Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Thermomechanical Stress Distributions in a Gas Turbine Blade Under the Effect of Cooling Flow Variations

Candelario Bolaina1
e-mail: cbolaina@prodigy.net.mx

Julio Teloxa
e-mail: j_teloxa@hotmail.com

Cesar Varela
e-mail: cvboydo@hotmail.com

Fernando Z. Sierra2
e-mail: fse@uaem.mx Centro de Investigacio n en Ingenier a y Ciencias Aplicadas, CIICAp, Universidad Auto noma del Estado de Morelos, UAEM, Av. Universidad 1001, Col. Chamilpa, C. P. 62209, Cuernavaca, Morelos, Mexico

Thermomechanical stresses in gas turbine blades are investigated. Attention is focused on effects caused by varying the cooling airow that runs through the blade interior, keeping constant a mainstream condition around the blade surface. Stress concentration was predicted numerically under engine real operating conditions. Temperature distributions in the metal blade surface produced by convective boundary conditions were linked with heat conduction within the blade using a conjugate solution. Results of stress concentration in the blade material for reduced cooling ow rate, blocked cooling ducts, and rotation rate were obtained. It is shown that temperature and stress distributions are a strong function of position in blade interior material and surface. Thermomechanical stress concentration was observed in the leading edge, with the endwall region affected by large stress concentration. Stress magnitude increments were found for combined cyclic thermal heating and sustained mechanical loads on specic planes of airfoil span for reduced cooling ow. Also, large stress gradients between leading and trailing regions of the blade were observed. The study reveals that blocking channels increase stresses in the central region of blade transversal cross section. [DOI: 10.1115/1.4023465]

Introduction
Gas turbines represent a major source of a signicant prime mover in modern life. They have been used during several decades for electricity generation and transportation. Their continuous development involves the use of higher operating temperatures, which is reected in higher turbine efciencies. The costs of increasing turbine efciency, however, are many, including better manufacturing techniques and materials of the blades, especially for rst stage, and more important, enhanced cooling techniques. A huge effort has been made in order to solve overheating using several cooling methods. However, risks of blade fracture may appear as a consequence of faulty engine operations. Deposits in
1 2

Present address: Dep. de Ing. Mec. Universidad Jua rez Auto noma de Tabasco. Corresponding author. Contributed by the International Gas Turbine Institute (IGTI) of ASME for publication in the JOURNAL OF TURBOMACHINERY. Manuscript received December 16, 2010; nal manuscript received August 8, 2012; published online September 13, 2013. Assoc. Editor: Matthew Montgomery.

