资源描述
Rapid Prototyping Versus Virtual Prototyping in Product Design and ManufacturingC. K. Chua1, S. H. Teh1 and R. K. L. Gay2School of Mechanical & Production Engineering; and 2Gintic Institute of Manufacturing Technology, Nanyang Technological University, SingaporeAbstractRapid prototyping (RP) is the production of a physical model from a computer model without the need for any jig or fixture or numerically controlled (NC) programming. This technology has also been referred to as layer manufacturing, material deposit manufacturing, material addition manufacturing, solid freeform manufacturing and three-dimensional printing. In the last decade, a number of RP techniques has been developed. These techniques use different approaches or materials in producing prototypes and they give varying shrinkage, surface finish and accuracy. Virtual prototyping (VP) is the analysis and simulation carried out on a fully developed computer model, therefore performing the same tests as those on the physical prototypes. It is also sometimes referred to as computer-aided engineering (CAE) or engineering analysis simulation. This paper describes a comparative study of the two prototyping technologies with respect to their relevance in product design and manufacture. The study investigates the suitability and effectiveness of both technologies in the various aspects of prototyping, which is part and parcel of an overall design and manufacturing cycle.Keywords: Product design; Rapid prototyping; Virtual prototyping1. IntroductionRapid prototyping (RP) is emerging as a key prototyping technology with its ability to produce even complicated parts virtually overnight. It enables product designers to shorten the product design and development process. The coming-of-age of this technology is clearly reflected in the inclusion of a stereolithography (STL) file generator in most, if not all, CAD. systems today. The STL file is the de facto standard used by RP systems in the representation of the solid 3D CAD models.While RP is a relatively young technology, virtual prototyping (VP) has been in steady development since the 1970s in many guises. Virtual prototyping is taken to mean the testing and analysis of 3D solid models on computing platforms. Today, VP is often tightly integrated with CAD/CAM software and sometimes referred to as CAE packages. It provides the ability to test part behaviour in a simulated context without the need to manufacture the part first 1.2. Definitions of RP and VPRapid prototyping (RP) is a widely used term in engineering, particularly in the computer software industry where it was first coined to describe rapid software development.This term has also been adopted by the manufacturing industry to characterise the construction of physical prototypes from a solid, powder, or liquid in a short period of time when compared to “traditional” subtractive machining methods. This technology has also been variously referred to as layer manufacturing, material deposit manufacturing, material addition manufacturing, solid freeform manufacturing and threedimensional printing 2.Virtual prototyping (VP) refers to the creation of a model in the computer, often referred to as CAD/CAM/CAE. Virtual or computational prototyping is generally understood to be the construction models of products for the purpose of realistic graphical simulation 1. In this paper, VP will refer to thesimulation, virtual reality and manufacturing process design domains 3.Nevertheless, there are many areas where the distinction between RP and VP is blurred. As RP systems rely on CAD systems to generate the files needed to produce the prototype, it would seem that RP is a downstream process from VP in the product or part development cycle. Indeed, Pratts definition of VP reveals the fact that VP is a term which is loosely used in the prototyping community. As such, it would be appropriate to clearly define both RP and VP.Rapid prototyping will be taken to mean, as above, the production of a physical model from a computer model without the need of any jig or fixture or NC programming. This also includes other related processes and applications which use RP-produced objects, such as rapid tooling.Similarly, VP is defined as the subsequent manipulation of a solid CAD model as a substitute for a physical prototype for the purposes of simulation and analysis, and is not inclusive of the construction of the solid 3D model. VP includes the following functions:1. Finite element analysis.2. Mechanical form, fit and interference checking.3. Mechanical simulation.4. Virtual reality applications.5. Cosmetic modelling.6. Assemblability.The relationships between RP and VP are shown in Fig. 1.Fig. 1. Classification of RP and VP3. Prototyping in SingaporeTwo selected multi-national companies (one American and one French) based in Singapore with significant product development activities showed differing approaches to both RP and VP. Both use RP in their