3D printing technique applied to rapid casting
The Authors
Elena Bassoli, Department of Mechanical and Civil Engineering, University of Modena and Reggio Emilia, Modena, Italy
Andrea Gatto, Department of Mechanical and Civil Engineering, University of Modena and Reggio Emilia, Modena, Italy
Luca Iuliano, Department of Production Systems and Business Economics, Polytechnique of Turin, Turin, Italy
Maria Grazia Violante, Department of Production Systems and Business Economics, Polytechnique of Turin, Turin, Italy
Abstract
Purpose – The purpose of this paper is to verify the feasibility and evaluate the dimensional accuracy of two rapid casting (RC) solutions based on 3D printing technology: investment casting starting from 3D-printed starch patterns and the ZCast process for the production of cavities for light-alloys castings.
Design/methodology/approach – Starting from the identification and design of a benchmark, technological prototypes were produced with the two RC processes. Measurements on a coordinate measuring machine allowed calculating the dimensional tolerances of the proposed technological chains. The predictive performances of computer aided engineering (CAE) software were verified when applied to the ZCast process modelling.
Findings – The research proved that both the investigated RC solutions are effective in obtaining cast technological prototypes in short times and with low costs, with dimensional tolerances that are completely consistent with metal casting processes.
Practical implications – The research assessed the feasibility and dimensional performances of two RC solutions, providing data that are extremely useful for the industrial application of the considered technologies.
Originality/value – The paper deals with experimental work on innovative techniques on which data are still lacking in literature. In particular, an original contribution to the determination of dimensional tolerances and the investigation on the predictive performances of commercial CAE software is provided.
Article Type:
Research paper
Keyword(s):
Rapid prototypes; Printers; Computer aided design.
Journal:
Rapid Prototyping Journal
Volume:
13
Number:
3
Year:
2007
pp:
148-155
Copyright ©
Emerald Group Publishing Limited
ISSN:
1355-2546
Introduction
The techniques based on layer-by-layer manufacturing are extending their fields of application, from the building of aesthetic and functional prototypes to the production of tools and moulds for technological prototypes or pre-series. In particular, additive construction applied to the production of dies and electrodes, directly from digital data, is defined as rapid tooling (RT). Patterns, cores and cavities for metal castings can be obtained through rapid casting (RC) techniques (Bernard et al., 2003; Rooks, 2002; Song et al., 2001). In both cases, since the tooling phase is highly onerous, great competitive advantages can be achieved thanks to solutions ensuring a short time-to-market. Moreover, RT and RC processes allow the simultaneous development and validation of the product and of the manufacturing process. Technological prototypes can constitute a strategic means, not only for functional and assembly tests or to obtain the customer's acceptance, but mainly to outline eventual critical points in the production process.
The relevance of RC techniques consists, above all, in a short time for parts availability. Traditionally, in order to produce cast prototypes a model and eventual cores have to be created, involving time and costs that hardly match the rules of the competitive market. For this reason, functional tests are typically performed on prototypes obtained by metal cutting (Rooks, 2002), which are not effective in outlining issues related to the manufacturing process. The possibility to verify the efficacy of a technological solution, in the early stages of the product development, ensures a Concurrent Engineering approach and minimizes the risk of late modifications of the definitive production tools (Ramos et al., 2003). The initial cost increase can thus be repaid through a reduction of costs and time for the following phases of development, engineering and production, as well as trough non-monetary advantages (Bernard et al., 2003). In particular, for relatively small and complex parts, the benefits of additive construction can be significant, thanks to its independence of geometry (Bak, 2003; Wang et al., 1999; Ramos et al., 2003).
In this field, innovative solutions are now available based on 3D printing process, which can extend RC possibilities thanks to the lower costs with respect to previous technologies such as selective laser sintering of sand (Gatto and Iuliano, 2001). One such technological solution consists in investment casting starting from starch patterns produced on 3D-printing conceptual modellers. A second solution is the ZCast™ process, in which 3D-printing technology with the use of a ceramic material allows the production of complex cavities and cores, suitable for casting light alloys.
A key issue regarding the investment casting process is the production of the expendable pattern in the case of a prototype casting, for which the traditional aluminium-alloy die is uneconomical. Rapid prototyping techniques can meet this requirement, producing single/few parts in short times and without tooling costs. The geometrical complexity achievable thanks to layer manufacturing matches the distinctive points of this casting process, suitable also for undercuts and hollow parts. Many solutions have been proposed, based on different technologies and materials. The present research regards starch patterns obtained by 3D printing on which, as in the conventional process, the ceramic shell can be built and then evacuated to obtain the cavity for pouring metal. Experimental studies regarding this solution are lacking in literature, in particular the technological feasibility in the case of thin-walled parts needs to be assessed and the dimensional tolerances calculated.