the compressor, for instance, may reduce compressed air discharge affecting the cooling ow that reaches the blades; partially blocked cooling channels may also contribute to increasing thermal stresses locally. Both conditions increase risks of thermomechanical stress concentration, affecting blade integrity through reducing its life. Thermal stress is important at high temperatures of the heating cycle because reduced low cycle fatigue life, which means that material sensitivity to high temperature exposition, produces cumulative fatigue damage. Thermal cyclic deformation producing structural displacement patterns may lead to dimensional changes every thermal cycle if plastic state is reached. Damage is accumulated, and both thermal and mechanical damage modes are combined in an overall damage evaluation using the Miners rule accumulation of damage [1]. Therefore, attention has been paid to the need of assessment methods for damage and life. Thus, a study of blade thermal stresses under ow reductions (thermal gradients perturbations) within the blade is essential. It can help to understanding how material blade temperature distributions are related with stress concentration in a 3D geometry of the blade. Furthermore, a study of cooling ow conditions may contribute to understanding turbine operation under faulty cooling and its cracking. Blade structure has been investigated intensely in recent years under a variety of operating conditions as contaminated channels with deposition of materials from several sources, or even blocked channels. Arnal et al. [2] studied the interaction between uid and structure. The stresses developed in the blade under thermal and inertia effects were investigated by modeling the heating process using nite volume and nite element numerical approaches. The research focused on simulating the transition from laminar to turbulent ow as the boundary layer develops. Ogata and Yamamoto [3] investigated several methods for life estimation of high temperature components in gas turbines based on the computation of stresses. Tensiontorsion biaxial thermomechanical fatigue tests were applied for a temperature range on a super alloy blade following Ieronymidis et al. [4]. It was found that a relationship between strain rate and fatigue exists, revealing that the startsteady state-stop cycles performed during the operation of the gas turbines are important. Several cooling schemes of a trailing edge were analyzed by Cuhna et al. [5], since this section of the blade has been identied to provide a degree of blockage in the gas ow path that may contribute to the overall degradation of turbine performance and efciency. The study revealed that the combination of different schemes of cooling may provide optimized arrangements in terms of structural requirements of the blade. Constraints of missions transients reected in internal turbine components were considered to develop a life prediction approach [6]. The method includes an evaluation of components transient metal temperatures, resolved maximum shear stresses and strain, and subsequent component life capability for fatigue and creep damage. The analysis focused on a simplied section cut at the blade midspan. Heat conduction in the solid material must be resolved for estimating the thermal mechanical stress on the blade surface. This requires a conjugate solution, which can be obtained by delimiting a thickness for walls, specifying a thermal resistance of the blade material. By meshing the solid in the normal direction to the wall, the heat conduction may be calculated. The coupled mesh blocks uid-solid include the cooling ow, the solid material, and the exterior mainstream, allowing the convection in the planar direction to the wall to be resolved [7]. Conjugated solutions have been applied widely: Bohn et al. [8] analyzed the effect of local heat transfer rates on secondary ow velocities within the lm cooling. It was emphasized that the cooling uid is heated convectively on the way through the supply channels and the cooling holes, which has an inuence on the external cooling performance. The same approach was used to investigate a closed circuit steam cooling, where external hot gas and steam ows through cooling tubes were predicted, taking into account the solid body vane [9]. NOVEMBER 2013, Vol. 135 / 064501-1

Journal of Turbomachinery

C 2013 by ASME Copyright V

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/ on 10/18/2013 Terms of Use: http://asme.org/terms

The aim of this paper concentrates on the study of blade stress due to both thermal and mechanical sources, and its behavior under cooling ow reductions. It describes a methodology by which thermomechanical stresses in a gas turbine blade are calculated under turbine real operating conditions in stationary conditions. Similar stress computations have been reported elsewhere without considering any cooling ow variations [10]. Since blade life may be reduced by low cycle fatigue, the objective of this investigation is quantifying blade stress distribution due to coolant airow variations and rotating rate. Flow variations are due to faulty reduced compressor air mass ow rate discharge and partially blocked cooling channels have been reported and caused by deposits from polluted air. A blade type D is studied following sponsor directions. The results obtained indicate that heat transfer rate is a strong function of position. It is found that some regions of the blade present stress concentration by combinations of cooling ow variations and rotation.

Methodology
Numerical Approach. The simulation of the process considers two main steps: rst, a nite volume conjugate solid-uid solution of the cooling and hot gases ows, which included heat conduction in the solid material using a rst computational grid, and second, a nite element solution of the strain rate in the solid structure using a second grid based on the temperature eld obtained in the rst solution as a boundary condition. An overview of the blade and uids considered in this study is shown in Fig. 1. The mesh of the real blade geometry used is illustrated in Figs. 2(a), 2(b), and 2(c). The mainstream of hot gases was represented by a ow of hot air at real engine operating temperature and pressure. It ows around the blade while coolant circulates through the cooling channels, which are shown in Fig. 2(c). Distributed cooling channels account eight from the leading edge of diameter 0.004 m, six of 0.003 m, and the last two of diameter 0.002 m. Simulations of coolant ow reductions and blocked nonblocked cooling channels (indicated as black points in Fig. 2(c)) were carried out. The simulations were conducted with commercial programs FLUENT and ANSYS, based on nite volume and nite element, respectively. In addition, 3D mass, momentum, and energy governing equations for a compressible ow through and around the blade were solved [7,11]. A Reynolds-averaged NavierStokes equations approximation, RANS, based on the Reynolds stress model was used to resolve the turbulence. This model has probed to render better results compared with two-equations models like the je. The temperature eld within the solid material was resolved by Fourier equation. Two body force terms were taken into account in the momentum equations: the Coriolis force and the centrifugal force. In these terms, a rotational speed,-, was used to simulate the real operating conditions of the engine. A 1.9 106 hexahedral elements mesh was built with coupled solid-uid domains. This structured mesh accelerates the convergence compared to an unstructured one, despite the greater effort required to build it due to the complexity of curvature of walls.

