An Experimental Investigation On Dimensional Stability

Journal of Materials Processing Technology 182 (2007) 580–587

An experimental investigation on dimensional stability of injected wax patterns of gas turbine blades S.A.M. Rezavand, A.H. Behravesh ∗ Department of Mechanical Engineering, Tarbiat Modarres University, Jalal-e-Al-e-Ahmad ExpWay, Tehran, Iran Received 17 October 2005; received in revised form 20 September 2006; accepted 26 September 2006

Abstract An experimental study on dimensional stability of simplified waxed models of gas turbine blade is presented. Gas turbine blades, made of a super alloy, have narrow dimensional and geometrical tolerances. Blades are manufactured by investment casting process consisting of wax injection, ceramic coating, wax removal, metal casting, and finishing. The dimensional accuracy of wax injection step introduces a great influence on the final blade dimension and thus on finishing process. The focus of this experimental work was on the injection stage, investigating the effects of processing parameters and blade geometrical features on the shrinkage of critical dimensions. To reduce the complexity of the analysis and mold manufacture, two designed models were extracted from the blade geometry. A mold was manufactured with two cavities (for two models). Injection temperature and holding time were chosen as variable processing parameters. The results indicated that the effect of blade curvature and non-uniform thickness are noticeably different. The effect of the holding time was found to be more dominant than that of the injection temperature. © 2006 Elsevier B.V. All rights reserved. Keywords: Gas turbine blades; Injection molding; Wax pattern; Shrinkage

1. Introduction Gas turbines transform heat energy to mechanical one, having applications in industrial sectors such as pumping, filtration, refinement, power plant, and transportation. The critical components of a gas turbine are blades consisting of rotor blades and stationary vanes. Blades function under severe service conditions such as high temperature, high mechanical stress, high heat fatigue, and corrosive environment. Gas turbine blades have close dimensional and geometrical tolerance, and are made of super alloys and manufactured by investment casting process. This process is used for producing high quality, net-shape complex parts. It is especially used when, due to the part geometry, application of other processes such as forging and machining are not, economically or practically, feasible. The major steps in investment casting process are: injection molding of a wax pattern, ceramic coating, removing wax, drying, and metal casting, followed by grinding. Each step introduces a certain effect on the final part dimensions, with wax injection and metal casting having the major influences.

∗

Corresponding author. Tel.: +98 21 88011001x3392; fax: +98 21 88005040. E-mail address: [email protected] (A.H. Behravesh).

0924-0136/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.jmatprotec.2006.09.029

The materials, used for pattern injection, must exhibit characteristics such as low viscosity, adequate solid strength, low dash, low shrinkage, high stability, chemical resistance to ceramic molding materials, ability to join, and not hazardous to health. Waxes are materials of choice introducing above characteristics. The final dimensions of wax pattern, in the injection step, are affected by: (i) type of wax; (ii) geometry and (iii) process parameters. On the other hand, the mere knowledge of the values of linear (or volumetric) shrinkage of employed wax is not sufficient to predict the resultant final dimensions [1]. Geometry and processing parameters have also considerable effects on the final dimensions. Waxes behave similar to semi-crystalline thermoplastic polymers. Besides, they have distinguished characteristics such as: (i) a very low melting point (below 100 ◦ C); (ii) a low heat conductivity; (iii) sensitivity to a high heating rate. Wax injection process consists of the following steps: (i) melting wax solids in a oil-heated vessel, (ii) conveying the melt into the injection barrel, (iii) injecting the melt into the mold using a ram injection machine, (iv) cooling the wax, and finally (v) ejecting the wax pattern (usually followed by calibrating step). If calibration is required to achieve desired dimensions, either the cycle time is considerably increased or more fixtures are

S.A.M. Rezavand, A.H. Behravesh / Journal of Materials Processing Technology 182 (2007) 580–587

581

3. Problem statement and objectives

Fig. 1. Schematic of wax injection machine.

