Development of a Sheet-Based Material Handling System for Layered Manufacturing

Size: px

Start display at page:

Download "Development of a Sheet-Based Material Handling System for Layered Manufacturing"

Harold Barnett
5 years ago
Views:

1 Proceedings of the 2001 IEEE International Conference on Robotics & Automation Seoul, Korea May 21-26, 2001 Development of a Sheet-Based Material Handling System for Layered Manufacturing Tao Wei, Sangeun Choi and Wyatt S. Newman Dept. of Electrical Engineering and Computer Science Case Western Reserve University (CWRU) Cleveland, Ohio Abstract Computer-Aided Manufacturing of Laminated Engineering Materials (CAM-LEM), developed at Case Western Reserve University, is a layered-manufacturing approach utilizing automated stacking of parts laser-cut from sheet materials. This paper presents the design of a materialhandling system and its integration with a modular lasercutting system, completing the most recent version of CAM-LEM. Our material-handling system is required to: transfer individual sheets from at least three material stacks to the cutting table of the laser-cutting system; extract singlelayer, laser-cut parts from the cutting table; and stack the parts onto a vertical-stack assembly, resulting in fabrication of a 3-D object. Design philosophy, details, and stacking performance measurements of the constructed system are presented. It is shown that the resulting system is capable of assembling objects within a 150mm cube with an assembly precision of approximately 0.050mm. Further, the assembly operations are relatively fast, so that part build rates are limited primarily by the laser cutter. 1. Introduction There are many existing approaches to building 3-D objects directly from an electronic description. Among these approaches, some of them perform incremental deposition, solidification, or fusion of material for layerby-layer growth in the vertical direction. Selective Laser Sintering (SLS), in which powdered materials are fused together, one thin layer at a time, using a high-powered laser [1]. Fused Deposition Modeling (FDM), which builds objects from filaments of plastic from a computercontrolled extrusion head. This method, described in [2] has extensions to alternative materials and multiple materials [3]. Stereolithography [4], in which parts are built up in layers of photo-curable resin, selectively exposed by a computer-controlled laser beam. 3-D Printing [5], in which a powder bed is selectively bonded using inkjet dispensing of an adhesive. Recent efforts have focused on application of loose powder compacts [6]. Different from these Solid Freeform Fabrication (SFF) approaches, a few other methods use sheet-based feedstock, and build up objects by cutting and stacking (or stacking then cutting) successive layers. Laminated Object Manufacturing [7], in which sheet materials are bonded to a stack, and each layer is laser cut in plane. ShapeMaker [8] uses a hot wire to cut Styrofoam layers, of which the latest version introduced a waterjet cutter to build largescale parts [9]. In our approach, Computer-Aided Manufacturing of Laminated Engineering Materials (CAM-LEM) [10][11][12], sheet materials are cut individually by a laser and the cut sections are robotically extracted and stacked to form assemblies. Since material removal around a perimeter can be significantly faster than material deposition throughout the corresponding enclosed area, the latter sheet-based methods can have higher build rates. Such build-speed advantage would be exaggerated for larger objects. Additionally, sheet-based materials can more easily accommodate alternative build materials. In this study, we describe the development of the CAM-LEM approach, specifically by designing an effective materialhandling system for grasping and stacking cut layers. Slice Cutting Lamination Computer Model Laser Conventional binder burnout and sintering Contour Representation Stacking Finished Component Fig 1: The CAM-LEM Process 2. Tangential-Cutting Machine The CAM-LEM laser-cutting machine should be capable of cutting boundary edges tangent to the physical 3D-model surface, which enables us to eliminate the staircased characteristic found in parts made using other RP/SFF technologies. The 4-axis tangential-cutting machine (more details in [14]) is shown in Fig 2. The x-y cutting table controls the perimeter of the contour to be laser cut, and the pitch and roll axes control the tangent angle of the edge, independent of the x-y motion. The 6 x 6 cutting table is cantilevered from a vertical support mounted on the x-y sled. This support structure permits the roll axis to swing the yoke below the cutting table, extending the range of accessible angles. The reachable range of roll and pitch angles is at least +/- 80 deg over the full 6 x 6 cutting table range of motion /01/$ IEEE 1352

The conveyor supports stacks of feedstock materials and moves those stacks in the Y direction. In our design, three material stacks are placed on the conveyor and moved under the gripper when needed.