As for the second process, the 3D printer consolidates a plaster-ceramic powder by selective jetting of polymeric glue. Complex cores and cavities can be produced directly from the CAD model, complete with the gating system and air vents, avoiding the construction of patterns and core boxes (Bak, 2003). After removal from the unconsolidated powder, the parts are thermal treated and assembled before pouring the molten metal. The process is marketed as ZCast™ Direct Metal Casting by Z-Corporation. Owing to the current limitation in the maximum temperature (about 1,000°C as stated by the producer), the technology is only suitable for casting light alloys, whereas the first described solution does not entail any limitation in the metal to be cast. With respect to traditional sand casting, limited by the pattern extractability, layer-by-layer construction allows obtaining complex parts, without any restrictions in terms of undercuts, provided only that the unconsolidated powder can be removed from the cavity.
The great advantages, in terms of relatively low costs and very low times for the casting availability, contrast with a very poor knowledge concerning the limits of application and the process performances. Bak (2003) only reports the data supplied by the producer, stating that the accuracy and surface finish are consistent with sand casting. The present research aims at calculating not only accuracy but the tolerance class, which is the most significant value for the assessment of the dimensional performances ensured by the technological chain. Tolerance calculation is particularly important in cases where the mould is obtained through the assembling of different parts, due to the limits of the 3D printer working volume. To this regard, the optimization of the mould design plays a key role in the best exploitation of the geometrical freedom ensured by layer construction. Wang et al. (1999) and Ramos et al. (2003) propose similar studies with regard to different solutions for the production of technological prototypes. Song et al. (2001) agree on the importance of identifying the dimensional errors introduced in the different steps of RT and RC processes, offering a numerical solution. Moreover, since the process involves pouring the molten alloy into a cavity built with unconventional techniques and materials, a relevant issue is the evaluation of the predictive performances of computer aided engineering (CAE) software applied to this innovative technology. CAE software is widely used to model the cavity filling and the solidification in traditional casting processes, allowing the detection of possible critical points and defects (Lewis et al., 2001). The applicability of those tools to new technologies like the ZCast™ process still needs to be proven. In particular, among all the commercially available codes, the low-cost ones appear particularly interesting since they are consistent with the needs of an approach aiming at time and cost compressions.
Experimental plan
The research regarded the concurrent product-process development and production of a technological metal prototype by means of the two described RC processes, with the following objectives:
- verifying the application limits of the ZCast™ technique and the feasibility of the investment casting process with 3D-printed patterns in the case of thin walls;
- evaluating eventual critical factors in the two technological chains;
- evaluating the predictive performances of a commercial low-cost CAE software applied to the ZCast™ technique, improving the poor knowledge regarding filling and solidification within a cavity produced with innovative materials; and
- calculating the class of dimensional tolerance ensured by the two proposed technological solutions.
For the described purposes, an Aluminium alloy casing was chosen as a benchmark, representative of the automotive field, where the application of RT and RC technologies is particularly relevant (Rooks, 2002; Ramos et al., 2003). The experimental procedure started with the CAD modelling of the benchmark (Figure 1), having a total volume of 214,000 mm3 and box dimensions of 250 × 180 × 70 mm3. The casing, characterized by uniform thickness, no sharp edges and a variety of geometrical entities, is suitable to be produced by casting processes and to outline their dimensional performances. Referring to the investment casting process starting from starch patterns, the benchmark geometry is particularly relevant to determine the application limits for the technology. In effect, the process feasibility had already been verified by the authors for previous applications, which were characterized by a much higher volume/surface ratio than the present case. Since, the starch pattern is infiltrated with cyanoacrylate (as in the present case) or wax before use, with a penetration depth of few millimetres, in case of thin walls the complete evacuation of the permeating agent without residues has to be proven.
Subsequently, the two technological prototypes were obtained through the steps described as follows. The first RC solution required:
- generation of the STL file of the benchmark, to be sent to the RP machine;
- printing of the expendable starch pattern on a Z402 machine; and
- investment casting following the traditional process configuration, through shell building on the pattern, pattern evacuation at 900°C, casting and extraction of a first technological prototype of the casing in AISI 304 steel.