required which, in any case, increases the total cost investment. The schematic of a typical wax injection machine is illustrated in Fig. 1. The degree of the influence of the processing parameters on the final dimensions, can be affected by the complex blade geometry. Hence, it is of research interest to investigate interacting effects of geometrical features and processing parameters. The results may introduce guidance to a cost-effective design or eliminate of the calibrating fixture. 2. Background Some research studies on wax behavior, both experimental and numerical, have been reported. Sabau and Viswanathan [2] have introduced the first computer program to simulate the behavior of an industrial wax called “CeritaTM 29-51”. Bonilla et al. [3] proposed a methodology to predict wax pattern shrinkage in investment casting process, using a computeraided heat transfer simulation and experimentally derived factors for injection parameters. They used a water emulsified wax type A7-TC2/E. Horacek and Lubos [1] studied the influence of injection parameters on the dimensional stability of wax patterns produced by injection molding process. They found an interrelationship between various injection parameters and their dependency on some dimensional parameters. Geometry of the parts was similar to a cross shape. Yarlagadda and Hock [4] determined the accuracy of wax patterns produced by hard (polyurethane mold) and soft (RTV mold) tooling and optimized the injection parameters used in a low-pressure injection molding. The geometry of the parts was similar to H-shape. Previous researches have mainly focused on the simple shapes of interest geometries. The given results can be used, to a certain extent, for some applications. However, due to the noticeable effect of part geometry on final dimensions, the data extrapolation from a basic geometry must be cautiously applied for other shapes, especially when the application is of great accuracy and complexity, such as turbine blades.

As mentioned earlier, the focus of this research was to study the effect of wax injection processing parameters on the final dimensions and (dimensional) stability of the produced wax pattern in relation to the critical blade geometry. It must be mentioned that the geometry has a great influence on how to apply dimensional corrections (here shrinkage factor). It is technically and scientifically inappropriate to apply a single shrinkage factor to all part dimensions. Usually for thermoplastic polymers, the material manufacturers introduce a range of shrinkage factor. This is because the processing parameters such as holding pressure, holding time, and part thickness have significant influences on the shrinkage, which can be explained by pressure–volume–temperature (PVT) diagram. Besides, part constraints such as cores and inserts could have considerable effects on the final dimensions, at various cooling time. Also, non-uniform shrinkage could promote warpage, which is a wellknown major defect in a molding process. Turbine blades have complex geometries. Thus, applying the results from simple geometries to a complex shape may not be fruitful. On the other hand, due to this complexity, obtaining an inter-relationship between any processing parameter and a specific geometrical feature is obviously difficult, if not impossible. The geometry of a typical turbine blade is shown in Fig. 2. In a turbine blade, non-uniform thickness of airfoil, thick root, curvature, and twist are critical features. To decouple the effects of geometry, it is proposed to examine models consisting of geometrical features close to those of real application. In this research, three characteristics of turbine blades, (excluding twist) and their inter-relationship with the processing parameters and dimensional stability were experimentally investigated. The twist feature was not included due to difficulty of machining. Hence, two models were designed and a mold was manufactured to produce waxed patterns. One design models airfoil curvature and the other design models non-uniform thickness (both consisting of the thick root). Therefore, the degree of the influence of each feature on the final dimensions could be examined.

Fig. 2. Configuration of a turbine blade.

582

S.A.M. Rezavand, A.H. Behravesh / Journal of Materials Processing Technology 182 (2007) 580–587 Table 1 Experimental parameters used for molding Injection temperature (◦ C)

Holding time (min)

60 64 68 72

1 2 3 4

Table 2 Experimental constant parameters Pressure Mold temperature Room temperature Injection course Holder pressure

55 bar 10 ± 2 ◦ C 27 ± 3 ◦ C 10 cm 80 bar

Design TP1 covers the airfoil lunate (curvature) and design TP2 covers the thickness difference of the airfoil (from its minimum value in the leading edge to the maximum value at about two-third of the airfoil width, followed by a decrease in the thickness to the trailing edge). For simplicity, cubic root and uniform airfoil width were consisted in the designs. As for dimensions, the first stage of GE Frame5 was chosen to design the models.

4.2. Experimental equipment and procedure Fig. 3. Designed pattern TP1 with given dimensions.

4. Methods and materials 4.1. Design of patterns As mentioned earlier, the blade geometry was decoupled in two models called: “TP1” and “TP2” (Figs. 3 and 4).

