shielding gas used in this case is helium. The factorial design is presented in Table 7. For each experiment, the volume of deposited wire per mm of weld was computed, to give a more complete image than wire feed rate alone for the influence of the added material on the response functions. As Fig. 18 shows, the number of cracks strongly depends on welding speed, this variation being exponential. With a welding speed of 3 mm/min, there are no open cracks. When the speed is 6 mm/min, however, open cracks, in the range of 20–100 in number, can be detected, this number being reduced slightly when the wire feed rate increases. The results of the factorial design as presented in Table 7 enable us to create, by multiple regression, a detailed second degree polynomial model (1), so that we may estimate, in this experimental field, the average tensile strength per mm of weld (Y) as a function of welding speed (x1 ) and wire volume deposited per mm of weld (x2 ). The coefficients of this model are presented in Table 8 and the graphical representation of this model response surface (for tensile strength) in Fig. 19. Y = b0 + b1 x1 + b2 x2 + b12 x1 x2 + b11 x12 + b22 x22
(1)
Table 7 Factorial design structure for the experimental model Experiment number
Welding speed (m/min)
Wire feed rate (m/min)
Wire volume/mm of weld (mm3 /min)
Experiment number
Welding speed (m/min)
Wire feed rate (m/min)
Wire volume/mm of weld (mm3 /min)
1 2 3 4 5 6 7 8
3 4 5 6 3 4 5 6
3 3 3 3 2 2 2 2
0.785 0.589 0.471 0.393 0.523 0.393 0.314 0.262
9 10 11 12 13 14 15 16
3 4 5 6 3 4 5 6
1 1 1 1 0 0 0 0
0.262 0.196 0.157 0.131 0 0 0 0
8
E. Cical˘a et al. / Materials Science and Engineering A 395 (2005) 1–9
Table 8 Coefficients of the polynomial modela b0 b1 b2 b12 b11 b22
6. Influence of phenomenological factors on hot cracking 43.3 −0.565 21.0 Non-significant Non-significant −18.3
a
For the computations using formula (1) welding speed (x1 ) should be in m/min, the wire volume (x2 ) in mm3 /mm and the estimated tensile strength (Y) results in daN/mm of weld.
Fig. 19. Variations in tensile strength relative to welding speed and wire feed rate.
The analysis of the degree to which tensile strength is dependent both on welding speed and on wire volume per weld mm (Fig. 19) shows that the mechanical strength of welds, caused by the increasing number of open cracks, decreases with welding speed and has an optimum value relative to the wire volume deposited per weld mm. This optimum value is slightly modified by the welding speed. As Fig. 20 shows, there are minor differences between the estimated and measured values of tensile strength per weld mm. This model was also employed for interpolations in a range of welding speeds ranging from 4.25 to 4.75 mm/min, the estimations being also close to measured values.
Fig. 20. Comparison between estimated and measured values for tensile strength/mm of weld.
By means of the factorial design presented above, the operating factors that influence hot cracking can be placed in order of importance. By considering the operating factors together with the phenomenological factors, it is then possible to determine the influence of phenomenological factors on hot cracking. The presence of filler wire with an appropriate chemical composition and feed rate is a principal factor, since it acts on the chemical composition of the molten pool, placing it outside the range in which hot cracking might occur. This phenomenon, which was already known, was verified in our experiments. Injecting supplementary material into the molten pool is, however, a delicate operation that can provoke defects in the weld. The materials to be added will depend on the operating factors relating to the filler wire. These are, for example, its longitudinal distance from the welded parts and its free length at the mouth of the nozzle. When the experimental configurations are optimal, the wire must reach the molten pool in a liquid state. This means that we have a continuous heat transfer into the molten pool and minimal reflectivity, as opposed to the situation when the wire reaches the molten pool in a solid state. Furthermore, optimum wire parameters allow for a better contact between the laser beam and the materials to be welded, the outcome being a stable molten pool with low values for the temperature gradient. The presence of filler wire also improves the weld’s morphological quality. For, as the 4047 filler wire has low viscosity, the superficial tension of the molten pool is thereby decreased. Welding speed is another important influence on solidification speed and the temperature gradient. Thus, the welding speed influences the deformation and tension of the molten pool during solidification. Since high welding speeds are desired, other factors must be adapted in order to balance the tensions and deformations undergone by the semi-solid welding alloy from the molten pool. Among these other factors, the fastening system plays an important role. Thus, to avoid cracks, a uniform compression fastening system must be designed. The solidification speed and the temperature gradient can also be reduced by the presence of a second heat source. The calorific supplement has an important influence, through its position and quantity, on hot cracking. Another study needs to be done on the relationship between laser beam configuration and hot cracking. The presence of gas shielding, acting on the oxidation and surface state of the molten pool, has a secondary importance for the hot cracking, relating to the chosen response functions of tensile strength and open cracks. Further, some fatigue tests made on welds obtained with nitrogen and helium as the shielding gas, with all other parameters being constant, revealed no differences. A more
E. Cical˘a et al. / Materials Science and Engineering A 395 (2005) 1–9
9
systematic study should enable us to understand the nitride role in hot cracking.
