Spanwise re-stacking techniques in turbo-machinery blades and application in Francis runner

Transcription

1 Spanwise re-stacking techniques in turbo-machinery blades and application in Francis runner Sailesh Chitrakar* 1, Binaya Baidar 1, Ravi Koirala 1 1 Turbine Testing Laboratory, Kathmandu University, Nepal Abstract This paper presents 3-D geometry modification techniques in turbo-machinery blades, implementation using a MATLAB code, and its applications in a Francis turbine runner. The modification can be carried out by aligning the sections of the blade from Leading Edge (LE) to Trailing Edge (TE) or from Suction Side (SS) to Pressure Side (PS) without changing the profile sections in spanwise direction. The development of a MATLAB code has been shown, which reads an input geometry in.curve format and by defining a transformation matrix, the modified geometry can be obtained in the same format, which can be used for further analysis in CFD. The code is used to analyse 7 different combinations of a Francis turbine runner blade. The result from the CFD analysis in ANSYS-CFX showed that the blade with a negative angle accounts for improved performance in terms of the total pressure loss along the streamwise direction of the blade. On comparing the total pressure contour downstream of the runner, it is seen that the ed blades minimizes the velocity deficit regions called as wake flows, which is a cause of hydraulic losses in turbines. Also, the secondary flows in runner blades are seen to have improved in the ed blade compared to the conventional one. The results of these analyses are supported by previous literatures and practical applications in current trend. Index Terms Re-stacking techniques, sweep,, bow, Francis turbine, total pressure loss I. INTRODUCTION The design of any rotor blade starts from preliminary 1-D design such as calculating or choosing the boundaries, rotation speed, inlet flow and velocity triangles at inlet and outlet. The shape of the blade is determined from the velocities at relative frame of reference and by defining an appropriate thickness, the 2-D profile of the blade is obtained. The distribution of the blade profile along the spanwise direction (hub to shroud) gives a complete 3-D shape of the blade. The process of combining all the 2-D profiles of the blade in given spanwise locations has been referred to as 'stacking'. The 2-D sections are mostly assumed to be best fitted when they are arranged straight on top of each other such that their respective centers of gravity would lie on one line. However, the spanwise profile sections can be arranged in different ways to achieve a certain distribution of flow parameters downstream of the blade row and/or to affect the flow in the passage in a certain way. Also, by maintaining a certain stacking condition, the mechanical integrity of the blades can be improved with respect to both steady and unsteady loadings. While stacking the blade profiles, following names and conditions are referred in this paper: *Corresponding author: sailesh@ku.edu.np Straight or base case i.e. normal stacking Sweep (inclination of the blade from Leading edge to Trailing edge) Lean (inclination of blade to either Pressure side or Suction side) Bow or compound (bowing of the blade in either Pressure side or Suction side) Combinations of above A swept blade was initially used [1] to decrease the noise level induced by shock waves. These types of blades then gained popularity in both turbines and compressors [2] with the improvement in 3D flow effects and performances. The influence of the blade on turbine losses was studied [3] and it was found that the blade had significant effect on blade loading, loss generation and state of boundary layers in blade suction surfaces and the endwalls. Also, a compound reduced the downstream mixing losses and off-design performances. While 3D blade modifications are very common in gas and steams turbines, compressors, wings and fans, it is also being studied and used these days in hydraulic turbines. After patenting a reversed leading edge blade for Francis runner [4] in 1982 by GE Hydro, this technology was later used in Three Gorges Project for China in 1998 and was named as X-Blade [5], which resulted in superior efficiency at peak and flatter hill-chart, cavitation-free runners in the field and wider range of stable operation. This type of ing also improved the mechanical integrity of the blades by reducing the maximum stresses in the shroud-trailing edge regions[6]. According to a theoretical study [7] of the flow in the runner blade channels related to the cross flow vortices from hub to shroud, it was seen that a negative blade at the inlet of the runner can reduce the cross flows, which can have a positive impact on hydraulic efficiency. Kathmandu University, Turbine Testing Lab has been using MATLAB design software (Khoj) for developing geometries and.curve files of high head Francis turbines. This software also allows some ing possibilities in the runner blade. A study [8] on blade ing between -5 o to 5 o was done using the same software and it was seen that the pressure distribution between pressure and suction side at hub was affected by the linear blade ing. In this paper, a standalone MATLAB code has been developed which reads the.curve file of any type of blade and by defining a modification definition, the re-stacking of the blades is done and.curve file of the modified blade is Rentech Symposium Compendium, Volume 4, September