As shown in Fig 3, the motor, encoder and transmission were mounted above the upper face of the conveyor. The conveyor was wider than required, and thus this mounting arrangement was not a problem.

Design Objectives of Material-Handling System The CAM-LEM system is based on a cut-and-stack approach, thus the material handling system is as important as the laser-cutting system.

Our handling-system objectives are as follows: To interface with the laser-cutting system, specifically: - feed at least three different 6 6 sheet material options to the cutting table without manual

2 The conveyor supports stacks of feedstock materials and moves those stacks in the Y direction. In our design, three material stacks are placed on the conveyor and moved under the gripper when needed. These stacks are mm x mm x mm (6 x 6 x 6 ), and the length of the conveyor is mm (48 ). As shown in Fig 3, the motor, encoder and transmission were mounted above the upper face of the conveyor. The conveyor was wider than required, and thus this mounting arrangement was not a problem. Alternative mounting options resulted in interference between the motor and the material-handling system frame. Fig 2: The 4-Axis Tangential Cutting Machine 3. Design Objectives of Material-Handling System The CAM-LEM system is based on a cut-and-stack approach, thus the material handling system is as important as the laser-cutting system. The handling system is needed to move materials onto and off of the cutting table and to stack useful parts on the assembly table with high precision and reliability. Our handling-system objectives are as follows: To interface with the laser-cutting system, specifically: - feed at least three different 6 6 sheet material options to the cutting table without manual intervention - extract desired cut regions from the cutting table, leaving waste material behind - avoid collisions with the laser-cutting system To assemble grasped, cut regions precisely (e. g., within +/- 25µm (1 )). To maximize build speed. - This includes not only achieving high gripper velocities, but also minimizing required motions (including releasing and re-grasping masks) and minimizing suspension of laser-cutting activity due to interference with material handing. To minimize size and cost. To include considerations of modularity and scalability. 4. System Components and Integration Our material-handling system design, shown in Fig 3, has three axes: two slides and one conveyor. The Conveyor Fig 4: Three Material Stacks and Servo- Controlled Conveyor On the conveyor, there are three stacks of sheet materials, as illustrated in Fig 4. One is the mask stack; masks are used to extract cut regions from the lasercutting table using the vacuum gripper. A second is the target stack, which is the chosen material of the desired 3- D object. The last stack is the fugitive stack, which is used to provide construction support for otherwise unsupported assembly sections. In front of the conveyor is the assembly table, which is fixed to ground via the frame. Because the gripper can only move in the X and Z directions (via the X and Z slides), the gripper and assembly table must be aligned in the Y direction. So, the position of the X-slide (and the gripper mounting-bracket) determines the position of the assembly table in the X-Y plane. The Gripper-Motion Slides Fig 3: Layout of the Material-Handling System 1353 Fig 5: Gripper Drive Mechanism (XZ Slide)

The X and Z slides of the material-handling system are shown in Fig 5. The X-slide was a critical component, as it required both a long reach and high repeatability.

001 ). However, our machine design only requires high repeatability for success. The selected unit had a quoted repeatability of +/- 0.01mm (0.0004 ), which fits our needs.

The Z-slide, made by Aerotech, Inc., also had high precision. It was mounted to the X-slide, and it carried the gripper.