Fig. 2 Computational grid of the blade used in the simulation; (a) nite volume; (b) nite element; (c) plan view of the blade with cooling channels: nine channels blocked and six channels blocked

Fig. 1

Computational domain of group blade-rotor

Converged nite volume solutions were obtained with a criterion of residuals less than 1 104. The grid independence was obtained following Celik et al. [12] with three grids of 3.13 105, 7.12 105, and a rened one of 1.9 106 cells. The error computation expressed by the convergence index from rened to midcoarse grid and from midcoarse to coarse grid was 1.07 and 1.18, respectively, within the range commonly accepted [12]. After the Transactions of the ASME

064501-2 / Vol. 135, NOVEMBER 2013

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/ on 10/18/2013 Terms of Use: http://asme.org/terms

Table 1

Boundary conditions of operating engine modeling Mainstream Rem Tm K P Pa 581 103 581 103 581 103 581 103 581 103 7,796.6 7,406.7 7,016.9 6,627.1 5,925.4 Rec Coolant Tc K 666.15 666.15 666.15 666.15 666.15

Run 1 2 3 4 5 1.35 105 1.35 105 1.35 105 1.35 105 1.35 105

1260.15 1260.15 1260.15 1260.15 1260.15

grid independence test, the rened grid was adopted for the rest of the computations. Boundary conditions applied during the simulations are given in Figs. 2(a) and 2(b) (see magnitudes in Table 1). The nite element computation is based on a combination of elements for thermal/elastic effects. Thermally, the model was based on the use of 46,791 SOLID 70 tridimensional elements with a total of 29,300 nodes. In the thermal analysis, the temperature was considered to represent each node with freedom degree of any element. For elastic displacements, SOLID 45 tridimensional elements of eight nodes and three degrees of freedom were used, which can be represented either by displacements or tractions. Stresses in the blade solid material were calculated by solving the equations that govern the elastic displacements within an isotropic body due to forces to which the solid structure is subjected using ANSYS [13]. Forces from rotational speed - and from thermal gradient due between mainstream and coolant were considered. The computational methodology described above was validated using data obtained with a simplied model of the blade, which was reproduced to form the cascade shown in Fig. 3. Temperature measurements of blade surface, in addition to PIV measurements around the blade, were obtained to describe the velocity eld and heat transfer. These measurements will be briey described below, but more details can be found elsewhere [14,15]. Temperature Validation. Measurements of metal surface temperature were conducted using temperature sensitive paint, TSP for a xed Rem 2 105 and variable Rec (for details see Sierra et al. [14,15]). A basis plate was connected with a coolant chamber underneath an instrumented stainless steel blade (0.2515 m height, 0.185 m chord, and 0.11 m pitch). An overall heat transfer effectiveness dened as h T Tm Tc Tm (1)

