Application of a Heat Conductor Technique in the Production of Single-Crystal Turbine Blades
DEXIN MA and A.BU HRIG-POLACZEK
The heat conductor(HC)technique was applied in the production of single-crystal turbine
blades with the aim of reducing grain defects originating from a geometrical feature.Becau of
its excellent thermal-physical properties,graphite is recommended as HC material.Numerical
simulation and temperature measurements were able to prove that the thermal condition in the
shroud region of turbine blades is significantly improved for single-crystal growth.The reduc-
tion in thermal undercooling entirely eliminated the high-angle boundary,resulting in an
effective improvement in the casting quality of components.
DOI:10.1007/s11663-009-9274-7
ÓThe Minerals,Metals&Materials Society and ASM International2009
I.INTRODUCTION
S INGLE-CRYSTAL solidification of turbine blades for advanced gas turbines is a key technology for the production of reliable and high-efficiency gas turbines. The excellent properties of single crystals originate from the abnce of grain boundaries,which eliminate sites liable to result in crack nucleation under thermome-chanical fatigue conditions.Under creep conditions,the abnce of grain boundaries not only removes the sites liable to host fracture initiation but also to remove a potential mechanism for high-temperature deformation, in particular grain boundary sliding.[1]Becau of ever-increasing demands on casting quality,the production of single-crystal turbine blades has been thoroughly investigated.Unfortunately,control of the industrial casting process is still not to a satisfactory level,so that a significant number of the components produced have to be rejected becau of the grain defects,which form a high-angle boundary with the primary crystal.[2]
Grain defects mostly occur as a conquence of the preferred solidification of geometrical features of the component.Experimental investigation has shown that the macroscopic curvature of the liquidus isotherm becomes markedly concave while traversing extreme enlargements in the cross ction of the , transition from the blade portion to the shroud portion.[3] This leads to the formation of an isolated,thermally undercooled region of melt,which may lead to heteroge-neous
nucleation and hence the formation of stray grains. As blade designs become more complex and the demand for larger single-crystal castings dramatically increas,the formation of grain defects caud by geometrical features becomes an increasingly rious challenge.This key problem cannot be effectively overcome by conventional process modification,such as optimization of withdrawal rate or baffle design. The authors have recently developed a heat conductor (HC)technique to improve the thermal condition at the transition point and thus to suppress stray grain forma-tion at the extremities of the platform.[4]The assumption was,as shown in Figure1(a),that a hot spot exists at the inner corner of the cross-ctional transition(position A),resulting from the poor local cooling condition.This hot spot hinders single-crystal growth from the blade portion into the platform.On contrast,the outer corner (position B)cools more rapidly becau the local shell mold is much thinner.As a result,becau of the concave curvature of the liquidus isotherm,the temperature at position B falls faster below the liquidus temperature, and the dendrite tips of the primary crystal are unable to reach this isolated undercooled region,where stray grains eventually occur.In the HC technique,as shown in Figure1(b),a heat conductor is inrted clo to the critical position A to improve the local cooling condi-tion.The other end of the HC protrudes from the mold to ensure effective heat radiation into the cold environ-ment in the Bridgman-furnace chamber.Efficient heat extraction minimizes the hot spot at position A.This promotes the successful transition of single-cryst
al growth from the blade body into the extremity B of the platform before local undercooling exceeds the critical value for heterogeneous nucleation.希望英语官网
white christmas歌词
The effectiveness of the HC technique has been confirmed in principle in previous projects by using a simple geometry with an abrupt ction enlargement.[5] The objective of the prent work is to apply the HC technique to the production of turbine blades and to confirm the effectiveness of the technique.
II.EXPERIMENTAL
A.Blade Geometry and Mold Preparation
The test component lected for the experiments was an actual turbine blade who shroud has both
DEXIN MA,Rearch Scientist,and A.BU HRIG-POLACZEK, Director,are with the Foundry Institute,RWTH University Aachen, 52072Aachen,Germany.Contact e-mail:d.ma@gi.rwth-aachen.de Manuscript submitted January20,2009.
Article published online August14,2009.
abrupt cross-ction enlargement from the aerofoil and an overhanging outer corner(Figure2(a)).This design was lected to provide a typical geometrical feature that is most prone to thermal undercooling and,hence,to subquent grain defects.In this ca,the application of the HC technique is especially desirable.
