INVESTIGATION OF CENTRIFUGAL CASTING CONDITIONS INFLUENCE ON PART QUALITY

Transcription

1 INVESTIGATION OF CENTRIFUGAL CASTING CONDITIONS INFLUENCE ON PART QUALITY A. N. Vassiliou 1, D. I. Pantelis 1, G. -C. Vosniakos 2 1. Department of Marine Structures, Shipbuilding Technology Laboratory, School of Naval Architecture and Marine Engineering, National Technical University of Athens, Greece. 2. Department of Manufacturing Technology, School of Mechanical Engineering, National Technical University of Athens, Greece ABSTRACT Centrifugal casting is widely used as an investment casting method for small parts of complex geometry. Considering that the selection of casting conditions is not a trivial process, casting simulation programs could contribute to the enhancement of the process knowledge. In order to obtain realistic results from simulation, the input parameters should be correctly defined. Determination of the interfacial heat transfer coefficient is the most crucial issue. The contribution of the present work is to provide information for a more realistic selection of these parameters via simulation and to highlight their effect in the centrifugal casting process. KEYWORDS: Centrifugal casting, investment casting, simulation, heat transfer coefficient. 1. INTRODUCTION Simulation of the casting process is widely used in modern foundry industry, as a tool in product design and process development. It helps in decision making, where industrial expertise is not enough or critical parameters of the process are missing, keeping the costs of development low to be competitive /1/, /2/. However, the result of a simulation depends on the correct definition of the problem. The initial conditions, the boundary conditions, the material properties and the run parameters should be correctly defined in order to get realistic results. A very crucial parameter is the heat transfer coefficient at the metal mould interface. Several previous works had focused on the determination of the heat transfer coefficient for gravity casting or continuous casting of simple shapes /3/, /4/. As far as determination of the heat transfer coefficient is concerned, in literature, numerical simulation and experiments, as well as some analytical models are equally employed. /1/, /3/, /4/, /5/. In practice, h.t.c varies according to the thermophysical properties of the contacting materials, the casting and mould geometry, the roughness of the mould contacting surface, mould coatings, contact pressure, melt superheat and initial temperature of the mould /5/. In addition, the formation of an air gap between the solidifying alloy and the mould defines, in general, three different cooling rates /5/, /6/. Different heat flow rates across the cast metal and mold surface regions affect the evolution of solidification and the micro-structural properties of the casting. Thus, the second issue is validation of the simulation results by comparing them with experimentally measured properties of the casting. This is the method used in the present work. Proceedings of the 3 rd International Conference on Manufacturing Engineering (ICMEN), 1-3 October 2008, Chalkidiki, Greece Edited by Prof. K.-D. Bouzakis, Director of the Laboratory for Machine Tools and Manufacturing Engineering (ΕΕΔΜ), Aristoteles University of Thessaloniki and of the Fraunhofer Project Center Coatings in Manufacturing (PCCM), a joint initiative by Fraunhofer-Gesellschaft and Centre for Research and Technology Hellas, Published by: ΕΕΔΜ and PCCM 347

2 Concerning centrifugal casting, previous works were focused on the prediction of defects, such as shrinkage porosity and on the pressure distribution in the cavity and how this affects the quality of the part. In all cases both experiments and simulations were conducted, but the determination of the heat transfer coefficient is not clearly expressed /7/. The present work focuses on the investigation of the effects of some casting parameters on part quality. To this end, a series of experiments and a large number of simulations were conducted for the determination of the heat transfer coefficient of brass during centrifugal casting at different rotational speeds (469rpm, 391rpm and 156rpm). The calculated head transfer coefficients were used in some typical casting scenarios that were simulated to investigate the effect of some casting parameters. 2. EXPERIMENTAL PROCEDURE The experiments were based upon the lost wax casting method. A centrifugal casting machine was used. Melting was conducted inductively in a ceramic crucible, of a max capacity of 70gr. The material used was brass composed of 66.7%Cu and 33.3%Zn. The geometry of the casting was rather simple. Each part consisted of two concentric cylinders with diameter/height: 4/8 and 10/8 mm respectively, as shown in Figure 1. The casting tree arrangement is also shown in Figure 1. It consists of a central sprue (6 mm diameter, 30mm height), and two parts, forming an angle of 45 ο with the central sprue. The centrifugation radius was 175 mm, see Figure 1. A 175 mm Figure 1: The casting tree. The objective of the experiments was to measure the variation of the temperature of the melt with time, as close to the mould metal interface as possible. For that reason, thermocouples had to be embedded in the mould. This is, in general, common practice used in similar cases. However, in the case of centrifugal casting this technique has intricacies. The mould and consequently the embedded thermocouple that are rotating had to be connected to a stationary D/A converter. Thus, the centrifugal casting machine had to be properly modified by employing slip rings mounted coaxially with the machine s axis to which the thermocouple ends made springloaded contact. The relevant part of the centrifugal casting machine is shown in Figure rd ICMEN 2008