gram of the R2IT mat´eriaux et proc´ed´es network. This network is supported by the French Industry Ministry.
7. Conclusion
References
The hot cracking phenomenon in laser welding of aluminium alloys is complex. In order to study the influence of operating factors, the factorial design approach seems to be an appropriate solution. From our study, the following conclusions can be formulated:
[1] E. Schubert, M. Klassen, I. Zerner, C. Walz, G. Sepold, J. Mater. Process. Technol. 115 (2001) 2–8. [2] A.P. Costa, L. Quintino, M. Greitmann, J. Mater. Process. Technol. 141 (2003) 163–173. [3] D. Fabr`egue, Rapport de Diplˆome d’ Etudes Approfondies DEA de Science et G´enie des Mat´eriaux, I. N. P. Grenoble, 2001. [4] A. Haboudou, P. Peyre, A.B. Vannes, G. Peix, Mater. Sci. Eng. A363 (2003) 40–52. [5] M. Braccini, M´emoire de th`ese I. N. P. Grenoble, 2000. [6] E. Schubert, M. Klassen, J. Skupin, G. Sepold, 6`eme Colloque International sur le Soudage et la Fusion par Faisceaux d’Electrons et Laser – CISFFEL’6, Toulon, 1998, pp. 195–203. [7] D. Boisselier, R. Lenoir, O. Wall´ee, J.M. Staerck, 6`eme Colloque International sur le Soudage et la Fusion par Faisceaux d’Electrons et Laser – CISFFEL’6, Toulon, 1998, pp. 205– 213. [8] B. Hohenberger, C.-L. Chang, C. Schinzel, F. Dausinger, H. Hˆugel, ICALEO’99, Section D 2 (1999) 167–176. [9] F. Dausinger, in: X. Chen, T. Fujioka, A. Matsunawa (Eds.), Proceedings of the SPIE-High-Power Lasers in Manufacturing, vol. 3888, 2000, Osaka, Japan, pp. 367–379. [10] G. Casalino, L.A.C. De Filippis, A.D. Ludovico, ICALEO’2001, Section E (2001) 524. [11] J. Eastman, D. Smith, ICALEO’99, Section D 2 (1999) 187– 196. [12] M. Grupp, T. Seefeld, G. Sepold, ICALEO’2001, Section G (2001) 1607. [13] S. Katayama, N. Seto, J.-D. Kim, A. Matsunawa, International Institute of Welding Doc. IX-1949-99 (1999) 1–10. [14] D. Lindenau, G. Ambrosy, P. Berger, H. Hˆugel, ICALEO’2001, Section A (2001) P308. [15] M. Naeem, ICALEO’99, Section D 2 (1999) 206–215. [16] D.G. Eskin, L. Suyitno, Katgerman, Progr. Mater. Sci. 49 (2004) 629–711. [17] E. Wettinck, Revue de la soudure 3 (1997) 10–13. [18] A. Matsunawa, J.-D. Kim, S. Katayama, International Institute of Welding Doc. IV-681-97 (1997) 1–9. [19] A.S. Salminen, ICALEO’2001, Section C (2001) 1704. [20] E.F. Cical˜a, Metode de prelucrare statistica a datelor experimentale, Politehnica, Timisoara, Romania, 1999. [21] D.C. Montgomery, Design and Analysis of Experiments, Wiley, Singapore, 1991. [22] A. Nichici, E.F. Cical˜a, in: Proceedings of the International Conference on LASER ’97, New Orleans, 1997, pp. 461–468. [23] J. Alexis, M. Taguchi, Practica Industriala, Tehnica, Bucarest, Romania, 1999.
