Proceedings of ISET2015 International Conference on Design, Manufacturing and Mechatronics ICDMM2015 February 11-13, 2015, Pune, Maharashtra, India ICDMM2015-010 WARPAGE IN A STEPPED WAX PATTERN - A SIMULATION APPROACH Sridhar S Email: sridhardmrl@gmail.com Alok S Chauhan Email: aloksinghchauhan@gmail.com Satyanarayana A Email: satyanarayana.a@gmail.com Pradyumna R * Email: r.pradyumna62@gmail.com ABSTRACT In the process of investment casting of precision blade/vane components for aero-engine applications, generation of a precison wax pattern is the first critical step. A significant part of the dimensional deviations observed in the casting can be attributed to the wax pattern. During the process of generation of wax pattern through injection moulding, two types of dimensional deviations occur, namely shrinkage and warpage. Shrinkage deviations are predictable to a an extent as they are predominantly an outcome of the injection temperature, which is controllable. Warpage on the other hand is difficult to predict as it depends on many other factors of injection molding process. However, control of distortion is critical in order to conform to the strict dimensional tolerances of aerospace standards imposed on the turbine blade/cane castings. Simulation based studies are emerging as the preferred route for prediction of extent and location of warpage in precision wax patterns. In the present study, Moldex3D based simulation package has been utilized to anlyze the effect of injection temperature, packing pressure, cooling time and flow rate on the distortion behavior of a stepped pattern. Two types of waxes with different rheological properties have also been included to gauge the effect of change in wax on the warpage. A full factorial design of experiments has been configured and results of the analysis of variance is presented. INTRODUCTION Turbine blade and vane components are made by the process of investment casting using Ni based superalloy materials in order to withstand high operating temperatures and stresses [1]. Injection moulding of wax to generate a sacrificial pattern in a metallic mold is the first critical step in the process. The dimensional accuracies of the aerofoil geometry has to be achieved in the casting process itself, as machining of these components is not permitted, considering possibilities for catastrophic failure in operation due to machining related defects and possibility of recrystallization during machining. In the process of injection moulding, wax pattern is subjected to shrinkage and warpage (S&W effects) leading to change in dimensions. The die designer has to precisely compensate for these effects in order to provide dimensionally acceptable patterns for the casting process. Precise determination and control of pattern S & W effects helps in achieving consistent dimensions on the castings and provides much needed cushion to the complex vacuum investment casting process. Shrinkage is defined as the difference between linear dimensions of the die/mold and that of the molded component part which occurs due to the process of cooling of molten injected material in the mold. Warpage, on the other hand, is the distortion in the shape of the component caused by differential volumetric shrinkage in molded components. Also, residual stresses present in the part tend to relax after the part is removed from the mold, resulting in further warpage [2,3,4]. Page 31
Studies have shown that injection pressure shows no effect on the warpage, however high melt and mold temperatures with high packing pressure showed optimized distortionfor an injection molded wind turbine blade [5]. Cooling time seems to have maximum effect on warpage, with lower total warpage reported for higher cooling time, although the in-mold warpage is higher [6]. Measurement of shrinkage in a wax pattern is straightforward based on linear dimensions in the three coordinate axes. It is generally expressed as anisotropic shrinkage in percentage. Warpage on the other hand is difficult to define and measure, as it occurs at each and every location of a product. Researchers have used tip deflection, average warp parameter, etc. to define extent of warpage [5]. In the present study, complete surface of the pattern is considered for the definition of warpage as is relevant for aerofoil shaped turbine