Development of a direct metal freeform fabrication technique using CO2 laser welding and milling technology

Development of a direct metal freeform fabrication technique using CO2 laser welding and milling technology

Journal of Materials Processing Technology 113 (2001) 273±279 Development of a direct metal freeform fabrication technique using CO2 laser welding an...

369KB Sizes 0 Downloads 19 Views

Journal of Materials Processing Technology 113 (2001) 273±279

Development of a direct metal freeform fabrication technique using CO2 laser welding and milling technology Doo-Sun Choia,*, S.H. Leeb, B.S. Shina, K.H. Whanga, Y.A. Songc, S.H. Parkc, H.S. Jeed a

Korea Institute of Machinery and Materials (KIMM), P.O. Box 101, Jangdong, Taejeon, South Korea b Department of Mechanical Engineering, Yonsei University, #134 Shinchon, Seoul, South Korea c Korea Institute of Science and Technology (KIST), P.O. Box 131, Cheongryang, Seoul136 791, South Korea d Department of Mechanical Engineering, Hong-Ik University, 72-1 Sangsudong, Seoul, South Korea

Abstract Since the ®rst introduction of rapid prototyping in 1986, several techniques have been developed and successfully commercialized in the market. However, most commercial systems currently use resins or waxes as the raw materials. Thus, the limited mechanical strength for functional testing is regarded as an obstacle towards broader application of rapid prototyping techniques. To overcome this problem, direct metal deposition methods are being investigated worldwide for rapid prototyping and even for rapid tooling applications. As a contribution to this development, a fundamental study on a process combination of wire welding technology using CO2 laser radiation with milling was carried out and is reported in this paper. Laser welding enables accurate deposition of metals and the subsequent milling increases the surface quality and accuracy to machining standard. Compared to powder, the use of wire is of advantage in terms of a simple feeding mechanism as well as a higher deposition rate. The main focus of the experimental investigation is to ®nd the basic process characteristics. For this purpose, basic parts were fabricated as a function of process parameters such as laser power, welding speed and bead distance. The microstructure, hardness and tensile strength are then examined as a function of these process parameters. In conclusion, the advantages and disadvantages of this process are discussed in comparison with other direct metal fabrication techniques. # 2001 Elsevier Science B.V. All rights reserved. Keywords: Rapid prototyping and tooling; Metal deposition; Laser welding; Mild steel; Metal fabrication

1. Introduction To remain competitive in the international market, it is important to respond quickly to the changing market and produce products that re¯ect these changes. This capability along with the quality and price of the products are the most important factors for survival. The lead time involved in product development is considered the most cost- and timeconsuming process of development, and therefore CAD/ CAM technology has been employed to meet these technical demands. CAD/CAM has revolutionized the time required for numerical data to be transferred into manufactured products. Rapid prototyping, which is a direct transfer of design data to an actual product, is a very important part of CAD/CAM technology. The fact that an actual model can be created and tested is concurrent to the technology of CAD/CAM, and for these reasons, rapid prototyping has received much attention from its initial development. Changes that occur in the design * Corresponding author. Tel.: ‡82-42-868-7124; fax: ‡82-42-868-7149. E-mail address: [email protected] (D.-S. Choi).

process will be least costly compared to changes made in other stages of product development. Because of these advantages there has been a constant explosion of interest in the research of rapid prototyping. The concept of automating the procedure of transferring design data to an actual product in an automated way is the critical component of development automations in use today [1]. Research in RPs date back to the mid 1980s. Since then there has been a rapid advance in its development. The technology involved can be classi®ed into two groups: those that are being in use today, and those that are under development. Commercialized technologies include laser utilized stereolithography, SLS, FDM which integrates extruded thermoplastic resin, LOM which integrates thin layers of the product's sections, and LENZ which is an advanced form of clading technology using laser melted metal particles [2±6]. Excluding the SLS technology by the DTM company, other widely used commercial rapid prototyping techniques such as FDM, LOM, etc. utilize plastic, paper or polymer and have the disadvantage of not being able to produce actual metal products [7].

