MEASUREMENTS IN BLADE CASCADE FLOW. David Krivánka

Transcription

1 SOUTĚŽNÍ PŘEHLÍDKA STUDENTSKÝCH A DOKTORSKÝCH PRACÍ FST 2007 MEASUREMENTS IN BLADE CASCADE FLOW David Krivánka ABSTRACT This paper deals with the the aeroelasticity problem which occurs when the turbine blades are operating. The blade cascade is defined and discussed. The main part of this paper is focused on the measurement problems in the blade cascade. The measurement of the velocity profile was described and the comparison of some methods for this measurement was made. The solution of the aeroelasticity problem using fast digital regulation was found and applied to the newly-built wind tunnel. KEYWORDS flow measurement, flow visualization, aeroelasticity, turbine blades, blade cascade 1. INTRODUCTION These days, the turbine engine is used in almost every power station, regardless of whether their primary energy source is conventional (coal, oil), renewable (water) or nuclear. The increase in performance requirements force us to develop new materials, blade shapes and construction techniques. However, new technology brings new problems related to measurements and safety standards. Therefore new measurement procedures are being developed. The main goal of this article is to describe some problems closely connected with the measurements, which occurs during the turbine (and especially blades) study. It is impossible to completely solve these problems in one paper, so only short overview of the current state in this area will be given. The blade cascade is used to measure fluid flow in the blades. The blade cascade consists of several blade channels (two blades together form a blade channel). Because the real blades are mounted to a high speed rotating turbine wheel and the blade cascade is usually stationary, it is necessary to use a wedge shape of the cascade to preserve the same boundary conditions. The angle of the shape is dependent on the velocity of the fluid flow. The typical shape of the subsonic blade cascade is shown in fig. 1. The shape of the transonic blade cascade is slightly different, especially the shape of boundary walls but the main difference is in the shape of the blades. The blade shape optimization is a key Fig. 1: Shape of the blade cascade issue for raising the efficiency of the whole turbine (as discussed for example in [1]) because even small changes of the blade efficiency can significantly increase the overall energy output. The most important factor for the blade optimization is the so called velocity profile. In principle, the velocity profile is a graph of fluid velocity dependency on the position of a probe across the blade channel or across the cascade. Because this profile is non-stationary (and therefore time-dependent), mean time-values are often measured. The velocity profile can be measured in front of or behind the cascade or directly inside the blade channel. Two examples of a subsonic velocity profile are drawn in the fig. 2. The velocity profile can be measured by pressure probes, by means of hot-wire anemometry or by some computer-aided, sophisticated visualization methods, such as Particle Image Velocimetry r y c h l o s t [ m / s ] l p o o h a [ m m ] Fig. 2: The velocity profile

2 (explained below). Measurement with pressure probes is very cheap and easily prepared, but can hardly measure big pressure fluctuations. The wire anemometry is much more precise but the probes are more expensive and very fragile. The greatest disadvantage of both methods is that they can be used only for spot measurements, so some kind of a traversing mechanism is always necessary. The most promising technology is the upcoming Particle Image Velocimetry (PIV). This brand new method uses smoke with small particles, laser sheet and high speed CCD cameras to take a set of pictures of whole measured area at once. From two successive pictures the motion of the particles (and therefore the motion of the fluid) is obtained. As such whole area can be measured at once. On the other hand, this equipment is extremely expensive and the measured area has to be well accessible, which is not very common. A flow separation example [2] measured by PIV method is shown in the fig. 3. Fig. 3: Flow separation measured by PIV method In recent years, another possibility to obtain data has arisen. The increasing computational power of modern processors gives us the possibility to simulate very complex fluid systems. The Computational Flow Dynamics (CFD) software, such as Fluent, can be used to model fluid flow, heat transfer and interactions between fluid and solid materials. The numeric simulations are very cheap compared to the measurements and they always provide a complete picture of the area, even in those part where measurement would be impossible. However, are these simulations precise? It depends on the chosen turbulent models, on the geometry, on the mesh quality and many other variables. A comparison of measured and computed data is shown in fig. 2; the measured velocity profile is drawn with the black line and the computed one with grey line. In this particular case, there is a slightly difference between those profiles but the agreement is still very good. So the numeric simulations must be always confirmed by experimental data. 2. DESCRIPTION OF THE WIND TUNNEL The brand new wind tunnel was built at the Department of Power System Engineering (PSE) at University of West Bohemia. A simple scheme of this tunnel is shown in the fig. 4. The compressor can be found in the upper left corner of the figure (1) and the measuring area with the airfoil cascade is in the right part of the scheme (7) in the suction part of the tunnel. This tunnel is purposed for the determination of the aerodynamic coupling forces. The flow velocity can be controlled either by the compressor revolutions regulation, or by three independent shutters (13, 17, 18). Fig. 4: Scheme of the PSE wind tunnel The purpose of this tunnel is to measure the influence of the flow on the airfoil models and to determine the aerodynamic coupling forces. The wind tunnel is therefore equipped with an airfoil cascade. A simple scheme of this cascade can be found in the fig. 1. The cascade consists of eight blade models. Four of these models are equipped by vibration units, which can be used either to excite the vibrations, or to measure the aerodynamic coupling forces. Each unit has two coils connected with parallelogram and therefore two degrees of freedom, so it can be used to excite either forward, or torsion oscillations. In addition, as we already know, this unit can also be used for measurement. This principle is based on the expectation that exciting force is proportional to current

