Applied Thermal Engineering 51 (2013) 953e962

Contents lists available at SciVerse ScienceDirect

Applied Thermal Engineering

Numerical evaluation of the blade cooling for the supercritical steam turbine
W1odzimierz Wrblewski*
Institute of Power Engineering and Turbomachinery, Silesian University of Technology, Konarskiego 18, 44-100 Gliwice, Poland

h i g h l i g h t s
< Numerical solution of the CHT problem for the blade channel of supercritical steam turbine. < The determination of the potential possibilities of using the martensitic steel for the rst blades. < Different cooling steam parameters, different turbulence levels of the main ow and the thermal barrier are examined. < The obtained results determine the thermal potential of the applied method of the blade cooling.

a r t i c l e i n f o
Article history: Received 11 June 2012 Accepted 26 October 2012 Available online 7 November 2012 Keywords: Steam turbine Blade cooling Conjugate heat transfer

a b s t r a c t
Steam turbine systems for supercritical parameters are one of the most important directions of the development of conventional power plants. A steam turbine working in such cycles has to be adjusted to operation with higher and higher steam temperatures. The aim of this analysis is to determine the cooling conditions of the blades of the steam turbine rst stage. The cooling is supposed to ensure that the same materials as those used for the live steam temperature of 873 K are used for the rst blades operating with the live steam temperature of 923 K. In the calculations the Conjugate Heat Transfer (CHT) model is used to simulate the following phenomena: the uid ow in the blade-to-blade cascade, the heat transfer in the blade and the coolant ow in the holes. Analyses of different cooling steam parameters and of different turbulence levels of the main ow are performed. The problem is analysed for the steam conditions corresponding to the rst blade of High Pressure (HP) and Intermediate Pressure (IP) turbines, respectively. For the HP turbine the on-blade thermal barrier is considered as an additional mean to reduce the metal temperature to the limit required. 2012 Elsevier Ltd. All rights reserved.

1. Introduction The development of modern power engineering technologies using gas and coal fuels is motivated mainly by the increase in the power generation efciency and a reduction in the environmental impact of such systems. One of the directions to achieve these objectives is the rise in the working medium temperature at the thermal turbine inlet. This concerns both the gas and steam turbine. Steam turbine systems for supercritical parameters are one of the most important directions of the development of conventional power plants. Due to such technologies, it is possible to obtain the efciency of electricity generation exceeding 50% [1]. Therefore, the steam turbine working in cycles with the steam supercritical pressure will have to face higher and higher thermal loads. It is

* Tel.: 48 32 2371971; fax: 48 32 2372680. E-mail address: [email protected]. 1359-4311/$ e see front matter 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.applthermaleng.2012.10.048

anticipated that the steam temperature in the planned supercritical power units will be by 50e100 K higher than in supercritical plants already in service. Achieving the respective temperature and pressure values exceeding 900 K and 30 MPa at the inlet calls for particular attention paid to the turbine nodes operating in these conditions. Designing them, the choice has to be made between using a new material which is more resistant to thermal loads but expensive, or introducing thermal screens and organising local cooling in the ow system. The former will require a wider application of materials made of nickel alloys. The latter might allow the use of materials which are currently used in facilities with lower steam parameters. The inuence of thermal screens and external cooling on the rotor thermal load in the live steam inlet area was investigated by Kosman [2]. Design solutions specic to supercritical steam turbines make it possible to keep the thermal load and stresses in the turbine components at a reasonable level [2]. The application of convective cooling to the blades of the rst stage of the HP and IP parts of the steam turbine for supercritical steam

954

W. Wrblewski / Applied Thermal Engineering 51 (2013) 953e962

List of symbols cp c cx H h k M p Re s t/c specic heat capacity at constant pressure, J/(kg K) chord, m axial distance of chord, m specic enthalpy, J/kg heat transfer coefcient, W/(m2 K) turbulence kinetic energy, J/kg Mach number static pressure, Pa Reynolds number curvilinear coordinate from the leading edge along the prole curve, m pitch to chord ratio