2 obtained. Application of the code has been shown for a Francis turbine runner blade and taking a base case, the modified blades at different angles of and bow have been compared. II. DESCRIPTION OF THE CODE A MATLAB code was developed from which the.curve file of the blade can be imported and all the transformations can be carried out which are based on displacing spanwise profile section in various directions, such as to achieve ed, swept or bowed blades. The transformed blades can be further exported in.curve format readable by ANSYS Turbo-grid, which is used to conduct meshing for flow simulations. The modifications are carried out in such a way that the profile section itself is preserved whereas the arrangement of the section in the profile is changed (i.e. restacking of the profile). Figure 1 shows an arbitrary blade profile with general locations and definition of the modifications carried out in this code. The modification can be carried out in two directions as shown, whereas bow or compound can be incorporated in the same code by giving different angles of in two or more spanwise locations. Fig. 1:Locations in a profile and definition of modifications (, sweep) The input of the geometry modification is specified by means of a text file. This file contains a matrix of 3 columns as mentioned below, with the first column specifying the number of spanwise locations (0 to 1, where 0 represents hub and 1 represents shroud), second column represents the respective component and the third column represents the respective sweep component. spanwise location(0 1) sweep component sweep component Lean and sweep are given as a fraction of the axial chord of hub. This text file also allows the selection of a suitable type of interpolation to be used for stacking, including linear, spline or cubic. Figure 3 explains an example of the code sequence using an arbitrary curve file of a blade and modification. Fig. 2:MATLAB code input and output 3.1 Advantages of the code This is a simple MATLAB code, which can be used as a function in standalone condition. It offers a wide range of spanwise profiles and modifications. The inputs and outputs of the code is in.curve format which is recognized by Turbogrid. No modifications are needed for hub and shroud profiles. It is applicable for all types of blades (fan, compressors, wings, wind/gas/steam turbines) having.curve format. III. APPLICATION IN FRANCIS TURBINE RUNNER Francis is a reaction type radial-axial flow turbine with a wide working range (speed number between 0.22 to 1.5). Most of the major hydro-power stations in Nepal use Francis turbines as the main conversion devices. Kaligandaki 'A', which is the biggest power station of Nepal (144 MW) till today, uses three 48 MW Francis turbines with the head of 115 meters. Similarly, other hydro-power stations such as Marsyangdi, Middle Marsyangdi, Bhotekoshi, Jhimruk etc. also use Francis turbines. The turbines used in these power plants are continuously facing the problem of sediment erosion. Various studies have been made so far related to the effect and possible solutions of sediment erosion. In this regard, a MATLAB design program was developed [9] to design and modify high head Francis turbines. Where these studies focused particularly on erosion related problems, the study of losses in the turbine is also a global need. These losses include secondary flow in Francis runner, rotor stator interaction, draft tube surge and cavitation. This paper focuses on the study of secondary losses in Francis runner and the effect of various shaped blades on them using the aforementioned code. The Francis runner of Jhimruk Hydropower Plant (head meters and flow 2.35 m 3 /s), which was used in previous studies have been used in this study too. Several combinations of the geometry modification were made and tested, however this paper has presented results from 7 combinations including the base case (conventional design). The combinations include and bow blades only, as the swept blades are more common in unshrouded blades and in this case, would require Rentech Symposium Compendium, Volume 4, September

3 changes in the shroud profile too. The combinations are listed in Figure 3. The modification parameters are in three locations i.e. hub, midspan and shroud. These blades are compared initially on the basis of the total pressure distribution from inlet to outlet using CFD (Computational Fluid Dynamics). The useful combinations are then compared further to discuss about the flow phenomena in the blade passages. Base case 10 in shroud 10 in hub 5-10 negative linear 10 bow in pressure side 3-10 negative 5-10 positive linear Fig. 3:Various shapes of the blade IV. RESULTS The 7 combinations of the blades as discussed in section III were analyzed using CFD in ANSYS CFX. The domain discretization, solver parameters, boundary conditions and inlet velocity vectors were chosen to be same for all the cases. A single blade passage was selected and a mesh element count of was chosen with boundary refinements. The simulations are done at Best Efficiency Conditions and RMS residual of 1E-4 was chosen as the convergence criteria. The comparison between the blades was done on the basis of the total pressure loss from inlet to outlet of the blade. On the streamwise location, 100 sample points were taken and the mass averaged total pressure on a particular location was plotted for each point. The values of the total pressure are normalized with the inlet value for each case such that the relative losses for each of the cases can be compared. The total pressure loss from inlet to outlet in these blades is shown in Figure 4. The losses are accumulated due to velocity deficit near endwalls characterized by a rapid reduction in total pressure. The losses could also be in the form of secondary flows like horse-shoe and cross-flow vortices present in the blade rows. It can be seen from the figure that compared to the base case, 5-10 o negative linear and 3-10 o negative linear showed improvements in terms of maintaining the total pressure compared to the inlet. Out of these two alternatives, the blade with 5-10 o negative linear at inlet showed remarkable result in CFX. The total pressure curve shifted upwards in the graph, which means less loss accumulation. The efficiency of the single blade passage increased from 97.62% to 98.02%. The head between the inlet and outlet of the passage increased from m to m. The shaft power increased from 4.75 MW to 4.79 MW. However, the sediment erosion pattern and quantity were not much affected due to the change in the design. 4.1 Effect on wake formation The loss generated in the runner blades can be explained further by observing the wake pattern in the downstream of the blade. In an incompressible fluid, wake means a disturbed flow behind a body and increases as the fluid flows downstream. Due to the boundary layer effect, the wake results in the decrease of the average velocity compared to the free stream, resulting in the total pressure loss. In a general turbomachinery term, this loss is named as drag, which affects the total lift generation and causing disturbances in the downstream components. In hydraulic turbine runners containing hydrofoils, the losses reduce the hydraulic efficiency and in case if the wakes are being induced from guide vanes, a phenomena called as rotorstator (runner-guide vane) interaction becomes more prominent. A comparative study has been made between the base case and the negative case with respect to the total pressure pattern as shown in Figure 5. This figure shows three consecutive runner blades and a contour plot of total pressure in the outlet of the runner. The lighter colored portion at the end of each blade shows the wake region. It indicates the losses developed due to velocity deficit in the end-walls. By changing the shape of the blade, the losses can be minimized,which was also shown in Figure 4. Rentech Symposium Compendium, Volume 4, September