prototyping activities.The first company, B, placed more emphasis on virtual prototyping. It manufactures telecommunications equipment such as pagers and handphones. It is moving all prototyping applications upstream, which is to move prototyping from RP to VP. At present, their RP models are used only for proof of concept and marketing purposes. Other prototyping activities are being carried out with VP.The second company, C, manufactures consumer electronics products such as television sets, video cassette recorders and telephones. It uses VP only as a tool to create a solid 3D model. From the solid 3D model, C generates the STL file needed to produce the RP prototype. Company C then uses the RP part as a master for silicone rubber moulds to produce a limited number of physical ABS (polyacrylonitrite butadienestyrene) prototypes for the various prototyping tests and simulation.Company B intends to move more prototyping to VP, rather than using physical models. Virtual prototyping allows for improvements in reliability and quality as well as reducing costs. Manipulation of virtual prototypes makes it easier for B to implement design improvements compared to an iterative cycle using physical prototypes.Company B drafts the CAD models in Pro/ENGINEER, then uses Patran to pre-process the models. Static finite-element analysis (FEA) is carried out with ABAQUS Standard whereas dynamic scenarios are analysed with ABAQUS Explicit. ALIAS/Wavefront is used for cosmetic modelling when presenting different conceptual and actual designs.The bulk of the VP carried out by B uses FEA, which typically takes 46 weeks for a pager design. Of all the FEA carried out, the majority are concentrated on structural strength (static) analysis and drop test (dynamic) analysis. Vibration tests are occasionally carried out. Some cosmetic modelling is carried out, but usually only for presentation purposes.Finite-element analysis is used to investigate the following problems:Relative comparison of different design options; to see how one design compares to another. Possible failure modes are:1. To evaluate a design change or design correction.2. To assess the possibility of failure, based on past experience.3. To make some educated-guess correlation with physical testing.4. To try to identify what initiated a failure.According to B, the drawback of VP is that it cannot simulate process problems efficiently and effectively. The accuracy of FEA is also limited because of the inconsistent behaviour ofmaterial. The amount of computing power also determines the accuracy of FEA.The application of RP is rather limited in B. The in-house laminated object manufacturing (LOM) RP system is used to produce design prototypes for proof of concept only, and notgeometrical prototypes. Company C uses RP heavily, but has very little VP. The parts produced using RP range from audio products to 29-in. television casings. Typically, it takes 1 year from the conception of the product to the sale of the product. Company C aims to prototype all (mostly plastic) parts by RP. A comparison between numerically controlled (NC) machining of prototypes from ABS against RP is shown in Table 1. Company C projected 50% savings using an in-house RP system versus an NC machining system.CAD models are created using I-DEAS. The .STL format is then created for production of the RP part. The main purpose of the RP parts is to verify the design. Rapid prototyping parts are used for the following functions:1. Form fitting.2. Ergonomics check.3. Proof of concept (to confirm design with industrialdesigners).4. Manufacturability (design for tooling, design for assemblability).5. Reliability check (whether part dislodges or breaks when force applied, especially snap-on covers).6. Kinematic check.Company C offers some insight into the limitations of VP, in that VP is unable to model:1. Tactile feeling (for buttons) not quantified; may be able to VP if able to quantify “pressing” force.2. Assemblability (e.g. PCBs inserted at an angle, difficult to visualise).4. Case Study 1: Prototyping of a Telephone HandsetThis case study investigates the design verification, assembly, interference check and form fitting aspects of both the RP and VP model. The production ABS, RP and VP parts or models were evaluated in the above aspects. The RP system used here is the stereolithography apparatus (SLA). Both the ABS and RP parts are shown in Fig. 2. Inspection of the RP parts reveal that:1. The surface finish was much poorer than in the ABS part.2. Warpage was clearly evident (see Fig. 3).4.1 Design VerificationAs a true dimensional physical part, the RP model is able to give the designer a sense of size estimation. The judgement of a VP part can be erroneous because parts are often automatically sized to fit the viewing window. Another advantage of a physical part is that it allows for ergonomic checks, ranging from the fit of a telecommunications device in a users palm to the