0924-0136/01/$ ± see front matter # 2001 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 4 - 0 1 3 6 ( 0 1 ) 0 0 6 5 2 - 5


D.-S. Choi et al. / Journal of Materials Processing Technology 113 (2001) 273±279

1.1. Metal deposition Unlike conventional rapid prototyping systems where non-metallic materials were used to form the products, current research focuses on integrating metal layers to form an actual metallic model. The metal deposition method that is in commercial use today is a by-product of research in this ®eld. However, accuracy problems that arise in the process of rapid prototyping, along with problems in the characteristics of the metal and the high cost of production, have impeded further development. Metallic integration techniques include a variety of methods such as laser cladding, ballistic particle manufacturing (BPM), droplet-based manufacturing (DBP), 3D printing, shape deposition manufacturing (SDM), direct metal deposition (DMD), and selective laser sintering (SLS) [8±13]. A few critical methods are introduced below. Steen and his associates have been used laser cladding to form 3D structures. This method is considered as a rapid prototyping method based on laser welding. Regular beads are formed using coaxial metal powder supply equipment through a control technique for bead height. Products are formed based on this technology [14]. Prinz and his associates have proposed microcasting, 5-axis CNC milling, shaping through electrical discharge machine (EDM), stress control through shot peening, or a metal cladding rapid prototyping process through an integrated sensing system in the SDM process. They have also proposed that cooling rate, metallic defaults and internal stress are the critical factors in determining the quality of the metal-composed product. Mazumder proposed direct metal deposition methods using cladding techniques along with applications for the direct production of products, low cost and fast casts for order and die production. He predicted 40% of the conventional die production time as well as millions of dollars in savings. He is also proposing that by using lasers for cladding, the thermal stress which occurs during metal integration could be greatly reduced [15,20,21]. On the same note, Kreutz and associates also claim that the use of the laser is a critical method for realizing metal integration processes. The laser is already widely used in cutting, welding and surface treating procedures as well as being recognized as a production process tool that can be usefully applied in expansive large scale production [8]. The DTM company, in association with the University of Texas, has further advanced the SLS process in developing a commercial rapid metal prototyping product called Sinterstation, which can be an alternative for liquid substance RP & M. In this process, metal powder was used instead of Acrylat. The metal powder was heated to just below the melting point, after which a laser scan would instantly melt the powder to form a layer of the product. The rest of the powder will remain to support the melted layer and prevent distortion. When a cycle is completed, a roller reapplies the powder and the process is repeated. Since the product is formed through deposition, precision of surface quality is

inevitably compromised. A post process is required. In this aspect, layer integration systems coupled with milling postprocessing are very realistic candidates for further research. The rapid prototyping system proposed by Song et al. [16] utilizes technology which has been in existence before current developments in rapid prototyping started, resulting in a cost effective and practical rapid prototyping system. This led to metal layer integration via welding, which is later post-processed through milling, and through this technique basic products have been successfully produced [17±19]. This study is based on the same line of thought. The main concept can be described as a rapid prototype mold or tooling in a wire welding method using laser techniques which are suitable for rapidly producing products such as large scale molds. Also, this system utilizes a simpli®ed wire feeding device compared to the delicate metal powder feeding systems required in previous rapid prototyping systems. This rapid prototyping system can be classi®ed as 3D welding technology in a broad sense, and its categorization is shown in Fig. 1 [8]. In this study, metallic materials are melted through laser heating and arc welding to form sections and integrated in layers. After the shape is formed, cutting is applied to enhance the surface roughness. By combining layer integration and cutting techniques into one and utilizing the good points of each, shapes that were impossible to manufacture with cutting alone can be formed. Also precision and surface accuracy, which were two weaknesses related to layer integration, could be fully attained.

Fig. 1. The RP technical tree [8].

