Rapid Investment Casting: Direct and Indirect Approaches Via Fused Deposition Modelling
Rapid Investment Casting: Direct and Indirect Approaches Via Fused Deposition Modelling
1007/s00170-003-1694-y
O R I GI N A L A R T IC L E
Rapid investment casting: direct and indirect approaches via fused deposition modelling
Received: 11 April 2002 / Accepted: 24 February 2003 / Published online: 25 October 2003 Springer-Verlag London Limited 2003
Abstract Investment casting (IC) oers an economical method for mass producing complex, shaped metal parts. However, high tooling cost and lead times associated with the fabrication of metal moulds for producing IC wax (sacricial) patterns result in cost justication problems for customised single casting, small- and medium-quantity production. Rapid prototyping (RP) techniques can reduce the costs associated with single-part or small-quantity production as they can be applied to the fabrication of sacricial IC patterns containing complex and intricate designs with signicant cost and lead-time savings. In this project, a benchmark model is designed to assess the fused deposition modelling (FDM) process for creating sacricial IC patterns. In addition, an indirect approach toward producing wax patterns via silicone rubber moulding is investigated. Cost and lead time comparisons between the two IC pattern production methods were carried out and presented. The dimensional accuracies of metal castings generated from the RP-produced patterns are also presented. Keywords Investment casting Rapid prototyping Fused deposition modelling Silicone rubber moulding
which can be dicult or impossible to create using other fabrication methods [1]. Despite the wide range of applications in many industries, the standard (conventional) IC process practiced in modern foundries has its drawbacks. High tooling costs and lengthy lead times are associated with the fabrication of metal moulds required for producing the sacricial wax patterns used in IC [2]. The high tooling costs involved in conventional IC result in cost justication problems when small numbers of castings are required. Rapid prototyping (RP) techniques are fast becoming standard tools in the product design and manufacturing industry. With the capability of rapidly fabricating 3-D physical objects, RP has become an indispensable tool employed for shortening new product design and development time cycles [3, 4, 5]. RP techniques are limited neither by the geometry nor by the complexity of the parts to be fabricated. In addition, RP techniques involve no tooling or xtures, resulting in simpler set-up, lower overhead cost and shorter production lead times compared to other fabrication methods. With RP, parts that were previously impossible or extremely costly and time-consuming to fabricate can be built with ease.
1 Introduction
1.1 Background Investment casting (IC) is a key technique among a range of modern metal casting techniques that is capable of providing an economical means of mass producing shaped metal parts containing complex features (e.g. thin walls, undercut contours and inaccessible spaces)
C. W. Lee C. K. Chua (&) C. M. Cheah L. H. Tan C. Feng School of Mechanical & Production Engineering, Nanyang Technological University, 50 Nanyang Avenue, 639798 Singapore, Singapore E-mail: [email protected]
1.2 Rapid investment casting processes The application of RP-fabricated patterns as substitutes for the traditional wax patterns employed in IC stems from the fact that RP materials can be melted and burned out from the ceramic shell (ceramic shell casting) without damaging it [6]. Most commercialised RP techniques are capable of producing such patterns that can be used directly in IC. Several foundries have utilised various RP techniques to produce IC patterns. For instance, Shellcast Foundries Inc. (Montreal, Quebec) researched and developed a process called solid model casting (SMC) that allows the direct conversion of RP models to functional prototype castings without the need for tooling [7]. Cercast Group (Quebec, Canada) has identied the most important variables
94
in designing RP models for IC applications and the advantages and disadvantages of RP models produced by dierent technologies [8, 9]. Nuclear Metals Inc. (Concord, MA) has done a comparison of RP techniques for casting the Beralcast family of alloys. [10]. Most of the limitations encountered with earlier applications of RP-fabricated IC patterns include damage to the ceramic shells due to excessive thermal expansion of the pattern or the release of corrosive degradation by-products during pattern burn out, which cracks or attacks the cavity surface of the shell. Ceramic shell cracking is attributed to a mismatch in the coecient of thermal expansion (CTE) between the RP and ceramic materials. Most RP materials have CTE values that are larger than the ceramic material and as such, the expansion of the pattern during heating imposes significant amounts of stress on the ceramic shell. Shell cracking occurs when the stresses imposed by the expanding pattern are greater than the modulus of