4 Fig.4: Graph of mass averaged total pressure distribution from inlet to outlet in all blades surface. It shows that the stagnation point (point where the flow meets the blade and contains maximum pressure) has shifted upstream, improving the flow pattern in both the pressure and the suction side. The velocity pattern in this case has been observed for best efficiency condition. In part load operations, it will be easier to classify and compare several forms of secondary flows. Fig. 5: Comparison of wake regions between conventional and ed designs 4.2 Effect on secondary flows Secondary flow implies any undesirable flow patterns in turbo-machines. Where the wake flows are also categorized under the secondary flows, several other patterns are also taken under this type. This includes rotor-stator interaction, leakage flow, draft tube surge, horse-shoe and cross flow vortices. These unwanted flow patterns result in efficiency loss, noise, vibrations, loss of fluid, increase in temperature etc. Hence, a lot of research activities focus on minimizing these characteristics. Figure 6 shows a comparison between the same two blades in terms of the flow pattern in the hub Fig. 6:Comparison of the flow pattern between conventional and ed designs Rentech Symposium Compendium, Volume 4, September

5 V. CONCLUSION Spanwise re-stacking of any turbo-machinery blades plays a significant role in modifying the turbine performances, whether in terms of efficiency, mechanical integrity or other characteristic features of a turbine. The MATLAB code developed in this study could be useful in implementing such modifications in any types of blades. In this study, the performance of a Francis turbine runner was studied on the basis of the fluid flow phenomena and a comparative study was done between the conventional design and 6 other combinations of the re-stacking, suitable for shrouded blades as of Francis runner. It was seen that blades with negative angle accounts for better performance with respect to the total pressure loss accumulating due to the end-wall effect resulting in the ultimate efficiency drop. This research can further continue for analyzing the structural integrity of such a blade and knowing influence on characteristic feature of reaction type hydro turbines such as cavitations. REFERENCES [1] Bliss, D.B., "Method of and Apparatus for preventing leading edge shock and shock related noise in Transonic and supersonic blades and like", US Patent , 1976 [2] Denton, J.D. and Xu, L., 1999, "The exploitation of three-dimensional flow in turbomachinery design", Proc Instn Mech Engrs,Vol 213, part C, pp [3] Harrison, S., "The Influence of Blade Lean on Turbine Losses",Journal of Turbomachinery, Volume 114, Issue 1, pp , 1992 [4] Holmes, D.G., "Blade configurations for Francis-type turbine runners", US Patent , 1984 [5] Coulson, S.T., "Runner for Francis type hydraulic turbine", EP Patent [6] Andritz Hydro, Hydro News, No. 15/5-2009, pp. 6-7 [7] Brekke, H., "A Review on Oscillatory Problems in Francis Turbines", 2010 [8] Gogstad, P.J., "Hydraulic design of Francis turbine exposed to sediment erosion", Master's thesis, 2012 [9] Eltvik, M., Thapa, B.S., Dahlhaug, O.G., and Gjosaeter, K., Numerical analysis of effect of design parameters and sediment erosion on a Francis runner, Fourth International Conference on Water Resources and Renewable Energy Development in Asia, Thailand, 2012 BIOGRAPHIES Mr. Sailesh Chitrakar is currently working as a Project Co-ordinator in Turbine Testing Lab. He completed his Masters from a Erasmus Mundus Program in Turbomachinery and Aeromechanics from Royal Institute of Technology, Sweden and University of Liege, Belgium. Mr. BinayaBaidar is currently working as a Research Fellow in Turbine Testing Lab. He completed his Masters from a Erasmus Mundus Program in Turbomachinery and Aeromechanics from Royal Institute of Technology, Sweden and Aristotle University, Greece. Mr. Ravi Koiralais a graduate of Mechanical Engineering from Kathmandu University. He is currently working as a Researcher in Turbine Testing Lab. Rentech Symposium Compendium, Volume 4, September