As for the second RC process, based on the ZCast™ technology, the following phases were planned:
- A design for manufacturing analysis was performed on the benchmark, with the aid of a commercial low-cost code for the simulation of the casting process (SOLIDCast®). The software had been preliminarily tested on a known component: an engine head for a sports car, with a well-established production know-how. The obtained results were consistent with industrial practice, even if a bit precautionary. The analysis on the casing lead to the definition of the feeding system and risers, in a concurrent product-process development (Lewis et al., 2001).
- The parts constituting the cavity were then CAD modelled starting from the benchmark, taking into account the limitations due to the machine working volume and defining the optimal assembly procedure.
- The parts were manufactured by 3D printing and treated following standard specifications (6 h isothermal at 200°C, heating ramp of 1.5°C/min).
- The inner surfaces of the cavity were air-blown and treated by foundry painting to improve the molten metal flow. Parts were assembled and an Aluminium alloy (GAlSi9MnMg-UNI3051, equivalent to ISO R 164 AlSi10Mg) was poured to obtain the technological prototype.
- A visual analysis of the defects in the cast model allowed the verifying of the predictive performances of the CAE software.
The two metal prototypes were then measured on a coordinate measuring machine (CMM), to calculate the overall dimensional tolerances given by the investigated RC techniques. The global accuracy depends on the errors introduced in all the subsequent steps: STL file generation, mould/pattern construction by 3D printing, thermal treatment of the mould, mould assembling for the ZCast™ technique, solidification shrinkage of the metal alloy in the cavity. For the first considered process, the starch expendable pattern was also measured before the shell construction, for a better quantification of the error sources.
The measurements have been made both on the internal and external surfaces of the three components: the starch pattern of the casing, the steel prototype produced by investment casting and the aluminium alloy part obtained by the ZCast™ technique. The procedure carried out in order to realize the dimensional control has been the following:
- definition of the control path to measure the internal and external surface of the pieces and control of the components with a CMM; and
- elaboration of the results to identify the tolerance class of the components.
Prototypes development and production
Investment casting with 3D-printed pattern
After exporting the STL model of the benchmark, the starch pattern was produced, a central riser with two runners was added and the whole part was covered with the ceramic shell; the pattern was then evacuated leaving the cavity in which steel was poured. The described procedure was carried out following the conventional know-how of investment casting operators. Figure 2 shows the broken shell after the casting removal and Figure 3 shows the cast technological prototype, after cutting of the feeding system.
The casting resulted in a good quality part, with only small porosities at the ribs base, but overall completely acceptable. No defects due to an incomplete pattern evacuation or residues were detected. Mean surface roughness R a was calculated from five measurements on a tracing length (L t) of 4.8 mm: a mean value of 4.0 μm was obtained, with a standard deviation of 0.67 μm. The values are quite good even if slightly higher than those typical of investment casting.
ZCast™ technique
With the aid of the software SOLIDCAST a design for manufacturing analysis was performed on the benchmark, taking into account different feeding systems and optimizing the cavity filling and the solidification process. The properties of UNI3051 were selected from the software database of casting alloys. The ZCast™ material was analysed though X-ray diffraction, revealing Calcium sulphate (CaSO4) and Forsterite (Mg2SiO4). Since, that innovative material could not be selected in the software database and specific data regarding its properties were lacking, traditional silica sand was chosen to model the mould. As to the heat transfer phenomena, the behaviour of the two materials was judged to be similar. Moreover, the analysis was targeted towards a comparison between different feeding configurations under the same materials conditions. For this reason, the simulation results were given only a qualitative significance that could not be affected by possible slight differences in the properties of the mould material.
The CAE software outputs were analysed in terms of preferred cooling direction, from the outer zones into the inlet runners, and absence of isolated zones solidifying without molten metal supply. In particular, the following parameters were considered:
- The Niyama factor, calculated with the temperature gradient divided by the square root of the cooling rate in a definite point. It is an indicator of the risk of shrinkage porosity.
- The plot of the critical fraction of solid (CFS), above which the metal is liquid enough to flow. It allows identifying areas that solidify without molten alloy supply.
- The FCC value, calculated from the local solidification time and wavefront speed. In moderately degassed Al alloy castings, it estimates the total per cent microporosity in a definite point.
Following the suggestions for Aluminium alloys, the CFS was set at 35 per cent, the fraction of solid for the Niyama calculation at 50 per cent and the default solidification shrinkage at 7 per cent. The analysis led to select a solution with a single riser and horizontal runners connected on the outer edge of the component. The corresponding software prevision in terms of CFS is shown in Figure 4.