rupture (MOR) of the shell material [11]. In some cases, residual ash can also cause defects in the nal castings when present in relatively signicant quantities. Also, due to the high operating and material costs of RP techniques, the utilisation of RP techniques has been reported to be benecial only when ve or fewer castings are required [10]. The application of RP techniques to produce sacricial IC patterns can be classied under two methods: the direct RP method, and the indirect RP method. The direct RP method covers the application of RP techniques for producing plastic, wax or paper IC patterns for fabricating ceramic moulds that can be employed directly for metal casting (i.e. direct shell production method) [2]. For the indirect RP method, alternative moulds produced via silicone rubber moulding in conjunction with an RP-fabricated master pattern of the nal desired casting are employed for the injection of IC patterns from foundry wax. In addition to silicone rubber moulds, moulds fabricated with RP techniques can be used to produce IC wax patterns. For the direct RP method, the RP-fabricated pattern can only be used to produce a single casting. On the other hand, for the indirect RP method, the silicone rubber mould or RP mould can generate multiple wax patterns to produce a number of castings. However, due to the cost of RP materials as mentioned, silicone rubber moulds are usually the preferred choice for indirect RP methods. The process chain for direct and indirect RP methods is illustrated in Fig. 1. Using either direct or indirect RP methods, hard tooling production can be avoided. This results in tremendous time and cost savings, especially for projects involving a low number of castings. In addition, design iterations can be accommodated without incurring the high cost of modifying or re-producing the hard tooling. The focus of the work presented in this paper is on the production of sacricial IC patterns using the fused deposition modelling (FDM) technique. To the authors knowledge, no attempts have been made so far to
benchmark the FDM process for the production of IC patterns and to assess the performance of FDM-fabricated IC patterns. In this project, a commercialised FDM system, FDM3000 (Stratasys Inc., Eden Prairie, MN) is evaluated for fabricating IC patterns following the direct RP method (Figure 1). A benchmark test part is designed to assess the performance of FDM-fabricated acrylonitrile butadiene styrene (ABS) patterns in the IC process. In addition, the indirect RP method for producing injected wax patterns via a silicone rubber mould is investigated and compared to the direct RP method.
2 Methodology
The implementation of a benchmark test part is an essential practice in most evaluation studies conducted on any manufacturing system or process, especially when high capital investments are involved. Such scenarios include the acquisition of expensive CAD/CAE/ CAM systems, CNC machine tools and rapid prototyping systems [12] or major revamping of an existing process line and implementing a new process chain. In general, system or process evaluation using a customised benchmark model that is tailored to test specic requirements will allow the strengths and weaknesses of the system or process to be assessed. For the evaluations conducted, a number of criteria are selected to assess the performance of the FDM process and the resulting ABS IC patterns. The list of criteria used in this benchmarking exercise is as follows:
95
1. Cost of RP fabrication 2. Time taken for part building 3. Accuracy of the fabricated part compared to the design model 4. Dimensional stability of RP part 5. Surface nishing of the RP part In order to facilitate smooth and accurate data acquisition, the benchmark model should be relatively simple in design but should contain the desired features. Such features must incur relatively low consumption of resources, low demand on material usage and facilitate the use of simple measuring instruments in generating analytical data. There are generally two primary types of measurements that can be taken from a benchmark model: main measurements and detailed measurements. The benchmark test part was designed using Pro/ ENGINEER software. The design of the test part (Fig. 2) consisted of a rectangular base with dimensions measuring 110 mm (L) 90 mm (W) 10 mm (H) to allow for the testing of linear accuracy, ve square bosses with 20 mm sides and 30 mm (H) for the testing
Fig. 2 Detailed drawing of benchmark test part
of squareness accuracy, four cylinders with varying diameters ranging from 3 to 20 mm for the testing of radial accuracy, three through holes with diameters between 3 and 20 mm to assess shrinkage rates, thin walls with thickness ranging from 0.52 mm to test for consistency of wall thickness, and ribs elevated at 30 and 60 to the base for testing angular accuracy. Table 1 summarises the purposes of each design feature on the test part as mentioned. The dimensions that were measured to assess the accuracy of the part are clearly indicated with appropriate denotations in Fig. 2.