represents a measurement of heat transfer between coolant and mainstream through the blade solid material made of steel. h is based on the temperature difference between the coolant, the mainstream, and the blade wall surface. For this case, h represents the temperature changes in the blade surface as a consequence of modifying the coolant ow rate. The surface temperature measured by the TSP is known as the local wall temperature, T. Heat transfer effectiveness h is calculated from the measured coolant temperature, Tc and mainstream temperature, Tm [16]. In practice, Tc must be smaller than Tm, allowing positive magnitudes of h. However, in laboratory experiments, a usual practice has been heating a coolant ow instead of the mainstream, which is many times larger than the rst [17]. The basis is that heat transfer rate behaves the same in one direction or another, provided temperature differences in Eq. (1) remain the same. Having the mainstream temperature lower than the coolants does not affect effectiveness as a function stated by Eq. (1). The experimental measurements are displayed as temperature contours in the suction side of the leading edge, as shown in Fig. 4(a), followed by numerical predictions as shown in Fig. 4(b). The results were plotted in the span direction as shown in Fig. 4(c) for coolant ow rates: Rec 10 103, 15 103, and 20 103. It is observed that the modeling results follow the experimental data for all three tests, with an acceptable discrepancy of 18% in the top of the blade surface and 25% in the bottom of the blade. The details of these experiments were published by Sierra et al. [14]. Flow Dynamics Validation. Vector maps overlapped with a color map of scalar magnitudes were obtained using particle image velocimetry, PIV, in order to characterize the ow. The ow in a vertical plane of the pressure side of the blade and close to the leading edge is shown in Fig. 5(a). As observed, the mainstream moves towards the wall, forming a region of high vortex activity located within a 10% of the span. A whole description of the experiment and results of PIV measurements is given elsewhere [12]. The computational domain was set up according to boundary conditions of PIV experiment. As observed in Fig. 5(b), the vortex ow is reproduced by the numerical solution obtained with a RSM model for turbulence [11]. Some details of the ow in the cavity of the blade squealer are observed in this gure, but this topic is out of the scope of this paper. Direct comparison of CFD predictions against PIV data is shown in Fig. 5(c) by a velocity prole that starts on the blade wall at a distance 0.02 S from tip. It is observed that the prediction follows the data with a maximum 18.5% discrepancy located a distance 15% of the chord from the wall. The experimental rig and data presented have been used as the basis of extended numerical simulations in both aspects, thermally and dynamically.

Results and Discussion


Thermal Stress Distributions With No Rotation. A rst solution for real engine operating temperature and pressure was obtained for a Reynolds cooling ow of Rec 7796, constant mainstream and no rotation, - 0. Afterwards, the Rec was varied according to conditions given in Table 2 and rotation was activated. The results of temperature for run 1 of this Table are shown by planes in Figs. 6(a)6(f) whose position appears in Fig. 6(g). Temperature contours reect the heating process along the blade. The rst three planes are not heated because the mainstream rounds only the blade span: the planes in Figs. 6(d) to 6(f) of higher temperatures observed. Strong temperature gradients are detected across pressure and suction sides of the blade in Figs. 6(d), 6(e), and 6(f). Inside the blade structure, the coolant effect is clear around channels that are at least 52 degrees cooler than the planes in Figs. 6(d) and 6(e). The higher temperature observed in NOVEMBER 2013, Vol. 135 / 064501-3

Fig. 3

Cascade of blades in a Perspex wind channel

Journal of Turbomachinery

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/ on 10/18/2013 Terms of Use: http://asme.org/terms

Fig. 5 Comparison of numerical prediction to experimental data for pressure side in leading edge: (a) schematic drawing of blade tip conguration and PIV velocity vectors; (b) CFD velocity vectors; (c) CFD against PIV velocity prole from the blade wall Fig. 4 Comparison of numerical to experimental results for leading edge suction side: (a) experimental contours of temperature; (b) numerical contours of temperature; (c) numerical against measurements of cooling effectiveness. Scale units are K.

the plane of Fig. 6(f) is attributed to blade tip leakage that leads to higher blade temperature at the blade tip. Temperature distributions observed in Figs. 6(a)6(f) produce strain in all directions in the material, and hence, stresses of 064501-4 / Vol. 135, NOVEMBER 2013

thermal origin entirely. The results of stress magnitude for the same conditions of Fig. 6 are shown in Figs. 7(a)7(f). Stress distributions in planes 1 to 6 follow the temperature distribution, with maximum stresses observed in planes 3 and 4 instead of planes 4, 5, and 6 of maximum temperature. Stress changes from the basis to the tip of the blade are a function of thermal gradients in solid material. Because of Hooks law, the strain or deformation rate varies according to the Transactions of the ASME

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/ on 10/18/2013 Terms of Use: http://asme.org/terms