3 The X and Z slides of the material-handling system are shown in Fig 5. The X-slide was a critical component, as it required both a long reach and high repeatability. After a survey of commercial options, this component was purchased from Techno-isel Co. Its accuracy (+/- 0.1mm (0.004 )/300mm (11.8 )) would not satisfy our stacking precision goal of +/- 25 µm (0.001 ). However, our machine design only requires high repeatability for success. The selected unit had a quoted repeatability of +/- 0.01mm ( ), which fits our needs. The stroke of the X-slide is mm (24.6 ), which can satisfy our X direction requirement of mm (21 ). The X-slide was mounted to a frame, which was constructed of extruded aluminum beams. The Z-slide, made by Aerotech, Inc., also had high precision. It was mounted to the X-slide, and it carried the gripper. We note that the Z- slide was an available component that exceeded our design requirements. For a lower-cost design, it may be possible to replace this servo-controlled axis with a simpler prismatic component. The chief requirement of this axis is straight-line motion over a sufficient (177.8 mm (7 )) stroke. We also note that the Z-axis and gripper mounting brackets were designed to be stiff, but high mounting precision was not a requirement. A leveler design, described below, enabled the required gripper alignment. The Assembly Table materials; and stack useful parts on the assembly table. We use two selective-area vacuum grippers and a pair of conventional vacuum cups for this purpose. The two selective-area vacuum grippers [13] each require two vacuum sources. The primary vacuum source generates suction through laser-drilled mask holes to selectively extract desired cut-outs from the cutting table. The second vacuum source is needed to acquire and hold the masks about the perimeter of the gripper. The masks must continue to be held when the primary vacuum source is disabled, e.g. when releasing grasped parts onto the assembly stack. The prototype of the triple gripper is shown in Fig 7. bottom view top view Fig 7: The Prototype Triple Gripper Figure 8 shows the 3-dof material-handling system integrated with our 4-dof laser-cutting system. To use the space more efficiently, the conveyor is located parallel to the Y-axis of the laser cutting for feeding multiple uncut sheet materials. It indexes these stacks into position perpendicular to the X-travel of the gripper. Using the conveyor makes the extended X direction shorter. Also we note that conveyor does not need to be particularly precise, since the precision requirements for the uncut sheets are not demanding. Fig 6: The Green Assembly Table The assembly table, shown in Fig 6, is located between the conveyor and the cutting table. The function of the assembly table is to provide a stable, level location on which to stack layers that have been cut. Four through holes on the assembly table are connected to a vacuum source. This feature provides vacuum clamping for a removable assembly base plate, which may be a sheet of fugitive material. The object being assembled may be glued to this sheet to prevent slipping or disturbances during assembly. After assembly, the object may be removed along with the fugitive base sheet. The Triple Gripper The gripper design is the centerpiece of the materialhandling system. This gripper should: feed mask, fugitive and target sheets to the cutting table; acquire and hold punched masks; extract the target and fugitive materials from the cutting table without picking up the waste Fig 8: The Material-Handling System Integrated with the Cutting System 5. Layout and Operating Sequence In Fig 9, the assembly site is a stationary table (fixed to ground). When we stack up the layers that have been cut, we need very good repeatability (our stated goal was +/- 1354

25µm (1 )). Therefore, we put the triple-gripper in one row in order to minimize the required movement in the Y direction.

down the layers on the assembly table using any of the three grippers without touching the other stacks.

But when gripper I puts down its layers on the assembly table, gripper II may interfere with the conveyor. So, we leave an empty space on the conveyor to which we index when stacking with gripper I.

Because the mask required for the fugitive and persistent materials are complementary, one cannot be used for handling the other. 6.

In order to get better measurement resolution of rotation errors, we chose a rectangular object, 76.2 mm x 12.7 mm (3 x 0.5 ).

The gripper moves in the X-direction and Z-direction, so we expect displacement errors mainly in the X-direction within the X-Y plane.

4 25µm (1 )). Therefore, we put the triple-gripper in one row in order to minimize the required movement in the Y direction. Fig 9: The Layout of Materials and Gripper (a) Punch mask (b) Feed target tape (c) Cut contours With the assembly table fixed to ground, there should be enough space for the triple-gripper to put down the layers on the assembly table using any of the three grippers without touching the other stacks. There is no problem when gripper II puts down its layers on the assembly table, because the cutting table can move away at that time. But when gripper I puts down its layers on the assembly table, gripper II may interfere with the conveyor. So, we leave an empty space on the conveyor to which we index when stacking with gripper I. The sequential material handling process is illustrated in Fig 10. Notably, one of the grippers handles the fugitive support material, and the other handles the persistent component material. Because the mask required for the fugitive and persistent materials are complementary, one cannot be used for handling the other. 6. Part-Handling Performances We have measured the performance in different ways. In the first experiment, we performed a pick-and-place repeatability test. In order to get better measurement resolution of rotation errors, we chose a rectangular object, 76.2 mm x 12.7 mm (3 x 0.5 ). The object was picked up from the assembly table, moved away and replaced on the assembly table for 15 times. The gripper moves in the X-direction and Z-direction, so we expect displacement errors mainly in the X-direction within the X-Y plane. For each handling operation, we took five pictures of the same image, averaged these five pictures and computed the centroid and second moments of the target in each averaged image. Fig 11 shows the centroids for 15 handling operations. During the 15 iterations, the total displacements in the X-direction and the Y-direction were mm ( ) and mm ( ), respectively. From trial to trial, the largest displacement was pixel, which is mm ( ). Thus, the repeatability of single-sheet handling is slightly worse than our goal of +/- 25 microns (0.001 ). (d) Pick and stack contour Fig 11: Displacement Errors for 15 Iterations Fig 10: Material-Handling Sequence (e) Feed, cut and stack fugitive Another experiment performed to test the repeatability of the material handling system was to stack up a tower with identical layers. 30mm x 30mm (1.18 x 1.18 ) squares were cut and stacked. Each layer was 2 mm (0.08 ) thick and the assembly was stacked up 40 layers, forming an 80 mm (3.15 ) high tower. From Fig 12, we 1355