3 Experimental procedures
3.1 IC pattern fabrication using FDM Upon nalising the design of the benchmark test part, a shrinkage factor was incorporated by scaling up the design by 1% to compensate for shrinkage encountered by the casting material (aluminium alloy, A356 series) during solidication. The shrinkage factor was determined largely from the experience of the local foundry
96 Table 1 Purposes of measurements General Feature Length, width Square area Cylinders and blind holes Thin walls and thickness of base Angular ribs Through holes Dimension d01, d02, d09, d11, d12, d13, d14, d17, d18, d20, d21, d23, d25, d26 d07d08, d10d19, d15d16, d22d24 d27, d28, d29, d30, d31, d32, d33 d03, d04, d05, d06, d37 d41, d42 d34, d35, d36, d38, d39 Purpose of measurement Linear accuracy Squareness accuracy Radial accuracy Thickness accuracy Angular accuracy Shrinkage
(Precision Products Singapore Pte Ltd). Aluminium alloy (A356 series) was selected for the current work as it is commonly employed in many IC applications. The test pattern was built on the FDM3000 RP system using P400 ABS modelling material (Stratasys Inc., Eden Prairie, MN). To facilitate the burning out of the ABS test pattern during the IC process, the pattern was built to be hollow using the sparse build style supplied with the FDM system software. With a hollow pattern, the material can expand inwards when heated, causing less pressure to be exerted on the ceramic shell and thereby reducing the risk of cracking. Figure 3 shows an example of a part built with the sparse build style. To test for repeatability, two benchmark test patterns, labelled BM02 and BM03 and with similar designs were built and subsequently sent to the foundry. A corner at the base of BM03 was chamfered to distinguish the castings produced from the two patterns. Figure 4 presents the designs of the two test parts. Figure 5 shows the FDM-fabricated ABS test patterns. 3.2 Fabrication of master pattern for silicone rubber moulding For the indirect RP approach, a master pattern that is similar to the benchmark test part illustrated in Fig. 2 was employed with shrinkage compensation factors for foundry wax (%1%) and aluminium alloy A356 series
(%1%) added. The master pattern was fabricated on the FDM3000 RP system using ABS material. Once completed, liquid silicone tooling material (MCP VTV750, HEK GmbH, Germany) was cast around the master pattern inside a mould box. When cured, the silicone block was parted into four pieces using a hand-held cutter to extract the master pattern. A gating system to allow the in-lling of the cavity within the silicone mould left by the master pattern was cut into the silicone block. Once ready, the silicone mould was re-assembled before liquid wax was injected into its cavity to produce a wax casting, which was subsequently employed as a sacricial IC pattern. The silicone rubber mould produced using the FDM-fabricated master pattern is presented in Fig. 6.
Fig. 3 Part with hollow interior built using the sparse build style
Fig. 4a,b Benchmark model a BM02 and b BM03 (with a chamfered corner)
97
Fig. 5a,b FDM-fabricated test patterns a BM02 and b BM03 (with a chamfered corner)
3.3 Investment casting (IC) process Two FDM-fabricated ABS patterns, BM02 and BM03, and one silicone rubber moulded wax pattern were sent to Precision Products Singapore Pte Ltd for IC. For the ABS patterns, the autoclave stage was by-passed and the models were burnt o inside a furnace set at a temperature of 1000C for one hour.