Table 2

Conditions for cooling channels ow in modeling Coolant Inlet, pressure, Pa Channels 1 to 8 594513.3 585852.84 577722.06 570106.9 557658.06 Channels 9 to 14 666989.15 650307.35 634764.85 620313.99 596914.11 Channels 15 and 16 909655.71 863.31.83 820436.6 781577.52 720208.45 Inlet temperature K 666.15 666.15 666.15 666.15 666.15

Run 1 2 3 4 5

Rec 7,796.6 7,406.7 7,016.9 6,627.1 5,925.4

displacement vector, which for a thermal source, represents thermal expansion associated to thermal gradients [1719]. It is worth noting that in our FEM model, all nodes that form part of the surface boundary located at the blade basis were declared with null freedom: ux uy uz 0. However, in order to avoid a totally static condition wherein the blade is joined with the rotor, subsequent nodes upwards were declared as partially static condition: ux uz 0 or uy uz 0. Nodes at wall cooling channels out of the basis were not constrained to the static condition at all, as well as all nodes of the blade span. Discontinuous material (like cooling channels) and/or reduced freedom may lead to over prediction of stress as we observe in Figs. 7(a) and 7(b). The case of high stress around cooling channels has been documented in literature [10], and stress drop from plane 5 to 6 is assumed to be due to the drop of thermal gradient. Since homogenous stainless steel material of constant properties [20] was considered, the tensor strain is a function only of space gradients, which in the case of no rotation are pure thermal ones due to the temperature difference between coolant and mainstream ows. By comparing Figs. 6 and 7, local thermal gradients produce local stress magnitude at each plane. Plane c, for instance, is more stressed than others because it is exposed to a drastic temperature change between blade basis and span, which leads to large displacement gradients, or expansion in the xy directions. In the static condition, thermal stress magnitude in the blade is due to the temperature gradient between the coolant and the mainstream, which increases vertically in the z direction until the span starts, close to plane c; after this plane, the thermal gradient starts to decrease because the coolant and blade temperature are governed by mainstream conditions. Maximum thermal gradient takes place in the region near plane c; hence it produces almost ve times more stress in this region than others. Mechanical Stress Distributions With Rotation. The computation of stress including blade rotation at - 3600 rpm at the normal temperature ambient condition was carried out with the aim of highlighting differences with thermal origin. The results of stress distributions in plane a to plane f are shown in Figs. 8(a)8(f). By considering rotation, stress distribution is given to solid material displacements produced by mechanical forces. Stress magnitude changed from a minimum in plane a, to a maximum in plane d, while moderate stresses occur in planes b, c, and e. The reason for this change stress magnitude is that centrifugal force takes place in radial direction, rather than in the xy directions. Furthermore, centrifugal force is a function of mass, which is larger in correspondence with plane 3. As observed, minimum stresses are located at the basis, plane a, and tip, plane f, because of minimum radius and mass, respectively. By comparing stress magnitude between Figs. 7 and 8, it is observed that stresses that are mechanically produced are in average one order smaller than those produced thermallyexcept in plane c where thermal stress is almost two orders larger than the mechanical one. Journal of Turbomachinery

Fig. 6 Numerical results of temperature distributions for run 1 of Table 1 with no rotation in: (a) plane a, 0.0 L; (b) plane b, 0.2 L; (c) plane c, 0.4 L; (d) plane d, 0.6 L; (e) plane e, 0.8 L; (f) plane f, L; and (g) planes location

Thermomechanical Stress Distributions. When stress was calculated under combined thermal and mechanical conditions, different distributions were found with effects from each source. However, as reported in the literature [10] maximum thermomechanical stresses have a primarily thermal origin due to NOVEMBER 2013, Vol. 135 / 064501-5

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/ on 10/18/2013 Terms of Use: http://asme.org/terms