D.-S. Choi et al. / Journal of Materials Processing Technology 113 (2001) 273±279


Therefore, in this study, optimal conditions for bead formation are determined through laser and arc welding. Using this condition, the shape of the product is formed and various mechanical characteristics are tested. Ultimately, a prototype for injection molding can be manufactured. 2. System 2.1. Hybrid RP rapid prototyping process A laser beam is projected onto the material surface while a wire is introduced to the focal point of the laser at a steady speed to induce a melt pool. The X±Y table is translated to form a one line bead which is then integrated to a multi-layer structure. The fundamental routine involved is a repetition of this process in Fig. 2. The one line bead is overlapped horizontally to form a two-dimensional layer, which is then integrated vertically to form a three-dimensional structure. After the bead or the 2D plane is constructed, a milling machine is used to process the top part of the beads by cutting. This provides a ¯at and stable surface on the top of the beads upon which additional layers can be stacked with greater stability. After the 3D structure is completed, the surface is processed by a ®nishing cutting process. (Fig. 2) The result is the ®nal product. 2.2. Hybrid RP rapid prototyping equipment Test equipment can be largely subdivided into three parts: CO2 laser or arc welding equipment, a regular milling machine and the wire feeding apparatus. For the CO2 laser equipment, a product from Ro®n±Sinar (maximum output: 1.5 KW), Germany was used. Conventional regular milling machines were inadequate because the control unit could not be applied to the study. Therefore, an integrated PC±NC based control unit suitable for rapid prototyping was developed. The new control unit is a 5-axis simultaneous control system which was designed to control laser welding, milling and wire feeding processes. The wire feeding system is actuated by servo motors for a steady feed of the wire into the framework. In actual testing, the direction of the laser beam in relation to the direction of the wire feed proved to be very critical, and so the wire feed direction was aligned with the direction of the shape forming process. The wire feed angle can also be adjusted manually for optimal effect. The

Fig. 2. The 3D laser welding and milling process.

Fig. 3. Schematic diagram of the rapid direct metal deposition machine.

equipment set-up is detailed in Fig. 3. The principal material used for processing is mild steel, also used as the wire element in experiments. Nitrogen gas was used to deter oxidation during layer integration, and the entire process was captured using a CCD camera and monitored through a monitor. 2.3. Software composition The data utilized in rapid prototyping is the two-dimensional data obtained from CLI (common layer interface) ®les. Since CLI ®les only contain information about the internal and external outlines of the product, further information besides the two-dimensional dissection information is required. The extra data in this case is de®ned as process variables. They include information of welding speed for ®lling the dissection gap, the distance between the bead and the welding bead, the cutting speed for leveraging the surface height and other information related to welding, cutting, dissection ®lling, welding layer integration direction, etc. Fig. 4 is a ¯owchart of the rapid NC process and Fig. 5 is a rapid NC composition chart. In rapid NC, the data collected from experiments is stored and managed in a collective database. Two-dimensional data is read from CLI ®les and stored in binary ®les called RPG ®les. Using the RPG ®les, the tool path data are generated and translated to NC data format which can later be used in rapid prototyping processes. 2.3.1. Process variables The variables that in¯uence the quality of the end product can be divided into ®xed and un®xed variables. Fixed

Fig. 4. Flowchart of rapid NC.


D.-S. Choi et al. / Journal of Materials Processing Technology 113 (2001) 273±279

Fig. 5. Display of rapid NC. Fig. 6. External and internal path of welding.

variables refer to values which cannot be altered for each process and include values such as welding wire type and size, type of gas used, principal base material, welding frequency, welding voltage, current, laser power and wire feed rate. Un®xed variables on the other hand are values which can be changed for each different process to achieve better quality and consume less time. The type of un®xed variable and its optimum value are determined by experimentation and are subdivided into technology, parallel, cutting and path process variables, depending on their function. 2.3.2. Technology process variables Technology process variables affecting the bead quality and size are the distance between the beads, the welding speed and compensation values. Since the bead thickness is inversely proportional to the welding speed, the thickness and width can be controlled by adjusting the welding speed. When welding the external surface, the bead size must be smaller than that applied to the dissection. This can be controlled by setting the external welding speed higher than that of the internal speed. Other problems include the bead ¯owing out towards the welding direction at the end of a weld. To compensate for this, the welding direction is slightly offset in the opposite direction near to the end. The compensation value determines the compensation amount. 2.3.3. Parallel process variable Parallel process variables are used to select the layer integration type. The values include outline welding and internal section welding options, the priority of welding processes, the scale value, the layer integration procedure during internal section welding and the integration direction selection between layers. Dissections formed in welding are divided into outlines and internal sections which can be selectively generated by manipulating the process values. Scale values generate welding paths by enlarging or reducing the outlines or internal sections actually formed in the process. When the outlines and internal sections require simultaneous welding, the scale value may be used to alleviate overlap. Fig. 6 shows the internal section and outline welding paths generated by rapid NC. In this case