inspection of potentially dangerous corners and edges. Also, it offers tactile inspection which is crucial in products for which ergonomics is important, such as touch buttons on audio or video products, which is not possible on VP systems.Rounded edges which appear innocuous on a VP model may prove to be unsafe upon scrutiny of the RP part. Above all, most RP parts are produced for aesthetic evaluation purposes. Aesthetic evaluation is also possible on VP models. All CAD software allows the model to be viewed in any spatial orientation, along with at least rudimentary rendering capabilities. It is then possible to view the part under the desired simulated lighting conditions with millions of shading and colour combinations. RP parts cannot be coloured, thus surface preparation and painting introduce additional finishing processes. Any visibly apparent design discrepancies could be immediately rectified without having to invest in a physical part. It also allows designers to evaluate the aesthetics of the design and make corrections, if necessary. In the case of most multi-national companies, the design and manufacturing facilities are often a considerable distance apart and in different countries and continents. The ease with which CAD files can be sent and received via electronic means greatly helps the design process, be it iterative or concurrent. With identical or compatible CAD software, the prototyping process can be swift and cheap. Any design change of the virtual prototype can bemade almost instantly available to all parties involved in the design process.4.2 AssemblyAssembly of RP parts must be carried out quickly, as warpage and shrinkage increases with time. Warpage is a function of both part geometry design and shrinkage. All but the bestdesigned parts suffer from varying degrees of warpage and shrinkage. Some RP material such as the SLA inherently shrinks and the part is actually built slightly larger to allow it to shrink to its proper dimensions. With such arrangements, assembly is possible but is often hampered by warpage and/or shrinkage. Some parts can be mated only with the application of some force. Assembly of RP parts allows the user not only to attempt different assembly sequences, but also if a part cannot be positioned in a linear movement, to insert the part, say, at an angle before being set into its proper location. The drawback in assembling RP parts is that for some RP parts such as SLA, the material is weak and brittle, and fails when attached using fasteners or under low to moderate loading (see Figs 4 and 5). CAD software allows for the assembly of parts and subassemblies in the form of 3D solid or surface models. Assembly in the virtual realm is very often used to check for interference and form fitting which will be discussed later. The ability ofCAD software to assemble parts and/or subassemblies allows a product designer to quickly check to see if he or she has designed the part or parts correctly, i.e. whether a boss is tall enough to accept a screw inserted through another part or if two slots are aligned to form a larger slot. The advantage of assembling in a virtual environment is that no physical parts need be produced and thus this reduces cost. The absence of physical parts also means that tooling time is eliminated. The assembly in a virtual environment can be done in a matter of minutes or up to a few days, but is much faster than producing the physical parts and then assembling them. The user can also build or change a part, or modify its attributes when all instances of the part will be changed accordingly. Assembly relationships can be written in engineering parameters, part dimensions and orientation dimensions. The equations are solved variationally to allow for flexibility while working with the assembly. Evaluation of the tolerance specifications of the design to optimise the engineering performance at the lowest possible cost can be carried out. This allows the user to measure the sensitivity of a critical dimension in an assembly to changes in individual constraints. Manufacturing cost can then be reduced by tightening the tolerances which contribute most to the overall variation of a critical dimension, and loosening tolerances that have little impact.4.3 Interference Check and Form FittingAgain, interference checking and form fitting is hampered by warpage and shrinkage of the RP part. Therefore, the problem of parts which interfere or fit poorly may be due to one or more of: warpage; shrinkage; or design error. Even when RP parts fit well, there is no assurance that the parts are dimensionally correct, as shrinkage of two or more parts in the same direction or directions could still produce a good fit. When such situations arise, CAD models are often used to determine whether the interference or poor fit is due to design flaws.The ability to check for interference as well as form fitting is very widely used in CAD systems. It gives the user