RAPID PROTOTYPING TECHNIQUE BASED ON 3D WELDING

R. Kovacevic
Southern Methodist University
Dallas, TX 75275
Phone: (214) 768-4865
e-mail: kovacevi@seas.smu.edu

Abstract: Rapid prototyping is one of the fastest growing automated manufacturing technologies that has significantly impacted the length of time between initial concept and actual part fabrication. However, to fully realize the potential cost and time savings associated with rapid prototyping, the capacity to go from CAD models directly to metal components and tooling is crucial. This paper will lay the foundation for developing a new rapid prototyping technique based on controlling the heat and mass transfer processes in gas metal arc welding. In 3D welding a droplet of the melted electrode wire is detached to form a metallic deposit. By properly depositing droplets, the metallic parts are made layer by layer. In order to accurately control the resultant shape of the part, as well as its mechanical and metallurgical properties, the drop size and detachment rate must be precisely controlled.

Introduction: Successfully responding to the ever changing and continually increasing high demands of today's global markets requires rapid product development and manufacture of new designs. Visualization tools often play a major role in taking an idea from the initial concept through the design phase, and into the final product development process. The rapid advancements in hardware and software in recent years has made computer graphics one of the leading tools for obtaining a two-dimensional or even three- dimensional representation of a product that is being created. Any designer will agree that three-dimensional graphical representations, often referred to as virtual prototypes, significantly expedite the design, evaluation, and production of a part. Thus, the old saying “A picture is worth a thousand words,” still stands. On this premise, it seems the next logical question would be "What is the value of a three-dimensional physical prototype?"

Substantial research in the area of rapid prototyping (RP) did not get started until the mid 1980's [1-3]. (The following terms are often used interchangeably when referring to rapid prototyping technology: solid free-form fabrication, desktop manufacturing, layered manufacturing, automated fabrication, and tool-less manufacturing.) The sudden boost in rapid prototyping research came with the availability of inexpensive computational equipment that allowed the three-dimensional geometry of a part to be transferred directly into control parameters for guiding rapid prototyping systems. The rapid prototyping approach to shape fabrication does not impose the constraints of conventional equipment, such as selection of part-specific cutting process tools, part- specific fixturing, and planning cutting trajectories for complex geometry. Rapid prototyping processes simplify the process planning and execution by decomposing three- dimensional geometry into layered two-dimensional representations and eliminate the need for part-specific tooling and fixturing.

Initially, rapid prototyping was viewed as a tool for the rapid development of three-dimensional models. The savings in terms of time and money mostly occurred during the development phase of a product. The driving force was the potential reduction in time to final design and prototype part due to the relative ease of doing the physical model iterations during the design stage. The prototype parts were simply three- dimensional models that were used to verify the feasibility of the design. Primarily the savings came from early detection of errors, which prevented costly redesigns. Current rapid prototyping techniques (Stereolithography, Selective Laser Sintering, Laminated Object Manufacturing, Fused Deposition Manufacturing, Droplet Deposition, and Three- Dimensional Printing) can only produce prototypes made from wax, plastic, nylon, and polycarbonate materials. The major rapid prototyping techniques and the names of the developers are listed in Table 1.

Metal parts with good surface quality, accurate dimensions, and high structural strength cannot be produced directly with the above named rapid prototyping techniques. Industry has expressed very strong interest to further develop current prototyping techniques or create new ones that are capable of directly producing metallic or ceramic parts. The transition from polymers to metals and ceramics is just beginning to take place. There are many more difficulties associated with producing a structurally sound, dimensionally accurate metallic part rather than simply a tangible three-dimensional model for visualization purposes. But if achieved, metallic parts will offer several advantages over plastic parts such as high strength, greater impact resistance and toughness, and greater durability. In addition, in situ testing of the formed prototype will be possible under practical working conditions.

