Computer simulation applied to jewellery casting: challenges, results and future possibilities

Size: px
Start display at page:

Download "Computer simulation applied to jewellery casting: challenges, results and future possibilities"

Transcription

1 IOP Conference Series: Materials Science and Engineering Computer simulation applied to jewellery casting: challenges, results and future possibilities To cite this article: Dario Tiberto and Ulrich E Klotz 2012 IOP Conf. Ser.: Mater. Sci. Eng View the article online for updates and enhancements. Related content - Numerical simulation of centrifugal casting of pipes E Kaschnitz - Modelling of heat transfer and solidification processes in horizontal twin-roll casting of magnesium AZ31 A Miehe and U Gross - The improvement of aluminium casting process control by application of the new CRIMSON process X Dai, M Jolly and B Zeng This content was downloaded from IP address on 12/09/2018 at 19:37

2 Computer simulation applied to jewellery casting: challenges, results and future possibilities Dario Tiberto * and Ulrich E. Klotz FEM Research Institute for Precious Metals & Metal Chemistry, Katharinenstrasse 17, Schwäbisch Gmünd, Germany * Corresponding author. Metallurgy Department, FEM, Germany. Tel: ; tiberto@fem-online.de Abstract. Computer simulation has been successfully applied in the past to several industrial processes (such as lost foam and die casting) by larger foundries and direct automotive suppliers, while for the jewelry sector it is a procedure which is not widespread, and which has been tested mainly in the context of research projects. On the basis of a recently concluded EU project, the authors here present the simulation of investment casting, using two different softwares: one for the filling step (Flow-3D ), the other one for the solidification (PoligonSoft ). A work on material characterization was conducted to obtain the necessary physical parameters for the investment (used for the mold) and for the gold alloys (through thermal analysis). A series of 18k and 14k gold alloys were cast in standard set-ups to have a series of benchmark trials with embedded thermocouples for temperature measurement, in order to compare and validate the software output in terms of the cooling curves for definite test parts. Results obtained with the simulation included the reduction of micro-porosity through an optimization of the feeding channels for a controlled solidification of the metal: examples of the predicted porosity in the cast parts (with metallographic comparison) will be shown. Considerations on the feasibility of applying the casting simulation in the jewelry sector will be reached, underlining the importance of the software parametrization necessary to obtain reliable results, and the discrepancies found with the experimental comparison. In addition an overview on further possibilities of application for the CFD in jewellery casting, such as the modeling of the centrifugal and tilting processes, will be presented. 1 Introduction CFD simulation is being commonly and successfully used in several industrial processes (such as lost foam and die casting) by larger foundries and direct automotive suppliers (such as BMW Leichtmetallgießerei [9-10]) while for the jewellery sector it is a procedure which is not widespread, and which has been tested mainly in the context of research projects for a few years now. Publications on casting simulation for jewellery applications date in fact back to 2005, [2-4,8,11] and focused on the behaviour of the metal flow and of the solidification in the static casting process, in which the filling of the mould occurs by gravity / applied pressure, and dealt with benchmark casting trials monitored via thermocouples or sensors and the comparison with the simulation output; these first steps used as test parts relatively thick objects, such as a coin or a ball ring, and compared the filling and solidification time with the experimentally measured ones (using as test materials gold and silver alloys). The work conducted by Fischer-Bühner was initially focused on sterling silver [4], and was conducted with the MAGMASOFT software: it covered several aspect of the casting process, from the turbulence during filling to the solidification of the parts and the influence of the feed sprue positioning. In a second phase of his research [1], the attention was shifted on gold alloys and the use Published under licence by Ltd 1

