5
Creep-Fatigue Tests on Full Scale Directionally Solidified Turbine Blades Xiaojun Yan Jingxu Nie Group 405, School of Jet Propulsion, Beijing University of Aeronautics and Astronautics, Beijing, 100083, P.R.C. A new experimental method, in which a full scale directionally solidified (DS) alloy turbine blade is loaded by a special design rig employing friction force and heated by eddy current induction, is proposed to conduct creep-fatigue life tests in this investigation. The method can take factors such as geometry, volume, especially cast procedures, etc., into creep-fatigue life assessment. Principle and design of the test rig are fully explained. Creep-fatigue tests of turbine blades made of DZ4 alloy (one type of DS alloys) were conducted and test data were analyzed. Life prediction based on test data of this investigation shows good agreement with actual flight experience of these blades. The method of this article pro- vides a new way to estimate the potential creep-fatigue or low cycle fatigue life for turbine blades. DOI: 10.1115/1.2901174 Keywords: turbine blade, directionally solidified alloy, creep, fatigue, life prediction 1 Introduction At present, almost all advanced civil and military aeroengines use directionally solidified DS or single crystal blades 1. Tur- bine blades of aeroengine are subjected to complex mechanical, thermal, and chemical loads during service. In particular, the blades may be damaged considerably due to creep-fatigue creep and fatigue coupled loadings. Various methods have been pro- posed in predicting creep-fatigue life of turbine blades 2. Traditional creep-fatigue life prediction methods of turbine blades are based purely on macroscopic descriptions of material behaviors and conventional finite element FE methods in con- junction with empirically developed failure laws. To reduce ex- perimental costs, in most cases, tested life data of plain specimen may be used to predict “component” behavior. However, the dif- ferences between blades and plain specimens, such as geometry, volume, and especially in cast techniques for DS blades, may cause discrepancy between their mechanical performance and life distribution, and therefore may greatly affect accuracy of tradi- tional life assessment methods 1. To overcome the above short- comings, the specimens can either be machined from blades or componentlike ones 3. Compared to specimens machined from blades or componentlike ones, full scale blades are more ideal, since the specimens own the same manufacture process with real flight blades. When selecting full scale blades as specimens in life assessment tests, the most challenge work is how to simulate op- erating conditions of blades in laboratory. The objective of this paper is to develop such an experimental life assessment method by conducting creep-fatigue tests on full scale DS blades. 2 Method of Creep-Fatigue Test Stress and temperature are two factors that dominate a turbine blade’s creep-fatigue life. The combination of stress and tempera- ture distribution will result in the most heavily damaged point, namely, the hot spot, or key section in this investigation where the hot spot locates on a blade. Generally, cracks always initiate from the location of the key section, and a blade’s life is limited by the life of its key section. Therefore, creep-fatigue life of a blade can be obtained by conducting tests on the key section with properly simulating stress and temperature conditions. For a gas turbine rotor blade with a complex geometry shape, the stress state of a point on the airfoil is mainly uniaxial condi- tion along the centrifugal direction 4, while the stress value along the direction varies from point to point even on the same section, which means that the stress value of a point located on the pressure concave side is generally different from that of on the suction convex side. During creep-fatigue tests of the key sec- tion, not only centrifugal load but also a special moment is needed to apply on the section. Thus, the method to apply the above loads is different from that of the plain specimens. As same as the stress, the temperature on the key section also has a distribution. It is difficult to provide a specific temperature distribution based on the present laboratory technology level. In this investigation, on the ground of scientific research, a uniform high temperature field around the dangerous area nearby the key section is provided and superposed on the stress field. Based on the above considerations, after getting the location of the key section, the key issue of the test scheme is how to simulate the stress field of the key section and at the same time provide this area with a uniform high temperature distribution. 2.1 Location of Key Section. The location of the key section can be determined by either numerical simulation results or service/flight failure experience of the blade. The DS blade made of DZ4 alloy selected in this investigation is shown in Fig. 1. Detailed numerical simulation of stress distribution and creep- fatigue life prediction based on the development of suitable ma- terial constitutive equations can be seen in Ref. 1. Combined with the results of numerical simulation and actual flight failures of the blade, the location of the key section is determined as shown in Fig. 1, in which the dark/red area around the key section represents a dangerous area. 2.2 Stress Field Simulation. A rig to apply load on the full scale DS turbine blade to simulate the stress distribution of the key section is designed, as shown in Fig. 2. The most distinctive characteristic of the rig is that the load from the hydraulic-servo system that transferred to the blade key section is achieved through the friction force between the rig and blade, which causes no damage to the blade. As illustrated in Fig. 2, the force from the hydraulic-servo system is firstly passed through upper Jointer 1 and adjusted pin on it, and then through upper Jointer 2, adjusted pin, out clamp, inner clamp, and last to the blade. It can be seen that there exist two contact pairs during the load transmitting path: contact between the outer and the inner clamp and contact be- tween the inner clamp and the test blade. These two contact pairs transfer the load by means of friction force caused by the normal pressure of four long bolts. The four long bolts drill through the outer clamps and grip inner clamps closely by normal compres- sion, which is achieved by applying torque on four nuts linked with four bolts. The required torque value applied on the nuts depends on the load acted on the blade’s key section and is cal- culated before installation of the rig. The blade root is inserted into the disk slots on a block, which is cut from a full size turbine disk. To ensure a complete contact between the inner clamp and the blade, two interior sides of the inner clamp are machined into a complicated surface shape just as contrary to surfaces of airfoil. Moreover, another contact pair, which is between the outer side of the inner clamp and the interior side of the outer clamp, is con- Manuscript received July 5, 2007; final manuscript received September 14, 2007; published online April 29, 2008. Review conducted by David Walls. Journal of Engineering for Gas Turbines and Power JULY 2008, Vol. 130 / 044501-1 Copyright © 2008 by ASME Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 05/08/2014 Terms of Use: http://asme.org/terms

