资源描述
Progress towards an international standard for data transfer in rapidprototyping and layered manufacturingM.J. Pratt*,*, A.D. Bhatt, D. Dutta, K.W. Lyonsb, L. Patilf, R.D. SriramAbstractThis paper discusses the informational requirements of rapid prototyping and layered manufacturing (RPLM). The study is motivated bythe recent decision to embark on the development of a new Application Protocol for the international standard ISO 10303, specifically tohandle layered manufacturing information.Keywords: Layered manufacturing; Rapid prototyping; Solid freeform fabrication; Standards; STEP; ISO 10303; Data exchange; Product data exchange1. IntroductionThe paper examines the informational requirements ofrapid prototyping and layered manufacturing (RPLM), anddescribes progress towards establishing a new electronicdata exchange standard for this mode of manufacturing.Current and proposed data exchange formats are reviewedand found to have insufficient representational capabilitiesin important developing areas of RPLM technology. Inparticular, it is noted that the capture and transfer of infor-mation about material property distribution is likely to be anessential need in the future, and that no formal proposal forRPLM data exchange has so far addressed this requirement.The study provides a preliminary survey of requirements forthe development of a new Application Protocol (AP) of theinternational standard ISO 10303, specifically for the trans-fer of RPLM data, which officially commenced in February2001. ISO 10303 1,2 is concerned with the electroniccapture and transfer of product data; it is unofficially knownas STEP (Standard for the Exchange of Product model data).The paper is organized as follows. First a brief review isgiven on the salient aspects of RPLM processes in general.The information needed during various stages of manufac-turing planning common to most of them is then discussed.Requirements for the manufacture of artefacts designed interms of inhomogeneous and/or anisotropic materials aregiven particular attention, because emerging technology inthis area appears to have great significance for the future ofRPLM. Several possible ways of representing non-uniformmaterials are briefly reviewed. Existing and proposed meansfor data exchange in RPLM are then examined and theiradvantages and disadvantages discussed. The resourcesprovided by the standard ISO 10303 are found to covermany of the RPLM data requirements established earlierin the paper, and the necessary extensions are identified.Finally, the paper outlines the current status of the recentlyestablished ISO 10303 project for RPLM data exchange.M. J. Pratt et al. / Computer-Aided Design 34 (2002) 1111l21N.O.2.Primary characteristics of RPLM processesRPLM, also known as solid freeform fabrication (SFF),represents a class of recently developed additive manufac-turing processes in which objects are constructed layer bylayer, usually in terms of a sequence of parallel planarlaminae. Many RPLM processes now exist, using a varietyof materials and layering methods 3,4. The methods maybe broadly classified as follows:. Photopolymer solidification (e.g. stereolithography, solidground curing), in which a liquid resin is hardened layerby layer with a laser or ultraviolet lamp. Material deposition (e.g. fused deposition modelling), inwhich drops or filaments of molten plastic or wax aredeposited to construct each layer. Powder solidification (e.g. selective laser sintering, 3Dprinting), in which powdered material layers are solidi-fied by adding a binder or by sintering with a laser. Partscan be built from ceramics, nylon, polycarbonate, wax ormetal composites. Lamination (laminated object manufacturing and solidground curing). The first of these methods uses lasersto cut layers from sheets of paper, cardboard, foil, plastic,or ceramic 5, stacks them and bonds them together. Thesecond uses a cut mask to expose regions of resin to besolidified by an ultraviolet lamp. Weld-based approaches. These methods produce a smallmelt pool from material delivered in the form of wire orpowder mix. Components manufactured with these pro-cesses use welding and cladding deposition techniquesthat result in fully dense material deposition, thoughfinish machining is usually required 6- 8.objects being built. For some RPLM processes, this econo-mizes on the wastage of expensive fabrication material byminimizing the amount that is unused and discarded aftereach run of the RPLM processor.A major prospective use of RPLM is for manufacturingheterogeneous objects. These may be made of more thanone material, contain embedded devices, or have continu-ously graded material properties. RPLM, unlike conven-tional manufacturing processes, in principle permitscomplete 3D control over material properties. New designmethods such as homogenization for structural topologydesign 12 specify varying material properties throughoutthe volume of the designed object; RPLM provides a