In the turbomachines field, friction losses are intensively studied due to their important influence in the overall efficiency of the machine. The parameter helping in quantifying these friction losses is the wall shear-stress. Its role is essential for the qualification of the boundary layer separation tendency and the losses prediction. Thus, the first aim of this PhD is to characterize the boundary layer development, in the cone of the Francis turbine. Afterwards, in the second part of this study, a new multidirectional wall shear-stress sensor is designed, manufactured and tested for the turbomachines applications. To develop this knowledge and the tools for flows prediction in the draft tube, EPFL joined major manufacturers in the context of the European initiative EUREKA project n° 1625. In the first part of the thesis, an experimental campaign is leaded in the cone of the nq 92 Francis turbine, to characterize the wall stress, using the hot-film technique. 6 operating points were investigated, covering a large operating range – from 70% to 110% from best efficiency point flow rate. For this specific draft tube, the efficiency characteristic has a sever drop, close to the best efficiency point, and the wall shear stress evolution in this region is pointed out. The calibration and measurement procedures are exposed and the accuracy study is performed. The evolution of the wall shear-stress steady values related to the spatial position of sensor – 16 positions were explored – and to the corresponding operating point is analyzed. A boundary layer separation tendency for the part load operating points is pointed out, as well as the bend influence on the spatial evolution of the wall shear-stress. These results were used to validate numerical calculation in the draft tube. Additional LDV measurements combined with the wall shear-stress results allowed to reconstruct the boundary layer. The best fit for representing the boundary layer is obtained with a composite power law. However the 3D boundary layer is complex and a profound knowledge is needed. From the unsteady point of view, in the runner outlet section, the amplitude of the wall shear-stress fluctuations obtained synchronous with the runner's rotating frequency is predominant. For the partial load operating points, the main fluctuations magnitude is obtained for the rope passage frequency and its amplitude depends on 2 parameters: the σ value and the proximity of the rope to the wall. To increase the knowledge for the boundary layers in turbomachines, it is necessary to explore fully 3D unsteady boundary layers, both in the fixed and rotating parts of the machine. Thus a multidirectional sensor with specific requirements is needed for the turbomachines application: a miniature hot film probe, which can be implemented in the complex geometry of the turbines, a multidirectional one, to take into account the complex character of the flow, mainly the strongly 3D flow, a sensor with a sensitivity and dynamic, allowing to obtain the main flow unsteady characteristics (guide vanes wake, runner blades wake, rope frequency, turbine-circuit interaction frequency, etc.), a good electrical isolation between the surface of the probe, which comes in contact with the water, and with the hot-film support. In this way, the second aim of this PhD becomes the design and development of a new multidirectional wall shear-stress sensor for turbulent boundary layer research for turbomachines applications. The development of the new multidirectional sensor implies technological developments using microtechnology, as MEMS offers opportunities for developing and manufacturing sensors with regard to complex applications, allowing, in the same time, a high accuracy at low cost. The new sensor represents a bridge between 2 different disciplines: micromechanical technology and fluid mechanics. Its concept is based on the heat transfer generated by a hot film with a general top-area of 1.12 × 0.1 mm and a thickness of 110 nm. The film, in platinum, is maintained at a constant temperature, of 65°C, by a feed-back electronic. Key process steps in fabrication of the new device are lithography, bulk micromachining, thin film deposition, surface micromachining, lift-off and chemical mechanical polishing. The manufacturing of new miniature wall shear-stress sensor is based on a combination of these techniques. A specific development is performed for the achievement of an insulating surface to reduce the heat conduction between the hot film and the sensor body, on which the hot film is deposed. This surface is obtained by manufacturing silicon dioxide layers, of 4 µm, by DRIE technique, in order to create high-aspect-ratio silicon pillars, which are then oxidized and/or refilled with LPCVD oxide or nitride. One of the major criteria for the trenches filling was the surface planarity at the end of the refill. 2 parameters are optimized: the thickness and the silicon pillars arrangement. Thermal numerical computations were carried out using Ansys and they allowed the achievement of the optimum thermal isolation thickness between the heated structures and the surrounding structures. During the development of the new wall shear-stress probe, the main topics, achievements and contributions can be categorized into: design of a new wall shear-stress sensor, fabrication steps development, validation and optimization of the design, numerical simulations of the thermal behavior of the heated-films, manufacturing of the new device. The new sensor is characterized in terms of time response, electrical insulation between the surface of the probe, which comes in contact with the water, and the hot-film, and reliability. The sensor is robust, with a good sensitivity for water measurements. The main improvements, which make the current device distinct, are its design for a directional response for 3D turbulent boundary layer study and the insulating surface for substantially reduction of the heat losses by conduction between the film and the surrounding substrate.
EPFL2007