T y x

static temperature, K dimensionless distance axial coordinate, m

Greek symbols l thermal conductivity, W/(mK) r density, kg/m3 Indices 0 1 2 c s

total parameters inlet outlet coolant isentropic

parameters is an additional option which may be considered for new designs. Only convection cooling should be taken into account for this case due to the parameters of steam which are available in the steam cycle and which may be used as the cooling agent. In the search for solutions to problems related to the cooling of steam turbine blades, the experience gained from research conducted on the cooling of gas turbines should be employed. Tracing back the evolution of cooling techniques in these turbines, it turns out that at rst convection cooling solutions were used. More effective cooling systems were introduced later, such as lm cooling or transpiration cooling. Dunn [3] analyses a wide spectrum of problems related to the blade prole aerodynamics and the heat transfer in various cooling methods applied in gas turbines. To protect the gas turbine blade thermally, an on-blade thermal barrier is used. The assessment of this solution for a selected gas turbine was carried out in [4] using advanced numerical models. This type of solution may also be considered as an option for steam turbine blades. In problems related to convection cooling of blades, the determination of the ow eld parameters is the basis for the calculation of the heat transfer in the turbine components. Due to the very complex ow structure and the high sensitivity of the heat transfer conditions to changes in this structure, the accuracy of the heat transfer coefcient calculations in the range of 10% is considered as good. In more complex areas, the differences between calculation results and measured values may reach 30% and more. The progress in the development of computational methods in the last dozens of years has allowed a transition from correlation relationships on a at plate, through the use of algorithms with coupled solutions for the boundary layer and the main ow for complex geometries, to methods based on solving NaviereStokes equations. Despite that, taking account of all the ow modelling phenomena is still a computational challenge. It is a difcult task because in such problems the weak points of the turbulence and the laminareturbulent transition models become more visible. Many heat transfer problems in the turbine are analysed assuming steady conditions. And this happens despite the fact that the blade operating conditions are highly unsteady and the blade is subject to the impact of the blade wake. Taking these phenomena into account calls for a much longer computational time. There is some hope for a more precise description of the ow phenomena in the use of the Direct Numerical Simulation (DNS) methods, with no application of procedures of averaging ow values. However, these are not the methods that could be used for complex geometrical systems in the very near future for the uid ow problems, especially in the coupled problems with the heat transfer in solids. A detailed determination of ow structures is much more signicant in the case of lm cooling. For these problems, the

conjugate algorithm methods are extended with the Large Eddy Simulation (LES) method, which however is much more demanding towards the numerical mesh than the methods based on the Reynolds-averaged NaviereStokes (RANS) equations and which requires much longer computation times [5,6]. Coupling the LES method with the heat transfer algorithm for a gas turbine blade with lm cooling is presented in [6]. However, the results of the performed simulations differ considerably in terms of quantity from the experimental data obtained from the RANS computations. Turbulence models thus remain, for the time being, the only effective way for most analyses of technical issues. The modelling methods which are most often used to model turbulence in uid ow and heat transfer problems in the blade-to-blade cascade are the two-equation models of eddy-viscosity. These models are a compromise between complexity and accuracy. Their usefulness in terms of the heat transfer coefcient determination was studied among others in [7] and [8]. In [8] comparative studies of the v2f model were conducted in conjugate computations for a blade with convective cooling. In [9] a comparative analysis is presented to study the ability of three turbulence models to predict lm cooling and to assess the inuence of the grid type on the solution. The results conrm that conjugate heat transfer models predict a signicant difference in the temperature predictions in comparison with adiabatic models. The realizable ke3 , and the shear stress transport keu (SST) turbulence models present a satisfactory agreement with experimental data. The interaction between the ow eld and temperature eld in the cooled blade makes it necessary to combine the ow problem with issues related to the heat transfer in a solid. The Conjugate Heat Transfer (CHT) calculations have their specicity and impose higher requirements, especially on the ow part calculations. This concerns, among others, the numerical mesh parameters near the wall to appropriately determine the temperature prole of the boundary layer. The publications presenting the solution to the Conjugate Heat Transfer problem for the turbine blade usually ignore the laminare turbulent transition (e.g. [10e12]). The variability of the transition models proves how difcult it is to model this phenomenon. A comparative study of bypass transition models was presented in [13]. The models were validated against experimental data from different test cases including the turbine channel ow. The model which has gained much popularity in engineering applications is the Gammaetheta model (e.g. [14,15]). The reason for this is its universality and availability in the ANSYS CFX software package. Using transition models in the CHT calculations makes it possible to obtain better results, both quantitatively and qualitatively. The gas turbine blade convection cooling technology may be used for supercritical steam turbines. The modelling of the ow

W. Wrblewski / Applied Thermal Engineering 51 (2013) 953e962

955

phenomena in such cases is more difcult due to the water vapour parameters. Taking account of the specicity of the working and cooling uids by employing the appropriate real gas state equation complicates nding the solution to the conjugate task and lengthens the time needed for the computations. The use of the CHT algorithm to analyse the gas turbine cooling with water vapour is presented in [16]. The use of the CHT algorithm for a closed steam cooling system in a state-of-the-art steam turbine is described in [10]. The works mentioned above use the perfect gas model with various adjustments. This paper formulates and solves the conjugate heat transfer simulation task for convectively cooled blades of the steam turbine. The aim of the analysis of the blade cooling conditions performed for a selected conguration of cooling holes is to provide a qualitative and quantitative assessment of the cooling conditions of the rst rings of the HP and IP parts of the turbine. Due to the water vapour properties, it is essential to determine the main ow eld parameters in detail. In this respect, the real gas state equation is used in the calculations and the impact of the turbulence level at the blade channel inlet on the heat transfer conditions is investigated. The calculations take account of the laminareturbulent transition model. Also, the quantitative impact of the change in the boundary layer character on temperature distributions on the blade is indicated. For the blade cooling purposes, steam with various parameters (pressure and temperature) which may be achieved at different points of the steam plant cycle is assumed. The change in the cooling agent velocity is also taken into consideration. Additionally, the effect of the application of an on-blade thermal barrier is analysed and the impact of the barrier on the thermal conditions is assessed. The obtained results allow an assessment of the possibility of applying the blade convective cooling for conditions in a supercritical steam turbine. The conclusions derived from the computations will make it possible to formulate the blade geometry optimisation task with the use of genetic algorithms according to the method described in [17]. The optimisation algorithm used to analyse the blade cooling conguration of the steam turbine in [18] will be extended using the CHT model. 2. General model description In order to determine the temperature eld of a blade washed by water vapour with high temperature and cooled convectively by water vapour owing in the blade passages, it is necessary to: - formulate the conjugate boundary value problem for the steam ow around the blade together with conductive heat transfer to the blade material and for the cooling steam ow in each of the passages separately, - to dene and discretise the computational area, - to include in the analysis the laminareturbulent transition conditions for the external ow, - to use the real gas model. The mathematical model that describes the uid ow and the heat transfer is composed of: - the the - the - the - the - the - the mass, momentum and energy conservation equations for uid, turbulence model, laminareturbulent transition model equations, energy conservation equation for a solid, gas state equation, relationships describing the material properties.