shows the result of the approach. The desired assembly (a square tower) is surrounded by waste sheet material. This approach can help limit error caused by unbalanced forces.

To guarantee tests along a repeatable straight line, the angle iron was moved along a rail. We moved the tower with the angle iron manually to test each layer for three times on both sides.

In the X direction, the shifting from layer to layer is small, but the perpendicularity is not very good.

Possible explanations for this observed behavior include the following: 1. Repeatability of the cutting table and gripper axes. 2.

Sheet stack thickness variations, which can accumulate with stacking 4. Non-uniform distribution of the spray adhesive, which can lead to non-parallel stacked layers. 5.

Displacements of parts on the cutting table, e.g. caused by cutting-table accelerations or the laser-cutter air jet. Some of the possibilities above are avoidable by design.

5 shows the result of the approach. The desired assembly (a square tower) is surrounded by waste sheet material. This approach can help limit error caused by unbalanced forces. The result is very satisfactory, as shown in Fig 13-(b). We set up the dial indicator and the tower clamped on an angle iron horizontally on a ground face. To guarantee tests along a repeatable straight line, the angle iron was moved along a rail. We moved the tower with the angle iron manually to test each layer for three times on both sides. The results are in Fig 14. X-direction Shifts Y-direction Shifts Fig 12: Stacking Performance for a 40-Layer Square Tower can see the distortion or shifting in the X and Y directions. In the X direction, the shifting from layer to layer is small, but the perpendicularity is not very good. In the Y direction, the layer-to-layer shifting is somewhat worse than in the X direction (making a rougher edge and face), but the perpendicularity is good. Possible explanations for this observed behavior include the following: 1. Repeatability of the cutting table and gripper axes. 2. Imperfect alignment of the cutting table, gripper and assembly table surfaces. Non-parallelism can cause a layer to shift horizontally when the gripper pushes down on the stack. 3. Sheet stack thickness variations, which can accumulate with stacking 4. Non-uniform distribution of the spray adhesive, which can lead to non-parallel stacked layers. 5. Compliance of the base sheet between the bottom layer and the assembly table, which could make the whole tower tilt due to uneven contact forces during assembly. 6. Displacements of parts on the cutting table, e.g. caused by cutting-table accelerations or the laser-cutter air jet. Some of the possibilities above are avoidable by design. A technique we investigated to help improve stacking precision was equivalent to using fugitives. Instead of picking up only the square cut-outs, we picked up both the square and the outside layer that was left on the cutting table. The outside layers behaved like fugitive layers to support the desired square tower assembly. Figure 13-(a) Fig 13: Improvement of Stacking Performance (a) Stacked Tower with the Support of Fugitives (b) Completed 40-Layer Square Tower Fig 14: Test Results of Experiment of the Stacked Layers with the Fugitive Support From Fig 14, we can see, in the X direction, the deviations change from microns ( ) to microns ( ) in 40 steps. The average standard deviation is microns ( ); in the Y direction, the deviations change from microns ( ) to mm (0.006 ) in 40 steps. The average standard deviation is microns ( ). Comparing Fig 12 to Fig 16, we see that the X-deviation from repeated handling of a single layer average (-7 microns) transfer, whereas the stacked tower had a slope of only 2.5 microns/layer. Apparently, the use of fugitive supports helped to suppress the influence of process-dependent disturbances. 7. Summary and Conclusion A novel CAM-LEM material-handling system had the objectives of integrating with an existing laser-cutting system, accomplishing precision assembly of laser-cut sheet materials, maximizing utilization of the laser cutter thereby minimizing build time, minimizing size and cost, and achieving modularity and scalability. The layout chosen enabled use of low-cost components, where possible, and achieved a relatively efficient utilization of floor space. Achieving precision assembly was the greatest challenge of the design. This was accomplished using 1356