blade/vane patterns. SIMULATION BY MOLDEX3D SOFTWARE Moldex3D is a popular software to simulate the plastic injection moulding process. Recent version of the software has been updated to include wax injection behavior modeling also. The software uses the following approaches to predict filling, packing, cooling and warpage behavior of waxes [7]: A. Viscous Behaviour : Viscosity is a measure of the resistance of a material for flow and relates the shear stress to the shear strain rate of a polymer/wax. Moldex3D uses Modified Cross Model 3 also known as Cross-WLF model for predicting the viscosity behavior of wax. B. Visco-elasticity : Polymers behave like amorphous glass at low temperatures, but at higher temperatures the melt behaves like a viscous fluid. For intermediate temperatures, the polymer behaves like rubber exhibiting a combination of elastic and viscous behavior, a condition known as visco-elastic behavior. The software uses White-Metzner model, relating stress, stress rate and strain rate to simulate this phenomenon. C. PVT Characteristics : Polymers undergo significant volumetric changes as temperature and pressures are varied. Pressure-Volume- Temperature (PVT) relationship depicts changes in specific volume at various combinations of pressure and temperature. Moldex3D uses Modified Tait model for PVT relationships. D. Warpage : The software assumes that the materials are linearly elastic, strains are small and deformation of structure is quasi-static. Total deformation calculated is based on two distinct steps : first step considers mold constraint from end of packing to end of cooling and the second step is free deformation from the pattern removal to the time pattern temperature reaches that of the room temperature. Inj. Temp T1 T2 T3 Table 1 Properties of Waxes DMR-A and DMR-B Property DMR-A DMR-B Melting Point ( 0 C) 65-75 62-67 Congeling Point ( 0 C) 66 + 1 65 + 1 Penetration @ 25 0 C (dmm) 5 + 1 5 + 1 Relative Density (g/cc) 1.05 + 0.01 1.02 + 0.01 Filler Content ( % w/w) 30 + 2 35 + 2 Viscosity @ 80 0 C (CPs) 500 1000 In the present study, two of the regularly used waxes at DMRL, namely DMR-A and DMR-B have been included in the simulation. Table 1 shows properties of waxes provided by the manufacturer and used in the present study. It can be observed from the table that the two waxes have similar properties but DMR-B exhibits much higher viscosity at rated temperature than DMR-A and has lower melting point range. EXPERIMENTAL DESIGN A stepped pattern design of a wax pattern has been selected for the study. The CAD model of the pattern with its dimensions is shown in Fig. 1. The pattern doesn t have any features to constrain the wax during cooling, except for the flat mold surfaces, which are assumed to be smooth. A full factorial design-of-experiment (DOE) approach has been selected, using three parameters of injection temperature, packing pressure and cooling time, each at 3 levels and flow rate at 2 levels, totaling 54 simulations. Considering two Inj. Press Cooling Time Table 2 Simulation design for DOE P1 P1 P3 Flow Rate Q1 Q2 Q1 Q2 Q1 Q2 A B A B A B A B A B A B Page 32
Fig. 1. CAD model of stepped pattern Table 3 Factors and levels for DOE types of waxes, DMR-A and DMR-B, the total number of simulations becomes 108. Parameters Abbreviation Levels Inj. Temp. ( 0 C) T1, T2 & T3 65, 75 & 85 Packing Press. (MPa) P1, P2 & P3 2.0, 2.5 & 3.0 Cooling Time (s), & 60, 90 & 120 Type of Wax A & B DMR-A & DMR-B Flow Rate (cc/s) Q1 & Q2 30 & 60 Mold Temp. (0C) - 23 (const) Table 2 shows the full factorial simulation design and Table 3 shows list of various parameters and their levels used in the simulation. It may be noted that a high temperature of 85 0 C is also included in the simulation, a value much beyond the melting range of waxes. While dealing with wax injection of thin-walled, hollow turbine blade/vane components, to prevent premature solidification of wax, especially at thin crosssections, higher temperatures are used. Also, high temperature of wax allows lower injection pressures, thereby preventing breakage of fragile ceramic cores used for generating complex internal geometries. Simulation runs have been made on Moldex3D software for all values as shown above for the modules of filling, packing, cooling and warpage. The flow module has predicted basically viscous flow