Gas Metal Arc Welding as a Process for Rapid Prototyping: The rapid prototyping techniques that could be used to directly make metallic parts could be divided into two groups: 3D cladding [4-6] and 3D welding [7-16]. In 3D cladding a laser beam creates a weld pool into which powder is injected and melted. The substrate is scanned by the laser/powder system in order to trace a cross-section. Upon solidification, the trace forms a cross-section of a part. Consecutive layers are then additively deposited, thereby producing a three-dimensional component. Sandia National Laboratories developed a technology known as LENS (Laser Engineering Net Shaping) [4], to fabricate metal components directly from CAD solid models and thus further reduce the lead times for metal part fabrication. A similar process named Directed Light Fabrication (DLF) [5] is under development at Los Alamos National Laboratory. The DLF process is more flexible because it has 5-axis positioning capability. This allows the manufacture more complex parts (i.e., overhangs) at the expense of increased cost and process planning complexity. A variant of these two approaches is under development at the Fraunhofer Institute for Production Technology (IPT) named Controlled Metal Build Up (CMB) [6]. In this process the high-speed 2-1/2 axis CNC milling operation ensures that the required levels of form and dimensional accuracy, as well as of edge sharpness, are met. Numerous metallic materials ranging from bronze through steel to the hard alloys, frequently used to protect against wear, can be processed using this method. The fact that a high-speed milling operation takes place after each application of a new layer makes it possible to produce narrow deep grooves, since the engagement depth of the milling tool remains at a constant, low level. By virtue of the generative, layer-by-layer nature of this technique, the CAD data can be processed more quickly and with considerably less effort than is required for conventional 5-axis milling.

Currently, rapid prototyping techniques based on CO2 and Nd:YAG lasers are more flexible than alternative non-laser RP techniques but still have shortcomings that sharply increase the total cost of fabricated parts. Among these shortcomings, existing laser based RP systems are expensive, bulky, and inefficient in conversion of electrical energy to thermal energy. If these shortcomings could be eliminated and thermal energy could be delivered in compact, efficient, low cost RP systems, the cost of fabricated parts would drop rapidly and precision parts manufacturers using this new technology would become strongly competitive in international markets.

The use of welding for creating free standing shapes was established in Germany under the process name Shape Welding in the 1960’s. This led to companies such as Krupp, Thyssen, and Shulzer developing welding techniques for the fabrication of large components of simple geometry, such as pressure vessels which could weigh up to 500 tons. Other work in this area has been undertaken by Babcock and Wilcox (Doyle, 1991) under the process name Shape Melting, that was used mainly for building large components made of austenitic materials. Also, work by Rolls-Royce has centered on investigating 3D welding as a means of reducing the waste levels of expensive high performance alloys which can occur in conventional processing. They have successfully produced various aircraft parts of nickel based and titanium based alloys. Research work on 3D welding has been in progress at the University of Nottingham, UK [7] (Fig. 1), University of Wollongong, Australia, and Southern Methodist University, Dallas, TX [11- 14]. Two new research groups, one from Korea [15] and another consisting of researchers from Indian Institute of Technology Bombay and Fraunhofer Institute of Production Technology and Automation [16] presented their conceptual ideas of combining a welding operation with milling at the 9th Solid Freeform Fabrication Symposium, Austin, TX, August 10-12, 1998. The Korean research group is proposing to combine welding and 5-axis CNC milling for direct prototyping of metallic parts. The other research group from Germany and India is proposing to combine welding with 2-1/2 axis milling, where complex shapes of the layers will be obtained by using angle cutters. The brazing process is proposed to deposit the masking material at the edges of each layer in order to allow the formation of overhangs.

One of the first systems designed to directly produce metallic parts was known as micro-casting deposition (MD) [17]. MD system uses a thermal spraying technique to fabricate three-dimensional multi-material structures having arbitrary geometric complexity. In an MD operation, each layer is sprayed using a disposable mask, which has the shape of the current cross-section. The masks are on average of 0.005 inches thick, and are usually created from paper using a CO2 laser. The following drawbacks are associated with the MD approach: stair-step surface texture which leads to poor precision and dimensional accuracy, porosity, low mechanical strength compared to cast or welded parts, and warpage and delamination caused by the build-up of residual stresses.

The improved rapid prototyping technique called Shape Deposition has been developed by the same research group who developed MD [18, 19]. In a Shape Deposition system each layer is deposited using a developed micro-casting process. In this process molten droplets are produced in a non-transferred mode characterized with low substrate temperature. However, the molten droplets contain enough heat to remelt the underlying material (approx. 10 microns deep). After the first layer is deposited, a 5- axis CNC milling machine removes unwanted material, forming the desired net shape of the present layer. In order to prevent the build-up of residual stresses, the part being formed is passed through a shot-peening based stress relief station. In the final step, the part returns to the microcasting station where complementary, sacrificial support material is added. While most other rapid prototyping techniques form a part by depositing a relatively large number of thin layers, Shape Deposition can use thicker layers and fewer passes. This process requires a relatively expensive setup and still does not have the capability to control the uniformity of the metal droplet size and its detachment rate.

