Sie sind auf Seite 1von 7

Journal of Materials Processing Technology 150 (2004) 201207

Some demands on rapid prototypes used as master patterns in rapid tooling for injection moulding
P. Dunne a , S.P. Soe b , G. Byrne a, , A. Venus c , A.R. Wheatley b
b

Department of Mechanical Engineering, University College Dublin, Dublin, Ireland Institute for Automotive and Manufacturing Advanced Practices, University of Sunderland, Sunderland, UK c Engineering Solutions, IBM (UK) Ltd., Warwick, UK Received 18 April 2001; received in revised form 1 February 2002; accepted 11 March 2003

Abstract Injection moulded components are nding vigorously increasing applications in numerous sectors. The exibility of the process allows complex components to be manufactured in high volumes in fully automated environments. Associated with such components are equally complex tools whose manufacture is both expensive and time consuming. It is therefore vital that the component design be evaluated thoroughly with regard to both performance and manufacturability before commitment is made to expensive production tooling. The use of a physical prototype often provides the best method of design evaluation. The major routes to prototype manufacture are; material removal, rapid prototyping and rapid tooling. Rapid tooling techniques provide the most representative route, with the ability to produce polymeric prototypes (often in nal material) by an injection moulding process. Research has been undertaken into two rapid tooling processes; enhanced silicone moulding and sand moulding. Both processes involve casting material against a rapid prototype (master pattern) to form a mould which is then used in the manufacture of product prototypes by injection moulding. The focus of this paper is on the demands of these tooling techniques placed on rapid prototypes when used as master patterns. The demands are discussed in terms of surface nish, surface porosity, dimensional accuracy, thermal resistance and thermal expansion. 2003 Published by Elsevier B.V.
Keywords: Rapid prototyping; Rapid tooling; Silicone moulding; Sand moulding

1. Introduction The injection moulding process is capable of producing high volumes of complex components in fully automated environments [1,2]. In addition, components are produced in a single operation with very little waste. If quality components are to be produced by the process, then manufacture of a precision mould is necessary. A complex component geometry will result in complicated mould cavities and perhaps moving cores. Runners, ejectors and location pins must also be incorporated. The machining of such features is complicated, e.g. by the fact that mould materials are typically hard metals which are difcult to machine. Finally, the addition of cooling channels and the hardening of inserts may be required such that the mould be capable of withstanding the high temperatures and pressures of the moulding cycle. These factors make the design and manufacture of a production mould

both costly and time consuming. Consequently, the decision to invest in a production mould requires utmost condence in the component design. In order to achieve such condence various methods of design evaluation are undertaken.

2. Design evaluation There are many ways of evaluating the performance of a design. These include the use of mathematical evaluation techniques and physical evaluation techniques. Some examples of mathematical evaluation techniques are shown in Fig. 1: (left) application of fundamental and engineering equations (e.g. mechanical and thermal) to predict signicant, but only fragmented aspects of a design; (middle) geometric models typically produced using CAD, components dened in 3D CAD, for example, can be assembled virtually, for interference checking and nally (right) nite element models which represent a unied approach. Many designers prefer physical simulation in combination with virtual models [3]. Physical prototypes can take

Corresponding author. Tel.: +353-1-7161883. E-mail address: gerald.byrne@ucd.ie (G. Byrne).

0924-0136/$ see front matter 2003 Published by Elsevier B.V. doi:10.1016/S0924-0136(03)00571-5

202

P. Dunne et al. / Journal of Materials Processing Technology 150 (2004) 201207

Fig. 1. Virtual evaluation techniques (LR), mathematical, geometric and nite element models of a snap tag.