Creep-Fatigue Tests on Full Scale Directionally Solidified Turbine Blades

  • Upload
    jingxu

  • View
    214

  • Download
    0

Embed Size (px)

Citation preview

CD

X

J

GSBB

AsriTcaoctflvc

Kf

1

ubtbap

bbjpmfvcdtccbsflaepb

p

J

Downloaded Fr

reep-Fatigue Tests on Full Scaleirectionally Solidified Turbine Blades

iaojun Yan

ingxu Nie

roup 405,chool of Jet Propulsion,eijing University of Aeronautics and Astronautics,eijing, 100083, P.R.C.

new experimental method, in which a full scale directionallyolidified (DS) alloy turbine blade is loaded by a special designig employing friction force and heated by eddy current induction,s proposed to conduct creep-fatigue life tests in this investigation.he method can take factors such as geometry, volume, especiallyast procedures, etc., into creep-fatigue life assessment. Principlend design of the test rig are fully explained. Creep-fatigue testsf turbine blades made of DZ4 alloy (one type of DS alloys) wereonducted and test data were analyzed. Life prediction based onest data of this investigation shows good agreement with actualight experience of these blades. The method of this article pro-ides a new way to estimate the potential creep-fatigue or lowycle fatigue life for turbine blades. �DOI: 10.1115/1.2901174�

eywords: turbine blade, directionally solidified alloy, creep,atigue, life prediction

IntroductionAt present, almost all advanced civil and military aeroengines

se directionally solidified �DS� or single crystal blades �1�. Tur-ine blades of aeroengine are subjected to complex mechanical,hermal, and chemical loads during service. In particular, thelades may be damaged considerably due to creep-fatigue �creepnd fatigue coupled� loadings. Various methods have been pro-osed in predicting creep-fatigue life of turbine blades �2�.

Traditional creep-fatigue life prediction methods of turbinelades are based purely on macroscopic descriptions of materialehaviors and conventional finite element �FE� methods in con-unction with empirically developed failure laws. To reduce ex-erimental costs, in most cases, tested life data of plain specimenay be used to predict “component” behavior. However, the dif-

erences between blades and plain specimens, such as geometry,olume, and especially in cast techniques for DS blades, mayause discrepancy between their mechanical performance and lifeistribution, and therefore may greatly affect accuracy of tradi-ional life assessment methods �1�. To overcome the above short-omings, the specimens can either be machined from blades oromponentlike ones �3�. Compared to specimens machined fromlades or componentlike ones, full scale blades are more ideal,ince the specimens own the same manufacture process with realight blades. When selecting full scale blades as specimens in lifessessment tests, the most challenge work is how to simulate op-rating conditions of blades in laboratory. The objective of thisaper is to develop such an experimental life assessment methody conducting creep-fatigue tests on full scale DS blades.