98
residual traces of ash or material from within the shell cavity, the shells were ushed with water and compressed air. The aluminium casting obtained from the ABS patterns is shown in Fig. 7. No major defects were observed on the aluminium castings and all the various features were well formed. To assess the quality of the castings, uorescent penetrant inspection was conducted. Casting defects detected on the aluminium casting produced from the BM02 pattern include the presence of gas holes (porosity), misrun and metal penetration. The gas holes formed were located inside the aluminium part and could not be observed visually. Gas holes are formed in the castings due to gas bubbles entrapped there during the solidication stage. The misrun defect occurred at the 0.5 mm thin wall (Fig. 8), causing a small portion of the wall to be missing. Misruns are usually caused by the inability of the molten material to completely ll up the cavities of the ceramic shell or by the presence of residual ash. A metal-penetration defect (Fig. 9) is observed as excess aluminium material formed at the base of the 30 angled rib. Metal penetration defects are due to improper formation of the ceramic shell or can be caused by damage to the cavity surface of the ceramic shell. For the casting produced from the BM03 pattern, the only casting defect observed is a misrun defect that was detected at one of the corners of the base (Fig. 10). This defect could have been caused by the presence of residual ash. It
should be noted that the defects encountered in the castings are common defects observed with the production of aluminium castings and are not only
99
restricted to cases where RP-fabricated patterns are applied in IC. The measurements of the various dimensions for the BM02 ABS pattern before and after the surface nishing process were recorded and the variations in the actual dimensions from their design values (with a compensation of 1% to oset the shrinkage of the aluminium alloy) were calculated. To allow changes in the dimensions to be monitored due to the surface nishing process, the averaged measured values before and after the surface nishing process were also acquired and tabulated. From the variations obtained in the dimensions, the percentage errors (before and after surface nish) can be calculated. The average percentage error in the dimensions before and after surface nishing was calculated to be 1.422% and 1.533% respectively. Similarly, the measurements conducted on BM03 ABS pattern were recorded and the average percentage error in the dimensions before and after surface nishing was calculated to be 0.750% and 1.178% respectively. The low percentage errors (%11.5%) observed in both ABS patterns allow users to ascertain the capability of the FDM process of producing accurate patterns for IC. The additional surface nishing process conducted on the benchmark test patterns did not aect the dimensional accuracy greatly as the percentage errors of the nished parts did not dier much (<0.5%). The measurements of the various dimensions of the nal aluminium BM02 and BM03 castings were recorded and the average percentage errors calculated for the BM02 and BM03 castings were about 1.5% and 1.2% respectively. These percentage errors are relatively low and can be corrected using minor compensating factors or additional machining process.
Both lead-time (Table 3) and cost -comparison (Table 4) studies were conducted to compare the lead-time and production cost taken for pattern production using the direct RP method (via FDM fabrication) and conventional wax injection moulding via hard tooling. In these studies, the fabrication of the benchmark test part (Fig. 2) was used. Comparing the lead-times required the fabrication of a sacricial pattern and the direct RP method allows a signicant amount of time-savings (89%) to be achieved. As for the production costs, the direct RP method allows 60% cost saving to be achieved over the conventional hard tooling method for pattern production. 4.2 Indirect approach of pattern fabrication for IC The surface roughness of the ABS master pattern (before and after surface nishing), the wax pattern and the nal metal casting were measured and recorded in Table 5. It must be noted that the values obtained were average values calculated from a number of readings (10 readings for each set). Figure 11a shows the wax pattern produced using the silicone rubber mould while Fig. 11b presents the resulting aluminium casting produced from the wax pattern. Visual inspection of the aluminium casting detected no major defect with all the various features properly formed. However, the presence of gas holes (porosity) was detected using uorescent penetrant inspection. Measurements of the master pattern before and after surface nishing were taken and recorded. Variations in the measured dimensions to the design values (with a compensation of 1% to oset the shrinkage of wax material and a further 1% to oset
CAD design of pattern 5 days CAD design of mould and inserts 3 weeks FDM fabrication of pattern 19 h Mould production + Fabrication 6 weeks Post processing of FDM 18 h of 1 wax pattern pattern (i.e., removal of support structures + surface nishing) Total time %1 week 9 weeks
FDM ABS pattern fabrication FDM-fabricated pattern (1) (Quoted by a local service bureau) S$800 %S$800
Conventional pattern fabrication via metal mould Aluminium mould + inserts (Quoted by a local foundry) Fabrication of 1 wax pattern S$2000 %S$2000
Total cost
100 Table 5 Surface roughness values of master pattern and metal casting Surface roughness (I 1/4 m Ra) ABS master before surface nish ABS master after surface nish Wax pattern Aluminium casting 19.93 0.38 0.32 5.79
the shrinkage of aluminium alloy) were calculated and tabulated. The average percentage error in the dimensions of the master pattern before and after surface nishing was calculated to be 1.701% and 2.187% respectively. The measurements of the wax pattern produced using the silicone rubber mould and the nal aluminium casting were recorded. The average dimensional error calculated for the wax pattern was about 2.064%, which is close to the value obtained for the master pattern after surface nish. For the aluminium casting, the average dimensional error was calculated to be 0.908%. Both the lead-time and production cost comparison studies for IC pattern production using the indirect RP method (via silicone rubber moulding) and conventional wax injection moulding via hard tooling were conducted and the results tabulated in Tables 6 and 7 respectively. Comparing the time and cost required to produce a wax pattern via silicone rubber moulding to employing a metal mould in the conventional IC process, signicant time (89%) and cost savings (50%) can be realised using the indirect RP method.
