Vous êtes sur la page 1sur 6

A Scientific Approach to the Process Development Bonded Attachments for High-Speed Rotor Application

The quest for increased work per stage of compression to reduce overall gas turbine engine system cost has placed extreme demands on the high-pressure turbine (HPT) system. As an example, the HPT is required to operate at unprecedented levels of AN2 (the product of turbine annulus area and mechanical speed squared) to enable compressor performance goals to be met. The typical approach of mechanically attaching blades via rtree or dovetail congured mechanical attachments, limits rotor speed because of the life limiting broach slots (stress concentrators) in the disk rim. Exacerbating this problem is the fact that the disk lugs, which react the blade loading, impose a dead load. Higher disk speed results in higher blade loading requiring a deeper or wider lug to support the blade. This in turn results in a wider disk bore to support the deeper, dead load lug region. The dilemma is that higher speed results in larger stress concentrations at the rim and a wider disk bore to support the added parasitic rim load. The answer to this dilemma lies in creating an integrally bladed rotor (IBR) in which the blades are integral with the disk. Since typically, for an HPT, the blades are single crystal and the disk equiaxed nickel alloys, the IBR design suggested precludes absolute machining as the fabrication approach. A solution lies in metallurgically bonding the blades to the disk rim. Bonded airfoil attachments have the potential to increase AN2 and component life by 910 percent by eliminating broach induced stress concentrations as noted. Moreover, bonded attachments can reduce external rim loading by upward of 15 percent with a corresponding reduction in disk weight. The key to the solution is a controlled, economical process to concurrently join a full complement of HPT blades in a repeatable manner. This paper discusses how a scientic approach and creative design practice can lead to such a process. Three alternative tooling concepts, and one universal tool that allows independent use of two of these concepts, were developed. Tool stresses and deections, tool load paths, and bond pressure proles were all quantied through ANSYS nite element analyses and closed-form analytical solutions. Prior experience has shown that joint strength is sensitive to the bond pressure level. Therefore, the tool materials and geometry were iterated upon until the pressure applied to the blade bond plane was as uniform as possible. Since absolute uniformity is elusive when deformable bodies are part of the bond load train, accurately determining the maximum and minimum bond plane pressure is absolutely essential for subsequent joint characterization and design allowable determination. This allows localized working stresses in the designed attachment to be compared to specic, bond pressure driven, allowable strengths rather than an average strength. This paper will show how applying a scientic approach to the development of a critical technology process can reduce both the cost and risk of process development. DOI: 10.1115/1.1414131

R. R. Cairo1
Pratt and Whitney P.O. Box 109600, M/S 714-03, West Palm Beach, FL 33410-9600

K. A. Sargent
Air Force Research Laboratory AFRL/PRTC 1950 Fifth Street, Wright-Patterson AFB, OH 45433-7251

Introduction
The emphasis in the gas turbine engine industry today is reduced cost: in procurement and operation. Although bonded airfoil attachments in the HPT have the potential to enable higher levels of AN2, reduce weight, and improve rotor system life, the fabrication process may render their rst cost prohibitive. A fabrication process that metallurgically joins a full complement of single crystal blades to an equiaxed nickel disk in a concurrent
1 Currently at GE Power Systems, Gas Turbine Technology Center, 300 Garlington Road, Greenville, SC 29602. Contributed by the International Gas Turbine Institute IGTI of THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS for publication in the ASME JOURNAL OF ENGINEERING FOR GAS TURBINES AND POWER. Paper presented at the International Gas Turbine and Aeroengine Congress and Exhibition, Munich, Germany, May 8 11, 2000; Paper 00-GT-355. Manuscript received by IGTI November 1999; nal revision received by ASME Headquarters February 2000. Associate Editor: D. R. Ballal.

manner is needed. The feasibility of metallurgically joining dissimilar alloys has been demonstrated with coupon joining, Fig. 1. Not only must the process produce concurrent bonds, but it must also do so with minimal material deformation upset to minimize the amount, hence cost, of post bond machining to blend laterally deformed material or notched boundaries. These constraints on the bond process are necessary to keep the cost of fabricating an HPT IBR to a minimum. The conundrum then focuses on processes that lend themselves to effecting simultaneous and precise bonds at upwards of 50 60 locations blade sites . Precise is a key word here because the resulting complement of blades must be precisely aligned with respect to a prescribed incidence angle and present a smoothly contoured inner and outer radial owpath for optimal aerodynamic performance. Moreover, since the blades must be cooled in service, the bonding process must append them in the prescribed orientation to allow full ingress of air to cool the interior of the blade. These constraints eliminate 1 forge bonding Transactions of the ASME