Starting from the CAD model of the casing, complete with feeding runners and risers, the mould was modelled with a wall thickness between 12 and 25 mm, minimizing the ZCast material to limit production time and cost. Vents were provided to eliminate gas due to the contact between molten metal and the binder. The CAD model of the complete mould is shown in Figure 5(a). Since, the overall dimensions exceeded the 3D printer format (250 × 200 × 200 mm3), the shape was split into four parts, ensuring the appropriate definition of a parting line and coupling surfaces. One of the four parts produced by 3D printing is shown in Figure 5(b).
The parts were heat-treated and assembled, then the Al alloy was poured obtaining the technological prototype shown in Figure 6. The result proved that the cavity was correctly filled, confirming the outputs of the qualitative CAE analysis. Only small porosities could be detected at the base of ribs and a defect on the upper surface, in a zone to be removed with the subsequent machining operations. The shape parting line caused an edge on the outer surface, but the casting was judged satisfying. Mean surface roughness R a was calculated from five measurements on a tracing length (L t) of 4.8 mm: a mean value of 10.0 μm was obtained, with a standard deviation of 2.91 μm.
Measurements results
The measurement paths for the internal and the external surfaces of the benchmark have been generated through the measurement software of the DEA Iota 0101 CMM (Ainsworth et al., 2000; Wolovich et al., 2002).
These paths direct the movements of the CMM probe along trajectories normal to the parts surface. About 100 points have been measured on the internal and external surface (Figure 7). For each point the machine software evaluates the deviations between the measured positions and the theoretical ones for the X, Y, Z coordinates.
The results of the dimensional measurements have been used to evaluate the tolerance unit (n) that derives starting from the standard tolerance factor i, defined in standard UNI EN 20286-1 (1995). The values of standard tolerances corresponding to IT5-IT18 grades, for nominal sizes up to 500 mm, are evaluated considering the standard tolerance factor i (in micrometers) indicated by the following formula, where D is the geometric mean of the range of nominal sizes in millimetres. Equation 1 In fact, the standard tolerances are not evaluated separately for each nominal size, but for a range of nominal sizes. For a generic nominal dimension D jn , the number of the tolerance units n is evaluated as follows: Equation 2 where D jm is a measured dimension.
The tolerance is expressed as a multiple of i: for example, IT14 corresponds to 400i with n=400.
The evaluation model adopted introduces the maximum tolerance grade for 95 per cent of the observations as a quality index, because the distribution is not log-normal and the tolerance grade sets up the maximum error allowed for each dimension (Ippolito et al., 1995).
The results obtained by the measurements have been grouped for each X-, Y- and Z-axis. Then for each coordinate the mean, the standard deviation and the 95th percentile of n (value corresponding to the 95 per cent of the measurements) were calculated, the latter taken as a reference index for the evaluation of the tolerance grade. The results are shown in the Tables I-III and the Figures 8-10.
The internal tolerance grades are IT15 for the starch pattern, IT16 for the steel casing and IT 15 for the Al alloy casing, while the external tolerance grades are IT13 for the starch pattern, IT15 for the steel casing and IT15 for the Al alloy casing. For the Al alloy casing, the tolerance grades of both the internal and external surfaces are almost the same for each axis X, Y, Z.
The technological chain based on investment casting, from the starch pattern to the steel casing, and the ZCast process ensure similar dimensional performances. It is important to notice that the tolerance grades calculated for the considered RC techniques are consistent with the values allowed for casting operations, between IT11 and IT18 (Chirone and Tornincasa, 2004). Both the technological prototypes obtained are thus completely acceptable in terms of dimensional tolerances.
Conclusions
The feasibility of investment casting starting from 3D-printed starch patterns was proven even in the case of thin walls, excluding problems of residues after the pattern burning out. This solution does not imply any limitation in the alloy to be cast.
The ZCast technique provided satisfactory results, limited at present to the field of light alloys. With respect to traditional sand casting it ensures a much higher geometrical freedom and permits the overcoming of the traditional shape definition concept. The research proved the possibility of realizing parts with overall dimensions exceeding the 3D printer working volume, through a modular mould. The process limits can be identified in the surface finish of castings, which will be the objective of future developments of the research. CAE simulations helped in the qualitative evaluation of different feeding systems, showing good predictive performances. A quantitative application of such tools would require the collection of comprehensive data about the innovative mould materials.