Table 6 Lead time comparison for IC pattern fabrication Indirect pattern fabrication via silicone rubber moulding CAD design of pattern FDM fabrication of master Post processing of FDM master (i.e., removal of support structures +surface nishing) Creation of silicone rubber mould + fabrication of 1 wax pattern %1 week
Fig. 11 a Wax pattern produced using silicone rubber mould. b Resulting aluminium casting
5 Conclusion
The results of this study showed substantial advantages when employing ABS models as direct IC patterns or as master patterns for producing silicone rubber moulds to
Conventional pattern fabrication via metal mould 5 days 24 h 18 h 12 h 9 weeks CAD design of mould and inserts Mould production + fabrication of 1 wax pattern 3 weeks 6 weeks
Total time
Table 7 Production cost comparison for IC pattern fabrication Indirect pattern fabrication via silicone rubber moulding Fabrication of FDM master Creation of silicone rubber mould (Quoted by a local service bureau) Fabrication of 1 wax pattern S$1000 %S$1000 Conventional pattern fabrication via metal mould Aluminium mould + inserts (Quoted by a local foundry) Fabrication of 1 wax pattern S$2000 %S$2000
Total cost
101
cast wax IC patterns. The advantages derived include signicant amounts of cost and time savings, relatively accurate nal castings (average dimensional error %1.5% for the direct RP method and 0.9% for the indirect RP method) with reasonable surface quality and the complete elimination of hard tooling required in conventional IC process. Thus, it is much more benecial for foundries to employ FDM-fabricated patterns in IC for single or small quantity production of castings (<5) and to employ the indirect RP approach of fabricating IC patterns via silicone rubber moulding for medium quantity (tens of castings) production. However, in the event of mass production, the conventional hard tooling method for producing IC wax patterns is still preferred as the initial cost of tooling can be oset gradually as the number of castings is increased. Nonetheless, RP can still be employed in the initial prototyping stage for design verication and production verication purposes.
Acknowledgements This project is a collaborative eort between the School of Mechanical and Production Engineering, Nanyang Technological University and Singapore Institute of Manufacturing Technology (SIMTech), with funding support from Singapores Agency for Science, Technology and Research.
References
1. Beeley PR, Smart RF (1995) Investment casting. The Institute of Materials
2. Sachs E, Cima M, Cornie J (1991) Three-dimensional printing: ceramic shells and cores for casting and other applications. In: Proceedings of 2nd international conference on rapid prototyping, Dayton, OH, 2326 June 1991, pp 3953 3. Wohlers T (2000) Wohlers report: rapid prototyping & tooling state of the industry annual worldwide progress report. Wohlers Associates, USA 4. Hilton PD, Jacobs PF (2000) Rapid tooling: technologies and industrial applications. Marcel Dekker, New York 5. Chua CK, Leong KF (1997) Rapid prototyping: principles and applications in manufacturing. Wiley, New York 6. Beaman JJ, Barlow JW, Bourell DL, Crawford RH, Marcus HL, McAlea KP (1997) Solid freeform fabrication: a new direction in manufacturing. Kluwer, Dordrecht 7. Greenbaum PY, Pearson R, Khan S (1993) Direct investment casting of rp parts: practical commercial experience. In: Proceedings of the 4th international conference on rapid prototyping, Dayton, OH, 17 June 1993, pp 4350 8. Sarkis BE, Kennerknecht S (1994) Rapid prototype casting (RPC): the fundamentals of producing functional metal parts from rapid prototype models using investment casting. In: Proceedings of the 5th international conference on rapid prototyping, Dayton, OH, 1215 June 1994, pp 291300 9. Sarkis BE (1994) rapid prototyping for non-ferrous investment casting. In: Proceedings of rapid prototyping and manufacturing 94 conference, Dearborn, MI, 2628 April 1994 10. Smith BJ, St. Jean P, Duquette ML (1996) A comparison of rapid prototype techniques for investment casting Be-Al. In: Proceedings of rapid prototyping and manufacturing 96 conference, Dearborn, MI, 2325 April 1996, pp 111 11. Yao WL, Leu MC (1999) Analysis of shell cracking in investment casting with laser stereolithography patterns. Rapid Prototyping J 5(1):1220 12. Chua CK, Chou SM, Low BC, Cheong A, Low KK, Neo SM (1997) Evaluation of rapid prototyping systems for Thomson multimedia mechanical development. TEC J, pp 1529