(1) The presence of filler wire and its dependencies represents a principal influence on the mechanical properties of welding in general and on hot cracking in particular. (2) The most influential factors for the response functions studied, namely, tensile strength and the number of open cracks, are: welding speed; fastening system; wire parameters. The best results were given by low welding speeds and a uniform compression fastening system. (3) There are optimum values for wire feed rate and position. High feed rates produce instabilities, whereas low feed rates do not sufficiently modify the chemical composition of the molten pool. (4) Shielding gas parameters have only a small influence on hot cracking. (5) Welding stability is also very important, as instability generates cracks and vitiates the weld’s mechanical properties. Temperature dispersion gives information related to the stability of the process. (6) Using the polynomial model, tensile strength can be estimated as a function of the two most important factors, namely, welding speed and filler wire quantity. (7) On the basis of the influence of the operating factors, the phenomenological influences on hot cracking can be placed in order of importance, leading to an understanding of how these factors may be modified to improve tensile strength.
Acknowledgements This paper is written within the framework of the ASA (All´egement des Structures en A´eronautique) research pro-
Hot cracking in Al–Mg–Si alloy laser welding – operating parameters and their effects E. Cical˘aa , G. Duffetb,∗ , H. Andrzejewskib , D. Greveyb , S. Ignata,b b
a Polytechnic University, Mechanical Faculty, 1 blv. Mihai Viteazu, 300222 Timisoara, Romania Universit´e de Bourgogne, Laboratoire Laser et Traitements des Materiaux (LTm), 12 rue de la Fonderie, 71200 Le Creusot, France
Received 1 June 2004; received in revised form 17 November 2004; accepted 18 November 2004
Abstract Hot cracking is a phenomenon that frequently occurs in the laser welding of some “special” alloys, such as the aluminium–magnesium–silicon type. Each occurrence of this phenomenon needs to be studied in itself, taking into account not only the individual, but also the interactive, influences of the various parameters. The advantage of using laser beams in welding processes lies in the speeds that can be reached. The disadvantage, however, is that, owing to the high cooling rates characteristic of the interaction between the laser beam and the material, the welding speed itself becomes a cause of hot cracking. The aim of this paper is to see how this disadvantage may be eliminated. We consider what the most important parameters may be, relating to tensile strength and the quantity of cracks produced, that might influence the presence or absence of hot cracking. The most influential factors in avoiding hot cracking are the welding speed and wire parameters. Also important is welding stability, as instability generates cracks. We can then determine a technological window, useful for industrial applications, which takes into account the values of these influential factors and stability. © 2004 Elsevier B.V. All rights reserved. Keywords: Laser welding; Aluminium welding; Hot cracking
1. Introduction Owing to their low density and good mechanical properties, aluminium alloys are increasingly employed in many important manufacturing areas, such as the automobile industry, aeronautics and the military. Using aluminium alloys implies the development of assembly processes, especially in welding, where conventional techniques have shown their limitations. Laser welding, therefore, has progressively attracted the attention of scientists during the last decade [1,2]. In the laser welding of some aluminium alloys, many types of defects have been revealed, such as porosity, cavities and hot cracking [3,4]. Hot cracking is a defect manifesting itself as a surface crack during the solidification of a metallic alloy. Metallic alloys do not have a good deforming capacity during the so∗
Corresponding author. Tel.: +33 3 85 42 43 17; fax: +33 3 85 42 43 29. E-mail address: [email protected] (G. Duffet).
0921-5093/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2004.11.026
lidification stage of a calorific process, and hot cracking takes place during the final stage of solidification [5,6], when the alloy is semi-solid. This cracking phenomenon occurs at a high percentage rate (85–95%). In the case of aluminium alloys, the cracks that appear during welding are produced by the direct interaction of many factors [7,8], such as: solidification shrinking and thermal tensions, which generate tensions and deformations; wide range of solidification; temperature and time-cycle of solidification speed; chemical composition of the alloy (a hot cracking domain) (Fig. 1) [9]); fastening system of the welding components, which can limit contraction. Current techniques for reducing hot cracking in the laser welding of aluminium alloys usually relate to the above factors [10–15]. In a recent review, Eskin et al. [16] presented results from the last 50 years on hot tearing of aluminium alloys. They conclude that a generic quantitative criterion that will predict hot cracking under varying conditions is still not available.
2
E. Cical˘a et al. / Materials Science and Engineering A 395 (2005) 1–9
Fig. 1. Hot cracking sensitivity of aluminium alloys dependent on the Si and Mg content.