many forms, from simple mock-ups which represent limited elements of a design to more complex prototypes which embody multiple elements of the intended design. Physical prototypes can be machined or made using rapid prototyping and tooling techniques. Machining techniques are both widely available and capable of producing components of high accuracy. Advances in the area of computer-aided manufacture (CAM) mean that machining information can be generated directly from a CAD model with relative ease. This eliminates the need for fully annotated engineering drawings. Furthermore, there is very little limitation on the size of the prototype that can be machined [4]. While it is possible to machine a prototype in nal material, typically the raw stock used is extruded rather than moulded. As a result, the ow orientation within the material will differ from that of the nal moulded component impacting on the overall mechanical properties of the prototype [5]. The main limitation of machining is that it is highly labour intensive. This is particularly true for complex components which may require several machining operations and multiple set-ups. This problem is often further compounded by the need for special clamps and xtures. In spite of this, table top computer numerically controlled (CNC) milling machines are nding application in the prototyping area and many rapid prototyping bureaux offer CNC machining as a viable route to prototype manufacture. Rapid prototyping techniques provide an alternate route with prototypes usually produced in a layered manufacture process. These techniques, introduced in the late 1980s, have matured and are now well established. The most common process, stereolithography (SLA), uses a laser to cure layers of liquid photopolymer. Based on information derived from the CAD model the laser is guided over the surface of a liquid photopolymer causing those areas to solidify and form a layer. The layer, which rests on a moving platform, is then lowered into the vat of liquid photopolymer on top of which subsequent layers are formed. Other commercially available systems include selective laser sintering (SLS), fused deposition modelling (FDM) and laminated object manufacture (LOM). Despite offering signicant leadtime reductions over conventional routes, rapid prototyping techniques have inherent technical limitations. First, the layered build

approach leads to stepped approximation of surfaces. Typically a degree of post-processing or nishing is required. The range of materials that can be processed is also limited. Stereolithography, for example, can only build components from photopolymers which are generally of low strength and brittle in nature. Where it is possible to build components in the intended nal material, processing differences will impact on the materials properties attained. Finally, rapid tooling techniques are nding application. These involve the manufacture of a low cost mould. Rapid tooling techniques can be classied as either direct or indirect. Direct techniques are those which employ layered manufacturing technologies to the manufacture of the mould. Laser sintering processes, for example, are capable of producing metal mould cavities in a range of materials. These cavities are suitable for producing prototypes in nal material by an injection moulding process [17]. Indirect techniques employ a combination of other established technologies with rapid prototyping technology. Typically the mould is produced by casting suitable material against a rapid prototype which serves as a master pattern. The cast material is allowed to cure following which the rapid prototype is removed. The resulting mould is then used to manufacture multiple prototypes. Examples of suitable casting materials are silicone rubber, epoxies and ceramics. Currently the main limitation regarding rapid tooling techniques is one of process economics. Rapid, low cost processes such as vacuum casting which uses silicone rubber are not capable of producing highly representative prototypes while those processes deemed technically adequate are more expensive and require greater lead times.

3. Rapid tooling processes In an effort to overcome such limitations, research has been undertaken into two indirect rapid tooling processes for injection moulding; enhanced silicone moulding and sand moulding. Both processes involve casting a material against a rapid prototype master pattern. The material is then cured to form a mould as shown in Fig. 2. Runners and gates are generally incorporated into the master pattern as necessary.

P. Dunne et al. / Journal of Materials Processing Technology 150 (2004) 201207

203

Fig. 2. Mould manufacture stages (LR), rapid prototype master pattern, material casting, material curing and pattern removal.

3.1. Enhanced silicone moulding The enhanced silicone moulding process is based on the widely employed vacuum casting process. Silicone rubber, impregnated with metal powder is used as the mould material. The addition of the metal powder improves the stiffness of the mould such that injection moulding of thermoplastics is possible. Hardness tests carried out at 25 C on the enhanced silicone rubber indicated a hardness of 63 Shore A compared with 39 Shore A for the standard silicone rubber [6]. The material is prepared by mixing silicone rubber with 50 wt.% iron powder, addition of catalyst which vulcanises the rubber, followed by degassing. The resulting mixture is then carefully cast against the master pattern and allowed to cure. This typically takes 24 h at room temperature. By increasing the temperature to the 40 C region, curing times can be reduced to between 4 and 8 h depending on mould geometry. Finally the master pattern is removed. No further processing is required (Table 1). 3.2. Foundry sand moulding The foundry sand process is a derivative of the traditional metal sand casting process. The material used is commercially available 3% phenolic resin-coated zircon sand. The material has a ne grain size, a higher thermal conductivity and lower thermal expansion than silicon sand. The material which is granular in nature requires no preparation and can be readily cast around a master pattern. The material is then cured by heating. Curing temperatures in the region of 150 C, applied for durations of approximately 24 h are typical [7]. The master pattern is removed from the mould and surface hardened, to allow it to be used for injection moulding. This is achieved by inltrating hardener into the
Table 1 Enhanced silicone and foundry sand moulding process details Enhanced silicone Material Preparation Silicone rubber and iron powder Catalyst to rubber, mixed iron powder, vacuum 23 C, 24 h None Foundry sand Resin-coated zircon sand None 150 C, 24 h Apply hardener, hand nishing

porous sand surface. Finally, slight hand nishing is required (Table 1).