meansfor producing such artefacts. Suggested applications of suchfunctionally graded materials include 13,14. Increased wear resistance for machine parts by use ofresistant surface material. Use of heat-resistant ceramics on the surface of gasturbine blades, grading to titanium in the interior forstrength. Manufacture of surgical bone implants with ceramicsurface material to ensure god bonding, grading tometal in the interior for strength. Design of electrical components by control of electricalproperties at a local scale. Time-dependent controlled release of ingested drugsthrough variation of the internal composition of pillsmanufactured by RPLM.RPLM was originally used. for rapid prototyping to helpthe designer verify part geometry, but is now increasinglyused to make production tooling, including moulds for cast-ings 9,10 and electrodes for electro- discharge machining(EDM) ll. It also has emergent use for the manufacture ofone- off and small batch production parts. RPLM processescan be used to build very complex artefacts, having intricategeometry, internal voids or multiple ready-assembled com-ponents, without any special tooling. Model transfer betweenCAD and the fabrication process is electronic, and the plan-ning of fabrication is largely automatic, demanding littlehuman intervention. However, because RPLM objects arebuilt layer by layer, their surfaces often have a staircaseappearance. They may also have inferior material propertieswhen compared with objects manufactured by other means.Poor surface quality is sometimes overcome by performingfinishing operations, such as grinding, polishing or NCmachining, after an object is built.Support structures are required by certain RPLM pro-cesses, when the object being built has overhangs or internalvoids, in which case part of any layer may have no under-lying layer to build it on. Supports are usually built togetherwith the part, and are later detached and discarded.Any RPLM process is characterized by a build volume.This determines the largest single artefact that can be fabri-cated at one time. In fact, it is a widespread practice tomanufacture multiple artefacts, simultaneously maximizingthe proportion of the build volume that is occupied by the3. Data transfer requirements for RPLMThe design and manufacture of an RPLM object are oftenperformed in separate organizations. Hence it must be pos-sible to transfer design data to the RPLM manufacturer,preferably electronically, to enable process planning andproduct fabrication. Since a single organization may manu-facture parts using several different RPLM technologies, aunified set of data transfer protocols is desirable for theentire class of processes. Although distinct RPLM processesuse specialized equipment, most of them use a tool that ismoved under computer control to add material to the part.Process planning for RPLM determines the tool paths andprocess parameter settings for the manufacture of a particu-lar object from a given material by a particular RPLMprocess 15. The basic steps, and their informationalrequirements, are as follows:1. Choice of object orientation during the build process.This requires knowledge of the 3D shape of the artefact.The objective may be to minimize build time or supportrequirements, maximize surface quality (e.g. avoid stair-case effects in functionally important regions), or to opti-mize in terms of some weighted combination of suchsimple criteria. More complex possibilities also exist.For example, the structural properties of the built objectare affected by the orientation and thickness of the layerswithin it, and by the spacing of deposition paths, since theRPLM process gives rise to small-scale material inhomo-geneity. These factors in turn will be affected by thechoice of process and material. The optimization ofstructural properties of an RPLM artefact could thereforealso be an objective at the orientation stage, but thiswould require material and process information in addi-tion to pure geometry. The output of the orientationtask is simply a transformation to orient the part modelappropriately.2. Support structure design. Here, the 3D object model isneeded. Support regions for any orientation can be foundby calculating the shadows it casts on the slicing planes.The output of this task may be a modified 3D shapemodel that includes the support volumes; alternatively,the supports may be represented only at the level of 2Dslice data.3. Part slicing. Given the oriented part model and itssupport volumes, this is a purely geometrical process ofdetermining the details of the layers to be laid down. Itsinput is the 3D object model, and its output is a set of 2Dslice contours, but if graded materials are used these mustalso contain 2D material distributions derived from aoriginal 3D distribution specified at the design stage.4. Path planning within each slice. The input to this stageis the shape of the slice, any 2D material distributionspecified for it, and