So, knowledge of the influence of the operating parameters on hot cracking during laser welding is an important step in establishing a criterion for prediction. The method most often employed to prevent the phenomenon of hot cracking consists in chemical modifications to the molten pool. Some alloys have a chemical composition within the range that presents a high risk of cracking [17]. To reduce this risk, wires having adequate alloying elements are added during the welding process. Nevertheless, welding may still be made difficult, or even impossible, unless a good correlation is achieved between the welding parameters [18,19]. The optimum correlation is hard to find, as the optimum zone, that is to say, the one in which no cracks are present, is very narrow [6]. Initially, butt-welding of plates was carried out, so that the operating parameters that influence hot cracking could be placed in order of importance. This process is fully described in this paper. Later, T-joint welding was carried out using two laser heads. The results of this process will be presented in a future paper.
2. Experimental strategy To clarify the phenomenon under review, any general approach should be designed with the aim of establishing from the outset the appropriate response functions and contributory factors. In our case, the “product”, that is, the final result of welding is the weld, which is the direct result of the solidification of the molten pool (Fig. 2). The characteristics of the
Fig. 2. Functional design of laser welding processes.
weld can be appreciated from different points of view: technological, aesthetic, economic, etc. As for laser welding, the factors influencing the response functions that characterise the weld can be put into two categories, namely, phenomenological factors and operating factors (Fig. 2). The phenomenological factors relating to the molten pool are: size, solidification speed, chemical composition, external tensions, gas shielding system, dynamics. The operating factors are: welding speed, laser power, nature of the shielding gas, focal point position, gap, gas flow. The operating factors are the welding setup parameters. Their values and interactions induce the phenomenological factors, which can be considered, in this context, as intermediate response functions (Fig. 2). From that, the phenomenological factors give the final result, namely, the weld. In order to control any phenomenon relating to laser welding and, in particular, hot cracking of aluminium alloys, the influence of the phenomenological factors needs to be understood. Among the operating factors influencing the shape and properties of the molten pool, and, consequently, the properties of the weld, we can identify absolute factors and relative factors. The absolute factors relate to the following: – The laser beam: its wavelength, power, power distribution, operating mode, spot size. – The welded materials: their chemical composition, properties, microstructure, geometry. – The shielding gas: its composition, flow rate, flow configuration. – The added material: its chemical composition, properties, geometry, shape. The relative factors are the following: – The position of the laser beam relative to the welded materials, to the shielding gas stream and to the added material. – The angle and distance relative to the welded materials. – The movement between the laser beam and weld material (welding speed). – The movement between the added material and the laser beam (feed rate of added material). We must emphasize here that, as in all phenomena relating to laser welding, and in particular the phenomenon of hot cracking, there are many parameters that also present internal interactions. When the parameters do not have precise levels, which are generally numerical, that will be a general example of the problem under review. When they have precise values and, consequently, an experiment takes place that will be an example of a particular case of the general problem. For every experiment, a value from the response area is obtained, which is specific to the studied phenomenon. In other words, the general case can be solved using experimental methods and data processing, usually based on statistical methods [20]. Therefore, using as parameters the influences on the phenomenon of hot cracking in the laser welding of aluminium alloys, a
E. Cical˘a et al. / Materials Science and Engineering A 395 (2005) 1–9
by the industrial world as an indication of internal and external weld quality, and, secondly, the number of open cracks, which can be considered proportional to the number of all cracks in the welded zone. The use of these types of response functions makes it possible, first, to establish and characterize the presence of cracks and, secondly, to investigate the effect of many diverse factors. The outcome will have practical consequences for manufacturing industries.
Fig. 3. Welding system design. Table 1 Chemical composition of plates to weld Alloy
Al 6056
3
Element (wt.%) Si
Fe
Cu
Mn
Mg
Zn
Ti
0.82
0.07
0.55
0.57
0.69
0.17
0.02
State
4. Design and results for the preliminary factorial experiment
T4
4.1. Factors that may influence hot cracking
factorial design method seems to be an appropriate solution [21,22]. The aim of the study presented in this paper therefore is to reveal, using a factorial design method, the influence of operating parameters on the response functions that are considered representative for hot cracking during Al–Mg–Si laser welding.