It has been recognized by several research groups, mainly abroad (Germany, England, India, Korea), that the principles of the Gas Metal Arc Welding process could be used in the development of a cost effective method for layered deposition manufacturing or Rapid Prototyping o full dense metallic parts and tools [8, 11-16]. Most work has focused on developing software to transfer CAD data into instructions for positioning systems. However, the importance of understanding and controlling the metal transfer process in GMAW was not recognized as one of the key issues in controlling the quality of the resultant weld and/or generated layer of metal for rapid prototyping. While extensive documentation and experience is available on selection and optimization of welding parameters for the production of welds with exceptional joint quality, this information cannot be successfully applied for the rapid prototyping process since the criteria for "good weld” and “good prototyped layer” differ significantly. For example, the requirements for build-up height, penetration depth into the previous layer, and the ratio of these two variables are very different for welding and rapid prototyping by welding. To our knowledge we are the first in an attempt to control the welding processes for the needs of rapid prototyping, i.e.; control of depth of penetration and heat buildup into the layer.

Currently, the PI with his research team has been developing a technique for sensing and control of the droplet size and the detachment rate in GMAW. This technique is based on using a high frame rate CCD digital camera as a sensor to monitor and control the size of the droplet and its detachment rate. More information about this metal transfer control technique is given in the next section.

Research and Development in Control of Mass and Heat Transfer in GMAW: One challenge for RP has been to develop the capability to directly create functional metal shapes which are dense, metallurgically bonded, geometrically accurate and exhibit good surface appearance. While functional metal parts have been built with RP through post- processing and/or conversion methods, it remains a goal to be able to directly build high performance metallic parts for such applications as fabricating custom tooling and functional production-ready prototypes. In the last several years a number of thermal deposition techniques (thermal spraying, welding, laser cladding) have been used to create fully dense metal parts. Key issues in applying thermal deposition techniques for RP of metallic parts is to control temperature gradients caused by fusing molten droplets onto previously deposited layers. Thermal spraying creates molten droplets with relatively small diameters (about 50 microns) and they do not possess enough heat to re-melt the underlying surface. The bond of layer to substrate is mechanical where adhesive and cohesive strength are relatively low [17]. In classical welding deposition approaches, such as GMAW and plasma welding, metallurgical bonds will be formed, but the large heat transfer will affect the shape of underlying material. Laser cladding is usually done with metal powder [4-6]. This process characteristically yields very thin build-up layer and a small heat affected zone. However, this process as a RP method is limited by the size of the parts and by its high cost. A thermal process must be developed that can provide a high metal deposition rate and metallurgical bonding between layers, but will not destroy the underlying geometry. The PI with his research team developed a technique for controlling the metal deposition rate and heat input in GMAW that will serve as an excellent base for the development of a new RP process [11 - 14].

Existing methods for metal transfer control in GMAW have two major flaws: uncertain detachment instant and inconsistent droplet size. These methods rely on the one-drop-per-pulse approach by properly selecting the duration of the peak current. To guarantee the detachment, the peak current has to be larger than the transition current (250-300 A). However, in addition to detaching the droplet, such a high current also generates excessive fumes due to the superheating of the droplet, results in a very high impact speed of the droplet, deforms the underlying geometry, and yields irregular droplet volume. The proposed detachment control approach eliminates the uncertainty in the detachment instant associated with the conventional method, as well as provides the capability to control the size of the metal droplet.

Proposed Approach to Control Metal Transfer in GMAW: A high frame rate digital camera (over 1,000 frames per second) assisted with a He-Ne laser and a real time image processing algorithm has been used to monitor the droplet formation (Fig. 2). A PCI frame grabber combined with a 400MHz PC Pentium II (120MB RAM memory) allows a real-time monitoring and control of about 60 droplets per second. It was found that switching the current from the peak level (about 200A, for selected electrode diameter and its material, as well as for selected electrode extension and shielding gas) to the base level (about 50A) will initiate an oscillation of the droplet at the tip of the electrode [23 - 25]. The developed machine vision system can monitor this oscillation (Fig. 3). When the droplet moves downwards, a signal is sent to the power source controller to raise the current to the peak level, which increases the electromagnetic force. The design of the current pulse depends on the required depth of bead penetration as well as on the allowed level of induced heat avoiding the geometrical distortion of the previously generated layer. The downward motion of the droplet in combination with the increased electromagnetic force generates a large enough detachment force to detach the droplet from the electrode. This new approach for controlling the GMAW process provides the following benefits : accurately controls the droplet size and its rate of detachment, eliminates the need for a high current to detach the droplet, reduces the heat energy input into substrate, reduces the fumes, eliminates spatter, improves the controllability of weld penetration. This is the first time that somebody can accurately control the height to width ratio of the bead layer generated by GMAW [11 - 14].