The ceramic shell molds were produced by a modified investment casting procedure,the principles of which are shown in Figures2(a)to(d).Afterfinishing thefirst ceramic layer on the wax pattern,as in standard investment casting,the graphite HC was bonded at the locations lected using a ceramic binder.By repeating the procedure of dipping in ceramic slurries and sanding with fud alumina,the shell molds were produced to thefinal thickness(Figure2(d)).In parallel,the exterior of the HC was cleaned with a toothbrush to maintain it free of ceramic accumulation.Figure2shows the principle of shell mold manufacture for a single blade.In fact,the blades were cast in clusters,in which the blades were arranged in a circle around a central rod (e also the mold cluster in Figure3).One half of the blades in each cluster were bonded with HC to inves-tigate the effectiveness of this new technique in compar-ison with the halves with no HC.
After dewaxing in a steam autoclave,the shell molds were subquently sintered at1200°C in a clod steel box in afiring furnace(Figure3).To protect the inrted graphite HC against oxidation,so
me pieces of waste graphite were placed in the box.This protective measure has been proven to ensure that no vacuum and inert gas atmosphere is necessary.
B.Selection of HC Material
Graphite is ud as HC material becau of its excellent thermal physical properties and very
good Fig.2—Procedure of shell mold manufacture for inrting HC:(a)wax model;(b)afterfirst layer;(c)attachment of HC;and(d)shell mold finished for dewaxing.
(a)(b)
shell mold
melt
智囊团英文Q A
heterogeneousundercooled
melt
crystal
A B
T L
Q B
heat conductor
B
A
Q A
Q B
Fig.1—(a)Schematic diagram shows the formation of the undercooled zone at the platform extremity B.(b)Schematic diagram of the HC technique for reducing undercooling at B by improving the cooling condition at region A.
resistance against high temperature in vacuum.Figure4 shows the relevant thermophysical properties of the ud graphite in comparison with tho of the superalloy and ceramic mold(Al2O3).The heat conductivity of graphite is about30times as high as that of ceramic (Figure4(a)).Thus,it can be estimated that the HC can extract correspondingly such more heat than the shell mold under the same condition.By lecting a suitable size of HC,it can extract enough of the heat from the critical position and effectively reduce the hot spot from hindering the single-crystal growth.The volumetric heat content q C p(q=density and C p= heat capacity)for graphite is lower than for ceramic material(Figure4(b)).This means,compared with ceramic mold,less heat is stored in the HC during the heating process,which has to be additionally extracted during the solidification pronaughty bear
cess.Graphite also has a high emission coefficient e(Figure4(c)),which ensures effective heat radiation into the cold chamber of the Bridgman-furnace.For the reasons,graphite becomes a very suitable material for the HC technique.
C.Casting with Temperature Measurement
To measure the development of the liquidus isotherm while traversing the critical shroud areas,thermocouples were placed at exactly defined positions.For this
purpo,the ceramic sheaths were inrted into the wax patterns before dipping and sanding(Figure5(a)). After dewaxing andfiring,the thermocouples of type ‘‘B’’(Pt-30pct Rh/Pt-6pct Rh)were installed into the ceramic sheaths in the shell mold(Figure5(b)).The casting experiments were performed as a standard Bridgman process.The alloy ud in this experiment is a Ni-bad,single-crystal superalloy CMSX-6(Cr9.8, Co5.0,Mo3.0,Al4.8,Ti4.7,and Ta2.0in wt pct)with a liquidus temperature of about1321°C.[6]A heater temperature of T H=1530°C,a pouring temperature of T P=1500°C and a withdrawal velocity of V=3mm/min are standard process parameters of the Bridgman process.
D.Computer Simulation
Ahead of casting experiments,numerical simulation was applied to predict the possible impact of HC on solidification in the shroud region.Macroscopic solid-ification modeling was executed using thefinite element code CASTS,[7,8]a well-established tool for simulation
steel box
shell
xpk
mold
HC
central
rod
四级报名时间waste
graphite
furnace chamber
Fig.3—Schematic diagram offiring shell mold in a furnace chamber.
of the Bridgman process.CASTS fully describes the heat transfer balance in the vacuum furnace,including calculation of view factor radiation.The experimental tup,including furnace,blades,and shell mold with HC,was generated as an finite element method (FEM)mesh.The geometric data were transformed into an appropriate file format,including additional elements at the interface between different materials.The elements realize the interfacial heat flow.The temperature-depen-dent thermophysical properties,including tho shown in Figure 4,were read.The initial and boundary condition were t to be the same as the experiment (i.e.,T H =1530°C,T P =1500°C,and V =3mm/min).All information was stored in a command file.Using both the modified geometry and the command file,the main program evaluated iteratively the temperature evolution,taking into account the relea of latent heat and position-dependent viewing factors.The transient temperature distribution as well as the macroscopic shape and progress of the liquidus isotherm were later visualized using the CASTS-tool COLOR3D.