In all cases under analysis, the SST version of the keu turbulence model is used. The SST turbulence model was modied using Katoe Launder formulation and curvature correction which are recommended for such a problem. For the ow in the blade channel the laminareturbulent transition is modelled using the Gammaetheta model [15]. This model is related to the SST turbulence model and consists two additional transport equations for the intermittency and transition onset Reynolds number. The governing equations are solved using the Ansys CFX software package procedures. The basic equation closing the system of the conservation equations is the gas state equation. For the validation performed for gas turbine the ideal gas model is used. For water vapour it is necessary to use the real gas state equation. As a result, many relationships between thermal and caloric parameters have a non-explicit mathematical form. This requires many additional iterative procedures that have to be used in the solving algorithms, which lengthens the time of computations. The steam state equations recommended by the IAPWS (International Association for the Properties of Water and Steam) according to standard IAPWS-IF97 are used in the calculations. Detailed formulae describing steam parameters may be found in [19]. In computational programs such as Ansys-CFX, procedures are used that make tables of steam properties in a specic range of parameters with a pre-set accuracy. In the CHT calculations the conduction equation in the solid domain is included in the solver. The conduction equation is treated as a degenerative energy equation of the uid, where all velocities are set to zero. This equation is solved numerically with the same algorithm as the ow equations with the use a nite volume approach. The CouranteFriedrichseLewy (CFL) time step conditions for solving the transport equations for solid and for uid domains can be adjusted individually. 3. Validation test case: cooling of the gas turbine blade Due to the lack of data for the steam turbine blade, the dened model is veried using experimental data for the gas turbine blade. In this case, the blade is cooled with air, using four simple holes with a circular cross-section (Fig. 1). This conguration was examined on the German Aerospace Center (DLR) test stand [20]. Owing to the announced experimental data, the selected prole is a good instance for validation.

Fig. 1. Mesh for validation calculations of the conjugate task for the blade with cooling channels description.

956

W. Wrblewski / Applied Thermal Engineering 51 (2013) 953e962

The basic turbine blade prole parameters are: the chord c 0.0834 m, the pitch to chord ratio t/c 0.71, the blade-angle setting in the cascade b 50 . The test stand with a cascade of proles is fed with gas with total temperature T0 794.5 K. The total pressure at the cascade inlet is p0 0.15376 MPa, and the static pressure at the outlet is p2 0.10217 MPa. These conditions correspond to the Mach number M1 0.328 at the cascade inlet. The inlet angle of the mass ow measured in the circumferential direction is a1 57. The ow uid used in the experiment is the gas obtained from the combustion of aircraft kerosene (C14H30). The physical properties of the aircraft kerosene exhaust gases are very similar to those of air, which results from the very high value of the air excess factor during combustion. Therefore, it may be assumed that the exhaust gases are replaced in the calculations with air treated as perfect gas. For air, the dependence of specic heat on temperature is taken into consideration. The blade under analysis is convectively cooled with air owing through four holes. According to the data given by Heselhaus [21], in the holes through which the cooling air ows, the boundary conditions are assumed based on the values given in Table 1. In the case under consideration, the cooling conditions are assumed as constant, which results from the fact that the heat transfer coefcient and the cooling air temperature values are assumed as constant, both on the hole perimeter and along the height of the passage. For the blade material, according to the literature data, the following constant values of material data are assumed: density: 7900 kg/m3, specic heat: 585.2 J/kg K, conductive heat transfer coefcient: 19 W/mK. For the numerical calculations, the mesh (Fig. 1), containing 582,660 nodes in the ow area and 72,072 nodes in the blade metal area is assumed. In the boundary layer for the ow conditions under analysis the value of y is smaller than 1. For the turbine blade under consideration, the SST turbulence model is used, and the laminareturbulent transition is modelled with the Gammaetheta model. The temperature distribution on the blade surface obtained in the experiment is used as validation data for the performed calculations. Fig. 2 presents a comparison of the calculated temperature distribution on the blade surface with experimental data. It can be seen that the temperature differences between the distributions on the pressure side and in the rst part of the suction side ranged from 15 to 30 K but the temperature differences in the second part of the pressure side ranged from 55 to 2 K. Additionally, the heat ux distribution on the blade is presented. On the suction side for s/ c w 0.65, an increase in the heat ux value is clearly visible. This rise corresponds to the location of the laminareturbulent transition. The curvilinear coordinate s is in this case dened from the leading edge along the prole curve. It may be assumed that the distributions obtained near the leading and the trailing edge feature a good concordance with the results of the experiment. Bigger differences appear on the suction side. They may result from the fact that the Gammaetheta model dened the beginning of the