In order to verify the effectiveness of the proposed approach extensive experimentation has been performed. As an example, a mild steel electrode with a diameter of 1.2 mm is used with an electrode extension of 16.5 mm. The shielding gas is a mixture of 95% Ar plus 5% CO2. The peak current and base current are 220 A and 40 A, respectively. Fig. 4 shows the results of these experiments. As could be seen from Fig. 4, the average current Im or the heat input is approximately kept constant for both experiments, 130 A. However, by adjusting the transfer frequency, the droplet size is changed. In the classical pulsed welding power sources, a change in the droplet size can be achieved only by changing the level of the average current or the wire speed. It is evident that in this case it is impossible to accurately control the droplet size. By designing the pulsed train through changing the duration of the droplet growing and droplet detaching periods, one is able to precisely control the droplet size and the detachment frequency, or heat and metal transfer in gas metal arc welding.

Fig. 5 shows the results of heat input control in GMAW. Fig. 5a shows that the current remains at the high level after the cut-off action. Because the cut-off action is applied as soon as the phase match condition is satisfied, the duration of the base current is short and actually uncontrolled. In this case, the average current is close to the level of the current that corresponds to the droplet growth phase. Thus, the detachment frequency is high and, consequently the heat input is high as well. The procedure to decrease the heat input is shown in Fig. 5b in this case the duration of the base current is controlled. As soon as the droplet is detached by increasing the current to the cut-off level, the current is returned back to the base current. By controlling the duration of the base current before initiating the next pulse will affect the droplet detachment frequency and the value of the average current, thereby controlling the heat input to the molten pool.

A mathematical model is developed to describe the globular transfer in GMAW. Using the Volume of Fluid (VOF) method, the fluid flow and heat transfer phenomena are dynamically studied in the impingement process of a droplet onto a solid structure, arc striking, the impingement process of multiple droplets on the molten pool, and finally the solidification process after the arc extinguishes [26]. Fig. 6a and Fig. 6b depict the behavior of a molten droplet (temperature and velocity distribution) impacting on the substrate. Fig. 6c shows that the experimental results are in good agreement with the modeling results.

Acknowledgment: This work is supported by National Science Foundation under contract DMI 9732848 and Raytheon Systems Co., Dallas, TX and Waco, TX.

References Cited:

[1] J.J. Beaman, J.W. Barlow, R.H. Bourell, H.L. Craford, H.L. Marcus, and K.P. McAlea, "Solid Freeform Fabrication: A New Development in Manufacturing," Kluwer Academic Publishers, Dordrecht/Boston/London, 1997.

[2] P.E. Jacobs, "Stereolithography and other RP&M Technologies," ASME Press, New York, 1996.

[3] L. Killander, "Future Direct Manufacturing of Metal Parts with Freeform Fabrication," Annals of the CIRP, Vol. 44/1, 1995.

[4] M.L. Grifith, D.M. Keicher, C.U. Atwood, J.A. Romero, J.E. Smugersky, L.D. Harwell, and D.L. Green, "Free Form Fabrication of Metallic Components Using Laser Engineered Net Shaping (LENS)," Proc. of the Solid Freeform Fabrication Symposium, The University of Texas at Austin, Austin, TX, Aug. 12-14, 1996.

[5] G.K. Lewis, "Direct Laser Metal Deposition Process Fabricates Near-Net-Shape Components Rapidly," Materials Tech., March 1995.

[6] F. Klocke, and U. Clemens, "Rapid Tooling Combining Laser Generating and High Speed Milling," Proc. of the 5th European Conference on Rapid Prototyping and Manufacturing, 1996.

[7] P.M. Dickens, R. Cobb, I. Gibson, and M.S. Pridham, "Rapid Prototyping Using 3D Welding," Journal of Design and Manufacturing, No.3, 1993.