190 Vol. 124, JANUARY 2002

Copyright 2002 by ASME

Downloaded 25 Jan 2012 to 182.72.197.106. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 1 Microstructure of metallurgical bond between single crystal and equiaxed nickel

because of the extent of material deformation; 2 diffusion bonding because of the duration of the bond cycle time cost inefcient ; and 3 transient liquid phase bonding due to insufcient property retention for highly stressed joints. This leaves the activated forge joining AFJ process 1 as the prime contender because of its combination of low material deformation, relatively short bond cycle time, and excellent property retention 2 . The strategy for applying the AFJ process for IBR fabrication relies on the differential thermal expansion between the bond tool xture and the blade and disk components being bonded. This is commonly referred to as the delta alpha tooling approach. Applying this strategy in an isothermal environment became the focus of this IBR development project and the thesis of this paper.

Fig. 3 Attempts to desensitize tooling with geometric compliance

Barriers to be Overcome
Although the delta alpha tooling approach has been used extensively to produce high-quality polymeric matrix composite components where the process is also referred to as elastomeric tooling , it has not been used extensively for high-temperature (2000 F(1093 C)) fabrication. The tooling is used to generate a bond pressure by the mutual interaction of deformable bodies in a contained environment. The deformable bodies in this case are the tool and the components being bonded. The preliminary tooling concept employed a full-annular containment ring the bull ring and radial blocks or shuttles, both of molybdenum.

These shuttles reacted against both the blade platforms and bull ring to generate pressure at the disk/blade root neck interface, as shown in the ANSYS model in Fig. 2. The large change in temperature in going from room temperature to bond temperature roughly 2000 F(1093 C) , coupled with the difference in expansion rates of the blade/disk assembly and bull ring, resulted in greater than yield strength stresses in the bull ring, which was unacceptable. Essentially, a small change in radial dimension produces a large change in bull ring hoop stress, which is one reason this approach has not been used extensively at temperatures in the 2000 F(1093 C) range. Attempts to desensitize the system by introducing compliance in the form of Belleville springs, leaf springs or features machined in the bull ring Figs. 3 a and 3 b proved futile. The dilemma for compliance is shown in Fig. 4 in which generated bond pressure is plotted against coefcient of thermal expansion with spring elastic modulus as parameter. As can be seen, dropping below a bond pressure of 65,000 psi 448 Mpa to a goal of less than 1000 psi 6.9 Mpa cannot be accomplished even with a leaf spring modulus of 0.5 Msi 3447 Mpa ! In addition to desensitization, another key barrier is control of the spatial positioning and orientation of the blades, especially with respect to their inner and outer owpath boundaries.

Candidate Solutions
In an effort to desensitize the bond tool system and control spatial positioning of the bonded blades, four alternative approaches were developed and evaluated: 1 blade shuttle; 2 platform hook and spring; 3 neck/pedestal fork; and 4 combined spring and wedge loaded neck/pedestal fork. Each approach utilizes the delta alpha concept and employs the massive, molybdenum bull ring to react the loads induced by thermal expansion. These approaches are listed in the order of which they were conceived, which also coincides with the order of increasing innovation. ANSYS nite element models were built for each approach to identify the uniformity of the bond pressure distribution at the blade/disk interface and to identify shortcomings and undesirable deformations/stresses. 1 Blade Shuttle Concept. The blade shuttle approach, shown in Fig. 5, utilizes a shuttle to position the blade and apply the reaction load that is imparted when the shuttle outer diameter OD contacts the bull ring and transfers load to the blade plat JANUARY 2002, Vol. 124 191

Fig. 2 Line-on-line contact results in unacceptably high stresses in delta-alpha tooling