A dimensional characterization has been performed on the obtained technological prototypes, through measurements on a CMM compared with the relative nominal positions. The results have been grouped for the different directions and for each coordinate the tolerance grade has been evaluated, considering as a reference index the 95th percentile of n. The overall dimensional tolerance ensured by both the proposed technological solutions belongs to classes IT15-16.
As a conclusion, both the proposed RC solutions proved to be effective for the production of cast technological prototypes, in very short times, avoiding any tooling phase and with dimensional tolerances that are completely consistent with metal casting processes.
Based on the results of this initial study, an interesting development of the research could be the assessment of the tolerance class of other parts produced with the two processes, aiming at the construction of a database for the accuracy and repeatability of RC solutions.
Equation 1
Equation 2
Figure 1CAD model of the automotive casing chosen as a benchmark
Figure 2Shell after casting removal
Figure 3Technological prototype obtained by investment casting with the starch pattern
Figure 4CAE analysis results: CFS
Figure 5(a) CAD model of the complete mould; (b) one of the four parts constituting the mould
Figure 6Technological prototype obtained by the ZCast technique
Figure 7Points measured on the external surface
Figure 8Tolerance grade of the internal and external parts of the starch pattern
Figure 9Tolerance grade of the internal and external parts of the steel casing
Figure 10Tolerance grade of the internal and external parts of the Al alloy casing
Table IElaboration of the dimensional measurements for the starch pattern
Table IIElaboration of the dimensional measurements for the steel casing
Table IIIElaboration of the dimensional measurements for the Al alloy casing
References
Ainsworth, I., Ristic, M., Brujic, D. (2000), "CAD-based measurement path planning for free-form shapes using contact probes", International Journal of Advanced Manufacturing Technology, Vol. 16 No.1, pp.23-31.
Bak, D. (2003), "Rapid prototyping or rapid production? 3D printing processes move industry towards the latter", Rapid Prototyping Journal, Vol. 23 No.4, pp.340-5.
Bernard, A., Delplace, J.C., Perry, N., Gabriel, S. (2003), "Integration of CAD and rapid manufacturing for sand casting optimisation", Rapid Prototyping Journal, Vol. 9 No.5, pp.327-33.
Chirone, E., Tornincasa, S. (2004), Disegno Tecnico Industriale, Il Capitello, Torino, Vol. Vol. 2.
Gatto, A., Iuliano, L. (2001), "Effect of the post curing process parameters choice on part produced by direct croning process", in Lonardo, P.M. (Eds),Proceedings of PRIME 2001 – 1st International Seminar on Progress in Innovative Manufacturing Engineering, Sestri Levante (Genoa), .
Ippolito, R., Iuliano, L., Gatto, A. (1995), "Benchmarking of rapid prototyping techniques in terms of dimensional accuracy and surface finish", Annals of the CIRP, STC E, Vol. 44 No.1, pp.157-60.
Lewis, R.W., Manzari, M.T., Gethin, D.T. (2001), "Thermal optimisation in the sand casting process", Engineering Computations, Vol. 3 No.4, pp.392-416.
Ramos, A.M., Relvas, C., Simões, J.A. (2003), "Vacuum casting with room temperature vulcanising rubber and aluminium moulds for rapid manufacturing of quality parts: a comparative study", Rapid Prototyping Journal, Vol. 9 No.2, pp.111-5.
Rooks, B. (2002), "Rapid tooling for casting prototypes", Rapid Prototyping Journal, Vol. 22 No.1, pp.40-5.
Song, Y., Yan, Y., Zhang, R., Lu, Q., Xu, D. (2001), "Three dimensional non-linear coupled thermo-mechanical FEM analysis of the dimensional accuracy for casting dies in rapid tooling", Finite Elements in Analysis and Design, Vol. 38 pp.79-91.
UNI EN 20286-1 (1995), "ISO system of limits and fits. Bases of tolerances, deviations and fits", .
Wang, W., Conley, J.G., Stoll, H.W. (1999), "Rapid tooling for sand casting using laminated object manufacturing process", Rapid Prototyping Journal, Vol. 5 No.3, pp.134-40.
Wolovich, W., Albakri, H., Yalcin, H. (2002), "The precise measurement of free-form surfaces", Journal of Manufacturing Science and Engineering, Vol. 124 No.2, pp.326-32.
Further Reading
Marutani, Y., Kamitani, T. (2004), "Manufacturing sacrificial patterns for casting by salt powder lamination", Rapid Prototyping Journal, Vol. 10 No.5, pp.281-7.
Corresponding author