Loyola Marymount University One LMU Dr. MS 8145 Los Angeles, CA, 90045 USA Abstract: - The project involves the evaluation of the effectiveness of a low-cost reverse engineering system. Recently, the reverse engineering technology has emerged as a valuable tool in making prototypes of parts for which there is no design data available for the reproduction of parts. While most reverse engineering systems are very expensive, this research concentrates on the evaluation of the effectiveness of a low-cost system (NextEngine s Desktop 3D Scanner). The specific objectives of this project are (1) to re-create physical models for which no design data exists, (2) to investigate the accuracy of the physical models and (3) to determine the surface finish of the physical models. Four different types of objects such as TV remote control, a sine plate with roller bars, as well as two diagnostic plastic Christmas trees (used to measure precision) were selected Multiple scans were conducted to input all the sides and viewable characteristics of each object. From these scans, a scaled down model of each object was generated and built by rapid prototyping machines. Both dimensions and surface roughness were measured to determine the margin of error between the original objects and the prototypes. While the reverse engineering system showed some limitations, it was able to recreate a similar (less than a 5% margin of error) prototype which was of similar dimensions to that of the original. Key-Words;- Accuracy, Computer-Aided design, Rapid Prototyping, Reverse Engineering, Surface Roughness 1. Introduction The world has already entered into a new era of global competition for providing products and services. Rapid acceleration of new and emerging technologies is fueling this growth in all aspects of business [1]. Companies engaged in product development and manufacturing are in tremendous competition to bring a product to market faster, cheaper, with both higher quality and functionality. Reducing the timeline for product development saves money in the overall time-to-market scenario. Reverse Engineering (RE) is the technology that helps companies reduce the cycle of product development and also facilitate making design improvements earlier in the process where changes are less expensive [2]. Reverse Engineering (RE) is the science of taking an existing physical model and reproducing its surface geometry in a three-dimensional (3 D) data file on a computer-aided design (CAD) system [3]. In many cases, only the physical model of an object is available. Examples of such situations include handmade prototype, crafts works, reproduction of old engineering objects, and sculptured bodies found in medical and dental applications [4-5]. In order to facilitate computer-aided manufacturing (CAM) operations of theses physical models, it is essential to establish their CAD models [6]. RE is the quickest way to get 3D data into any computer system. It is like having a low cost but accurate X-ray machine for parts or having a 3D copier. The surface data of any object can be collected by means of direct (Coordinate Measuring Machine (CMM)) or indirect measurements using laser-based systems [1]. However, CMM systems are Numerical Control (NC) driven and inherently slow in acquiring point data as it has to make contact with the part surface for every point. It is also inefficient and difficult to measure a freeform part. Therefore, an intelligent reverse engineering approach can be ISSN: 1790-5109 122 ISBN: 978-960-474-013-0
very important for a product development when freeform geometric shapes are applied in the product design. The general strategies used to reconstruct a true-form surface model for product development involve laser scanner, vision system and robots [3-5]. While these strategies have worked for many companies, they are expensive and time-consuming. Most of the laser-based reverse engineering equipment such as Z Scanner and Konica Minolta costs about $45K and $55K, respectively. In an effort to use a low-cost system, the Mechanical Engineering Department has very recently purchased a desktop 3D scanner from NextEngine which costs only $2,800. In this project, we propose a much simpler methodology that will take advantage of the low cost but breakthrough technology to reconstruct surface data of fairly complex geometries with accuracy and surface finish. Our goal is to evaluate the effectiveness of this low cost reverse engineering system. 