the Nth layer is being integrated in a cellular phone mold core. 2.3.4. Cutting, path process variables Cutting process variables determine the cutting process types. Cutting is administered after every layer integration to ensure that the dissection surface is level. The values include the tool radius, cutting speed, z-axis direction or side cutting selection and cutting path in the z-direction. Path process variables include the operation initiation and conclusion height along with layer rotation selection values pertaining to the welding process. 2.4. Experiment variables Laser power determines the output needed to melt the wire. Table speed and wire feed rates determine the bead thickness and width, while the protective gas prevents oxidation of the bead. The key values in the experiment are laser power, table speed and wire feed rate. The variables required during the layer integration and composition of the wire are listed in Tables 1 and 2. A single-line experiment was conducted to optimize these values. The same values were used in generating the basic geometric model. Table 1 Parameters and range of experiments Parameter


Laser power (P) Table speed (Ft) Wire feed rate (Fw) Shield gas pressure Spot size (diameter) Angle of wire feeding

600±700 W 200±600 mm/min 400±600 mm/min 10±15 l/min 2.5±3.0 mm 20±30 8

Table 2 Composition of the wire (wt.%) C










D.-S. Choi et al. / Journal of Materials Processing Technology 113 (2001) 273±279

Fig. 7. Flowchart of the experiments.

2.5. Experimental procedure Single-line experiments and layer integration experiments were conducted separately. In the single-line experiment, the feed angle, the laser power and the wire feed rate were controlled while the table speed was varied to determine its effects on bead formation. In the next stage, the laser power and table speed were controlled to determine the effects of the wire feed rate variation. Laser power variation experiments were conducted in the same way. Repeated experiments were conducted along with investigations into the effects of the laser spot size and the pressure exerted from protective gases. Fig. 7 shows the ¯owchart of the experiments. In these experiments, properly formed beads were isolated and measured in terms of height and width and relations to variables. In layer integration experiments, the optimal conditions and bead-to-bead distances were varied in order to obtain conditions for generating a level surface. 3. Experiment results and discussion 3.1. Dissection experiment In this experimental set-up, the wire feeding mechanism supplies the wire to the focal point of the laser. Therefore the layer integration direction can only be formed in the xdirection on the x±y plane. Due to these characteristics, feeding cannot occur along the z-axis and beads will be formed along the line between the point of the wire feed and the melt pool formed by the laser. Layer integration has similar directional characteristics so that bead formation along with layer integration occurs only when the wire feed direction coincides with the table translation direction in the x±y plane. In experiments, this occurred in the x direction. To select optimum melting conditions, a single-line test was conducted. Laser power and wire feed rates were controlled, whereas the table speed was varied for investigation. As the table speed was increased, the bead thickness and width decreased, while a decrease in table speed led to an increase in bead thickness and width. Additional experiments were conducted where the laser power and table speeds were controlled and the wire feed rate varied. An increase in wire feed rate led to increased bead thickness and


Fig. 8. Single line of the bead.

width. A decrease in wire feed rate resulted in decreased bead thickness and width. In conclusion, table speed varied inversely to bead size, while the wire feed rate varied proportionally. Similar experiments were conducted with laser power, table speed, wire feed rate, laser focal point size, and feed angle. The optimal conditions from these tests are listed in Fig. 8. 3.2. Layer integration experiment The results of the layer integration experiments show that basic beads form walls, which ultimately form hexagonal structures. These fundamental structures become the building blocks for the actual mold being manufactured. When data collected from dissection experiments was used in layer integration tests, the layer thickness and line width began to differ between layers due to differences in heat transfer conditions. For example, the laser power for the melting and welding processes in the ®rst layer was the highest, while the subsequent layers required less power in succession. This can be explained by the differences in heat-transfer rates. When beads are formed, the top part has a curvature. This led to the problem of the melted metal sliding down the sides when a new layer was being integrated. To make matters worse, some layers had poor surface ®nishing, leading to bigger defects as a result of integration. To provide a solution, a cutting process was implemented before adding a layer by removing 2 mm off the top of the previous layer. Beads with regular heights could be formed through this procedure. Also, any surface defects could be alleviated by removing them completely. Cutting off the top portion of the bead also prevented melted metal from sliding down the side. Fig. 9 shows the optimal conditions in layer integration and Fig. 10 shows the ®nal product.