3 Figure 2: Part of the centrifugal machine. Calibrated K-type thermocouples were used. The thermocouples were inserted in location A, as shown in Figure CENTRIFUGAL CASTING SIMULATION Centrifugal casting experiments were conducted for three different rotational speeds (156, 291 and 469 rpm) that concluded to three cooling curves. Determination of the heat transfer coefficient has been achieved by a trial and error approach, assuming h. t. c values to obtain simulated cooling curves with the use of casting simulation software, and comparing the experimental cooling curves. ProCAST by ESI Group was used as the simulation tool performing coupled thermal - flow analysis. The input parameters for the simulation were the experimental conditions (Table 1). The material properties were taken from literature. Table 1: Input parameters for the simulation. Experiment #1 469 rpm Experiment #2 391 rpm Experiment #3 156 rpm Tmould Tmetal Flow rate kg/sec kg/sec kg/sec The flow rate depends on the rotational speed, the geometry of the crucible and the melt density: m o = Q * ρ where: o m : mass flow rate (kg/sec) Q =ν A : volumetric flow rate(m 3 /sec) ν : inlet velocity Α : inlet cross section area (m 2 ) ρ : melt density (7,7 *10 3 kg/ m 3 ) 2lα ν = R : velocity at inlet (as described in /4/ ) α R = ω 2 R : centrifugal acceleration ω : speed of rotation Casting, Thixo - Forming, Welding 349

4 Note that in the simulation, the behaviour of the interface must be modelled by introducing a single appropriate value for the heat transfer coefficient, either constant or as a function of time. 4. CALCULATION OF THE HEAT TRANSFER COEFFICIENT At first, several random values were assigned to the h. t. c. At the end of each simulation the simulated cooling curve at the location A (Figure 1) was plotted, and compared to the experimental one. A large number of h. t. c. values were tested, until satisfactory agreement was achieved. Thus, three time dependant heat transfer coefficients (one for each rotational speed) were determined, as depicted in Figures 4 to 6. The values of these h. t. c. are shown in Figure 3. h.t.c. (W/m 2 K) h.t.c. (W/m 2 K) h.t.c. (W/m 2 K) (c) Figure 3: h. t. c. as a function of time for 469 rpm, 391 rpm and (c) 156 rpm. In general, according to the literature, the time variation of h.t.c can be distinguished into three stages, /5/: stage I, where the metal remains in liquid phase and h.t.c (h1) decreases rapidly from a high initial value, stage II, which starts when temperature drops below the liquidus line and h.t.c (h2) has a constant value, and stage III, where the value of h.t.c (h3) either increases, due to an increase in contact pressure, or decreases due to the increase in the gap between the casting and the mould. Alternatively, h.t.c may be constant, if the pressure is constant. The same type of variation of the h.t.c. was observed in this work rd ICMEN 2008

5 Figure 4: Experimental and simulated cooling curves at 469 rpm sec 0-20 sec. Figure 5: Experimental and simulated cooling curves at 391 rpm sec 0-20 sec. Figure 6: Experimental and simulated cooling curves at 156 rpm sec 0-20 sec. 5. DETERMINATION OF THE EFFECT OF CASTING PARAMETERS Having determined the heat transfer coefficient for centrifugal casting at 156, 391 and 469 rpm, simulation runs of the centrifugal casting scenarios were conducted. The effect of casting parameters, such as the mould temperature, the liquid metal temperature and the rotation speed, Casting, Thixo - Forming, Welding 351