3. Problem definition The experimental conditions for studying the problem were as follows (Fig. 3): AA 6056 T4 aluminium alloy (Table 1) plates, 1.6 mm thick; butt-joint configuration; compression fastening system; full penetration welding; 3 kW cw Nd:YAG type laser beam; 150 mm focusing lens; 0.45 mm diameter focalized beam. For this case, the response functions that will give a quantitative picture of the dimensions and number of cracks in the welded zone are, first, tensile strength, which is accepted
Previous experiments run in our laboratory allowed us to choose the operating factors for the preliminary experiments and to study their influence on the response functions. These operating factors are: the type and feed rate of the added material (wire); the power of the beam; the nature and flow rate of the shielding gas; the gap between the materials to be welded; the welding speed; the position of the shielding gas and the wire relative to the laser beam and to the materials to be welded; the position of the focal point of the laser beam. To be able to place the analysed factors in order of importance, the principal response function employed was the tensile strength obtained with 30 mm wide samples. At the same time, the number of open cracks was revealed by penetrative inspection and by measuring the molten pool temperature with a pyrometer. 4.2. Structure of the factorial experiments Two Taguchi [23] type factorial experiments were designed to study the effects of the operating factors. The levels are given in Tables 2 and 3. The results given
Table 2 Influential factors for the first factorial experiment Factor level
Wire (–)
1 2 3 4
No 4047 – –
Shielding gas Position (–)
Angle
Ahead Behind – –
39 49 – –
(◦ )
Type (–)
Nature (–)
Feed rate (l/min)
Direct Direct + indirect – –
He N2 – –
20 30 – –
Laser power (kW)
Welding speed (m/min)
Focal point position (mm)
2.4 2.6 2.8 3.0
3.0 4.0 5.0 6.0
+0.5 0 −0.5 −1.0
Table 3 Influential factors for the second factorial experiment Factor level
Wire (–)
Gap (mm)
Wire position (mm)
Focal point position (mm)
Shielding gas nature (–)
Shielding gas flow (direct) (l/min)
Laser power (kW)
Welding speed (m/min)
Wire feed rate (mm3 /min)a
1 2 3 4
4043 5356 – –
0 0.2 – –
0.5 1.5 – –
0 −0.5 – –
He N2 – –
20 30 – –
2.4 2.6 2.8 3.0
3.0 4.0 5.0 6.0
0.785 1.570 2.355 3.140
a
For 4043 the wire diameter is 1 and 0.8 mm for 5356.
4
E. Cical˘a et al. / Materials Science and Engineering A 395 (2005) 1–9
Table 4 Chemical composition of added material Alloy type
4043 4047 5356
Wire diameter (mm)
1 1 0.8
Element (wt.%) Si
Fe
Cu
Mn
Mg
Cr
Zn
4.5–4.6 11–13 0.25
0.8 0.8 0.4
0.3 0.3 0.1
0.05 0.15 0.05–0.2
0.05 0.1 4.5–5.5
– – 0.05–0.2
0.1 0.2 0.1
Fig. 4. Temperature instabilities.
Table 4 relate to the chemical composition of the added material.
Fig. 6. Cracks.
4.3. Analysis of results In order to gain a precise value for tensile strength, three samples obtained in the same conditions were analysed for each value of tensile strength. The data processing for these preliminary experiments led us to the following conclusions. The operating parameters under investigation influence the shape and properties of the weld zone. The importance of their influence depends on the particular operating parameters and their variation range. The most regular welds were obtained with wireless welding and high energy. The filler wire produces perturbation in the welding process. The range that corresponds to regular welds is reduced by the presence of added wire, the level of reduction being dependent on the wire’s parameters. The variation in the temperature of the molten pool gives a picture of the stability of the welding process and of the uniformity of the weld (Figs. 4 and 5). The instabilities are often associated with defects such as cracks and porosity (Figs. 5 and 6). For the welds obtained without wire, or with a small quantity of filler wire, the solidification is different and leads more
Fig. 5. Weld instabilities.
Fig. 7. Wireless weld.
rapidly to cracks (Figs. 7 and 8) than when there is a large amount of filler wire (Fig. 9). The presence of wire mitigates the tendency towards hot cracking. Figs. 10 and 11 show the effects of the relevant factors. These may be summarized as follows. The presence of wire,
Fig. 8. Wire weld (0.196 mm3 /mm).
E. Cical˘a et al. / Materials Science and Engineering A 395 (2005) 1–9
5
Table 5 Parameters of the best weld derived from the preliminary experiments Wire (–) Wire feed rate (m/min) Wire position (mm) Gap (mm) Laser power (kW) Focal point position (mm) Welding speed (m/min) Gas nature (–) Gas position (–) Gas angle (◦ ) Gas flow – direct (l/min) Wire diameter (mm)
4047 3 1.5 0 2.8 −1 3 N2 Behind 39 30 1
Fig. 9. Wire weld (0.628 mm3 /mm).
After analysing this first group of results, we were able to establish a starting-point for the next series of experiments, shown as Table 5, which presents the parameters giving the best weld.