for all simulations for lower flow rate of 30 cc/s, but quasi-viscous (with jetting) for higher flow rate of 60 cc/s as shown in Fig. 2, which is for an injection temperature of 85 0 C. No cooling channels have been incorporated in the simulation as wax pattern tooling does not generally require forced cooling. However, running the cooling module takes care of the drop in temperature during the cooling time for a more accurate prediction of shrinkage and in-mold behaviour of the pattern. The warpage module calculates the dimensional deviations occurring in the pattern after its removal from the die and till the time the pattern reaches the room temperature. The warped model was exported as an.stl file and matched against the mold cavity model in the CAD system to analyze precise shrinkage values in X,Y and Z ( width, length and thickness) directions for each run. CAD model of the mold/die cavity was shrunk using the anisotropic X-Y-Z values for each run and matched against the warpage model. This process neutralizes the effect of shrinkage and highlights only the total warpage of the pattern. The warped model and shrunk CAD model were superiposed based on 3D best-fit method in Geomagic software and a point-cloud deviation profile was obtained along with their statistical profiles ( Refer Fig. 3). Considering the stringent dimensional deviation limits placed on wax patterns, a full surface envelop of + 0.025 mm was fixed as the level of acceptance and all deviation values above Fig. 3. 3D best-fit analysis of warped data against shrinkage compensated CAD Model METHODOLOGY OF SIMULATION AND ANALYSIS Fig. 2. Jetting of flow at 85 0 C and 60 cc/s this level were considered as warpage for the present study. The outcome of the study is evaluated as this response. Statistical analysis have been carried out to study the effects of various parameters based on main effects and interaction effects on warpage. ANOVA has been performed to calculate the percentage contributions of each factor and the statistical significance at a confidence level of 95 %, i.e. α = 0.05. Page 33
MAIN EFFECTS OF PARAMETERS ON WARPAGE In the main effects graphs, levels of a parameter are scaled to the X-axis and the total warpage (mean) observed in terms of percentage of measured points on the entire surface falling beyond the +0.025 mm envelop of the pattern, is shown as response in the Y-axis. Fig. 4. Main effects of parameters on response (a) Flow Rate and (b) Packing Pressure Variations in flow rate and packing pressure have no effect on the response of warpage for the pattern (Refer Fig. 4 (a) and 4(b) ). Flow of wax into the mold takes place within a fraction of a second as compared to the total cycle time. Its effect on warpage is thus insignificant. Predominant factor affecting warpage seems to the cooling time as shown in Fig. 5(a). At a low cooling time of 60 s, the distortion is maximum at nearly 10 %, which falls to < 1% for a cooling time of 120 s. Giving higher time for the pattern to stay in the mold leads to complete solidification of pattern before removal with minimum or no post-ejection distortion. Among the interactions of variables, wax type seems to interact with cooling time and injection temperature influencing the warpage (Refer Fig. 6(a) & 6(b)). Highest interaction mean effect is produced by wax DMR-B and a low cooling time of 30 s with a value of 14.35%. For interaction of wax with injection termperature, highest mean interaction effect is again produced by DMR-B for a termperature of 85 0 C with a value of 8.12. In both cases, the reponse drops sharply as the wax is changed to DMR-A, as already seen in the main effects. Cooling time interacts with injection temperature significantly affecting the mean warpage response, as shown in Fig. 7. Low cooling time of 30 s results in maximum warpage and as cooling time is increased, a drastic fall is seen. Fig. 6. Interaction effects of parameters on response (a) wax Type Vs Cooing Time and (b) Wax Type Vs Inj. Temp. Fig. 5. Main effects of parameters on response (a) Cooling Time (b) Injection Temperature and (c) wax Type As temperature increases, warpage also increases as shown in Fig. 5(b), indicating that higher temperature injection is not conducive for control of pattern warpage. Higher temperature injection infuses higher heat into the pattern leading