Fig. 6

(Continued)

temperature gradients. In fact, cracks found in the blade cooling holes at midheight blade span have been attributed to thermal condition rather than mechanical condition [10]. The results in this study are given in Figs. 9(a)9(f) for conditions of run 1, Table 2. It is observed in this gure that stress remains on the same level in planes a, b, and c. However, by comparing Figs. 7 (thermal), 8 (mechanical), and 9, (thermomechanical) stress increases from the pure thermal to the thermomechanical condition, especially in planes d, e, and f, as observed in Fig. 9. This result is due to tensor components in all directions when temperature and rotation are combined: effects do not follow one or another source. The combined effect produces large gradients of stress along the blade span from plane c to f. In addition, thermomechanical sources lead to large stress gradients between leading and trailing regions of the blade. This last condition may have a strong inuence on part life, leading to reduced blade life, especially near plane d. By focusing the analysis on planes e and f, we observe that blade tip is also affected when rotation is activated, producing large temperature gradients between the leading and trailing edges in the tip too, as shown by Figs. 10(a), 10(b), and 10(c). The tip is affected by leakage ows from pressure to suction sides as well. However, high temperature concentrates in the leading edge as a consequence of complex ow and convective-conductive heat transfer processes. Effects of Blocked Cooling Channels. Effects of blocked cooling channels were investigated. Results for plane d are being used for discussion. Temperature distributions in plane d are rst presented in Fig. 11(a) for no blocking, for six blocked channels in Fig. 11(b), and nally for nine blocked channels in Fig. 11(c). Effects are noticed in these results like a central region of stress drop around the camber line. This region reduces as blocked channels increase. Another effect is that a large temperature gradient is observed between the leading edge and the rest of the blade for the three conditions, with the thermal gradient increasing as the number of blocked channels increases. An explanation for high temperature leading edge concentration is that rotation modies the ow patterns around span through combined Coriolis and centrifugal forces, which drive convective streams. Effects of cooling ow on blade stress distribution are 064501-6 / Vol. 135, NOVEMBER 2013

Fig. 7 Numerical results of thermal stress distributions for run 1 of Table 1 with no rotation in: (a) plane a; (b) plane b; (c) plane c; (d) plane d; (e) plane e; (f) plane f; Planes location as in Fig. 6(g)

summarized by averaging stress magnitude over the whole cross section area of plane d. Averaged stress magnitude was then plotted against cooling ow Reynolds number, Rec, for a number of blocked channels. The results are shown in Fig. 12, where it can Transactions of the ASME

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/ on 10/18/2013 Terms of Use: http://asme.org/terms

Fig. 8 Numerical results of mechanical stress distributions for rotation rate - 5 3600 rpm in: (a) plane a; (b) plane b; (c) plane c; (d) plane d; (e) plane e; (f) plane f; planes location as in Fig. 6(g)

Fig. 9 Numerical results of thermomechanical stress distributions for run 1 of Tables 1 and 2 in: (a) plane a; (b) plane b; (c) plane c; (d) plane d; (e) plane e; (f) plane f; planes location as in Fig. 6(g)

be observed that blocked cooling channels produce an increment of blade stresses compared to the nonblocking condition. It is observed that blocking six channels increases stress magnitude by 3.4%. Furthermore, nine blocked channels increase stresses by 6%. These stress magnitude differences for varying blocking condition remain almost the same for the whole range of Rec investigated. It is worth saying that these numbers reect average effects, while stress concentration due to blockage may be Journal of Turbomachinery

even worse in terms of fatigue for specic regions. The results of Figs. 10 and 11 are shown in transversal cross sections, and emphasize that stress concentration is a local effect, more than average surface effect. Figure 12 shows that stress magnitude increases for minimum Rec and nonblocking condition, compared with the maximum Rec studied. Results in this gure show that the stress increment obtained with low Rec reduces as the number of cooling channels NOVEMBER 2013, Vol. 135 / 064501-7

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/ on 10/18/2013 Terms of Use: http://asme.org/terms

Fig. 12 Thermomechanical stress distributions in plane d, as a function cooling Reynolds number, Rec, runs 1 to 6 of Tables 1 and 2, and blocking conditions

Fig. 10 Numerical results of temperature distributions for no blocked channels and rotation rate - 5 3600 rpm in: (a) plane d; (b) plane e; (c) plane f; location as in Fig. 6(g)

Fig. 11 Numerical results of temperature distributions for plane d, run 1 of Tables 1 and 2, and rotation rate - 5 3600 rpm, in: (a) no blocked channels; (b) six blocked channels; (c) nine blocked channels; plane location as in Fig. 6(g). Blocking according to Fig. 2(c).

