Aerothermal optimization of axial gas turbine blade tip geometry

A gap must be maintained between the rotating blades and the stationary casing to allow the relative motion of the blades in order to avoid mechanical failures due to the rubbing and the thermal expansion in turbo machines. However, tip clearance results in a leakage flow because of the pressure difference between the pressure side (PS) and the suction side (SS). The fluid flow through the tip gap is highly complex and 3-dimensional. The leakage flow leaving the tip gap rolls into a vortex called tip leakage vortex and interacts with the main passage flow. The leakage flow through the tip gap, the tip leakage vortex and the interaction with the main passage flow are significant loss mechanisms for the turbo machines. Almost one-third of the overall aerodynamic loss is due to the flow structures related to the tip leakage flow in axial flow turbine stage. Besides, the leakage flow does not contribute to the work extracted from the turbine since leakage flow is not turned as the main passage flow. In real axial turbines, the blade tip surfaces are exposed to hot gas stream because of the leakage flow passing over them. The heat transfer between the hot fluid flow and the blade tip surfaces causes thermal stresses on the blade tip surfaces. Thus, it becomes necessary to cool the blade tip surface. Various passive control methods have been offered in order to improve the aero-thermal performance of the axial turbine by reducing undesired effects of the tip leakage flow. Squealer and winglet designs have been widely investigated both numerically and experimentally in the literature as passive control methods. In these methods, blade tip surface is modified in spanwise direction or tangential direction to enhance axial turbine rotor blade performance. Also, combination of squealer and winglet designs have been studied recently. A better understanding of tip leakage flow is essential in order to reduce the adverse effects of it on turbine aero-thermal performance. The flow inside a turbo machine is highly complex, 3-dimensional and turbulent. Also, many parameters in turbo machine designs make it harder to clarify the effects of the tip leakage flow. It is possible to encounter different outcomes, even contradictory results are available in the literature. Distinct results are accessible for the squealer and partial squealer due to the dimensions and the location of the squealer rims. Also, aerodynamic and thermal investigations of winglet designs have led to controversial results. Disputed results for the winglet designs on the PS and SS of the blade are obtained from various researches. Thus, the first step in this study is to clarify the tip leakage flow physics. Flow throughout a rotor blade in a linear cascade arrangement and the leakage flow are investigated using numerical calculations. The tip clearance is a narrow gap, thus it becomes difficult even impossible to quantify the leakage flow in this tiny gap using measurement techniques. In the field of turbo machines, experimental measurements may also become expensive and time consuming. CFD is a powerful way of clarifying the flow inside the gap. Developments in computer capacities and speed increase the use of CFD in turbo machines. The flow field for an axial turbine blade in linear cascade is calculated using CFD methods. The computational predictions are performed for Axial Flow Turbine Research Facility (AFTRF) rotor blade tip profile. AFTRF is a rotational test rig in Turbomachinery Laboratory in Penn-State University. The solid model in current calculations is formed by extruding the AFTRF rotor blade tip profile to be 123 mm in height. Then, the computational domain is formed for a single blade passage in a linear cascade to be periodic in pitchwise direction. Numerical calculations on the effects of tip clearance height and the relative casing motion are carried out. The results show that increase in the tip gap height gives rise to stronger tip leakage vortex and higher total pressure loss. Relative casing motion is found to weaken the tip leakage vortex and reduces the aerodynamic loss. ICEM CFD, a general purpose grid generator, is used for the grid generation in this step. The primary focus in this study is to enhance the aero-thermal performance of the axial turbine blade by tip carving. Genetic algorithm integrated with meta-models are utilized to develop carved blade tips offering higher performance. Building a meta-model function evaluation requires an initial database for training, testing and validation. The choice of samples for preparing a database is highly important for the success of the predictions. The number of the points in initial database is also a crucial parameter for successful predictions. Coarse-to-fine approach method is used in optimization. Considering the number of CFD results, a parametric assessment becomes an obligation. The computational domain and the grid generation should be defined in a parametric way so as to reduce the time cost. In case of a parametric approach, differences due to the grid structures are reduced substantially. Besides, preparing a parametric working layout enable to investigate the effects of squealer and winglet on the tip leakage flow. Based on the requirements, a parametric working environment is created on ANSYS Workbench interface. ANSYS Meshing module is used in calculations instead of ICEM CFD. The next issue is to carry out a comprehensive numerical investigation that involves totally 28 different squealer blade tip designs for 7 different squealer width and 4 different squealer height. Limited studies on the influence of squealer width and height are available in the literature. The most studied parameter is the squealer height, however most of the studies deals with a narrow range. The literature survey indicates that insufficient studies focus on the squealer width. Hence, a systematic study is carried out on squealer dimensions in a wide range. It is believed that this study will provide a serious contribution to the literature. CFD results reveal that squealer designs can reduce the overall aerodynamic loss and average heat transfer coefficient to the blade tip surfaces effectively. In addition to conventional passive control methods, casing treatments, which are widely used in compressors, are studied in the present study using CFD methods. In order to improve the stall margin of the compressors, modifications having various geometrical designs are introduced to the casing surface for axial turbine blade. CFD results show that grooves on casing surface is capable of the reducing the tip leakage flow rate and weakening the tip leakage vortex and the passage vortex. The numerical results up to this point are obtained for the AFTRF rotor blade tip profile. The future studies use the blade solid model provided by TAI (Turkish Aerospace Industries) for CFD calculations. In addition to squealer designs, winglet designs, which aims to reduce the driving pressure difference, are also studied on the rotor blade obtained from TAI. The winglet designs on PS and SS of the blade are found to reduce the leakage flow rate by reducing the driving pressure difference. The investigation reveals that length and position of the winglet geometries on the PS and the SS affect the aero-thermal performance considerably. Finally, a numerical investigation is conducted on carved blade tip designs, a relatively novel approach in axial turbines. Few studies are available in the literature for the carved blade tip geometries. In the current study, novel blade tip designs having higher aero-thermal performance are explored using genetic algorithm coupled with artificial neural network (ANN). In addition to ANN, Support Vector Machine (SVM) and Extreme Learning Machine (ELM) are used in training, testing and validation. Firstly, 2-dimensional optimization methods are used to carve the blade tip surface. The flow through the tip gap is assumed to be 2-dimensional. However, it is obtained that this assumption is not valid longer since there have been deviations in the leakage flow direction. Therefore, 2-d optimization studies are terminated, then 3-d optimization methods are applied. The initial database consists of 55 carved blade tips which are obtained randomly. The carved blocks which will be removed from the blade tip surface are formed using parametric approach based on the Bezier points at prescribed sections in streamwise direction. The number of the solutions in initial database corresponds to the number required to define a second order response surface for a function including 9 variables. Meta-models are utilized in order to predict the fitness values of the designs. A coarse-to-fine approach is followed to improve the prediction capability of the meta-models, thus decrease the prediction errors. Once the prediction errors are kept at a reasonable level, the process is ended. The results indicate that tip carving approach enables to design novel blade tips offering higher aero-thermal performance. Tip carving approach carries a great potential in field of axial turbines since it allows to improve undiscovered blade tips.