6 on the quality of the final product, was examined. Τhe solidification time, the fraction solid and the shrinkage porosity were the parameters linked to product quality. 5.1 Effect of the mold temperature For a melt temperature (Tmetal = 940 ) five different mold temperatures (450, 500, 550, 600, 650 ) were examined. In Figure 7 the minimum and the maximum solidification time of the cylindrical part as well as the maximum shrinkage porosity are shown. Solidification time is the time from the beginning until the end of solidification throughout the part volume. Solidification time for Tmold = 450 varies between 6.19 sec and sec, whereas for Tmold = 650, it varies between sec and sec, at 469 rpm. Therefore differences in microstructure of the parts are potentially expected. Actually, the maximum porosity that appeared in the part for Tmold = 450 was 0.294%, whereas for Tmold = 650 it was 0.124%. The porosity traced was macroporosity. Thus, higher mold temperature results in smaller value of porosity. The distance between the curve of maximum solidification time and the curve of minimum solidification time is a hint to uniformity of the solidification of the part. In Figure 7 it is noted that higher mold temperatures result in more homogenous solidification of the part. Tmold ( ) Tmold ( ) Figure 7: Impact of Tmold at 469 rpm on solidification time on shrinkage porosity. The procedure described above was repeated in order to examine the effect of the mould temperature on solidification time and shrinkage porosity at 391 rpm. The results are displayed in Figure 8. The solidification time for Tmold = 450 varies between 5sec and sec, whereas for Tmold=650 it varies between and sec. It is obvious that for lower mold temperatures solidification of the part is uneven. The shrinkage porosity is 0.025% when Tmold = 450, 500 and 550, but when Tmold = 650, shrinkage porosity is 0.012%. Tmold ( ) Tmold ( ) Figure 8: Impact of Tmold at 391 rpm on solidification time on shrinkage porosity rd ICMEN 2008

7 At 156 rpm, analogous results are displayed in Figure 9. It can be seen that in all cases, the distance between the two curves almost remains constant. When Tmold = 450, solidification time varies between 14.79sec and 28.15sec, whereas when Tmold=650 it varies between 42.47sec and sec. The shrinkage porosity reduces from 0,42% to half (0.25%). In both cases, the result is characterized as macroporosity. Tmold ( ) Tmold ( ) Figure 9: Impact of Tmold at 156 rpm on solidification time on shrinkage porosity. Finally, comparing the values of the maximum solidification time for different speeds of rotation, it can be seen that as mold temperature increases, solidification time increases as well, the gradient being the same, regardless of the rotational speed, see Figure 10. Solidification time (sec) Tmold Figure 10: Impact of Tmold on the maximum solidification time, for 156, 391 and 469 rpm. 5.2 Effect of the pouring temperature For a mould temperature of T mold =550 five different pouring temperatures (930, 940, 950, 960, 970 ) were examined. The results are displayed in Figures 11 to 13. Solidification time for 469 rpm and Tmetal = 930 varies between 5.57 sec and sec, whereas for Tmetal = 970 it varies between sec and sec, therefore differences in microstructure of the parts are potentially expected, see Figure 11. Actually, the maximum porosity that appeared in the part for Tmetal = 930 was 0.025%, whereas for Tmetal = 970 it was 0,0023%. For Tmetal =930 and 940 the porosity traced was macroporosity, while for Tmetal = 950, 960 and 970 it was microporosity. The significant reduction of the shrinkage porosity is clear. However, it should be mentioned that a high pouring temperature (above 955 ) causes surface oxidation, so it is probably not advisable. It is worth mentioning than in Fig- Casting, Thixo - Forming, Welding 353

8 ure 11, the distance between the minimum solidification time curve and the maximum solidification time curve almost remains constant. Tmold ( ) Tmold ( ) Figure 11: Impact of Tmetal at 469 rpm on solidification time on shrinkage porosity. Tmold ( ) Tmold ( ) Figure 12: Impact of Tmetal at 391 rpm on solidification time on shrinkage porosity At 391 rpm, maximum solidification time increases almost linearly with the pouring metal temperature, while the minimum solidification time remains constant, see Figure 12. At 156 rpm, the maximum solidification time is almost unaffected by the pouring metal temperature, see figure 13. Tmold ( ) Tmold ( ) Figure 13: Impact of Tmetal at 156 rpm on solidification time on shrinkage porosity rd ICMEN 2008