5. Experimental modeling 5.1. Exploration of the optimal zone
Fig. 10. Influential factors – experiment 1.
and its position and feed rate, have the greatest influence on tensile strength. Welding speed is also an important influence on tensile strength. The parameters of the shielding gas, its nature, position, protection type, flow rate, angle, have the lowest importance, of the studied parameters, for tensile strength. From a mechanical angle, and for producing more regular welds without open cracks, the best results were obtained using a low welding speed (3 m/min), high laser power (3 kW) and a focal point placed either on or just under the surface with zero gap.
Fig. 11. Influential factors – experiment 2.
The preliminary experiments, referred to in the first part of this paper, enabled us to place the factors relevant to hot cracking in order of importance. The experiments also established the basis for supplementary factorial experiments. These were designed to study the influence on the response functions of the following factors: welding speed, wire feed rate, laser power, focal point position, wire position and fastening system. The experiments were conducted in the following conditions: 4047 type filler wire, 1 mm diameter, placed in front of the molten pool, at 22◦ to the horizontal; N2 shielding gas, direct – 30 l/min, and indirect (under the weld) – 15 l/min, placed behind the molten pool, at 39◦ to the horizontal; zero gap. Table 6 summarizes the different levels of operating factors. Tests for tensile strength were conducted, as before, on three identical welds to obtain more accurate values. After data processing, the following conclusions can be formulated: (1) The fastening system has a major influence on the weld’s stability and mechanical properties. The best results are obtained with a uniform compression system (two fastening points) – see Figs. 12 and 13. (2) Having the focal point at the surface gives the best results – see Fig. 14. (3) There is an optimal position for the wire relative to the laser beam and welding plates. In the configurations of this experiment, that position is at 1.5 mm in front of the laser spot and at 0.3 mm above the welded plates. This position guarantees maximum stability and the best mechanical properties for the weld (Fig. 15).
6
E. Cical˘a et al. / Materials Science and Engineering A 395 (2005) 1–9
Table 6 Influential factors for the second factorial experiment Factor level
Welding speed (m/min)
Laser power (kW)
Wire feed rate (m/min)
Focal point position (mm)
Wire position (mm)
Fastening points (–)
1 2 3 4
2.5 3.0 3.5 –
2.7 2.8 2.9 3.0
2.5 3.0 3.5 –
0 −0.5 −1.0 –
0.5 1.0 1.5 –
1 2 – –
Fig. 12. Influence of fastening system on tensile strength.
Fig. 14. Influence of focal point position.
(4) When the wire feed rate is greater than about 3 m/min, tensile strength is attenuated, leading to a substantial dispersion of values. The cause of this is instability in the wire’s movements (Fig. 16). (5) Welding speed greatly affects the weld’s tensile strength. High speeds, of over 5 m/min, lead to fewer cracks and are not detrimental to mechanical operation. The effects induced by the welding speed, wire speed and laser power, as well as the interactions between them, can be analysed from Fig. 17. The most important, in this experimental field, are the filler wire and the welding speed together with the interaction between the laser power and the wire feed rate.
Fig. 15. Influence of wire position.
Fig. 13. Welds obtained with the same operating parameters: using two fastening points (a) and one fastening point (b), respectively.
E. Cical˘a et al. / Materials Science and Engineering A 395 (2005) 1–9
Fig. 16. Wire feed rate influence on tensile strength.
Fig. 17. Factors, and their interactions, affecting tensile strength.
5.2. Experimental model for the estimation of tensile strength The results obtained allow the influence of the two most important factors to be studied, that is to say, that the response functions that characterize hot cracking in laser welding are the welding speed and the wire feed rate.
7
Fig. 18. Variations in quantity of open cracks relative to welding speed and wire feed rate.