Fig. 2. Distributions of the temperature and the heat transfer on the blade surface for validation calculations.

laminareturbulent transition too early. The laminareturbulent transition takes place on the suction side only. On the pressure side the ow is laminar. 4. Model description for the steam turbine blade 4.1. Geometrical data Due to the limitations on the availability of explicit experimental and geometrical data for the real steam turbine, the task is analysed based on the blade prole assumed for validation, adjusting its geometry to the size which is typical of the steam turbine rst stage blade. The 0.6 scale is assumed for the prole. In each case the blade is analysed as a cylindrical one, with a constant cross-section in the radial direction. The aim of the analysis conducted in this paper is to indicate the potential possibilities of the steam turbine blade cooling and to emphasise the essential problems related to this solution. The simplications employed at this level of thoroughness should be considered as sufcient. This is also justied by the relatively small Mach number and the slight change in the steam temperature in the blade channel, which is presented below. The computational area of the CHT task is divided into the following (Fig. 3): the steam main ow area (the blade channel), the blade area (metal) and the areas of the cooling passages (passages

Table 1 Boundary conditions in the cooling holes. Hole number 1 2 3 4 Nu 101.7 99.5 87.1 98.0 Coolant average temperature Tav, K 343.1 345.4 343.8 341.1 Re 1.58$104 1.54$104 1.83$104 2.10$104 Pr 0.715 0.715 0.715 0.715 Heat transfer coefcient h, W/m2 K 278.79 273.86 287.01 375.48 Fig. 3. Domains for the steam turbine blade cooling CHT task.

W. Wrblewski / Applied Thermal Engineering 51 (2013) 953e962

957

in the blade and their extension), in which the heat transfer problem is formulated separately. The geometrical data for the blade are dened in the AutoCAD program. The blade system geometry prepared in this way may be directly used in programs generating a computational mesh. The meshes are generated using the ANSYS ICEM program. The hexa dominant type of mesh is employed in computational areas, assuming structural meshes with clearly distinguished hexahedron layers and elements near the walls. The meshes in the main ow area are constructed according to the recommendations for the appropriate boundary area modelling, to satisfy the needs of both the laminareturbulent transition and the heat transfer task. Meshes with the value of y smaller than 1, with the mesh increment rate in the range of 1.1e1.2 together with 20e25 mesh lines in the layer area are assumed. Although a developed turbulent ow is assumed in the cooling passages, due to the requirements of the temperature prole, computational meshes with similar parameters are used in the boundary layer. An extension is added to the cooling passages at the inlet to allow the development of the turbulent ow before the blade passage, where the heat transfer with the solid domain of the blade takes place. This makes it possible to improve the accuracy of the results in the lower part of the blade. In order to reduce the task, the top and bottom surface of the ow area, as well as the blade metal surface are assumed as surfaces of symmetry. This makes it possible to eliminate the boundary layer on these areas in the ow area. The calculations are performed using the Ansys-CFX software package. The geometry with a conguration of ve holes: four noncircular ones and one circular e GEO2 is considered. 4.2. Selected physical properties of the blade material The selection of the assumed physical properties of the blade material, apart from the correct determination of the uid thermodynamic properties, is of fundamental importance in the calculations of the ow eld with heat transfer. In the conjugate heat transfer task under consideration it is assumed that the blades are made of martensitic steel. The material data for this type of steel are assumed based on the average values given for steel used for turbine construction and designed for operation in the analysed temperature range [22]. The martensitic steel density is assumed as r 7850 kg/m3, and the specic heat is dened by the dependence:

Table 2 List of variants for the HP turbine blade. No. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. Variant WP_GEO2_50_310_5 WP_GEO2_50_327_5 WP_GEO2_64_330_5 WP_GEO2_64_390_5 WP_GEO2_95_360_5 WP_GEO2_95_450_5 WP_GEO2_95_360_5_Ma075 WP_GEO2_95_360_5_Tu01 WP_GEO2_95_360_5_Tu02 WP_GEO2_95_360_5_Tu05 WP_GEO2_95_360_5_Tu12 WP_GEO2_95_315_5_Tu12_Ma070 T0c,  C 310 327 330 390 360 450 360 360 360 360 360 315 p0c, bar 50 50 64 64 95 95 95 95 95 95 95 95 p2c, bar 42 42 54.2 54.2 80.1 80.5 65.6 80.1 80.1 80.1 80.1 80.1

c 0:46 0:000177T 4:67e 7T 2 ;