A two-cavity injection mold was manufactured to produce both patterns. The mold was made of AL5050 with 88 HB hardness. Fig. 5 shows two halves of the molds. The gates, of the same design and dimension, were machined at the middle of the roots. In the present research, the influences of injection temperature and holding time were studied. Previous researches have shown that injection temperature and holding time have the most influences on the final dimensions [1,4]. Because waxes are injected in low pressure, injection pressure has no disputable effect on the dimensions, except in parts with ceramic core or narrow cross-sections [1]. The wax patterns were then produced at the injection conditions specified in Table 1. Other parameters were maintained constant in all experiments as specified in Table 2. The used wax type was Filled Wax B417 (DUSSEK C.) with properties given in Table 3. Three samples of each experimental point were taken and measured, to deliver confident results. After injection, the patterns were placed on the root for 24 h as shown in Fig. 6. The corresponding dimensions, as shown in Fig. 7, of the wax patterns were measured using CMM. Six sections of the patterns, at the airfoil and the root, were measured. Cloud points of CMM were then produced to simulate the desired sections and to extract the critical dimensions, as described below (Fig. 8): • Dim a: airfoil chord length (Fig. 9). Chord length is one of the most important dimensions in turbine blades, because in the grinding step, the blades are clamped at the leading and trailing edges of the airfoil. • Dim b: the lateral deviation of the second and third sections with respect to the first section (Fig. 10). This deviation appears rather in all turbine blades. • Dim c: root middle thickness (Fig. 11) • Dim d: airfoil length. Table 3 Characteristics of wax used in the experiments

Fig. 4. Designed pattern TP2 with given dimensions.

Name

Filled wax B417

Viscosity at 80 ◦ C

1000 cPa

Producer Filler type Melting point Conglition point

REMET Polystyrene 75 ◦ C 61 ◦ C

Filler quantity Penetration at 25 ◦ C Dash quantity Color

38% 3 dmm Maximum 0.03% Green

S.A.M. Rezavand, A.H. Behravesh / Journal of Materials Processing Technology 182 (2007) 580–587

583

Fig. 5. Pattern injection mold: (a) top half and (b) lower half.

Fig. 6. Wax patterns were stored for 24 h as shown after injection. Fig. 8. Cloud points extracted from CMM data for: (a) TP1 and (b) TP2. The corresponding dimensions on the mold were also measured using CMM to calculate shrinkages.

5. Results and discussion The experimental results are shown in Figs. 12–19. Figs. 12–14 depict the chord length variation with holding time at various melt temperatures, for three airfoil sections, respectively. In overall, an increase in holding time causes a decrease in the shrinkage, which is an expected outcome. The effect of temperature is more evident for the second and third sections,

since they are more free to shrink with respect to the first section which is attached to the root (that is considered as a constrain). In general, an increase in melt temperature decreases the shrinkage, probably due to lower viscosity of the melt that promotes more uniform pressure distribution throughout the mold. Therefore, a less pressure drop is promoted and especially for the sections further from the injection point (for instance Sections 2 and 3) a higher pressure is felt, consequently, a lower shrinkage is resulted.

Fig. 7. Part sections measured by CMM.

584

S.A.M. Rezavand, A.H. Behravesh / Journal of Materials Processing Technology 182 (2007) 580–587

Fig. 9. Measured Dim “a” in two models.

Fig. 10. Measured Dim “b” in two models.

Fig. 11. Measured Dim “c” in two models.

One important result is that the amount of shrinkage for pattern TP1 is much higher than that of pattern TP2 (almost twice for the second and third sections). This can be due to the curvature feature of pattern TP1. For this design, it can be inferred that the resulted shrinkage consists of two types: thermal shrinkage and differential shrinkage. The first is caused by the temperature difference in the cooling period and the latter is caused by the shrinkage difference (of the adjacent points) called “warpage” [5]. This is due to the difference in the cooling rate of the two sides of the curved shapes (cooling rate at the concave side is lower than that of convex side) (Fig. 15). Figs. 16 and 17 illustrate the degree of the deviation of the sections two and three with respect to the first section at different holding times and melt temperatures. It is evident that an increase in holding time causes a decrease in deviation. The effect of temperature is rather noticeable, so that an increase in melt temperature causes a decrease in the deviation. The reasons were explained earlier as both parameters signify the pressure effect. However, the interesting result is that the direction of deviation for pattern TP2 is opposite to that of pattern TP1. The amount of deviation is larger and the sensitivity to melt temperature is greater so that the direction of the deviation changes at the highest temperature of 72 ◦ C. It is clearly seen that the deviation in pattern TP1 (consisting of curvature) is always existing and well behaved, while in pattern TP2, it is not stable, so that by a small increase in melt temperature, it can vanish.