3 of the PoligonSoft, again to investigate the solidification and the formation of porosity, providing a satisfactory experimental comparison with thermocouple measurement and metallography of the cast parts. While the work from J. Fischer-Bühner was mainly focused on the thermal aspects and cooling times of the parts, the work carried out by Actis-Grande [5-6] et al. analysed in detail the filling of the mould, starting with simple test parts, and then moving on to more complex parts such as filigree objects. The results of the software obtained with the software Flow-3D (in terms of filling time) were validated with a high frequency (1000 Hz) acquisition system, necessary because of the high speed of the metal (the total filling is in fact complete in less than seconds). The software proved reliable in predicting the lack of filling in the thin sections of the filigree when the mould temperature was too low. Additionally, a comparison between the three softwares Flow-3D, ProCAST, and MAGMASOFT was carried out analysing their behaviour on the same test tree [7]. With this work (and following the steps of the previous research) the authors had the objective of extending the horizons on the CFD possibilities in jewellery investment casting, testing the softwares with different alloys, in order to find out the eventual limitations and discrepancies with the experimental data, and to investigate further possibilities (different processes, like the centrifugal or tilting ones) 2 Simulations 2.1 Alloy and investment properties Three different measuring facilities have been applied for determination of the thermophysical properties of the selected gold alloys: for the the thermal expansion a high temperature dilatometer (Netzsch DIL 402 C), for the investigation of the melting range, the heat of solidification and the specific heat capacity a Differential Scanning Calorimeter (Netzsch DSC 404 C Pegasus), for the measurement of the thermal diffusivity a Laser Flash Apparatus (Netzsch LFA 427) By using the equation [λ(t) = ρ c p (T) a(t)] that links the density ρ, the specific heat c p, and the thermal diffusivity a, the temperature-dependent thermal conductivities for all selected alloys were calculated (see Figure 1). Because the heat transfer through the investment material during the casting process has a large influence on the cooling behaviour of the cast part, the thermal conductivity of the gypsum bonded investment Silk, selected as standard material, was determined. For this purpose the comparative method was applied: using this method the thermal conductivity of the material to be investigated is measured relatively to the well-known conductivity of a certified reference material. Heat conductivity [W/(m K)] Sample Form-3 1 Aufheizen heating Sample Form-3 1 Abkühlen cooling Sample Form-4 2 Aufheizen heating Sample 2 cooling Form-4 Abkühlen Temperature [ C] b) a) Figure 1 Thermal conductivities a) 4 selected 18kt gold alloys b) 2 samples of the investment material Silk 2

4 Figure 2 - Wax tree with 16 thermocouples Figure 3 - Different types of acquired cooling profiles 2.2 Experimental casting trials In order to investigate non-equilibrium cooling conditions as benchmarking for the simulation, casting trials with thermocouple measurements were realized by using of 4 different 18k gold alloys (yellow, red, Ni-white and Pd-white) and application of standard casting patterns. Thereby solidification time profiles for various locations could be acquired during each experiment. Furthermore comprehensive metallographic investigations were performed on the as-cast test pattern regarding the determination of porosity levels and appropriate locations. Figure 2 shows one of the typical wax trees with 16 integrated thermocouples at different positions, where a ball ring has been used as a standard pattern. 2.3 Filling and cooling simulations The simulations were carried out using two different softwares: Flow3D 9.4 and Poligon Flow3D was used to obtain a temperature profile of the casting at the end of the filling step, and then such profile was transferred to Poligon for the cooling step. Flow-3D uses the Finite Difference Model, a method which discretizes the spatial domain into small cells to form a volume mesh (or grid), and the software has its own internal multi-block meshing system, which allows to increase the detail on specified areas of the geometry (see Figure 4b), and it adopts then a control volume approach to solve the Navier-Stokes equations: for the examples here presented a grid with cells was adopted, with a cell size between and 0.009mm; Poligon uses instead the Finite Element Method and requires the grid to be fitted on the geometries (both cast and mould), then to be processed externally, in order to obtain a tetrahedral solid mesh of the model that can be used for the computational step (see Figure 4c): the model here adopted has nodes and elements, with sizes that range between 0.5 and 1mm. The benchmarking was initially performed on the tree used for the casting of the 18k simple alloys, and the model for such tree was assembled in the 3D-CAD Rhino 4.0 (see Figure 4a): for the casting experiments the same 4 selected 18k gold alloys were used which had been investigated concerning their thermophysical properties in detail before (yellow gold, red gold, white gold and Pd white gold). After the casting trials all the measured cooling curves were compared with the software output with reference to the particular position at the tree and / or the ball ring (Figure 5 shows an example). 2.4 Results obtained with CFD The ability of the CFD to predict the flowing of a fluid in a cavity was exploited to optimize the sprue system of a cast part, to reduce turbulence during filling and most of all to insure a proper temperature distribution and a directional solidification, in order to reduce the formation of shrinkage porosity. 3