3 induced in the coils. In fig. 5 the measuring area is closely displayed. The flow direction is on case from right to left. The cascade is mounted to a massive steel frame to prevent influence of the tunnel body vibrations to the blade vibrations. Velocity profile can be measured by two shifted pressure probes mounted to a traversing mechanism. Two probe types are used for measuring either in-front or behind the cascade. Fig. 5: Photos of the measuring area 3. VELOCITY PROFILE EQUALIZATION As was already mentioned, the blade cascade is wedge shaped to preserve the boundary conditions of the real blades mounted to the rotating wheel. This shape causes the non-uniform velocity profile and thus different input conditions for each blade. In order to equalize the velocity profile in the PSE tunnel, the walls of the cascade are porous. The suction chambers behind these porous walls are used for suction of the boundary layer and therefore for the velocity profile equalization. Five suction chambers are in the PSE cascade, one on the left side (from the flow direction) and four on the right side, see fig. 1. Each suction chamber is equipped with a pressure sensor and an independent shutter, so the volume of the sucked air can be easily regulated. The influence of the suction to the shape of the velocity profile was studied by means of CFD methods and some configurations of the shutters were proposed in [3]. However, these simulations had to be confirmed with some experimental measurements. Generally, the velocity profile can be measured by pressure probes or hot-wire anemometry, as mentioned above and the whole velocity field can be measured, or rather visualized by the PIV method or so called interferometry. Both methods are used for the visualization of whole velocity field at once. The interferometry is based on the combination of two different harmonic signals, see [4]. It is one of the basic methods of flow visualization and it is still very popular. The typical transonic interferogram is shown in the fig. 6. Although the interferometry can not be used for precise measurements, it is still possible to use this method for the velocity field visualization and therefore for the equalization of the velocity profile. The main disadvantage of both methods is that they can be applied only on well accessible blade cascades with transparent walls. However, the PSE walls are metal, so this methods can not be used directly and only comparison between the PSE cascade simulations and other simulations confirmed by these methods is possible. Without these methods, the pressure probes were the only option to measure the velocity profile in the PSE blade cascade. The PSE cascade was therefore equipped with the pressure probes which are used to measure the velocity profile behind the blades. However, the CFD simulations proved that the measurement of the velocity profile behind the cascade is insufficient because this profile is not affected by the suction of the boundary layer. To solve this problem, new pressure probes were developed to measure the velocity profile in front of the cascade. To prevent the influence of these new probes on the aeroelasticity measurements, these probes were mounted to original traversing mechanism behind the cascade and were used only when the electrodynamic vibration units were stopped. These new measurements finally confirmed the numeric Fig. 6: Interferogram simulations and the velocity profile was equalized.