J=kg K

(1)

assumption is made for the inlet cross-section. The average total parameters in the cross-section are constant. Different values of total pressure and total temperature at the inlet are assumed for the cooling passages in order to determine the potential of the used steam for the purposes of blade cooling. At the inlet, static pressure values are selected as they ensure moderate velocity values in the cooling passage of approximately M 0.5. The impact of the steam velocity in the cooling passage is analysed below. Table 2 lists all the variants of the cooling passage geometry and of the cooling steam parameters for the conditions corresponding to the rst blade row of the HP turbine. The following marking convention is adopted for the variants: WP_GEO2_yy_zzz_n_aaaawhere WP stands for the HP turbine, GEO2 is the geometry symbol, yy is the total pressure at the cooling passage inlet (bar), zzz is the total temperature at the cooling passage inlet ( C), n is the number of holes and aaaa provides additional information concerning ow conditions. For the rst blade row of the IP part, the following total parameters are assumed at the inlet: pressure p0 5.9 MPa and temperature t0 670  C, which correspond to the parameters of a supercritical thermal cycle. At the outlet, the selected static pressure value is p1 5.695 MPa and it is assumed that this is the average value in the outlet cross-section. This results from the assumption that the isentropic drop in enthalpy at the stator is the same as for the HP turbine DHs 15 kJ/kg. In the inlet crosssection, the total parameters do not change along the blade height. Table 3 lists all the variants of the cooling steam parameters for the conditions corresponding to the rst blade row of the IP turbine. SP stands for the intermediate pressure part and the other elements of the marking are the same as those for the HP turbine. 5. Calculation results for the HP turbine blade The distributions of the main ow parameters in the blade-toblade channel of HP turbine do not vary signicantly during the testing. There are bigger changes only in the turbulence level resulting from changes in the boundary conditions. The obtained maximum values of the Mach number are about 0.3, which indicates a subsonic ow with relatively small velocities. Assuming that the turbulence level is 1%, at the passage inlet the laminareturbulent transition is located quite early on the suction
Table 3 List of variants calculated for the IP turbine blade. No. 1. 2. 3. Variant SP_GEO2_95_360_5 SP_GEO2_50_327_5 SP_GEO2_50_327_5_Tu12 T0c,  C 360 327 327 p0c, bar 95 50 50 p2c, bar 80.1 42 42

The heat conductivity coefcient is approximated with the function:

l 0:01801T 23:88;

W=mK

(2)

4.3. Formulation of boundary conditions The conditions for the blade channel ow domain in the HP turbine row are selected assuming that the isentropic drop in enthalpy in the stator is DHs 15 kJ/kg. This results from the distribution of the enthalpy drop in the HP turbine into 18 stages and from the assumption that a stage reactivity is at the level of 0.4. Taking account of the following steam parameters at the turbine inlet: total pressure: p0 30 MPa and total temperature t0 650  C, the value of static pressure after the blade row is assumed as p1 28.82 MPa. It is assumed in the calculations that the pressure value in the outlet cross-section is an average value. A similar

958

W. Wrblewski / Applied Thermal Engineering 51 (2013) 953e962

Fig. 4. The turbulence level Tu contour (no cooling).

side of the blade (for x/cx about 0.2) (Fig. 4). Also on the pressure side of the blade, the laminareturbulent transition is observed in the front part of the blade. The character of the steam ow for the assumed parameters differs from what is obtained from the validation calculations described in Section 3. Fig. 5 presents the contours of the steam temperature distribution in the blade channel during expansion. A slight difference in temperature can be noticed, which is included in the range of 6 K. This points to very similar thermal conditions on the whole blade perimeter, which translates into strict requirements concerning the blade cooling organisation in the entire cross-section. After preliminary calculations, a conguration of 5 cooling passages with an irregular cross-section is assumed for a further analysis. The conguration of passages and the computational mesh are presented in Fig. 6. The passages are placed in a relatively uniform manner along the prole to achieve an appropriate cooling effect in the entire cross-section. Passage 1 is moved closer to the prole leading edge to ensure more intense cooling in this area. The circular passage is located as close to the trailing edge as possible to limit the temperature rise in the trailing edge area. The numerical mesh has 630,318 nodes in the main ow area, 488,649 nodes for the ve passages of the cooling steam area and 66,429 nodes in the blade metal area. 5.1. The impact of the cooling steam parameters Six variants of parameters of cooling steam with high pressure are analysed. The steam pressure is selected based on a power plant cycle for steam parameters of 30 MPa and 650  C. The selected variants are those with possibly highest pressure and possibly low temperature values. The temperature selection is limited by the need to guarantee superheated steam in the cooling passages, i.e. to eliminate the possibility of steam condensation. The variants under analysis are listed in Table 4. The blade temperature contour for the selected WP_GEO2_95_ 360_5 variant is presented in Fig. 7. The chart in Fig. 8 compares the temperatures on the wall blade in its tip cross-section. The differences between the maximum temperatures are not big and amount to several kelvins. The laminareturbulent transition occurs in each case at the same place, but for variants where the lowest temperatures are achieved it is observed that the temperature curve after the laminareturbulent transition, despite the occurrence of the turbulent boundary layer, is lowered further at a certain section of

Fig. 5. Distribution of the steam temperature in the blade-to-blade channel (no cooling).