The particular problem is defined by all the conditions mentioned above. Supplementary conditions, which guarantee the best welds, are: a two-point compression fastening system; a focal point positioned on the surface of the welding parts; an optimum wire position of 1.5 mm; laser power of 3 kW. The shielding gas used in this case is helium. The factorial design is presented in Table 7. For each experiment, the volume of deposited wire per mm of weld was computed, to give a more complete image than wire feed rate alone for the influence of the added material on the response functions. As Fig. 18 shows, the number of cracks strongly depends on welding speed, this variation being exponential. With a welding speed of 3 mm/min, there are no open cracks. When the speed is 6 mm/min, however, open cracks, in the range of 20–100 in number, can be detected, this number being reduced slightly when the wire feed rate increases. The results of the factorial design as presented in Table 7 enable us to create, by multiple regression, a detailed second degree polynomial model (1), so that we may estimate, in this experimental field, the average tensile strength per mm of weld (Y) as a function of welding speed (x1 ) and wire volume deposited per mm of weld (x2 ). The coefficients of this model are presented in Table 8 and the graphical representation of this model response surface (for tensile strength) in Fig. 19. Y = b0 + b1 x1 + b2 x2 + b12 x1 x2 + b11 x12 + b22 x22
(1)
Table 7 Factorial design structure for the experimental model Experiment number
Welding speed (m/min)
Wire feed rate (m/min)
Wire volume/mm of weld (mm3 /min)
Experiment number
Welding speed (m/min)
Wire feed rate (m/min)
Wire volume/mm of weld (mm3 /min)
1 2 3 4 5 6 7 8
3 4 5 6 3 4 5 6
3 3 3 3 2 2 2 2
0.785 0.589 0.471 0.393 0.523 0.393 0.314 0.262
9 10 11 12 13 14 15 16
3 4 5 6 3 4 5 6
1 1 1 1 0 0 0 0
0.262 0.196 0.157 0.131 0 0 0 0
8
E. Cical˘a et al. / Materials Science and Engineering A 395 (2005) 1–9
Table 8 Coefficients of the polynomial modela b0 b1 b2 b12 b11 b22
6. Influence of phenomenological factors on hot cracking 43.3 −0.565 21.0 Non-significant Non-significant −18.3
a
For the computations using formula (1) welding speed (x1 ) should be in m/min, the wire volume (x2 ) in mm3 /mm and the estimated tensile strength (Y) results in daN/mm of weld.
Fig. 19. Variations in tensile strength relative to welding speed and wire feed rate.
The analysis of the degree to which tensile strength is dependent both on welding speed and on wire volume per weld mm (Fig. 19) shows that the mechanical strength of welds, caused by the increasing number of open cracks, decreases with welding speed and has an optimum value relative to the wire volume deposited per weld mm. This optimum value is slightly modified by the welding speed. As Fig. 20 shows, there are minor differences between the estimated and measured values of tensile strength per weld mm. This model was also employed for interpolations in a range of welding speeds ranging from 4.25 to 4.75 mm/min, the estimations being also close to measured values.
Fig. 20. Comparison between estimated and measured values for tensile strength/mm of weld.
By means of the factorial design presented above, the operating factors that influence hot cracking can be placed in order of importance. By considering the operating factors together with the phenomenological factors, it is then possible to determine the influence of phenomenological factors on hot cracking. The presence of filler wire with an appropriate chemical composition and feed rate is a principal factor, since it acts on the chemical composition of the molten pool, placing it outside the range in which hot cracking might occur. This phenomenon, which was already known, was verified in our experiments. Injecting supplementary material into the molten pool is, however, a delicate operation that can provoke defects in the weld. The materials to be added will depend on the operating factors relating to the filler wire. These are, for example, its longitudinal distance from the welded parts and its free length at the mouth of the nozzle. When the experimental configurations are optimal, the wire must reach the molten pool in a liquid state. This means that we have a continuous heat transfer into the molten pool and minimal reflectivity, as opposed to the situation when the wire reaches the molten pool in a solid state. Furthermore, optimum wire parameters allow for a better contact between the laser beam and the materials to be welded, the outcome being a stable molten pool with low values for the temperature gradient. The presence of filler wire also improves the weld’s morphological quality. For, as the 4047 filler wire has low viscosity, the superficial tension of the molten pool is thereby decreased. Welding speed is another important influence on solidification speed and the temperature gradient. Thus, the welding speed influences the deformation and tension of the molten pool during solidification. Since high welding speeds are desired, other factors must be adapted in order to balance the tensions and deformations undergone by the semi-solid welding alloy from the molten pool. Among these other factors, the fastening system plays an important role. Thus, to avoid cracks, a uniform compression fastening system must be designed. The solidification speed and the temperature gradient can also be reduced by the presence of a second heat source. The calorific supplement has an important influence, through its position and quantity, on hot cracking. Another study needs to be done on the relationship between laser beam configuration and hot cracking. The presence of gas shielding, acting on the oxidation and surface state of the molten pool, has a secondary importance for the hot cracking, relating to the chosen response functions of tensile strength and open cracks. Further, some fatigue tests made on welds obtained with nitrogen and helium as the shielding gas, with all other parameters being constant, revealed no differences. A more
E. Cical˘a et al. / Materials Science and Engineering A 395 (2005) 1–9
9
systematic study should enable us to understand the nitride role in hot cracking.
gram of the R2IT mat´eriaux et proc´ed´es network. This network is supported by the French Industry Ministry.