4 4. THE AEROELASTICITY PROBLEM As was already said, the main purpose of the PSE tunnel is the to study the aeroelasticity and the influence of the flow on the blade vibrations. The equalization of the velocity profile was the essential part of this work but yet this is only the beginning. The blades in the gas turbines are exposed to enormous aerodynamic and mechanic loading. In fact, the speed of the blades differs from several hundreds to thousands meters per second. In such velocities, the centrifugal forces acting on the blades are several hundreds millions of newtons. In these conditions, blade vibrations are inevitable. However, blade vibrations influence the fluid flow and the flow disturbances influence not only the same blade but also the other blades. So each blade is influenced not only by the flow disturbances and immediate neighboring blades but also by all other blades and by itself. Small vibrations of one blade can easily lead to catastrophic failure and destruction of the whole turbine. Because all blades are connected it is not possible to measure only one blade or use only one vibration unit although many studies were made using only one blade model. These experiments are necessary to provide basic data about the blade behavior but can not solve the problem of stability of blade vibrations. First experiments with the blade cascades were made using simple vibration unit on the first blade and force sensors on the neighboring blade. This concept has several disadvantages. The main disadvantage is that vibrations of the first blade are influenced by the flow and thus the displacement of this blade is not corresponding to the exciter signal. The other disadvantage is the stationary position of the second blade which is necessary to precise measurements of the force. Another way of measuring the vibrations is to use the PIV method. The principle is measurement of the blade displacements instead of force. Although great progress has been made in the last few years, this method is still premature and can not be used for precise measurements. However, it is almost the only possibility how to obtain the complex picture of the flow velocities in the area. The Computational Flow Dynamics simulations are very promising in this area and results from the simulations of the blade cascades were confirmed many times, but can we use them to solve the problem of aeroelastic stability? This question is surprisingly hard to answer. The main problem is that CFD software can compute only the fluid flow. So it is possible to obtain the pressure and velocity in every point around the blade and even the forces acting on the blade but it can not compute the reaction of the blade. As mentioned earlier, the blade vibrations are closely coupled and this disadvantage makes current CFD calculations completely unusable. However, the algorithms describing the blade behavior are being developed [5] and their integration in some CFD software is only a matter of time but in the present, the CFD is only usable for partial calculations and for measurement validation and confirmation. So, is there any possibility to measure the coupling forces precisely enough? Fortunately, with the development of fast industrial computers and digital regulators arise new method of measurement. These regulators controlling the vibration units are used either to excite the vibrations, or to measure them. The advantage of this regulator is that it maintains the desired input signal (track regulation) and can suppress the influence of the flow and vibrations of the neighboring blades. Moreover, the same regulator is used for the measurements of the coupling forces too. If it is fast enough, it can be used to suppress the vibrations of the neighboring blade (constant regulation) we can easily measure the electric current in the coils of the vibration unit. The force induced in the vibration units is proportional to this current and it equals the same force with which the fluid flow influences the blade. As was already mentioned before, the PSE tunnel has four blades equipped with the vibration units and each vibrating units consists of two independent coils. The simple scheme of the vibration unit is shown in the fig. 7. The blade (1) is mounted to the parallelogram (2) with two separate coils (3). So, there are eight independent coils to regulate in the whole cascade. The precise regulation is therefore very time demanding and thus enormous fast digital regulators are needed [6]. In the PSE, these regulators were already constructed and they are now being prepared for final testing and hopefully they will provide the solution of the aeroelasticity Fig. 7: Vibration unit with the blade problem of the blades. 5. CONCLUSION Such a complex topic as the aeroelasticity problem of the blades can not be solved or even fully described in this short article. So only brief introduction was given. To sum up, several problems with measurements were mentioned, such as the measurement and equalization of the velocity profile, measurement of the coupling forces and problems related with numeric simulations.

5 Several methods were also described and some suggestions were made. The measurement of the flow in the whole area of the blade cascade is still very problematic but the PIV method could be the solution in the very near future. The CFD simulations combined with sophisticated blade models are being developed, so the precise simulations of the coupling forces will be soon possible. On the other hand, even the most precise simulation can not fully substitute the measurements. The extensive development of the discrete regulation gives us the opportunity to obtain data with high precision and thus to better understand the aeroelasticity problem, which will lead to additional improvements of the performance of turbine engines and whole turbo machinery branch. BIBLIOGRAPHY [1] Vomela, J.; Micak, P.; The development of high-efficiency turbine blades. In Cieplne maszyny przeplywowe. Łódź: Technical University of Łódź, p ISSN [2] University of Kentucky; Turbine Blade Separation Flow Control, [3] Placek, T.; Linhart J. Equalization of velocity profile in front of the blade cascade In Powersystems - temomechanics - fluid mechanics Pilsen: University of West Bohemia, p ISBN [4] Kozel, K; Prihoda, J.; Fort, J.; Safarik, P. Transonic flow through turbine cascade: Experimental and numerical results In QNET-CFD Network Newsletter. Brussel: QNET-CFD, p [5] Rasmussen, F. Present status of aeroelasticity of wind turbines In Wind energy. Roskilde: Rise National Laboratory, p [6] Schlegel, M.; Cech, M.; Balda, P.; Mertl, J. Active vibration damping of vanes in wind tunel In Process control Pardubice: University of Pardubice, p ISBN Ing. David Krivánka, University of West Bohemia in Pilsen, Univerzitní 22, Plzeň, tel.: , krivanka@kke.zcu.cz