9 5.3 Effect of rotation speed Rotation speed affects both inlet flow (kg/min) of the melt and interfacial heat transfer coefficient. In addition, note, in Figure 3, that the value of the h.t.c. during the first stage (h1) is higher at higher rotational speeds. During the second stage, h.t.c. (h2) reduces at high speeds, whereas h.t.c (h3) during the third stage invariably keeps almost the same (very high) value. For T mold =550, T metal =940 at higher r.p.m., solidification of the part occurs faster, see Figure 14 presenting the fraction solid for 156rpm and 469rpm, at two different points in time (3 sec and 28 sec). Note that for the same point in time solidification at 469rpm is more advanced. 6. CONCLUSIONS The first goal of the present work was the determination of the heat transfer coefficient of brass during centrifugal casting. Combining an experimental procedure for centrifugal casting with the results from casting simulation, the heat transfer coefficient (as a function of time) of brass has been determined, for three speeds of rotation (469, 391 and 156 rpm). The three stages, followed by time variation of h. t. c. according to the literature, have been observed in this case as well. The value of the h. t. c. especially during the third stage is rather high, pointing at the high pressure in the cavity during centrifugal casting. (c) (d) Figure 14: Fraction solid for t=3 sec at 156 rpm, t=3 sec at 469 rpm, (c) t=28 sec at 156 rpm, (d) t=28 sec at 469 rpm. Casting, Thixo - Forming, Welding 355

10 The second goal was the investigation of the effect of process parameters on part quality, using the h. t. c. values determined. The effects of process parameters on solidification time and shrinkage porosity have been studied. Simulation for a number of combinations of input parameter values (Tmold, Tmetal, rpm) and results (solidification time, shrinkage porosity) were obtained and the following conclusions have been reached. The speed of rotation affects the h. t. c. and alters the way the metal solidifies. The shrinkage porosity decreases at higher speeds of rotation. As the mold temperature increases, maximum solidification time rises, whereas minimum solidification time remains low. Shrinkage porosity is rather high when solidification concludes fast, while, when solidification happens more gradually, shrinkage porosity falls. The only restriction in that case is the melt temperature at which brass oxidizes, resulting in a bad surface quality. Generally, the speed of rotation in centrifugal casting, the mold temperature and the initial temperature of the liquid metal influence the part quality. Their relative importance is being studied in the continuation of this work using Taguchi DoE method. 7. ACKNOWLEDGEMENTS This work was partially funded by ELKA S.A. Neilas Bros company is thankfully acknowledged for providing the centrifugal casting equipment. 8. REFERENCES 1. N Pagratis, N Karagiannis, G-C Vosniakos, D Pantelis, P Benardos, An holistic approach to the exploitation of simulation in solid investment casting, Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture, Volume 221, Number 6, pp , A. S. Sabau, Numerical simulation of the investment casting process, AFS Trancactions, Li Changyun, Wu Shiping, Guo Jingjie, Su Yanqing, Bi Weisheng, Fu Hengzhi, Model experiment of mold filling process in vertical centrifugal casting, Journal of Materials Processing Technology 176, pp , Menghuai Wu, Michael Augthun, JuÈ rgen SchaÈdlich-Stubenrauch, Peter R. Sahm, Hubertus Spiekerman, Numerical simulation of porosity free titanium dental castings, Eur J Oral Sci, vol. 107, pp , T.K. Vaidyanathan, A.. Schulman, J.P. Nielsen, S. Shalita, Correlation between macroscopic porosity location and liquid metal pressure in centrifugal casting technique, J Dent Res, vol 60, pp.59-66, January G. Chirita, D. Soares, F.S. Silva, Advantages of the centrifugal casting technique for the production of structural components with Al Si alloys, Materials and Design, vol. 29, 20 27, Ho K., Pehlke PD., Transient methods for determination of metal-mold interfacial heat transfer during solidification simulation and its use in the optimal feeding design of castings, Metal Trans B 29B: Ho K., Pehlke PD., Metal-mold interfacial heat transfer, Metal Trans B 1985;16B: K.A. Woodbury, Y. Chen, J.K. Parker, T.S. Piwonka, Measurement of interfacial heat transfer coefficients between aluminium castings and resin-bonded sand molds, AFS Trans, 106:705, Nishida Y, Droste W, Engler S., The air gap formation process at the casting-mold interface, and the heat transfer mechanism through the gap, Metal Trans B 1986;17B: rd ICMEN 2008