4. Demands on rapid prototypes An important consideration when preparing rapid tools is the achievement of good dimensional accuracy and surface nish of the cavity. These are largely affected by the rapid prototype master pattern and the ability of the mould material to replicate that master. For the purposes of this work we are only concerned with rapid prototype aspects. The demands placed on rapid prototypes can be broadly categorised into three areas: (i) pre-processing demands (ii) processing demands and (iii) post-processing demands. 4.1. Pre-processing demands The surface nish of the master pattern will be replicated to some extent on both the mould surface and subsequently moulded components. Silicone rubber is capable of replicating surface features/defects in the order of 1 m [8]. Foundry sand is more limited as a result of its granular nature, an average surface roughness Ra in the order of 6 m has been reported [7]. In either case much of the pattern detail, be it feature or defect, may be replicated. Consequently, master patterns need to be thoroughly inspected and nished to remove traces of the support structure and stepped build effects before use. This is both time consuming and detrimental to the dimensional accuracy of the pattern. A number of possible solutions have been investigated across the range of rapid prototyping systems at both process and post-process levels. These include materials developments, improved build strategies and investigation of sanding, tumbling and abrasive ow machining as possible nishing techniques. In total these investigations have yet to yield signicant results which will eliminate the problem and provide a consistent, predictable and acceptable surface nish [9,18]. A comparison between unnished and nished stereolithography average surface roughness Ra is given in Fig. 3. The error bars indicate the maximum and minimum values recorded. The two prototypes were manufactured on a SLA-250 machine in SL 5170 material with a layer thickness of 0.15 mm using an ACES (accurate clear epoxy solid) build style. One of the prototypes was left unnished

Curing regime Surface treatment

204

P. Dunne et al. / Journal of Materials Processing Technology 150 (2004) 201207

Fig. 3. Rapid prototype surface nish results. Inset shows surface nomenclature.

while the other was microblasted at 0.55 MPa (80 psi) using Honite 13 grit. The average surface roughness Ra of each prototype was then measured using a contact stylus prolometer. The top surface is effectively the nal layer of the build, the bottom surface is the underside of the component while the side surface is made up of the edges of each layer. The results for the unnished stereolithography prototype show a large variation across the component with average surface roughness Ra values of 0.25, 3.94 and 5.38 m being recorded for the top, bottom and side surfaces, respectively. Visually this equates to a component with a very inconsistent surface nish throughout. The top surface has a glass-like texture, the bottom surface is characterised by traces of both the support structure and striations resulting from the laser scanning, while the side surface is characterised by the edges of the stacked layers. For the nished stereolithography prototype average surface roughness Ra values of 0.44, 1.05 and 1.30 m were recorded for the top, bottom and side surfaces, respectively. Specically, these results represent a reduction in surface roughness on the bottom and side surfaces to the 1 m region and a relatively minor increase in the top surface roughness from 0.25 to 0.44 m. In more general terms these results are indicative of a more consistent surface nish across the part. For comparative purposes the surface roughness achievable with milling and grinding are in the 110 and 0.12 m ranges, respectively [21]. Porosity can also pose a problem, particularly when combined with a mould material capable of a high level of surface replication. The effect is one of the mechanical locking with the mould material lling and then curing in the pore. Separation of the pattern from the mould becomes difcult and can result in tearing of either component. Often this renders the mould unusable. Porosity is a particular problem for prototypes produced from powders, selectively laser sintered components, for example [10,24]. This is illustrated in Fig. 4 where the sintering operation results in a quasi-solid porous material. The density of sintered polycarbonate, for example, is approximately 85% that of moulded polycarbonate, while for PVC the corresponding value is approximately 60%. Typical particle sizes used in the process range from 50 to 120 m [15]. This problem can be overcome by seal-