details of the path planning strategyif any choice is available. For lamination methods asdescribed earlier, it is only necessary for the path totraverse the boundary of the slice. As they use uniformlaminae, these methods cannot handle graded materials.However, the use of fibre composite layers will requirematerial orientation to be specified. For other methods,an area fill strategy must be used with or without aboundary traversal. The path planning output is a geo-metric representation of the generated paths, with infor-mation on material composition if this varies within theslice. Criteria for path planning may include minimiza-tion of build time and requirements on part stiffnessand strength- -different scanning strategies give differentmaterial properties.The tasks described are common to most RPLM pro-cesses. Their common informational requirements establishthe possibility of standardizing RPLM data transfer.Other planning tasks in RPLM include process parameterselection, a very method-specific task, less suitable for astandard approach. Further, most RPLM techniques requirepost-processing of the built part (e.g. finishing operationsto improve surface quality, or curing following stereolitho-graphy to remove unsolidified material). Most such process-specific activities are currently not automated and involveno electronic data transfer; they are not further consideredhere. The exceptions are those methods that employ NCmachining for part finishing. However, they can in principleuse standards that already exist in that area.The following sections analyse various aspects of RPLMinformation in more detail.3. Geometry-related input needed for RPLMThe most basic input to a RPLM process planning systemis the description of the shape of the object to be manufac-tured. Additionally, tolerance information, surface finish,material data, etc. should ideally be used in performingcertain process planning tasks. However, the process plan-ning modules of most RPLM systems currently accept onlyshape information. Several types of shape representationexist. Computer aided design (CAD) models used to generateinput for RPLM are usually surface models or boundaryrepresentation (Brep) solid models 16. Either of thesemay be regarded as composed of bounded regions ofdifferent surfaces forming a large composite surface,and may in general be composed of combinations ofplanar and curved surface elements. Polygonal or faceted models are defined in terms ofplanar surfaces alone, the object boundary being repre-sented by a mesh of polygonal facets. If the facets areall triangular, the model is consistent with STL, themost commonly used input format for RPLM processplanning systems. Most CAD systems can compute andoutput such approximate triangulated surface or solidrepresentations (specifically, STL files), the chord-heightdeviation between the true object surface and thetriangular facets being controlled according to the usersrequirements on approximation accuracy.4. Existing and proposed RPLM data transfer protocolsWe now give a brief survey of some current and proposedformats for data transfer in RPLM. Ideally, any such formatshould be comprehensive in terms of information coverage,efficient in terms of data volume, and free from errors orambiguity. So far, we have not attained that desirable situa-tion, as will be shown. In particular, the transfer of materialdistribution information is not possible with any of thecurrent formats. Data transfer is becoming increasinglycritical for RPLM as accuracy requirements become morestringent 30.4.l. The STL formatAlthough it was developed specifically for the stereo-lithography process, STL has become the de facto industrialstandard for the transfer of data to RPLM process planningsystems. It represents a 3D shape in terms of a triangulatedapproximation of its boundary.The advantages and disadvantages of STL have beendiscussed exhaustively elsewhere, for example, see Refs.31,32. To summarize, STL files are conceptually simpleand easy to generate, but they have disadvantages of file sizeand numerical accuracy, are subject to a variety of errorsituations and lead to inefficient slicing algorithms.However, despite its proprietary nature and its associatedproblems, STL is currently almost universally used as aneutral (system-independent) format for the transfer ofRPLM shape models. Most STL file errors arise from thelimitations of triangulation algorithms, and can now becorrected by automatic or semi-automatic means. Thelarge file size does not pose severe difficulty in the present4.2. STEP Standard for the exchange of product modeldata. ISO 10303-44 (Product structure) specifies means formodelling assembly structures, which are relevant inRPLM since assemblies can be fabricated with all com-ponents in situ. ISO a 10303-45 (Materials) provides representationsfor material properties, covering requirements identifiedearlier. ISO 10303-47 (Shape variation tolerances) coversanother