Fig. 9. Thin wall.


D.-S. Choi et al. / Journal of Materials Processing Technology 113 (2001) 273±279

Fig. 10. Block.

3.3. Tensile specimen creation and tensile test Along with the basic geometry production, the rapid prototyped metallic part was also tested according to ASTM standards in tensile tests in order to evaluate the mechanical characteristics of the product. A specimen for the test was created. By creating the specimen, the precision and strength of the product could be assessed. Hence, whether this procedure would be adequate for a rapid prototyping mold process could also be determined. To test the strength, a standard ASTM tensile test specimen was created, as shown in Fig. 11. The wire used in laser welding was a AWE ER 70S-6 carbon steel wire of 0.9 mm diameter. The tensile test results showed a tensile strength of 55.66 kg/mm2 and an elongation percentage of 53%. The elongation percentage in this case is almost identical to that of conventional carbon steels. Table 3 below lists the comparison between conventional carbon steels and the specimen tested in the experiment. 3.4. Microstructure Fig. 12 shows an SEM image of a layer integrated bead. There is a clear indication of a general mild steel structure at the base. Beads integrated by welding base level can be

Fig. 11. Test specimen.

Fig. 12. Microstructure of a wall.

categorized as the same type of mild steel, although their compositions will vary slightly. [A] shows the base level, while Fig. 12 in [B] shows the HAZ (heat affected zone) directly under the beads in the initial layer of integration. [C][D][E] show the structure of the welded beads. [F] shows the martensite structure formed by rapid cooling at the top. [G] is an example of a bead from a different integration layer where the welding has not been completed properly leading to gaps between the beads. Such defects occurred when melting power was insuf®cient or when melting materials were insuf®ciently provided. The key focus of the experiment was determining the link structure between the beads and whether it would provide suf®cient strength and ductility. Oxidation composition charts at the dissection segment do not contain martensite structures. This is because the martensite structure is removed by the cutting that takes place after each layer has been integrated. Microstructure photos indicate that the structures formed at the intersection area of the beads (20±30 m) are larger than those formed inside a bead (5±10 mm). This growth in structure is assumed to be a result of the tampering effect when a new layer is added. The larger structure displays less strength in comparison to the internal portion of the bead [E], but it can be assumed to be less brittle and thus less prone to cracks, etc. The upper-most layer contains martensite structure due to rapid cooling as clearly shown in the picture. It is also important to note that bubbles were not observed during the layer integration process, which is an advantage when manufacturing molds. 3.5. Hardening

Table 3 Comparison between mild steel and test specimen of mechanical properties

Mild steel Test specimen

Yield strength (kg/mm2)

Tensile strength (kg/mm2)

Elongation (%)

47 *

56 55.7

30 52

Lasers, much like arc welding, allow effective control over heat-induced effects in comparison to other heat-related processes. Nevertheless, local hardening caused by heat is unavoidable in this case also. When injection molds are manufactured using this technique, the hardening effect may be bene®cial since the mold must have enough strength to withstand wear and crack formation. Heat-induced hardening is closely related to the

D.-S. Choi et al. / Journal of Materials Processing Technology 113 (2001) 273±279


5. Future research Rapid prototyping with a laser welding process allow the manufacture of many products which were previously dif®cult to produce with conventional cutting techniques. The following areas provide guidelines for future research:

Fig. 13. Micro-hardness.