ing the pattern. For the enhanced silicone rubber process any appropriate sealant can be used. However for the foundry sand process, the high temperatures needed for material curing limits the range of sealants that can be used. The use of adhesives has been investigated in previous work. Composite nylon selective laser sintered specimens were sealed using cyanoacrylate adhesives, dried in air and hand nished using sandpaper. The investigation concluded that adhesives were suitable for use as sealants. An improvement in surface nish was also noted. An average surface roughness Ra of 1.65 m was recorded for the sealed pattern after nishing against an equivalent value of 2.73 m for the nished unsealed pattern. The average surface roughness Ra of the unnished unsealed pattern was 13.25 m. Further to these surface issues are demands on the dimensional accuracy of the prototype. Despite continued improvements over the past decade, obtainable accuracies remain limited when compared with CNC machining [20]. The selective laser sintering and stereolithography processes, for example, have claimed accuracies of 0.05 and 0.1 mm, respectively [11]. Achieving such accuracies in reality however requires a thorough understanding of the process and considerable experience. A dimensional investigation was undertaken on the unnished and nished stereolithography prototypes used for the surface trials. The average results are presented in Fig. 5. The error bars represent maximum and minimum values recorded. The prototypes are rectangular in shape, nominal length 150 mm, width 25.4 mm and depth 3.175 mm, and were built such that the depth measurement corresponds to the direction of build or Z-direction as shown in Fig. 3 (inset).

Fig. 4. Surface porosity as a result of laser sintering, unsintered (left) and sintered (right).

P. Dunne et al. / Journal of Materials Processing Technology 150 (2004) 201207

205

Fig. 5. Rapid prototype dimensional accuracy results.

Measurement of the unnished prototype indicated an oversized component with an average length of 150.69 mm, width of 25.52 mm and depth of 3.181 mm. These gures represent relative dimensional errors of +0.46, +0.47 and +1.89%, respectively. These results show that stereolithography is less accuracy in the Z-direction (depth measurement) than in the XY laser scanning direction (length and width measurement). This characteristic is widely reported in the literature [22]. Reduced accuracy in the Z build direction is also characteristic to other rapid prototyping systems including selective laser sintering and laminated object manufacturing [14,23]. Measurement of the nished prototype showed a reduction in all dimensions as would be expected after mechanical nishing. Specically the average dimensional changes recorded were 0.06 mm over the length, 0.15 mm across the width and 0.05 mm over the depth. At rst sight the similarity between the length and depth reductions perhaps suggest a constant removal rate. Closer inspection however suggests that the similarity is more coincidental on account of the wide spread of results for the depth measurement. The results for the length measurement show an average reduction of 0.06 mm with 0.05 mm spread. The nishing has not signicantly improved the accuracy over the length, reducing the dimensional error from +0.69 to +0.63 mm. The width measurements show the greatest actual variation with dimensional changes of up to 0.2 mm in places. This dimensional change corresponds to high material removal from the side surfaces to remove the stepped build effects. This high material removal was not as evident on the stepped surfaces across which the length was measured. These surfaces were signicantly smaller in surface area and perhaps received less attention from the operator during the nishing operation. Finally, the depth measurements show an average reduction of only 0.05 mm with a spread of 0.21 mm. While the spread of results is signicant, comparison with the unnished results indicates that the variation is independent of nishing. In general the overall results correspond with the literature which reports dimensional changes in the order of 0.1 to 0.3 mm after nishing [19].