2. Proposed work The experimental system used for this research consists of NexEngine s 3D Scanner [7], Z Corp s color rapid prototyping system, Stratasys s Fused Deposition Modeling (FDM) rapid prototyping machine (model: 1650), a coordinate measuring machine (CMM), a surface roughness measuring equipment and digital calipers. Figure 1 shows the experimental setup. Fig. 1: Experimental Setup The goal of this research is to study the building of physical models (reverse engineering) using a low cost RE system. The proposed work has three objectives: (1) to recreate a physical model for which no design data exists, (2) to investigate the accuracy of the physical model and (3) to determine the surface roughness of the physical model. In the first part of the research, four different types of objects were reverse engineered in order to determine the effectiveness of the reverse engineering system using the NextEngine s Desktop 3D Scanner. The objects of interest were a TV remote control, a sine plate with roller bars, as well as two diagnostic plastic Christmas trees. Multiple scans were conducted to input all the sides and viewable characteristics of each object. From these scans, a CAD file was created. A Stereolithography (stl) file was then created from the CAD data. The stl file was then prototyped using the Z Corp s color rapid prototyping system and Stratasys FDM-1650 rapid prototype machines to fabricate the physical model. Some of the models were scaled down to fit into the rapid prototyping system and to save time. Both dimensions and surface roughness were measured to determine the margin of error between the original objects and the reverse engineered prototypes. The accuracy and surface roughness measurements were used to validate the effectiveness of the low cost reverse engineering system. 3. Results and Discussions The measurements of the surface roughness and the dimensions of the original and prototyped objects are shown below. Table 1 shows the prototyped dimensions as well as the mean and the percentage error for the TV remote control. Table 1 - TV Remote Control Dimensions Length Dimensions, Plastic TV Remote Nominal (in) (Mean, in) 5 4.95 10% Length Dimensions, Powder TV Remote 5 5.00 0% Width Dimensions, Plastic TV Remote 2 2.04 2% Width Dimensions, Powder TV Remote ISSN: 1790-5109 123 ISBN: 978-960-474-013-0
2 2.06 3% As the table illustrates, the NextEngine Desktop 3D Scanner was very accurate in retrieving data from the original object and relaying it to both types of rapid prototyping machines. Every margin of error was below five percent for both length and width dimensions. However, there were some physical differences between the original and the prototypes. Figure 2 shows the original and the prototypes of the objects. Fig. 2 From Left, Original Remote Control, and Plastic and Powder Prototypes As the figures demonstrate, the surface the plastic prototype resembled a layered cake. Each layer was visible, which made the surface feel much coarser than the powder prototype. The powder prototype, as seen in Figure 3, had a much smoother and more consistent surface, which made it more like the original remote control. Table 2 shows the prototyped dimensions as well as the difference from the original for the sine roller block. Table 3 displays the same dimensions for the sine roller bars. Table 2 - Sine Roller Dimensions Length Dimensions, Plastic Sine Roller 4 3.94 1.5% Length Dimensions, Powder Sine Roller 4 4.02 0.5% Width Dimensions, Plastic Sine Roller 4 3.98.5% Width Dimensions, Powder Sine Roller 4 4.00 0% Depth Dimensions, Plastic Sine Roller 1 0.97 3% Depth Dimensions, Powder Sine Roller 1 1.01 1% Table 3 - Sine Roller Bar Dimensions Diameter Dimensions, Plastic Sine Roller Bar 1 0.5 0.40 20% Diameter Dimensions, Powder Sine Roller Bar 1 0.5 0.44 12% Diameter Dimensions, Plastic Sine Roller Bar 2 ISSN: 1790-5109 124 ISBN: 978-960-474-013-0