Fig. 6. Cooling passages geometry and numerical mesh for the GEO2 conguration.

W. Wrblewski / Applied Thermal Engineering 51 (2013) 953e962 Table 4 The set of cooling steam parameters for the GEO2 passage conguration. No. 1 2 3 4 5 6 Variant WP_GEO2_50_310_5 WP_GEO2_50_327_5 WP_GEO2_64_330_5 WP_GEO2_64_390_5 WP_GEO2_95_360_5 WP_GEO2_95_450_5 T0c,  C 310 327 330 390 360 450 p0c, bar 50 50 64 64 95 95 p2c, bar 42 42 54.2 54.2 80.1 80.5

959

the contour. This is evidence of intense cooling in this area of the blade. The steam parameters of 95 bar and 360  C turn out to be the most favourable ones in terms of the obtained maximum blade temperature. For these cooling steam parameters the maximum blade surface temperature of 880 K is obtained. 5.2. The impact of the cooling steam velocity Assuming an increased steam ow in the cooling passages, calculations are performed for the cooling steam parameters: p0c 95 bar, T0c 360  C, selected in the previous analysis. The static pressure at the outlet is lowered from p1c 80.1 bar to the value of p1c 65.6 bar. The coolant ow is faster, reaching the Mach number values of 0.6e0.75, depending on the heat transfer in individual passages. Fig. 9 compares the distribution curves of the temperature on the blade surface for different conditions of the cooling steam ow in the passages. An increase in the mass ow resulted in a slight reduction in the surface temperature by a few kelvins only. In the trailing edge area, the obtained drop in temperature is about 7 K. The maximum temperature on the blade surface is on the suction side with the coordinate of about x/cx 0.55. In terms of improvement in the cooling conditions, the effect of a substantial rise in the coolant mass ow velocity is not too big. 5.3. The impact of the main ow turbulence level In the next step, the impact of the main ow turbulence level on the blade cooling conditions is studied. The calculations
Fig. 8. Distribution of temperature on the blade prole for different parameters of the coolant.

are performed for the following cooling steam parameters: p0c 95 bar, T0c 360  C and p1c 80.1 bar. The basic case, in which the turbulence level before the passage is about 1%, and cases with the turbulence level at the inlet of 2%, 5% and 12% are also considered are shown in Fig. 10. For the turbulence level of 12% the temperature is higher in the front part of the blade near the edge. The temperature in the remaining fragments of the cross-section changes only slightly. Increasing turbulence to just 2% results in a shift in the laminare turbulent transition towards the blade leading edge and a rise in the minimum temperature by approximately 30 K. For higher turbulence level values before the blade, the boundary layer is turbulent on the whole perimeter of the blade. This causes a rise in temperature on the leading edge by 20e25 K. The temperature of the blade contour in the trailing part is only slightly higher, by a few kelvins. Fig. 11 compares the heat transfer coefcient distribution on the blade surface for the variants with the turbulence level of Tu 1%

Fig. 7. Distribution of the blade temperature (variant WP_GEO2_95_360_5).

960

W. Wrblewski / Applied Thermal Engineering 51 (2013) 953e962

Fig. 9. Distribution of temperature on the blade prole for different mass ow values of the coolant.

Fig. 12. Temperature distributions WP_GEO2_95_360_Tu12_coating.

for

variants

WP_GEO2_95_360_Tu12

and

between the coefcient distributions is signicant and reaches 30,000 W/m2 K. 5.4. The impact of the thermal barrier As presented above, the cooling of the HP turbine rst blade is not an easy task. The cooling has to be rather intense in the entire cross-section, and the steam and metal properties cause substantial temperature gradients in the blade. Therefore, the application of the thermal barrier on the blade surface is considered. One of the methods of blade protection used in gas turbines is a thermal barrier in the form of a ZrO2 coating. Zirconium dioxide, as a ceramic material, features a low heat transfer coefcient of 0.8e3 W/(mK). It is assumed for a comparative analysis that on the blade surface there is a zirconium dioxide coating which creates heat transfer resistance
Fig. 10. Distribution of temperature on the blade prole for different turbulence levels.

R d=l 5e 5
and Tu 12%. The coefcient is determined based on the heat ux in the blade surface at the assumed value of the uid temperature of Tbulk 923 K. The heat transfer coefcient for steam is very high; for the Tu 12% variant it reaches values of approximately 65,000 W/m2 K in the leading edge area. The difference

m2 K

. W;

(3)

where d is the coating thickness. The calculations are performed for the variant WP_GEO2_95_360_Tu12. A much better temperature equalizing is visible than in the case of the variants analysed previously. The metal maximum temperature is substantially lower. This can be seen in Fig. 12, which compares the blade surface

Fig. 11. The heat transfer coefcient distribution on the blade surface (variants WP_GEO2_95_360_5, WP_GEO2_95_360_Tu12).