Manuscript received July 5, 2007; final manuscript received September 14, 2007;

ublished online April 29, 2008. Review conducted by David Walls.

ournal of Engineering for Gas Turbines and PowerCopyright © 20

om: http://gasturbinespower.asmedigitalcollection.asme.org/ on 05/08/201

2 Method of Creep-Fatigue TestStress and temperature are two factors that dominate a turbine

blade’s creep-fatigue life. The combination of stress and tempera-ture distribution will result in the most heavily damaged point,namely, the hot spot, or key section in this investigation where thehot spot locates on a blade. Generally, cracks always initiate fromthe location of the key section, and a blade’s life is limited by thelife of its key section. Therefore, creep-fatigue life of a blade canbe obtained by conducting tests on the key section with properlysimulating stress and temperature conditions.

For a gas turbine rotor blade with a complex geometry shape,the stress state of a point on the airfoil is mainly uniaxial condi-tion along the centrifugal direction �4�, while the stress valuealong the direction varies from point to point even on the samesection, which means that the stress value of a point located on thepressure �concave� side is generally different from that of on thesuction �convex� side. During creep-fatigue tests of the key sec-tion, not only centrifugal load but also a special moment is neededto apply on the section. Thus, the method to apply the above loadsis different from that of the plain specimens.

As same as the stress, the temperature on the key section alsohas a distribution. It is difficult to provide a specific temperaturedistribution based on the present laboratory technology level. Inthis investigation, on the ground of scientific research, a uniformhigh temperature field around the dangerous area nearby the keysection is provided and superposed on the stress field.

Based on the above considerations, after getting the location ofthe key section, the key issue of the test scheme is how to simulatethe stress field of the key section and at the same time provide thisarea with a uniform high temperature distribution.

2.1 Location of Key Section. The location of the key sectioncan be determined by either numerical simulation results orservice/flight failure experience of the blade. The DS blade madeof DZ4 alloy selected in this investigation is shown in Fig. 1.Detailed numerical simulation of stress distribution and creep-fatigue life prediction based on the development of suitable ma-terial constitutive equations can be seen in Ref. �1�. Combinedwith the results of numerical simulation and actual flight failuresof the blade, the location of the key section is determined asshown in Fig. 1, in which the dark/red area around the key sectionrepresents a dangerous area.

2.2 Stress Field Simulation. A rig to apply load on the fullscale DS turbine blade to simulate the stress distribution of thekey section is designed, as shown in Fig. 2. The most distinctivecharacteristic of the rig is that the load from the hydraulic-servosystem that transferred to the blade key section is achievedthrough the friction force between the rig and blade, which causesno damage to the blade. As illustrated in Fig. 2, the force from thehydraulic-servo system is firstly passed through upper Jointer 1and adjusted pin on it, and then through upper Jointer 2, adjustedpin, out clamp, inner clamp, and last to the blade. It can be seenthat there exist two contact pairs during the load transmitting path:contact between the outer and the inner clamp and contact be-tween the inner clamp and the test blade. These two contact pairstransfer the load by means of friction force caused by the normalpressure of four long bolts. The four long bolts drill through theouter clamps and grip inner clamps closely by normal compres-sion, which is achieved by applying torque on four nuts linkedwith four bolts. The required torque value applied on the nutsdepends on the load acted on the blade’s key section and is cal-culated before installation of the rig. The blade root is insertedinto the disk slots on a block, which is cut from a full size turbinedisk.

To ensure a complete contact between the inner clamp and theblade, two interior sides of the inner clamp are machined into acomplicated surface shape just as contrary to surfaces of airfoil.Moreover, another contact pair, which is between the outer side of

the inner clamp and the interior side of the outer clamp, is con-

JULY 2008, Vol. 130 / 044501-108 by ASME

4 Terms of Use: http://asme.org/terms

ttp

0

Downloaded Fr

acted with inclined planes, which makes them more closely con-act with increasing load. The goal of all these measures is torevent the specimen slipping out of the rig.

Besides loading through the friction force between contact

Fig. 1 Key section location

Fig. 2 Illustration of the load rig

44501-2 / Vol. 130, JULY 2008

om: http://gasturbinespower.asmedigitalcollection.asme.org/ on 05/08/201

pairs, another distinctive feature of the rig is that it can applyvariable moment on the blade key section by adjusting pins. Asseen in Fig. 2, there are totally four adjusting pins. The two upperjointers are linked together by two adjusting pins, which are lo-cated in two perpendicular directions, so do the two lower joint-ers. The jointers can move along the axis of adjusting pins, i.e.,different bending moments can be obtained and applied on the keysection of the blade by different combinations of four adjustingpin positions.