7. Conclusion
References
The hot cracking phenomenon in laser welding of aluminium alloys is complex. In order to study the influence of operating factors, the factorial design approach seems to be an appropriate solution. From our study, the following conclusions can be formulated:
[1] E. Schubert, M. Klassen, I. Zerner, C. Walz, G. Sepold, J. Mater. Process. Technol. 115 (2001) 2–8. [2] A.P. Costa, L. Quintino, M. Greitmann, J. Mater. Process. Technol. 141 (2003) 163–173. [3] D. Fabr`egue, Rapport de Diplˆome d’ Etudes Approfondies DEA de Science et G´enie des Mat´eriaux, I. N. P. Grenoble, 2001. [4] A. Haboudou, P. Peyre, A.B. Vannes, G. Peix, Mater. Sci. Eng. A363 (2003) 40–52. [5] M. Braccini, M´emoire de th`ese I. N. P. Grenoble, 2000. [6] E. Schubert, M. Klassen, J. Skupin, G. Sepold, 6`eme Colloque International sur le Soudage et la Fusion par Faisceaux d’Electrons et Laser – CISFFEL’6, Toulon, 1998, pp. 195–203. [7] D. Boisselier, R. Lenoir, O. Wall´ee, J.M. Staerck, 6`eme Colloque International sur le Soudage et la Fusion par Faisceaux d’Electrons et Laser – CISFFEL’6, Toulon, 1998, pp. 205– 213. [8] B. Hohenberger, C.-L. Chang, C. Schinzel, F. Dausinger, H. Hˆugel, ICALEO’99, Section D 2 (1999) 167–176. [9] F. Dausinger, in: X. Chen, T. Fujioka, A. Matsunawa (Eds.), Proceedings of the SPIE-High-Power Lasers in Manufacturing, vol. 3888, 2000, Osaka, Japan, pp. 367–379. [10] G. Casalino, L.A.C. De Filippis, A.D. Ludovico, ICALEO’2001, Section E (2001) 524. [11] J. Eastman, D. Smith, ICALEO’99, Section D 2 (1999) 187– 196. [12] M. Grupp, T. Seefeld, G. Sepold, ICALEO’2001, Section G (2001) 1607. [13] S. Katayama, N. Seto, J.-D. Kim, A. Matsunawa, International Institute of Welding Doc. IX-1949-99 (1999) 1–10. [14] D. Lindenau, G. Ambrosy, P. Berger, H. Hˆugel, ICALEO’2001, Section A (2001) P308. [15] M. Naeem, ICALEO’99, Section D 2 (1999) 206–215. [16] D.G. Eskin, L. Suyitno, Katgerman, Progr. Mater. Sci. 49 (2004) 629–711. [17] E. Wettinck, Revue de la soudure 3 (1997) 10–13. [18] A. Matsunawa, J.-D. Kim, S. Katayama, International Institute of Welding Doc. IV-681-97 (1997) 1–9. [19] A.S. Salminen, ICALEO’2001, Section C (2001) 1704. [20] E.F. Cical˜a, Metode de prelucrare statistica a datelor experimentale, Politehnica, Timisoara, Romania, 1999. [21] D.C. Montgomery, Design and Analysis of Experiments, Wiley, Singapore, 1991. [22] A. Nichici, E.F. Cical˜a, in: Proceedings of the International Conference on LASER ’97, New Orleans, 1997, pp. 461–468. [23] J. Alexis, M. Taguchi, Practica Industriala, Tehnica, Bucarest, Romania, 1999.
(1) The presence of filler wire and its dependencies represents a principal influence on the mechanical properties of welding in general and on hot cracking in particular. (2) The most influential factors for the response functions studied, namely, tensile strength and the number of open cracks, are: welding speed; fastening system; wire parameters. The best results were given by low welding speeds and a uniform compression fastening system. (3) There are optimum values for wire feed rate and position. High feed rates produce instabilities, whereas low feed rates do not sufficiently modify the chemical composition of the molten pool. (4) Shielding gas parameters have only a small influence on hot cracking. (5) Welding stability is also very important, as instability generates cracks and vitiates the weld’s mechanical properties. Temperature dispersion gives information related to the stability of the process. (6) Using the polynomial model, tensile strength can be estimated as a function of the two most important factors, namely, welding speed and filler wire quantity. (7) On the basis of the influence of the operating factors, the phenomenological influences on hot cracking can be placed in order of importance, leading to an understanding of how these factors may be modified to improve tensile strength.
Acknowledgements This paper is written within the framework of the ASA (All´egement des Structures en A´eronautique) research pro-