heat distribution characteristics, and therefore, research in heat exchanges and distribution is of critical importance. However, experimental techniques are limited in handling high temperatures and heat distribution so that alternative methods utilizing numerical analysis are expected to provide better solutions in determining heat distribution. It is important to note that the melted intersection part of the layers has hardness values of about 20 Hv less than that of a regular layer. This softening effect can be attributed to the tampering effect caused by heat when a layer is integrated with another layer beneath it. This effect is most clearly observed in the lower portion of the beads. In terms of the entire bead stack, the hardness values decrease and converge as the height increases. At the very top, the martensite structure formed by rapid cooling provides a very hard structure. Fig. 13 shows the micro-hardness. 4. Conclusions 1. With fundamental experiments in rapid prototyping, the optimum variables were determined. Initial bead formation, fundamental vertical walls, hexagonal formations and tensile test specimens were created. From these basic parts, test data for manufacturing precision molds or metallic prototypes could be determined. 2. The characteristics of the entire part could be determined from investigating the microstructures on the integrated layers. A reliable mechanical connection between layers could be observed. Also the fact that bubble formation did not occur throughout the entire bead will be of bene®t when molds are actually manufactured. 3. Hardness and tensile tests on rapid prototype mold parts show that they have suf®cient hardness and dimensional precision to be applied to prototype mold production through the injection process.

1. The production time may be reduced by 50±70% by improving table speeds, processing at higher power and single processing being applied after the integration of multiple layers. 2. Improved precision: the precision needs to be improved suf®ciently for the process to be used for dental work or ®ne molds. 3. Reduction of thermal distortion and FEM analysis is required to test for residual stress. 4. Molds designed by rapid prototyping will be applied to injection processes and the processing time compared to that of conventional cutting methods. References [1] M. Murphy, C. Lee, W.M. Steen, ICALEO, 1993, p. 882. [2] L. Ahmad, L. Eckstrand, J. Pantarotto, Can. Ceramics Quart. 66 (2) (May 1997) 104. [3] Rapid Prototyping Report, CAD/CAM Publishing Inc., January, 1999, p. 4. [4] B.K. Paul, S. Baskaran, J. Mater. Process. Technol. 61 (1996) 168. [5] S. Ashley, Mech. Eng. 117 (7) (1995) 63. [6] D.A. Belforte, Laser Focus World (June 1993) 126. [7] M. Murphy, W.M. Steen, C. Lee, ICALEO, 1994, p. 31. [8] E.W. Kreutz, G. Backes, A. Gasser, K. Wissenbach, Appl. Surf. Sci. 86 (1995) 310. [9] G.K. Lewis, Mater. Technol. 10 (3-4) (1995) 51. [10] X. Yan, P. Gu, Comput. Aided Des. 28 (1996) 307. [11] J.H. Chun, C.H. Passow, CIRP Ann. 42 (1993) 235. [12] M.L. Murphy, W.M. Steen, C. Lee, ICALEO'94, 1994, p. 31. [13] S.A. Morgan, M.D.T. Fox, M.A. McLean, D.P. Hand, F.M. Haran, D. Su, W.M. Steen, J.D.C. Jones, ICALEO, 1997, p. 290. [14] J. Lin, W.M. Steen, ICALEO Sec. A, 1996, p. 27. [15] C.H. Amon, K.S. Schmaltz, R. Merz, F.B. Prinz, T. ASME 118 (1996) 164. [16] Y.A. Song, S.H. Park, J.K. Cho, D.S. Choi, K.H. Whang, B.S. Shin, H.S. Jee, KSPE Conference, 1998, p. 940. [17] D.S. Choi, K.H. Whang, B.S. Shin, Mach. Mater. 10 (1998) 130. [18] D.H. Kim, Manufacturing of Laser, Kyung Moon Sa, 1992. [19] N. Ichikawa, M. Misawa, S. Kano, N. Aya, H. Iwamoto, Y. Enomoto, ICALEO Sec. C, 1997, p. 216. [20] J. Mazumder, J. Choi, K. Nagarathnam, J. Koch, D. Hetzner, Summary of the Direct metal deposition of H13 tool steel for 3D components, JOM, 1997. [21] P.F. Jacobs, ICALEO'95, 1995, p. 194.