0.5 0.42 16% Diameter Dimensions, Powder Sine Roller Bar 2 Nominal (in) (Mean, in), 0.5 0.44 12% Length Dimensions, Plastic Sine Roller Bar 1 (Mean, in), 4 4.01 0.25% Length Dimensions, Powder Sine Roller Bar 1 (Mean, in) 4 4.05 1.25% Length Dimensions, Plastic Sine Roller Bar 2 (Mean, in) 4 4.02 0.5% Length Dimensions, Powder Sine Roller Bar 2 (Mean, in) 4 4.04 1% As seen in Table 2, the percent errors between the original sine roller block and the prototypes were still below 5%. The overall shapes of the prototypes were identical to those of the original. However, the sharp edges of the block were dull and the ends of the slits were rounded. Table 3 had some high percent errors in regards to the diameter of the rods. This is due to the limitation of the scanner and its field of view. Regardless of the number of attempts that were made to scan the rods correctly, the scanner was not able to recreate the smooth cylindrical surface. Figure 3 displays the original and the prototyped objects. Table 4 shows the measurements of surface roughness of all the objects. As Table 4 illustrates, the Sine Roller was actually coarser in both prototypes (the powder was softer than the plastic). The Sine Roller Bar prototypes were also much coarser than the originals due to the limitations of the scanner. The TV Remote prototypes were also much coarser than the original because the plastic of the original remote control was very smooth. The powder prototypes were generally smoother due to the texture of the material. Unfortunately, the NextEngine s Desktop 3D Scanner could not scan the diagnostic trees correctly. The scanner was not able to scan the ceilings of each limb of Christmas trees and, therefore, the 3D scan was not completed and the prototype was not constructed. Table 4 - Surface Roughness July 23, 2008 1:00PM Object (Original) Sine Roller 31.96 Sine Roller Bar 1 12.92 Sine Roller Bar 2 10.88 TV Remote 0.86 Object (Plastic) Sine Roller 24.56 Sine Roller Bar 1 21.92 Sine Roller Bar 2 22.1 TV Remote 23 Object (Powder) Sine Roller 22 Sine Roller Bar 1 24.32 Sine Roller Bar 2 20.92 TV Remote 19.2 Fig. 3 - From Left, Original Roller and Bar, then Plastic and Powder Prototypes ISSN: 1790-5109 125 ISBN: 978-960-474-013-0
4. Conclusions and Recommendations The following conclusions are drawn from the research: - The NextEngine s Desktop 3D Scanner is a lowcost system but powerful enough to reverse engineer many objects within the limits of its size and geometry. -One limitation of the NextEngine s Desktop 3D Scanner is that it has difficulty retrieving data from objects which contain many features. - Depending on the settings of the scanner, curves become somewhat rigid and sharp edges become dull while creating a 3D scan of an object. - Some other shortcomings are the inability to detect semi-reflective objects, the inability to detect the color black, the small area of detection, and the long time taken for each scan. - This scanner and software is viable for small-scale objects, however, the large objects will need an alternative form of reverse engineering. - The scanner was able to recreate a similar (less than a 5% margin of error), if not completely identical, prototype which was of similar dimensions to that of the original. - The Z Corporation s Spectrum Z510 rapid prototyping machine using powder produced prototypes which were softer in texture when compared to the Stratasys FDM-1650 rapid prototype machine that uses plastic. The recommendations for the future research are the following: - An alternate scanner should be used in order to compare speed, efficiency, and constraints of the process. - More objects of various sizes and details should be used in order to determine how viable reverse engineering scanners really are. References [1] Noorani, Rafiq., Rapid Prototyping: Principles and Applications. Wiley; Oct., 2006 [2] S. Singh, Rapid Reverse Engineering to Rapid Prototyping: A Case Study. Proceedings on Reverse Engineering, SME, Newport Beach, CA, (December 9-10, 1998). [3] G. Lin and L. Chen, A Vision-Aided Reverse Engineering Approach to Reconstruct Free-Form Surfaces. Proceedings of CAD/CAM, Robotics and Factories of the Future, London, (August 14-16, 1996) 854-859 [4] G. Lin and L. Chen, An Intelligent Surface Reconstruction Approach for Rapid Prototyping Manufacturing, The Fourth International Conference on Control, Automation, Robotics and Vision, Singapore, (December, 3-6, 1996) 43-47. [5] C. Schoene and J. Hoffmann, Reverse Engineering Based on Multi Axis Digitized Data, Proceedings of the International Conference on Manufacturing Automation, Hong Kong, (April 28-30, 1997) v909-914. [6] N. Tsang, K. Chan, and S. Tan, Reverse Engineering: From Discrete Point Data to 3D CAD Model, Proceedings of the International Conference on Manufacturing Automation, Hong Kong, (April 28-30, 1997) 921-927. [7] NextEngine Preliminary User s Guide: Model 2020i Desktop 3D Scanner, NextEngine, USA, 2007 [8] Hibbeler, R.C., Mechanics of Materials, Sixth Edition. Pearson Prentice Hall: New Jersey, 2005. ISSN: 1790-5109 126 ISBN: 978-960-474-013-0