Journal of Engineering for Gas Turbines and Power

Downloaded 25 Jan 2012 to 182.72.197.106. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 4 The sensitivity of bond pressure to material compliance

form. As can be seen, the shuttle reacts the load through four pedestals; two to the fore and two to the aft platform regions. Available platform surface area dictates load application points. This approach was discounted because a it is complex to fabricate; b it lacked blade-to-blade positioning control; and c it resulted in extremely nonuniform bond pressures that peaked at the fore and aft locations and were near zero at the center. 2 Platform Hook and Spring Concept. Figure 6 shows the platform hook approach, which addresses a few of the deciencies of the shuttle approach relative to positioning control and fabrication complexity. Unlike the blade shuttle, this concept utilizes an open structure consisting of fore and aft rods attached to load application blocks. The blocks also position two adjacent platforms to control the platform inner ow path uniformity. Also present are ceramic coil springs to desensitize the system. Coil springs allow more axial travel than leaf springs, hence enabling the desired degree of desensitization for the system. Ceramics were selected because of the high temperatures the entire load train will be subjected to during bonding. Once again, the outer bull ring provides all reaction forces and the springs are key to

maintaining ring hoop stresses below the yield point of the molybdenum. Although the platform hook and spring solves a few of the blade shuttle problems, it was discounted because a it still produced an extremely nonuniform bond pressures that peaked at the fore and aft locations and were near zero at the center and b the many critical engagement points for load application added risk to assembly operations. It did, however, highlight the fact that the platforms were simply too thin and exible for effective load transfer to the bond plane. 3 NeckPedestal Fork Concept. This innovative approach utilizes the best features of the platform hook and spring concept, eliminating the platform locating slots. It applies the load, very close to the bond plane, on a solid foundation instead of a cantilevered platform edge. The approach, shown in Fig. 7, utilizes fore and aft rods with ceramic springs to apply the load to loadapplication blocks. These, in turn, apply load to the fore and aft ends of a fork that reacts against a step that is machined near the root of the attachment. Notches in the load application blocks control adjacent platform alignment without imparting signicant load to them. Also visible in Fig. 7 is a vertical ridge in the left load application block that engages a slot in a base plate to guide the motion of the spring/blade/load application block assembly in the radial direction. This helps to control the blade alignment. Imparting the bond load to a step that is more rigid than a blade platform has the potential for more uniform pressure distribution

Fig. 5 Shuttle concept

Fig. 6 Platform hook and spring concept

192 Vol. 124, JANUARY 2002

Transactions of the ASME

Downloaded 25 Jan 2012 to 182.72.197.106. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 8 Fork bending

Fig. 7 Neckpedestal fork concept

in that region. This created high Hertzian contact pressures in the center of the bond plane Fig. 9 and very minimal contact pressure outside the middle third region. The next iteration involved the introduction of a at spot in the center of the cylindrically cambered region. This reduced the peak pressure as shown in Fig. 10. A map was generated, seen in Fig. 11, to determine the status of contact. The regions of contact, sliding, and no contact are labeled in the gure. Subsequent iterations to ne-tune the length of the at spot enabled the goal bond pressure to be achieved with an acceptable pressure distribution. Given the nature of deformable bodies, uniform bond pressure is unattainable, but tuning to an acceptable level by geometric manipulation is economically possible when a scientic approach is utilized. 4 Wedge Loaded NeckPedestal Fork Concept and Combined Spring and Wedge Loaded NeckPedestal Fork Concept. In principle, the spring loaded neck/pedestal fork concept will work but there is a risk using the brittle ceramic springs. As an alternative to pure delta alpha loading, an approach relying on external, hydraulic loading was developed. The tooling concept shown in Fig. 12, loads the bond plane with the same load application rod and fork arrangement described previously, but does so by wedge rather than spring action. The wedges are loaded axially by an external hydraulic ram. The wedge travels along a mating surface on the inner boundary of the bull ring generating a radially inward component of load to the load application block/fork arrangement. Spacer bumps, integral to the radial sides of the wedges, ensure parallel movement for consistent bond pressure

on the bond plane. However, the bending exibility of the fork caused the fore and aft regions of the step to receive higher load than the central third. This effect was examined using the ANSYS model shown in Fig. 8. Contact elements were used to model the interaction of the fork and blade root. As can be seen, there is actually a tendency for the central third zone of the bond plane to lift. This effect underscores the need to address tool design using the principles of mechanics for deformable bodies and using either closed form or nite element analyses. This tooling concept had enough merit to warrant continued renement to improve the bond stress distribution. The approach taken was geometric stiffening. The depth of the fork cross section was increased increasing section modulus and cylindrical camber was introduced in the central third of the fork, to force contact