5 The filling simulation was used to test different sprue solutions on the parts: an example can be seen in Figure 7a and Figure 7b, which show the evaluation of two different kind of feeding possibilities, and the first one (Figure 7a), though asymmetrical, seems to provide a more uniform temperature distribution in the part at the end of the filling. Subsequently the solidification simulation was applied to identify in advance which geometries could lead to potential formation of porosity, and which places were more likely to be critical, and used to optimize the sprue system consequently (see Figure 9); Figure 8 shows the matching between the software output and the metallography performed on the part. The heating up of the investment right after the casting was also calculated for two different mould temperatures (500 C - Figure 10a, and 600 C - Figure 10b), because a very hot mould will facilitate the formation of porosity due to the chemical reaction between the investment and the molten metal, but this effect cannot be taken into account by the software, and its influence must be therefore evaluated according to the operator s experience. This can be generalized to the entire approach that one should have with the simulation in order to apply it in a feasible and efficient way: as a tool able to point out certain problems (wrong feeding system, or mould temperatures for example) which need to be solved according to the operator s experience. In this sense, the CFD can be used as a method to optimize certain aspects of the process, without having the expectation of finding out the exact machine parameters to obtain a porosity free and perfectly filled casting. a) b) c) Figure 4 a) 3D Model of the tree / b) Flow3D meshing / c) Detail of the solid meshing of the model for the Poligon software a) b) Figure 5 - Comparison between simulation and experimental data a) test part at top of the tree / b) test part at the bottom of the tree 4

6 a) b) c) Figure 6 a) Casting tree with positions of temperature measurement / b) Filling simulation / c) Solidification profile a) b) Figure 7 - Example of filling simulation of a ring a) with feeding sprue on the side, and b) with symmetrical feeding system a) b) Figure 8 - Example of b) predicted porosity and a) matching metallography on the cast part. 5

7 Figure 9 - Example of predicted porosity for two different sprues. Red and yellow colour indicate highest porosity levels. a) b) Figure 10 - Example of investment heating with a flask temperature of a) 500 C and b) 600 C 2.5 Matching and refining of the results, and the importance of parametrization The various temperature curves which were acquired during the casting experiments allowed to compare 1) different position on the trees (at the top and at the bottom) and 2) different type of cooling profiles: this allowed to evaluate the response of the software and its limitations. Two different cooling profiles were observed (shown in Figure 3): a first one with a short plateau followed by a constant cooling rate, and a second one with a long plateau followed by fast cooling. The software offers a good matching both with the first type (as already shown in Figure 5) and also with the second one, but in this case a particular attention was needed to tune-up the material parameters and the internal fit parameters of the software to get a proper response from the calculation, and match the long plateau registered by the thermocouples (Figure 13a, shows the mismatch before the optimization, and Figure 13b shows the optimized output). The parameters that needed to be optimized are related to the heat transfer: in the Poligon software one can define the most significant regions by selecting element faces of the solid mesh (both for the mold and for the casting), and then assign one or more heat transfer coefficients to each area: by varying the values assigned to each region the cooling of the cast part can be controlled, and the output fine tuned. The basic regions defined in this case were: the surface of the cast part and the inner surface of the 6

8 mould (to take into account the heat exchange between the hot metal and the investment), the outer surface of the mould (to simulate the cooling in air or protective atmosphere such as Ar) and the top of the casting (where the riser is exposed, and a significant heat flow takes place); the different areas can be seen in Figure 12. It was then observed that the software (after the proper optimization) offers a good matching with the measured cooling curves on test parts positioned at the tip of the tree, while for the pieces which are placed near the bottom, close to the riser (where the heat flow towards the outside is more consistent) the temperature mismatch between reality and simulation is more pronounced, and a proper tuning could not be reached (see Figure 5b): the software tends in fact to overestimate the solidification time in this part of the casting, and the experimentally measured curves are steeper than the output obtained from the simulation. The adjustment and refinement of the results through the software parameterization is (most of the times) possible when a series of experiments with thermocouples and metallographical investigations have been run in advance, and can offer a basis for benchmarking. In normal cases (to approach an industrial everyday situation) the background work of benchmarking tests is not always available, and thinking of optimizing the software for every different alloy and investment (which has a great influence on the heat transfer and therefore on the cooling of the metal) is obviously not possible, both for time and costs reasons. To this it must be noted that the investment casting procedure is influenced by several factors: performance of the refractory material (which can vary from batch to batch), chemical reaction between the hot metal and the mould, relevant influence of the furnace during the burn-out of the flask (e.g. two flasks prepared in two different ovens can have a significantly different performance). All these aspects cannot be taken into account in the simulation, and therefore it is better to know how to employ it in a meaningful way: for example being able to evaluate the effect of the mould temperature (as in the example shown in Figure 10) and the possible influence on the investment reaction that could lead to gas porosity. Figure 11 - Inner surface of the mould (visible through a virtual cross-section) 7