Fig. 12. Variation of chord length at the first section for: (a) TP1 and (b) TP2.

S.A.M. Rezavand, A.H. Behravesh / Journal of Materials Processing Technology 182 (2007) 580–587

585

Fig. 13. Variation of chord length at the second section for: (a) TP1 and (b) TP2.

Fig. 14. Variation of chord length at the third section for: (a) TP1 and (b) TP2.

Fig. 15. Warpage promotion of TP1 pattern airfoil.

Fig. 18 shows the effect of holding time and melt temperature on the root width at the second (middle) section. It is clear that the amount of root shrinkage is relatively high, up to 4%, due to the root thick section. The effects of melt temperature and holding time are seen to be negligible, although the influence is rather unexpected. It is well known that, an increase in the holding time causes a decrease in shrinkage. The obtained results for the thick root show an opposite outcome, but the degree of change is not intense. Fig. 19 shows the variation of airfoil length with the injection parameters. It is well shown that an increase in holding time causes a decrease in the airfoil length while the effect of temperature is rather negligible. It is also noticed that the amount of shrinkage is different for both patterns. This could be explained by the hydraulic radius of both sections which is in relevant to the ease of flow and

Fig. 16. Variation of Dim b (deviation of the second section from the first section) for: (a) TP1 and (b) TP2.

586

S.A.M. Rezavand, A.H. Behravesh / Journal of Materials Processing Technology 182 (2007) 580–587

Fig. 17. Variation of Dim b (deviation of the third section from the first section) for: (a) TP1 and (b) TP2.

Fig. 18. Variation of shrinkage percent at the second (middle) section of the root for: (a) TP1 and (b) TP2.

Fig. 19. Shrinkage of the airfoil length for: (a) TP1 and (b) TP2.

thus the degree of pressure drop. According to the dimensions, the hydraulic radius of TP1 is 13.2 which is lower than that 16.1 of TP2. A lower hydraulic radius represents a higher flow restriction. Thus, it is expected that a higher pressure drop is yielded which consequently, could cause a higher shrinkage. However, it cannot be concluded that this amount of difference in hydraulic pressure could be the principle reason for difference in shrinkage. 6. Conclusion The experimental results indicated that: • Both melt temperature and holding time have great influences on the final dimensions of injected wax pattern.

• Each blade geometrical feature has particular influences on the final dimensions. • Chord length and airfoil deviation are most influenced by the curvature of the blade. These issues cannot be eliminated by only variations of the processing parameters. A calibration step is seemed to be necessary. • The airfoil deviation appeared always toward the convex side affected by the curvature. It is expected, the larger the curvature, the further the deviation. • The large thickness of the root produces significant sink marks. Thus, the molding process to produce the large thickness is unacceptable. This result verifies the importance of inserting a chiller (of the same wax material) into the mold before injection. Therefore, the final shrinkage and sink mark could be significantly reduced.

S.A.M. Rezavand, A.H. Behravesh / Journal of Materials Processing Technology 182 (2007) 580–587

Acknowledgements The authors would like to thank MavadKaran (Jahed Noavar) Engineering Company staff, especially Mr. Foroughi, for their useful assistance and accommodating technical equipments and Mr. Godsi (wax injection machine operator). Also, the authors would like to extend their gratitude to Shahab Engineering Company for providing measuring machines. References [1] M. Horacek, S. Lubos, Influence of injection parameters to the dimensional stability of wax patterns, in: Proceedings of the Ninth World Conference on Investment Casting, San Francisco, California, USA, 1996, pp. 1–20.

587

[2] A. Sabau, S. Viswanathan, Prediction of wax pattern dimensions in investment casting, AFS Transact. (2002) 733–746. [3] W. Bonilla, S.H. Masood, P. Iovenitti, An investigation of wax patterns for accuracy improvement in investment cast parts, Int. J. Adv. Manuf. Technol. 18 (2001) 348–356. [4] P.K.D.V. Yarlagadda, T.S. Hock, Statistical analysis on accuracy of wax patterns used in investment casting process, J. Mater. Process. Technol. 138 (2003) 75–81. [5] A.H. Behravesh, M. Moalemi, Investigation of warpage phenomenon on injection molding of plastic parts, in: Proceedings of the Fifth Iranian National Conference on Manufacturing Engineering, Tehran, 2002, pp. 673–683 (in Persian).