III.RESULTS AND DISCUSSION
A.Solidification Simulation
As stated,numerical simulation was performed ahead of the casting experiments.Figure 6shows the
simulated temperature development in two shrouds,without (left)and with (right)application of HC,respectively.Becau of the symmetrical arrangement,both blades have the same external thermal condition.The difference in temperature history can then be attributed solely to the application of HC.the revolution
At time t1,the temperature at the outer corner of both shrouds decreas below T L and the effect of HC has not yet been clearly revealed.At times t2and t3,the isolated outer corner of the original shroud becomes increasingly undercooled.In contrast,the bonded HC acts as a heat sink for the casting and effectively removes the hot spot at the inner corner of the shroud.The undercooled
extremity soon comes into contact with the primary crystal in the blade body,from which an epitaxial crystal transition should be introduced.With no HC applica-tion,the undercooled extremity remains thermally isolated for a much longer time,over which local undercooling continues to increa.
The simulation result clearly shows that thermal undercooling at the shroud extremities is unavoidable and,at the same time,the effectiveness of the HC technique.Heat extraction from the cross-ctional transition is significantly improved.The solidification front in the blade body can move into the shroud more quickly,before the melt in the outer corner becomes deeply undercooled.From t
his prediction,it can be assumed that the risk of heterogeneous nucleation of stray grain is dramatically reduced,although it remains impossible to eliminate undercooling entirely.
alacrity
The simulation was performed under equilibrium conditions and thermal undercooling at the extremities was practically underestimated.[9]A consideration of the kinetics of dendrite growth,especially the transition from vertical to lateral growth,as well as nucleation kinetics,leads to the conclusion that the thermal undercooling measured will be greater than predicted by the simulation.
B.Temperature Measurement
Figure 7shows the cooling curves measured by the thermocouples at the marked positions on the shrouds.With no HC (Figure 7(a)),the extremity of the platform (point 4and 3)cools more deeply than the inner point 1.The measurement shows a maximum undercooling D T M of approximately 29°C below an approximate T L of 1321°C at the outer corner (thermocouple 4)prior to the start of local solidification.The corresponding undercooling time D t M is approximately 244s.
With the application of HC,a notable change in the cooling curves was measured,indicating a significant improvement in cooling condition at the inner corner of the platform (Figure 7(b)).The tem
perature at point 1becomes even lower than that of point 2.Approximately 45s after point 4,point 1had cooled to T L ,whereas with no HC the same cooling took approximately 120s,as shown in Figure 7(a).Thus,the undercooled zone at the extremity is no longer so isolated from the solidi-fication front in the blade body.The maximum under-cooling at point 4,D T M ,and the corresponding undercooling time D t M are reduced from approximately 29°C to 18°C and 244s to 150s,respectively.Since the probability of grain nucleation in an alloy melt is dependent on the value of D T max and D t max ,the conclusion must be that the notable reduction achieved by application of HC will significantly reduce the probability of stray grain formation.
From the temperature measurement,it is possible to reconstruct the liquidus isotherm passing through the platform (Figure 8).With no HC,a significantly con-cave-shaped isotherm is obtained (except on the outer side where the isotherm becomes convex becau in a position favorable to radiation by the heaters).Through the application of HC,the T L -isotherm in the
platform
Fig.5—Placement of alumina insulator sheaths in wax pattern (a )and instrumentation of thermal couples in mold (b ).
becomes significantly smoothed.At the time point t1=2105s,the T L-isotherm reaches the inner corne
r of the platform(point1).With no HC,this was delayed to t2=2190s,approximately85s later,resulting in an increasingly undercooled extremity.Thus,as stated,the application of HC can significantly reduce the time and the extent of the isolated undercooling and,con-quently,the possibility of grain
defects. Fig.6—Simulated temperature development in the blade platform(left:without HC,right:with HC).once in the long ago