of the desired RPLM capabilities. ISO 10303-49 (Process structure and properties) pro-vides a basis for the transfer of process-related informa-tion in RPLM (if that proves possible or desirable). . ISO 10303- 50 (Mathematical constructs) contains themeans for transferring mathematical functions for therepresentation of such physical phenomena as electro-magnetic or flow fields, or more specifically in theRPLM context, material distributions.6. ConclusionsRPLM processes involve multiple tasks, each withspecific data requirements. V arious file formats are in useor have been proposed for transferring data between thosetasks. Most are restricted to the exchange of geometry data,though there is an increasing need for the transfer of addi-tional types of information. STL, the current de facto stan-dard, allows the transfer of approximate 3D shapeinformation alone. This is inadequate for the future, becauseof the increasing importance of non- shape information andmodel accuracy in RPLM. In the authors view, extendingthe STL specification will not give a good long-term solu-tion. The ultimate survival of STL is called into question by. The prospect that RPLM will become suitable forproduction use (apart from its current applications formoulds and dies) in the medium-term future. Such usewill place a premium on the transfer of exact productgeometry from the CAD system, together with informa-tion on tolerances, surface finish and materials. The possibility that some phases of RPLM planning willmigrate into the CAD system, raising the need for analternative standardized slice format. The prospect of using RPLM for the generation of partsThese considerations have, led us to initiate the develop-ment of a STEP AP for the transfer of RPLM data. ExistingSTEP resources cover most necessary types of information,including shape representation, assembly structure, materialspecification, process parameters, tolerance information andmathematical functions. 3D shape data can be transferredusing STEPs faceted Brep (a direct replacement for STL),or as an advanced Brep solid with topology information andgeneral curved surfaces. Additionally, STEP resources canbe used to represent slice contours in terms of lines orgeneral curves; thus both 3D and slice data can be definedin the same standard. The major required capability lackingin STEP is that for modelling inhomogeneous and non-isotropic materials, but suggestions have already beenmade for one new resource suitable for this purpose 27 .The development of a new AP for ISO 10303, aimedat the transfer of RPLM data, will be a timely expansionof the coverage of the STEP standard into an importantemerging area of manufacturing. Many of the STEPresources needed for this effort already exist as componentsof an international standard that is increasingly beingadopted by industry. They have the advantage that theyhave been tested and refined in practical use, and can beReferences1 ISO. Industrial automation systems and integration product datarepresentation and exchange, ISO 10303: 1994, 1994 (InternationalStandard ISO 10303; the ISO catalogue is at http:/www.iso.ch/cate/cat.html, search on 10303 for a listing of parts of the standard).2 Owen J. STEP: an introduction. 2nd ed. Winchester, UK: InformationGeometers, 1997.3 Burns M. Automated fabrication. Englewood Cliffs, NJ: Prentice-Hall, 1992.4 Wright P. 21st century manufacturing. Englewood Cliffs, NJ:Prentice-Hall, 2001.5 Klosterman DA, et al. Proceedings of the 1998 Solid Freeform Fabri-cation Symposium, Austin, TX; 1998. p. 671-80.6 Mazumder J, Koch J, Nagarathnam K, Choi J. Rapid manufacturingby laser aided direct deposition of metals, Technical report, Cham-paign, IL: Department of Mechanical Engineering, University ofIllinois, 1996.7 Griffith ML, Harwell LD, Romero JT, Schlienger E, Atwood CL,Smugeresky JE. Multi-material processing by LENS. Proceedingsof the 1997 Solid Freeform Fabrication Symposium, Austin, TX;1997. p. 387- -93.8 Brown S. How hot lasers are taming titanium. Fortune 2000;141(4):240. MI.J. rrall el al. 1 Compuler-Al9 Jacobs PF. Stereolithography and other RP and M technologies, fromrapid prototyping to rapid tooling. New York: SME Press, 1996.10 Karapatis NP, van Griethuysen J-PS, Glardon R. Direct rapid tooling:a review of current research. Rapid Prototyping J 1998;4(2):77- -89.11 Drr H, Pilz R, Eleser NS. Rapid tooling of EDM electrodes by meansof selective laser sintering. Comput Ind 1999;39(1):35- -45.12 Bendsoe MP, Diaz A, Kikuchi N. Topology and generalized layoutoptimization of elastic structures. In: Bendsoe MP, MotaSoares CA,editors. Topology design of structures, Amsterdam: Kluwer, 1993.13 Jackson TR, Patrikalakis NM, Sachs EM, Cima MJ. Modeling anddesigning components with locally controlled composition. Proceed-ings of the 1998 Solid Freeform Fabrication Symposium, Austin, TX;1998.14 Park S-M, Crawford RH, Beaman JJ. Functionally gradient materialdesign and modeling usin
展开阅读全文