9 Figure 12 Different regions: 1- top of the riser (red) 2-surface of the casting (purple) 3- external surface of the mould (blue) and 4- top of the mould a) b) Figure 13 Comparison between experiment and software output a) before and b) after the parameters tune-up) 2.6 Further applications currently in development Further applications, which are currently being tested and will be applied in a future research project, are the centrifugal and the tilting casting processes. The CFD can be applied to the centrifugal process to establish how the centrifugation influences the filling of the parts according to their positioning, or to the type of adopted main sprue (Figure 14a). The simulation can be also applied to the tilting process, and first tests have been focused on the rotation angle and the speed of rotation, and how these parameters condition the flowing of the metal out of the crucible (see Figure 14b and Figure 15). Different crucible geometries could also be tested to choose the best one for the pouring of the metal. For instance such tests could be used during the machine development to achieve a proper design of the tilt mechanism and the machine geometry. This case shows the potential of the CFD to predict the effect of the turbulences during the pouring of the melt in tilt casting. 8

10 a) b) Figure 14 - Example of a) centrifugal casting / b) tilt casting a) b) Figure 15 - Example of two different rotation speeds in a tilt casting machine, with corresponding turbulence of the liquid metal in the crucible 3 Summary and Outlook The work here presented demonstrates the applicability of the CFD simulation in the jewellery sector to a wide range of processes (static, centrifugal and tilt casting), exploiting the different possibilities offered by the softwares commercially available: mold filling, solidification and prediction of porosity, investment heat-up. The quality of the results depends of course on the work previously conducted on the material characterization (investment and alloy properties) and then on the software tune-up and parametrization: good results have been obtained in terms of porosity prediction and sprue optimization. Variations can be although found in the matching of the output when a wide range of conditions is taken into account (different alloys, investments, position on the casting tree, etc), and in the everyday industrial reality the simulation could be disadvantaged when compared to an experimental trial (in terms of quick problem solving); it remains a solution to be exploited in the design and planning phase, where its potential of virtual laboratory would assume more weight to test different solutions. Further research will be focused on deepening the possibilities of the CFD for the centrifugal and tilting processes (applied for example to precious metals as Pt), also as a valuable help to design and engineer the casting machines and the crucible geometries. Acknowledgements The European Commission is acknowledged for financial support of this study under the contract no INTOGOLD. The authors thank the partners of the INTOGOLD project for the good co- 9

11 operation and the provision of materials. Special thanks are to the members of the metallurgy department of FEM, and to Mr. Marco Actis Grande (Politecnico di Torino, Italy) and Mr. Jörg Fischer-Bühner (Legor, Italy) for their precious co-operation. References [1] Fischer-Bühner J 2007 Advances in the prevention of investment casting defects assisted by computer simulation The Santa Fe Symposium on Jewelry Manufacturing Technology [2] Actis-Grande M 2005 Computer simulation of the investment casting process: experimental validation TCN CAE Conference [3] Actis-Grande M 2006 Computer simulation of the investment casting process: experimental validation 4th International Jewelry Symposium [4] Fischer-Bühner J 2006 Computer simulation of jewelry investment casting The Santa Fe Symposium on Jewelry Manufacturing Technology [5] Actis-Grande M, Porta L and Tiberto D 2007 Computer simulation of the investment casting process: widening of the filling step The Santa Fe Symposium on Jewelry Manufacturing Technology 1-18 [6] Actis-Grande M and Wannarumon S 2009 Numerical simulation of Investment casting of Gold Jewelry: Experiments and validations World Academy of Science, Engineering and Technology [7] Actis-Grande M and Wannarumon S 2009 Comparison of Computer Fluid Dynamic software programs applied to jewelry investment casting process World Academy of Science, Engineering and Technology [8] Wright J 2005 Computer simulation and jewellery production The Santa Fe Symposium on Jewelry Manufacturing Technology [9] Lang H 2010 Use of modern flow simulation tools for process optimization Flow-3D User Meeting [10] Lang H 2011 Improvement of the simulation through validation of boundary conditions and inlet sizes Flow-3D User Meeting [11] Alonso A and Franco L 2005 Computer optimization of vertical gating system designs Foundry Management and Technology 133 (10)