Fig. 13 Numerical results of thermomechanical stress distributions in the blade as a function cooling Reynolds number, Rec, runs 1 to 6 of Tables 1 and 2, and blocking conditions: (a) planes a, b, and c; (b) planes d, e and f

064501-8 / Vol. 135, NOVEMBER 2013

Transactions of the ASME

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/ on 10/18/2013 Terms of Use: http://asme.org/terms

augments. However, the nine blocked cooling channels condition presents minimal variations for all Rec under study. The effects of Rec on other planes are displayed as follows. First, averaged stress magnitude in planes a, b, and c, are shown in Fig. 13(a). Based on this gure, stress magnitude in plane c is one order larger compared to planes a and b. Cuhna et al. [5] have reported high centrifugal stresses in lower regions of blades, and suggested using centerline discharge with cooling holes in lower portions of the blade in order to release thermomechanical stress concentration. Second, results in planes d, e, and f, given in Fig. 13(b), indicate that blade tip is where largest effect of Rec was found, while planes d and e show that stress magnitude is almost constant for all Rec. The remainder of this paper is directed to recall that averaged stress inplane seems to be constant in part of blade, while stress concentrations occur in specic regions within all planes analyzed, and these are related to heating. Ogata and Yamamoto [3] have addressed the dependency of fatigue life on strain ratio, shear to normal, for a part subjected to thermomechanical fatigue loading. It is shown that maximum strain occurs at maximum temperature during the start-down cycle, depicted through a blade waveform plot in time. From there, it follows that the number of cycles to fatigue depends on stress magnitude to which the part is subjected. It is thus clear that temperature increments reduce life.

P Rem Rec S T Tc Tm Tref Twall T0 uc um x, y, z  m r q h -

static pressure, Pa Reynolds number mainstream: Rem um C=m Reynolds number coolant ow: Rec uc d=c blade span, 0.25 m (experimental), 0.12 m (real geometry) local wall temperature, K coolant temperature, K mainstream temperature, K reference ambient temperature, K local surface temperature, K temperature set as boundary condition coolant velocity at the inlet of cooling holes, m/s mainstream velocity at inlet, m/s Cartesian coordinates, m kinematic viscosity of air at ambient temperature, m2/s kinematic viscosity of air at operating temperature, m2/s stress tensor magnitude (von Misses) density, kg/m3 overall effectiveness, dimensionless rotation rate, rpm