The issue of dimensional accuracy is receiving attention at many levels, both commercial and academic, with much research being undertaken into such effects as warpage, distortion, swelling and shrinkage [12,15,16]. 4.2. Processing demands During processing, the master pattern can be subjected to thermal, chemical and gravitational effects. For the case of the enhanced silicone moulding process, the pattern is encapsulated by the curing rubber mixture. Fortunately no signicant temperature rise results during vulcanisation, allowing patterns of virtually any material to be used. By comparison, the curing regime for the foundry sand process requires the addition of heat. The rapid prototype must therefore possess good heat resistance in order to at least survive the curing process. Experimental trials undertaken have indicated that this heating can lead to unacceptable distortion of the pattern (sintered trueform prototype) or de-lamination (laminated prototype). While stereolithography and sintered nylon prototypes survived the curing regime, it was found that thermal expansion resulted in loss of dimensional accuracy [7]. These prototypes, made from polymeric materials have thermal expansion coefcients in the order of 60100 106 K. This corresponds to a linear expansion of 0.61% when raised through 100 C. This would compare with an equivalent linear expansion of around 0.1% in most metals. Because of this high expansion coefcient of polymers, it is desirable that the cure of the sand binder resin is achieved at the lowest temperature possible to minimise expansion of the master within the cavity. One possible solution is to employ a two stage curing regime. The material is rst cured at 125 C for 1 h with the master pattern in place, resulting in a mould with good handling strength. The master pattern is then removed and the sand mould further cured to full strength (Table 2). Ultimately these expansions which result in an oversized cavity may not be a problem. Mould cavities are generally oversized to counteract shrinkage of the injected polymer. All that is therefore required is a design calculation to en-

206

P. Dunne et al. / Journal of Materials Processing Technology 150 (2004) 201207

Table 2 Thermal test results for the foundry sand process Process SLS Material Temperature ( C) Time (h) Comment Trueform 125 1 Distorted SLS Nylon 200 2 Good SLA SL 5170 125 1 Expansion LOM LPS 038 150 2 Peel off

sure that the rapid prototype model is of the appropriate dimensions. This is routine in many moulding application. 4.3. Post-processing demands Once the curing regime has been completed, it remains to remove the master pattern such that the mould can be assembled and used for injection moulding. For the enhanced silicone rubber process this is simply achieved as the mould material is exible in nature. The mould is simply exed and the pattern removed. For the rigid foundry sand mould this stage can be more problematic. During the experimental trials it was found that the rougher mould surface made removal of the pattern quite difcult, in some cases areas of the mould were found to break away. This problem can be minimised by ensuring that adequate draft has been included in the component design and that the rapid prototype is nished to a satisfactory level before use.

resin and process developments such as the introduction of adaptive slicing to improve both accuracy and surface nish. Commercially such developments are easily justied. It has been predicted that the rapid prototyping industry worldwide will have a revenue of $1.1 billion in the year 2000. In addition over 50% of prototypes produced are used in applications where materials properties and/or dimensional accuracy are critical [13,20]. Technical development therefore, will undoubtedly receive much attention from system manufacturers wanting to survive in an increasingly competitive industry.

Acknowledgements The authors wish to acknowledge the support of AMT Ireland, the Science and Innovation Directorate at Enterprise Ireland, Schenectady Europe Ltd., Hepworth Minerals and Chemicals (HMC) Ltd. and the School of Computing, Engineering and Technology of the University of Sunderland.

References
[1] V.J. Stokes, Thermoplastics as engineering materials: the mechanics, materials, design, processing link, Trans. ASME 117 (1995) 448455. [2] D. Bryce, Plastic Injection Moulding: Material Selection and Product Design Fundamentals, SME Publication, USA. ISBN 0-87263-488-4, 1996. [3] W. Pollman, Prototyping at Daimler Benz: state of the art and fundamental requirements, in: Proceedings of the IMS International Conference on Rapid Product Development, 1994. [4] A. Lennings, CNC offers rapid prototyping on the desk top, in: Proceedings of the Prototyping Technology International97, pp. 297301. [5] W. Michaeli, S. Brinkmann, M. Langen, Sind formteile aus RT-werkzeugen seriennah? Kunstoffe 86 (11) (1996) 16741676. [6] P. Dunne, Young, G. Byrne, Dimensional stability in rapid tooling processes, in: Proceedings of the 15th Conference of the Irish Manufacturing Committee, 1998, pp. 485494. [7] S.P. Soe, A. Venus, J. Elder, A. Wheatley, The development of injection mould tooling processes by using resin coated sand, in: Proceedings of the Seventh European Conference on Rapid Prototyping, November 1998. [8] W. Hechtl, R. Kastner, Silicone rubber, in: Mold-Making Handbook for the Plastics Engineer, Carl Hanser, Munich, 1983, pp. 270280. ISBN 3-446-13629-0. [9] P. Reeves, R. Cobb, Surface roughness investigation of stereolithography ACES components, in: Proceedings of the Second National Conference on Developments in Rapid Prototyping and Tooling, 1996, pp. 1725. [10] P. Dickens, R. Stangroom, et al., Conversion of RP models to investment castings, Rapid Prototyping J. 1 (4) (1995) 411. [11] D. Pham, S. Dimov, F. Lacen, Techniques for rm tooling using rapid prototyping, J. Eng. Manuf. B 212 (4) (1998) 269277. [12] S. Onuh, K. Hon, Optimizing build parameters for part accuracy in stereolithography using Taguchi methods, in: Proceedings of the Second National Conference on Developments in Rapid Prototyping and Tooling, 1996, pp. 115. [13] T. Wohlers, Rapid prototyping and tooling: state of the industry, 1998 Worldwide Progress Report, Wohlers Associates, 1998.