Fig. 13. Distribution of temperature on the blade prole for different ow parameters (IP turbine).

W. Wrblewski / Applied Thermal Engineering 51 (2013) 953e962

961

temperature distributions in the tip area for the variants with and without the thermal barrier. The use of a ceramic coating causes a drop in the blade surface temperature by 60e80 K. The obtained temperature level after the application of the thermal barrier for the assumed cooling parameters turns out to be lower than required. Consequently, it is possible to limit the blade cooling by: reducing the cooling steam mass ow, raising the cooling steam temperature or changing the thermal barrier parameters. The cooling steam temperature of 733 K makes it possible to obtain maximum temperatures lower than 870 K. 6. Calculation results for the IP turbine blade The calculations for the IP turbine blade are conducted for the GEO2 geometrical conguration of the cooling passages. The IP turbine blade is about four times higher than the HP turbine blade. The boundary conditions for the calculations are presented in Section 4, and the cooling steam parameters are listed in Table 3. The assumed drop in enthalpy in the stage is the same as the enthalpy drop in the HP turbine stage. Because of that, the distributions of relative ow parameters in the blade channel are very similar. The change in the pressure value is included in the range of 0.3 MPa, and the maximum Mach number in the channel reaches a moderate value of 0.29. Three variants of the ow conditions analysed for the IP part blade are selected as characteristic after the testing conducted for the HP turbine. Consequently, in the rst variant the comparative conditions for the cooling steam assumed for the HP turbine, namely p0c 50 bar, T0c 327  C and the turbulence level Tu 1%, are considered. The next analysed variant is the one with p0c 95 bar, T0c 360  C and Tu 1%, and then the rst variant with the turbulence level at the inlet increased to 12%. Quantitatively changes in temperature can be examined in more detail in Fig. 13. For ows in which the laminareturbulent transition occurs on the blade, the blade surface reaches the lowest temperature values at the level of 710 K. The part of the blade with a laminar ow is cooled down very intensely. In the trailing edge area, the maximum temperature reaches 835 K. Such temperature values are lower than 873 K e the value assumed as permissible for metals. It can be seen that in all these cases the blade material temperature is relatively low. So, it is much easier to cool the IP turbine blade than the HP turbine one. Therefore, the appropriate level of maximum temperature may be obtained reducing the cooling steam mass ow. In the temperature distributions locations can easily be noticed where the laminareturbulent transition occurs: x/cx 0.1 for the pressure side and x/cx 0.6 for the suction side. For the variant with a high level of turbulence, the blade surface temperature is higher and reaches the maximum value of 851 K on the leading edge of the prole. 7. Conclusions The subject matter of this analysis is the determination of the potential possibilities of using the same materials for the rst blades of a steam turbine with the live steam temperature of 923 K as those used for the live steam temperature of 873 K. The employed computational conjugate heat transfer simulation model for the steam ow in the blade channel, of the heat transfer in the blade and of the cooling steam ow in the passages allows a comprehensive assessment of the phenomena which occur during the heat transfer. Analyses of different cooling steam parameters and of different turbulence levels of the main ow are performed. The presented results of the analyses indicate that the steam expansion conditions in the rst blade row cause that the blade operates in very balanced