References
[1] Cunha, F. J., Seetharaman, V., and Chyu, M. K., 2006, Thermal-Mechanical Life Prediction System for Coated Anisotropic Turbine Components, ASME Paper No. GT2006-91017. [2] Arnal, M., Precht, Ch., and Sprunk, T., 2007, Fluid Structure Interaction for Cool Gas Turbine Blades, ANSYS Advantage, 1, pp. 68. [3] Ogata, T., and Yamamoto, M., 2006, Biaxial Thermomechanical Fatigue Life Property of a Ni Base DS Super Alloy, ASME Paper No. GT2006-90758. [4] Ieronymidis, I., Gillespie, R. H., and Ireland, P. T., 2006, Detailed Heat Transfer Measurements in a Model of an Integrally Cast Cooling Passage, ASME Paper No. GT2006-91231. [5] Cuhna, F. J., Dahmer, M. T., and Chyu, M. K., 2005, Analysis of Airfoil Trailing Edge Heat Transfer and Its Signicance in Thermal-Mechanical Design and Durability, ASME Paper No. GT2005-68108. [6] Cuhna, F. J., Dahmer, M. T., and Chyu, M. K., 2005, Thermal-Mechanical Life Prediction System for Anisotropic Turbine Components, ASME Paper No. GT2005-68107. [7] Fluent, Inc., 2001, Fluent V6.3 Users Guide, Vol. 3, Fluent Inc., Canterra Resource Park, Lebanon, NH. [8] Bohn, D., Ren, J., and Kusterer, K., 2003, Conjugate Heat Transfer Analysis for Film Cooling Congurations for Different Hole Diameters, ASME Paper No. GT2003-38369. [9] Bohn, D., Wolff, A., Wolff, M., and Kusterer, K., 2002, Experimental and Numerical Investigation of a Steam-Cooled Vane, ASME Paper No. GT2002-30210. [10] Campos-Amezcua, A., Mazur-Czerwiec, Z., and Gallegos-Mun oz, A., 2011, Thermomechanical Transient Analysis and Conceptual Optimization of a First Stage Bucket, ASME J. Turbomach., 133, p. 011031. [11] Gatski, T. B., Hussaini, M. Y., and Lumley, J. L., 1996, Simulation and Modeling of Turbulent Flows, Oxford University Press, New York. [12] Celick, I. B., Ghia, U., Roache, P. J., and Christopher, R., 2008, Procedure for Estimation and Reporting of Uncertainty Due to Discretization in CFD Applications, ASME J. Fluids Eng., 130, p. 078001. [13] Ansys Inc., 2008, ANSYS Users Guide, Ansys Inc., Ann Arbor, MI. [14] Sierra, F. Z., Narzary, D., Bolaina, C., Han, J. C., Kubiak, J., and Nebradt, J., 2009, Heat Transfer and Thermal Mechanical Stress Distributions in Gas Turbine Blades, ASME Paper No. GT2009-59194. [15] Teloxa, J., Carrillo, F., Bolaina, C., Varela, C., and Sierra, F. Z., 2013, Experimental and Numerical Study of Gap Size and Cooling Flow Rate Effects on the Tip Flow of Gas Turbine Blades, Int. J. Turbo Jet Engines, 30(1), pp. 111122. [16] Han, J. C., Dutta, S., and Ekkad, S. V., 2000, Gas Turbine Heat Transfer and Cooling Technology, Taylor & Francis, New York. [17] Mase, G., and Mase, T., 1992, Continuum Mechanics for Engineers, CRS Press, Boca Raton, FL. [18] Lai, M., Rubin, D., and Krempl, E., 1978, Introduction to Continuum Mechanics, Pergamon Inc, Oxford, UK. [19] Bolaina, C., 2010, Analysis of the Thermo-Mechanical Stress Distributions in a Gas Turbine Blade (in Spanish), Ph.D. thesis, Universidad Auto noma del Estado de Morelos, Cuernavaca, Mexico. [20] ASM International, 1990, ASM Handbook Vol. 1. Properties and Selection: Irons Steels and High Performance Alloys, ASM International, Novelty, OH.

Conclusions
Coolant ow rate in a rst stage gas turbine blade was varied, keeping the mainstream ow rate constant. Temperature sensitive paint and particle image velocimetry laboratory data were used to validate numerical predictions, which were extended for real engine operating conditions. Thermomechanical stress strain was calculated numerically and discussed as a function of cooling ow rate. Results indicate that high temperature concentration mainly affects the leading edge, with the endwall region being affected by large stress concentration under normal rotation rate. Cooling Reynolds number Rec variation effects on stress generation indicate that stress magnitude increases for decreasing cooling ows. It is shown that stress magnitude increases for low Rec, compared with the high Rec studied. Results show that the stress increment obtained with low Rec reduces as the number of blocked cooling channels augments. However, the nine blocked cooling channels condition presents minimal variations for all Rec under study. The maximum effect of cooling ow rate on stress magnitude is observed for the nonblocking condition, while blocking six cooling channels from a total of 16 produce a 3.4% stress increment; nine blocked channels leads to a 6% stress increment.

Acknowledgment
Thanks to the Comisio n Federal de Electricidad for nancial support through CONACYT, under Grant No. CFE-2004-C01-6. The rst author thanks PROMEP-SEP, for the founding assistantship during his Ph.D. studies at UAEM.

Nomenclature
C D keff K L blade chord, 0.185 m effective cooling hole diameter, m effective thermal conductivity of mainstream air thermal conductivity of blade material, 15.6 W/m K total blade length, m

Journal of Turbomachinery

NOVEMBER 2013, Vol. 135 / 064501-9

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/ on 10/18/2013 Terms of Use: http://asme.org/terms

You might also like