5. Conclusions The technical demands of both the enhanced silicone and foundry sand processes have been described. These include demands on surface nish, porosity, dimensional accuracy and heat resistance of rapid prototypes. The surface nish has been shown to be more critical for the enhanced silicone moulding process which is capable of replicating features of the order of 1 m. The porosity of rapid prototypes effects both processes with mould material lling and curing in the pore. This problem can be overcome by sealing the rapid prototype. For the case of the foundry sand process the sealant must also possess heat resistant characteristics, cyanoacrylate adhesives have been used successfully on selective laser sintered prototypes. The dimensional accuracy of rapid prototypes also effects both processes. In particular, the effect of nishing of the rapid prototype has been shown to effect accuracy. Dimensional changes as large as 0.2 mm as a result of nishing of stereolithography specimens were recorded. Finally, heat resistance of rapid prototypes has been shown to be necessary for the foundry sand process. Currently no rapid prototyping systems offer a satisfactory solution to these demands. The literature does however report on many developments being undertaken in these areas. These include materials developments such as the introduction of high temperature SL-5530HT stereolithography

P. Dunne et al. / Journal of Materials Processing Technology 150 (2004) 201207 [14] M. Durham, T. Grimm, J. Rollins, SLS and SLA: Different Technologies for Different Applications, Promotional Literature, Accelerated Technologies Inc., Austin, TX, April 1996. [15] P. Jacobs, The role of rapid prototyping and manufacturing, in: C. Machover (Ed.), The CAD/CAM Handbook, McGraw-Hill, New York, Chapter 10, pp. 97112. ISBN 0-07-039375-3. [16] J. Kruth, M. Leu, T. Nakagawa, Progress in additive manufacturing and rapid prototyping, Ann. CIRP 47 (2) (1998) 525540. [17] E. Fritz, Laser-sintering on its way up, in: Proceedings of the Prototyping Technology International98, 1998, pp. 186189. ISSN 1367-2436. [18] R. Williams, V. Melton, Abrasive ow nishing of stereolithography prototypes, Rapid Prototyping J. 4 (2) (1998) 5667. [19] J. Male, N. Lewis, G. Bennett, The accuracy and surface roughness of wax investment casting patterns from resin and silicone rubber tooling using a stereolithography master, in: Proceedings of the

207

[20]

[21] [22] [23]

[24]

Second National Conference on Developments in Rapid Prototyping and Tooling, 1996, pp. 4352. Y. Yan, G. Hong, Development and tendency of rapid prototyping systems, in: Proceedings of the SPIE on Laser Process of Materials and Industrial Applications II, vol. 3550, 1998, pp. 298-306. E. Doebelin, Measurement Systems: Applications and Design, 4th ed., McGraw-Hill, New York. ISBN 0-07-100697-4. J. McClurkin, D. Rosen, Computer-aided build style decision support for stereolithography, Rapid Prototyping J. 4 (1) (1998) 413. T. Lowe, O. Johnson, Integrating RP technology, in: Proceedings of the Prototyping Technology International98, International Press, UK, 1998, pp. 115118. ISSN 1367-2436. J. Kruth, L. Froyen, B. Morren, J. Bonse, Selective laser sintering of WCCo hard metal parts, in: Proceedings of the Eighth International Conference on Production Engineering, Rapid Product Development, 1997, pp. 149156.

Das könnte Ihnen auch gefallen