Fig. 9 Bond pressure distribution due to fork stiffening and cylindrical curvature

Journal of Engineering for Gas Turbines and Power

JANUARY 2002, Vol. 124 193

Downloaded 25 Jan 2012 to 182.72.197.106. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 10 Bond pressure distribution realignment in center region of fork

due

to

geometric

Fig. 13 Combined spring and wedge loaded neckpedestal

Fig. 11 Contact zones on bond plane

proles at each bond site. Again, in principle, the wedge loaded neck/pedestal fork approach will work but there is a risk that the wedges will bind due to excessive friction. To mitigate the risk of committing to either approach too early in the component development effort, a combined tool design Fig. 13 was conceived that could load the bond plane with the previously described spring loaded approach or, should the springs prove problematic, be converted to the wedge loaded approach. The tool would have a wedge shaped insert piece with a right circular cylinder outer contour and a conical interior boundary to match the conical contour of the bull ring. For pure delta alpha tooling, the springs react against the conical surface of the bull ring insert. For hydraulic load augmentation, the springs are removed and radially deeper wedges are installed. A hydraulic ram presses on the axial face of the wedges which move along the conical surface of the bull ring insert imparting a radially inward load to the loading block fork assembly. Spacer bumps integral to

Fig. 14 Universal tool assembly allowing the spring or wedge approach

the radial sides of the wedges, ensure parallel movement for consistent bond pressure proles on at each bond site. The bond tool assembly for the spring and wedge loaded neck/pedestal fork approach is shown in Fig. 14. This also shows a closure ring to hold the wedges in place under thermal growth.

Summary
IBRs utilizing bonded attachments are an enabling technology to achieve the higher rotational speeds of advanced high-pressure turbine systems. Metallurgically bonding the blades to the disk rim can increase AN2 and component life by 910 percent. The key to the solution is a controlled, economical process to concurrently join a full complement of HPT blades in a repeatable manner. The process must be capable of the accurate spatial positioning of the blades with respect to their angle of attack and their inner and outer owpath boundaries. This study showed that applying nite element analyses and creative design practice to tool design is very benecial. It demonstrated that modifying the tooling geometry, based on nite element results, produced a bonding pressure that was within acceptable, dened limits across the entire bond area. Prior experience has shown that joint strength is sensitive to the bond pressure level. Since absolute uniformity is elusive when deformable bodies are part of the bond load train, accurately determining the Transactions of the ASME

Fig. 12 Wedge loaded neckpedestal

194 Vol. 124, JANUARY 2002

Downloaded 25 Jan 2012 to 182.72.197.106. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

maximum and minimum bond plane pressure is absolutely essential for subsequent joint characterization and design allowable determination. Taking this approach towards tool design allows these pressures to be determined, thus providing the actual allowable attachment strengths. A process that precisely controls blade position and imparts the required material response cannot be efciently developed by trial and error, and shouldnt be. The blade and disk components are too expensive, the variables are too vast, and the design requirements are too unique for an empirical approach. This paper has demonstrated how the use of the scientic approach to process development and tool design cannot only lead to such a process but is absolutely essential in todays scal environment.

Acknowledgment
The authors wish to thank Advanced Rotor System Team members David Cunningham, Chad Garner, Johnny Wang, and Beatriz Suarez-Crosse for their input of ideas, gures, and ANSYS analyses.

References
1 Cairo, R. R., 1999, Composite Ring Reinforced Turbine Program Final Report, Contract F33615-92-C-2201, Report Number AFRL-PR-WP-TR-19992050. 2 Cairo, R. R. and Sargent, K. A., 1998, Twin Web DiskA Step Beyond Convention, ASME Paper No. 88-GT-505.

Journal of Engineering for Gas Turbines and Power

JANUARY 2002, Vol. 124 195

Downloaded 25 Jan 2012 to 182.72.197.106. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Vous aimerez peut-être aussi