thermal conditions on the entire surface. The cooling of the blade of the HP turbine using 5 cooling passages is a complex task due to the relatively large uxes of transferred heat and to the material properties of water vapour and metal: very high heat transfer coefcients for steam and a high heat conductivity coefcient for metal. It is shown that it is possible, from the thermal point of view, to select such geometry of passages and such steam parameters that the temperature in the blade is kept below 873 K. The most critical points that need particular attention in the blade cooling are the leading edge and the trailing edge. The cooling of the IP part blades is a much easier task compared to the cooling of the HP turbine blades. The required cooling effect may be achieved with more moderate parameters of the cooling steam. This allows a much easier matching of the cooling conditions to the temperature level, avoiding signicant temperature gradients. The character of the ow in the blade channel, the location of the laminareturbulent transition and the turbulence level all have a substantial impact on the heat transfer conditions. The obtained results determine the thermal potential of the applied method of the blade cooling. However, due to signicant temperature gradients, they require a strength verication. In order to reduce the temperature gradients, solutions should be found to supplement the cooling, such as a replacement of the material with one that features a reduced heat conductivity coefcient (in the calculations the average value from the different material data is assumed) or an introduction of the thermal barrier at the places most exposed to high temperatures, i.e. in the leading and trailing edge areas. The results of the analysis can be use to formulate assumptions and boundary conditions for the blade geometry optimisation task. Acknowledgements The results presented in this paper were obtained from research work co-nanced by the Polish National Centre of Research and Development in the framework of Contract SP/E/1/67484/10 e Strategic Research Programme e Advanced technologies for obtaining energy: Development of a technology for highly efcient zero-emission coal-red power units integrated with CO2 capture. References
[1] B.G. Miller, Clean Cowal Engineering Technology, Elsevier Ltd, Oxford United Kingdom, Butterworth Heinemann, 2010. [2] W. Kosman, Thermal analysis of cooled supercritical steam turbine components, Energy 35 (2010) 1181e1187. [3] M.G. Dunn, Convective heat transfer and aerodynamics in axial ow turbines, transactions of the ASME, Journal of Turbomachinery 123 (2001) 637e686. [4] A.H. Rossette, Z. Mazur, A. Demeulenaere, J.A. Roque Lpez Hernndez, The effect of start-up cycle in ceramic coating used as thermal barrier for a gas turbine bucket, Applied Thermal Engineering 29 (2009) 3056e3065. [5] V.K. Garg, Heat transfer research on gas turbine airfoils at NASA GRC, International Journal of Heat and Fluid Flow 23 (2002) 109e136. [6] F. Duchaine, A. Corpron, L. Pons, V. Moureau, F. Nicoud, T. Poinsot, Development and assessment of a coupled strategy for conjugate heat transfer with Large Eddy Simulation: application to a cooled turbine blade, International Journal of Heat and Fluid Flow 30 (2009) 1129e1141. [7] V.K. Garg, A.A. Ameri, Two-equation turbulence models for prediction of heat transfer on a transonic turbine blade, International Journal of Heat and Fluid Flow 22 (2001) 593e602. [8] J. Luo, E.H. Razinsky, Conjugate heat transfer analysis of a cooled turbine vane using the V2F turbulence model, transactions of the ASME, Journal of Turbomachinery 129 (2007) 773e781. [9] M. Silieti, A.J. Kassab, E. Divo, Film cooling effectiveness: comparison of adiabatic and conjugate heat transfer CFD models, International Journal of Thermal Sciences 48 (2009) 2237e2248. [10] D. Bohn, J. Ren, K. Kusterer, Cooling performance of the steam-cooled vane in a steam turbine cascade, ASME Paper GT2005e68148 3 (2005) 217e226. [11] Y. Okita, S. Yamawaki, Conjugate heat transfer analysis of turbine rotor-stator system, ASME Paper GT2002e30615 3 (2002) 1103e1113.

962

W. Wrblewski / Applied Thermal Engineering 51 (2013) 953e962 [17] G. Nowak, W. Wrblewski, Cooling system optimisation of turbine guide vane, Applied Thermal Engineering 29 (2009) 567e572. [18] G. Nowak, W. Wrblewski, I. Nowak, Convective cooling optimization of a blade for a supercritical steam turbine, International Journal of Heat and Mass Transfer 55 (2012) 4511e4520. [19] W. Wagner, J.R. Cooper, A. Dittmann, J. Kijima, H.-J. Kretzschmar, A. Kruse, R. Mares, K. Oguchi, H. Sato, I. Stoecker, O. Sifner, Y. Takaishi, I. Tanishita, J. Truebenbach, Th. Willkommen, The IAPWS industrial formulation 1997 for the thermodynamic properties of water and steam, Journal of Engineering for Gas Turbines and Power 122 (2000) 150e182. [20] M. Elfert, The Research on a Film Cooling on a Surface of a Hot Gas Blade. DLRBericht IB 325-19-86, Cologne, Germany, 1986 (in German). [21] A. Heselhaus, The hybrid method to conjugate solution of hot gas ow and solid temperature based on example cooled turbine blades, PhD thesis, RuhrUniversitt Bochum, Germany, 1997 (in German). [22] G. Kosman, Steel Properties, Personal Communications, Gliwice, Poland, 2010.

[12] E. Divo, E. Steinthorsson, A.J. Kassab, R. Bia1ecki, An iterative BEM/FEM protocol for steady-state multi-dimensional conjugate heat transfer in compressible ows, Engineering Analysis with Boundary Elements 26 (2002) 447e454. [13] L. Cutrone, P. De Paloma, G. Pascazio, M. Napolitano, An evaluation of bypass transition models for turbomachinery ows, International Journal of Heat and Fluid Flow 28 (2007) 161e177. [14] F.R. Menter, T. Esch, S. Kubacki, Transition modelling based on local variables, engineering turbulence modelling and experiments 5, in: W. Rodi, N. Fueyo (Eds.), Proceedings of 5th International Symposium on Engineering Turbulence Modelling and Measurements, Mallorca, 2002, pp. 555e564. [15] F.R. Menter, R.B. Langtry, S.R. Likki, Y.B. Suzen, P.G. Huang, S. Vlker, A correlation based transition model using local variables, part 1 e model formulation, ASME Paper GT2004e53452 4 (2004) 57e67. [16] D. Bohn, A. Wolff, M. Wolff, K. Kusterer, Experimental and numerical investigation of a steam-cooled vane, ASME Paper GT-2002e30210 3 (2002) 477e484.