To verify the load capacity of the above designed rig, series ofexperiments were conducted. As illustrated in Fig. 3, strain gaugeswere pasted surrounded the key section. Before tests, positions offour adjusting pins were adjusted and then fixed by inserting thinmetal spacers between two jointers. After applying a load, a strainvalue along the centrifugal direction of every paste point aroundthe key section surface was recorded by a computer. Figure 4shows a typical strain distribution around the key section surfacewith one of four pin position combinations. In Fig. 4, the horizon-tal axis represents the absolute distance from the paste gauge tothe start point, which locates at the trailing edge of the key sectionnear Gauge 1, as shown in Fig. 3. Combining positions of fouradjusting pins, different moments or strain distributions can beapplied on the key section. The strain distributions got from ex-periments can be compared to those calculated under engine’sdifferent operating conditions. The solid line in Fig. 4 shows astrain distribution of the key section under the engine’s maximumaerodynamic conditions. When a specific engine’s operating con-dition is selected to conduct a creep-fatigue test, the positions offour pins need to be adjusted until the strain distribution of the keysection agrees well with that of a specific engine’s operating con-dition, and the pins can then be fixed and begin to conduct creep-fatigue tests. It is worth pointing out that the strain value of Fig. 4was measured at room temperature; therefore, the calculated strainvalues under engine operating condition have been processed toeliminate the effect of temperature.

2.3 Temperature Field Simulation. High-frequency eddycurrent induction furnace was selected in this investigation. Asseen in Fig. 5, part of the blade was heated with induction coils.One advantage of the heating method is that it facilitates visualobservation of the blade crack growth. Another merit is that it caneasily be assembled with the loading rig, as shown in Fig. 2. Theheated area is part of airfoil around the key section, and in thisinvestigation, a uniform temperature distribution near the key sec-tion was simulated.

3 Tests and Results

3.1 Test Conditions. Creep-fatigue tests on full scale DS

Fig. 3 Positions of paste strain gages

blades were conducted using a 10 ton hydraulic-servo system con-

Transactions of the ASME

4 Terms of Use: http://asme.org/terms

tbaartb

ribu

J

Downloaded Fr

rolled with MTS 407 panel and computer software. The full scalelade made of DZ4 superalloy as mentioned before is selected asspecimen. Furthermore, the experimental setup is just as stated

bove. Isothermal cycle tests with hold times �t=28 s were car-ied out at 880°C. The load spectrum is shown in Fig. 6. Theemperature of 880°C is a significant temperature for the DZ4 DSlade in this investigation. Since the creep-fatigue test requires

Fig. 4 Strain dist

Fig. 5 Picture of conducting creep-fatigue test

ournal of Engineering for Gas Turbines and Power

om: http://gasturbinespower.asmedigitalcollection.asme.org/ on 05/08/201

long time to get a group of life data of blades, completing acreep-fatigue test of a blade may need months of time. An accel-erated load spectrum T=900°C was used for later blade tests. Thehold times of �t at different temperatures are calculated based onthe principle of equivalent damage in one cycle. �t at 900°C is13 s.

3.2 Results and Analysis. Altogether, 19 creep-fatigue testedblades were conducted with a load spectrum of Fig. 6 and most ofthem are shown in Fig. 7, among which the number of new bladesis 12 and the used ones is 7. It can be seen from Fig. 7 that allblades fractured at the location of the key section.

The life data of creep-fatigue tests on DZ4 blades are given inTable 1. It is noted that tests of new blade Nos. 1–6 were con-ducted at 880°C and the others were at 900°C, and sample num-bers are reordered based on lifetime. For convenience of the dataprocess, life data conducted at 900°C have been multiplied by acoefficient, which is obtained by dividing a mean value of the lifedata at 880°C by 900°C and converging into the correspondingdata at 880°C. From Table 1, the following can be observed.