to higher distortion when cooling time is insufficient. DMR-A wax seems to give lower warpage when compared to DMR-B (Ref. Fig. 5(c)). DMR-B being a higher viscosity wax develops higher level of thermally induced stresses resulting in higher warpage. INTERACTION EFFECTS OF PARAMETERS ON WARPAGE Fig. 7. Interaction effects of parameters on response Cooling Time Vs Inj. Temperature. ANALYSIS OF VARIANCE ANOVA has been performed for all the variables and their interactions at a confidence level of 95 % or α = 0.05 to crosscheck the graphical results and also to understand the level of contribution of parameters and their interactions on warpage. The results corrobarate the graphical analysis closely. Flow rate and packing pressure do not show significant effect on the warpage response. These parameters along with other interactions showing very low effect on response have been Page 34
Table 3 Selected results of ANOVA showing % contribution deep involvement and guidance given by Shri M A H Baig, Shri. Niranjan Das, Shri Satyapal Singh and all their colleagues for their unstinted support and cooperation extended for the work presented in the paper. REFERENCES pooled. Parameters and interactions having considerable effects have been retained and Table 3 shows some of the selected results. Cooling time displays maximum effect on the warpage of the stepped pattern at a percentage contribution level of 64.5. The factor having the next highest influence is the type of wax, contributing nearly 15.5 %. Interaction between wax type and cooling time has a contribution level of 13.0 %. Injection temperature, which is a major factor for controlling shrinkage of a pattern, has very little effect of just 3.2 %. Interaction between cooling time and injection temperature shows ~ 2% contribution, rest all the interactions are non-significant. CONCLUSIONS A simulation based analysis has been carried out to study the effects of various injection molding parameters and their interactions on the warpage of a stepped wax pattern, using Moldex3D software. A total of 108 simulations based on 3 levels each of injection temperature, packing pressure and cooling time, along with 2 levels of flow rate and type of waxes have been studied. Results indicate that the most significant parameter affecting warpage is the cooling time. The longer the pattern is retained in the mold, the lower is the warpage. Type of wax seems to affect the warpage significantly, with high viscosity wax giving higher warpages. A practical outcome of this result is that in a production environment, the practice of using different waxes for the same tooling based on availability, is to be avoided. Wax type interacts with cooling time to affect warpage significantly, indicating that conclusions should not be made based on observations of individual factors. [1] Campbell F. C., Manufacturing Technology for Aerospace Structural Materials, Published by Elsevier Ltd., 2006 [2] Handbook of Molded Part Shrinkage and Warpage, Jerry M Fischer, Published by Plastics Design Library/William Andrew, Inc., 2003 [3] Qualification Method for Powder Injection Molded Components, Donald F Heaney, P/M Science & Technology Briefs, Vol 6, No. 3, 2004, p 21-27 [4] Jafarian A R and Shakeri M, Investigating the Influence on Different Process Parameters on Shrinkage of Injection molded Parts, American Journal of Applied Sciences 2(3), 688-700, 2005, ISSN 1546-9239 [5] John T Tester and Ty Hargroder, Reducing distortion in simulated injection-molded wind turbine blades, Proceedings of 42 nd AIAA Aerospace Seminar meeting and Exhibiion, American Institute of Aeronautics and Astornautics, 5-8 Jan 2004, Reno, Nevada. [6] Yi-Hui Peng, et al, The warpage simulation with inmod constraint effect in injection molding, ANTEC 2004, Proceedings of 62 nd Annual Tech. Conf., Chicago, IL, May 16-20, pp 524-529 [7] Andreas Ostergreen, Prediction of residual stresses in injection molded parts, Master Thesis, Dept. of Applied Mechanics, Chalmers Univ. of Tech., Goteborg, Sweden, 2013 ACKNOWLEDGEMENT Authors would like to acknowledge the guidance and support given by Director, DMRL and are thankful for according permission to publish the paper. Thanks are also due to DRDO for funding various projects under which the study have been carried out. Authors would like to place on record Page 35