6 prismatic axes, meeting widely-varying design specifications with separate components, introducing integrated calibration mechanisms, and invoking design techniques yielding stiff and precise components. The resulting assembly performance was measured to be within approximately +/-0.050mm--twice as large as the ambitious goal of 0.025mm. However, these errors include influences beyond material handling, including laser-cutting and feedstock irregularities. 8. Acknowledgments This work was supported by the National Science Foundation under NSF grant DMI This support is gratefully acknowledged. 9. References 1. Zong, G., Wu, Y., Tran, N., Lee, I., Bourell, D. L., Beaman, J. J. and Marcus, H. L., "Direct Selective Laser Sintering of High Temperature Metals", Texas, pp.72-85, Kumar, C., Jones, L., and Roscoe, L., Support Generation for Fused Deposition Modeling, Texas, pp , Gasdaska, C., Jamalabad, V. R., "Direct Manufacture of Spatially Engineered Components for Aerospace Applications by Fused Decomposition of Ceramics", Texas, pp , Jacobs, P. F., 1993, Stereolithography 1993: Epoxy Resins, Improved Accuracy and Investment Casting, Conference Proceedings of the Fourth International Conference on Rapid Prototyping, Dayton, OH, pp , Sachs, E., Cima, M. J., and Cornie, J., Three Dimensional Printing: Rapid Tooling and Prototypes Directly from a CAD Model, CIRP Annals, 39, [1], pp , Sachs, E. M., Allen, S., Guo, Banos, Cima, M., Serby, and Brancazio., "Progress on Tooling by 3D Printing: Conformal Cooling, Dimensional Control, Surface Finish and Hardness", Proceedings of the Solid Freeform Fabrication Symposium, University of Texas at Austin, Austin, Texas, pp , Griffin, C., Daufenbach, J., and McMillin, S., Solid Freeform Fabrication of Functional Ceramic Components Using a Laminated Object Manufacturing Technique, proceedings of the Solid Freeform Fabrication Symposium, University of Texas at Austin, Austin, Texas, pp , Lee, C. H., Gaffney, T. M., Thomas, C. L., "Paradigms for Rapid Prototyping", Proceedings of the 6-th International Conference on Rapid Prototyping, Dayton, OH, pp , Chamberlain, P. B., Thomas, C. L., "Direct Thick Layer Rapid Prototyping from Medical Images", Texas, pp , Choi, S., Hebbar, R., Zheng, Y., Newman, W.S., CAD and Control Technologies for Computer-Aided Manufacturing of Laminated Engineering Materials, Texas, pp , Aug Choi, S., and Newman, W. S, Design and Evaluation of a Laser-Cutting Robot for Laminated, Solid Freeform Fabrication, Proc. IEEE Int. Conf. on Robotics and Automation San Francisco, California, Zheng, Y., Choi, S., Mathewson, B. B., Newman, W. S., "Progress in Computer - Aided Manufacturing of Laminated Engineering Materials Utilizing Thick, Tangent-Cut Layers", Proceedings of the Solid Freeform Fabrication Symposium, University of Texas at Austin, Austin, Texas, pp , Newman, W. S., Mathewson, B. B., Zheng, Y., and Choi, S., A Novel Selective-Area Gripper for Layered Assembly of Laminated Objects, Robotics and Computer-Integrated Manufacturing, Vol. 12, No. 4, pp , Choi, S., Design and Evaluation of an Automated Laser-Machining System for Layered Manufacturing, Ph. D. thesis, Department of Mechanical Engineering, Case Western Reserve University,

Design Analysis Process

Prototype Design Analysis Process Rapid Prototyping What is rapid prototyping? A process that generates physical objects directly from geometric data without traditional tools Rapid Prototyping What is