[8] K. Kussmaul, F.W. Schoch, and H. Luckow, "High Quality Large Components Shape Welded by a SAW", Welding Journal, Sept. 1983.

[9] G. Gnirss, "Shape Welding: the Current State of the Art in the Federal Republic of Germany," Welding in the World, 1987.

[10] T.E. Doyle, "Shape Melting Technology," 3rd Intl. Conf. Desktop Manufacturing, 1991.

[11] R. Kovacevic, and H.E. Beardsley, "Process Control of 3D Welding as a Droplet- Based Rapid Prototyping Technique," Proc. of the 9th Annual Solid Freeform Fabrication Symposium, Aug. 10-12, 1998, Austin, TX.

[12] H.E. Beardsley, and R. Kovacevic, "Controlling Heat Input and Metal Transfer for 3D Welding-Based Rapid Prototyping," Proc. of the 5th International Conference on Trends in Welding Research, June 1-5, 1998, Pine Mountain, GA.

[13] H.E. Beradsley, and R. Kovacevic, "New Rapid Prototyping Technique Based on 3D Welding," Proc. of the 31st CIRP International Seminar on Manufacturing Systems, May 26-28, 1998, Berkeley, CA.

[14] H.E. Berdasley, and R. Kovacevic, "Controlling Heat and Metal Transfer for Droplet-Based Rapid Prototyping," Abstract of Papers of the 79th Annual AWS Convention, Detroit, April, 1998.

[15] Y. Song, S. Park, K. Hwang, D. Choi, and H. Jee, "3D Welding and Milling for Direct Prototyping of Metallic Parts," Proc. of the 9th Annual Solid Freeform Fabrication Symposium, Austin, TX, Aug. 10-12, 1998.

[16] K. Karunakaran, P. Shanmuganathan, S. Roth-Koch, and U. Koch, "Direct Rapid Prototyping of Tools," Proc. of the 9th Annual Solid Freeform Fabrication Symposium, Austin, TX, Aug. 10-12, 1998.

[17] L.E. Weiss, F.B. Prinz, D.A. Adams, and D.P. Siewiorek, "Thermal Spray Shape Deposition," Journal of Thermal Spray Technology, Vol. 1, 1992.

[18] R. Merz, F.B. Prinz, K. Ramaswami, M. Terk, and L.E. Weiss, "Shape Deposition Manufacturing," Proc. of the Solid Freeform Fabrication Symposium, Austin, TX, Aug. 8-10, 1994.

[19] R. Merz, and F.B. Prinz, "Method and apparatus for Depositing Molten Metal," U.S. Patent No. 5,281,789, 1994.

[20] L.A. Jones, T.W. Eager, and J.H. Lang, J.H., "Investigation of Drop Detachment Control in Gas Metal Arc Welding," Proc. of the 4th International Conference on Trends in Welding Research, Gatlinburg, TN, June 5-8, 1995.

[21] J.A. Johnson, N.M. Corlson, H.B. Smart, and D.E. Clark, "Process Control of GMAW: Sensing of Metal Transfer Mode," Welding Journal, 70(4), 1991.

[22] S. Rhee and E. Kennatey-Asibu, "Observation of Metal Transfer During Gas Metal Arc Welding," Welding Journal, 71(10), 1992.

[23] R. Kovacevic, Y.M. Zhang, H.E. Beardsley, and M. Erlenbeck, "On-Line Sensing of Metal Transfer for Adaptive Control of GMAW," Proc. of the 4th International Conference on Trends in Welding Research, June 5-8, 1995, Gatlinburg, TN.

[24] R. Kovacecvic, M. Erlenbeck, H.E. Beardsley, and Y.M. Zhang, "Real-Time Image Acquisition and Processing of Metal Transfer," The 77th Annual AWS Convention, Chicago, IL, April 23-25, 1996.

[25] Y.M. Zhang, E. Liguo, and R. Kovacevic, "Active Metal Transfer Control by Monitoring Excited Droplet Oscillation," Submitted to the Welding Journal, 1997.

[26] H.G. Fan and R. Kovacevic, "Dynamic Analysis of Globular Metal Transfer in Gas Metal Arc Welding - Comparison of Numerical and Experimental Results," accepted by the Journal of Physics D: Applied Physics, 1998.