�1� The creep-fatigue life of full scale DS blades determined bythe above experimental method is greatly shorter �nearly50%� than the one predicted by the numerical simulationbased on the test data of the plain specimens �1�. Therefore,the experimental method such as this investigation shouldbe developed so that factors such as geometry, differentcasting procedures, etc., can be considered when predictingcreep-fatigue life of DS turbine blades.

tion comparison

Fig. 6 Load spectrum of creep-fatigue tests

JULY 2008, Vol. 130 / 044501-3

4 Terms of Use: http://asme.org/terms

B

Fig. 7 Fracture blades of creep-fatigue tests

0

Downloaded Fr

Table 1 Creep-fatigue life data of DS blades at temperature 880°C

ladetype Sample No.

Life of creep fatigue �h, cycle�Arithmetic

average1 2 3 4 5 6 7 8 9 10 11 12

New t /h 90.40 104.88 138.05 175.15 284.27 322.55 85.83 148.61 163.31 182.87 255.80 343.76 191.29N /cycle 10,848 12,586 16,567 20,070 31,410 35,116 9,273 16,066 18,003 19,848 27,763 37,308 21,238

Used t /h 47.45 54.11 165.30 173.31 192.83 235.65 294.58 166.18N /cycle 5,166 5,896 17,986 18,843 20,988 25,643 32,092 18,088

Fig. 8 Histogram and PDF of creep life „new blades…

44501-4 / Vol. 130, JULY 2008 Transactions of the ASME

om: http://gasturbinespower.asmedigitalcollection.asme.org/ on 05/08/2014 Terms of Use: http://asme.org/terms

cbflhb

naurtcsim

4

t

J

Downloaded Fr

�2� The life data of the used blades have a big scatter than thenew ones. This is reasonable since the used blades haveexperienced different operating conditions and the accumu-lated damage for each used blade before tests is different.

�3� For new blades, the life scatter of normal test at 800°C�new Blades Nos. 1–6� is similar to that of the acceleratedtest at 900°C.

Based on the test data of Table 1, different probability modelsan be selected to process the data, and based on the relationshipetween the flight and the experimental load spectrum, the safeight life of the tested blades can be obtained. Figure 8 showsistogram and probability density function �PDF� of new bladesased on the creep lifetime using Weibull distribution model.

It is worth pointing out that the ratio of the safe creep life ofew blades to used blades agrees well with the actual flight situ-tion: The new blades fractured at flight hours two times of thesed blades; moreover, after conversion between experimental andeal operation conditions, the remaining life of the used blade plusheir flight life is nearly equal to the life of new blades. Thisonsistency between experimental results and flight practice canupport that the experimental life obtained by the method of thisnvestigation will play an important role in DS blade life assess-

ents.

ConclusionsWith a special design of the test rig, operating stress field and

emperature of the key section on the DS turbine blade airfoil can

ournal of Engineering for Gas Turbines and Power

om: http://gasturbinespower.asmedigitalcollection.asme.org/ on 05/08/201

be simulated in bench environment. Consistency between experi-mental results and flight practice of the tested DS blades supportsthat the experimental life obtained by the method will play animportant role in DS blade life assessments. The method of thisarticle provides a new way to estimate the potential creep-fatigueor low cycle fatigue �LCF� life for DS turbine blades, and thismethod can also be applicable to blades made of single crystal andgenerally high temperature superalloys.

AcknowledgmentThis work was supported by a Foundation for the Author of

National Excellent Doctoral Dissertation �FANEDD, ApprovedNo. 200351� and a Program for New Century Excellent Talents inUniversity of PR China �NCET-06-0178�.

References�1� Li, H. Y., 2002, “Research and Application of Anisotropic Viscoplastic Dam-

age Uniform Constitutive Mode,” Ph.D. thesis, Beijing University of Aeronau-tics and Astronautics, Beijing.

�2� Harrison, G. F., Tranter, P. H., Shepherd, D. P., and Ward, T., 2004, “Applica-tion of Multi-Scale Modeling in Aeroengine Component Life Assessment,”Mater. Sci. Eng., A, 365, pp. 247–256.

�3� Issler, S., and Roosb, E., 2003, “Numerical and Experimental InvestigationsInto Life Assessment of Blade-Disc Connections of Gas Turbines,” Nucl. Eng.Des., 226, pp. 155–164.

�4� Cross, C. J., 2000, “Multi-Axial Testing of Gas Turbine Engine Blades,” 36thAIAA/ASME/SAE/ASEE Joint Propulsion Conference, Huntsville, AL, PaperNo. AIAA2000-3641.

JULY 2008, Vol. 130 / 044501-5

4 Terms of Use: http://asme.org/terms