the ability to fit two parts together and check for interference without having to produce a part or parts which are potentially dimensionally incorrect, thereby increasing cost.The interactive nature of the process in a CAD system also frees the user or designer from the need to manually interpret engineering drawings to detect interference. This process also allows the user to establish tolerances which are crucial in the manufacturing process. The advantage of interference checking on a CAD system is not evident when an assembly consists of a small number of parts. For complex assemblies with alarge number of parts, there are often many features on a particular part that must be mated or aligned with features on one or more other parts. CAD systems allow not only the detection of any misalignment or interference but also immediate rectification of the problem. Interference checking is performed by the CAD system on an assembly when required by the user, and is relatively faster and more accurate and precise than other methods. The CAD system would also identify and list the features which interfere. The user can then view the entities to rectify the situation.5. Case Study 2: Prototyping of a Knee Prosthesis5.1 BackgroundRapid prototyping has applications in the field of medicine. However, in this application the STL file is no longer obtainable from a CAD model. There is a need to generate the necessary STL files from data acquired by medical equipment. Swaelens and Kruth 4 proposed three approaches to producing an RP part from computer assisted tomography (CT) scanner data (see Fig. 6). In most cases, STL-interfacing was used. In STL-interfacing, a CT scanner maps the contour of a 3D surface. This data is then converted into triangular file format which is then converted into the STL format required by RP machines. There is a direct conversion of data from the CT scanner to the RP machines. In effect, the scanned surface is faithfully reproduced by the RP machine.When used in this fashion, VP plays an almost negligible role, in RP-assisted surgery prototyping, as a viewer to verify the contour of the surface. Jacob et al. 5 constructed 3D models from CT scanner data using CTrans from Proform. They reported that the decisive advantage lies in the clearness and manual “getting in touch” as the surgery proper is elaborate manual craftsmanship. The model can be viewed and palpated from any angle and could even be operated upon. In that way, surgeons could literally grasp the problem. This study shows VP as a viewer for a 3D model. While the study did not state whether the 3D model was a solid model, it opened the possibility of integrating CAD software into the process, data exchange problems notwithstanding.This contrasts with the CAD system route shown above. Researchers in the University of Leuven, Belgium identified contours from CT scanner data and introduced them into CAD software to generate surface models. The physical model of a hip was produced with much effort, and the whole procedure took several working weeks.The procedure of converting CT scanner data to a solid 3D model is tedious and prone to error. Given the triangulation points from the CT scanner, they must be joined to the appropriate adjacent points to form curves. Confusion sometimes occurs when a surface folds back; while a point “below”is the nearest point, it may not be an adjacent point.These curves must then be individually and manually selected to define surfaces. Again, care must be taken to ensure that the appropriate surfaces which approximate the original surfaces are formed. After the surfaces are formed, they are connected to form patches or quilts. These quilts are then combined to form a surface model. If the surface model is fully enclosed, the CAD system may then convert it into a shell or solid 3D model.The complexity and shape of the human body also presents problems. Most of the extracted outlines are represented as complicated Bezier curves. A mapping algorithm sometimes fails to combine these Bezier outlines to form 3D data. So, it is necessary that this process be supported by hand 6. Human supervision is also required where software is unable to recognise features such as joints where bone structures abut. The data must be separated into individual components (disarticulation) in order to evaluate a function (e.g. a jaw joint).Direct interfacing has two major problems. The data from the CT scanner are in the form of shaded images and are automatically segmented. While it is possible to calculate triangles from the images, they do not contain enough surface information. Therefore, it is difficult for the RP system which requires supports to construct the appropriate support structures. Secondly, the interpolation from successive contours obtainedfrom the segmentation is not evident. The CT scanner has a threshold filter to isolate regions within the desired density range. When tissue density r
展开阅读全文