A new family of subsonic compressor airfoils, which are characterized by low losses and wide operating ranges, has been designed for use in heavy-duty gas turbines. In particular the influence of the higher airfoil Reynolds numbers compared to aeroengine compressors and the impact of these differences on the location of transition are taken into account. The design process itself is carried out by the combination of a geometrical code for the airfoil description, with a blade-to-blade solver and a numerical optimization algorithm. The optimization process includes the design-point losses for a specified Q3D flow problem and the off-design performance for the entire operating range. The family covers a wide range of inlet flow angle, Mach number, flow turning, blade thickness, solidity and AVDR in order to consider the entire range of flow conditions which occur in practical compressor design. The superior performance of the new airfoil family is demonstrated by a comparison with conventional controlled diffusion airfoils (CDA). The advantage in performance has been confirmed by detailed experimental investigations, which will be presented in Part II of the paper. This leads to the conclusion that CDA airfoils which have been primarily developed for aero engine applications are not the optimum solution, if directly transferred to heavy-duty gas turbines. A significant improvement in compressor efficiency is possible, if the new profiles are used instead of conventional airfoils.

In Part I of this paper a family of numerically optimized subsonic compressor airfoils for heavy-duty gas turbines, covering a wide range of flow properties, is presented. The objective of the optimization was 10 create profiles with a wide low loss incidence range. Therefore, design point and off-design performance had to be considered in an objective function. The special flow conditions in large scale gas turbines have been taken into account by performing the numerical optimization procedure at high Reynolds numbers and high turbulence levels. The objective of Part II is to examine some of the characteristics describing the new airfoils, as well as to prove the reliability of the design process and the flow solver applied. Therefore, some characteristic members of the new airfoil series have been extensively investigated in the cascade windtunnel of DLR Cologne. Experimental and numerical results show profile Mach number distributions, total pressure losses, flow turning and static pressure rise for the entire incidence range. The design goal with low losses and especially a wide operating range could be confirmed, as well as a mild stall behavior. Boundary layer development, particularly near stall condition, is discussed using surface flow visualization and the results of boundary layer calculations. An additional experimental study, using liquid crystal coating, provides necessary information on suction surface boundary-layer transition at high Reynolds numbers. Finally, results of Navier-Stokes simulations are presented which enlighten the total pressure loss development and flow turning behavior especially at high incidence in relation to the results of the design tool.

A single-stage transonic axial compressor was equipped with a casing treatment (CT), consisting of 3.5 axial slots per rotor pitch in order to investigate the predicted extension of the stall margin characteristics both numerically and experimentally. Contrary to most other studies, the CT was designed especially accounting for an optimized optical access in the immediate vicinity of the CT, rather than giving maximum benefit in terms of stall margin extension. Part I of this two-part contribution describes the experimental investigation of the blade tip interaction with casing treatment using particle image velocimetry (PIV). The nearly rectangular geometry of the CT cavities allowed a portion of it to be made of quartz glass with curvatures matching the casing. Thus, the flow phenomena could be observed with essentially no disturbance caused by the optical access. Two periscope light sheet probes were specifically designed for this application to allow for precise alignment of the laser light sheet at three different radial positions in the rotor passage (87.5%, 95%, and 99%). For the outermost radial position, the light sheet probe was placed behind the rotor and aligned to pass the light sheet through the blade tip clearance. It was demonstrated that the PIV technique is capable of providing velocity information of high quality even in the tip clearance region of the rotor blades. The chosen type of smoke-based seeding with very small particles (about 0.5 μm in diameter) supported data evaluation with high spatial resolution, resulting in a final grid size of 0.5×0.5 mm2. The PIV database established in this project forms the basis for further detailed evaluations of the flow phenomena present in the transonic compressor stage with CT and allows validation of accompanying computational fluid dynamics (CFD) calculations using the TRACE code. Based on the combined results of PIV measurements and CFD calculations of the same compressor and CT geometry, a better understanding of the complex flow characteristics can be achieved, as detailed in Part II of this paper.

A single stage transonic axial compressor was equipped with a casing treatment consisting of 3.5 axial slots per rotor pitch in order to investigate its influence on stall margin characteristics, as well as on the rotor near tip flow field, both numerically and experimentally. Contrary to most other studies, a generic casing treatment (CT) was designed to provide optimal optical access in the immediate vicinity of the CT, rather than for maximum benefit in terms of stall margin extension. The second part of this two-part paper deals with the numerical developments and their validation, carried out in order to efficiently perform time-accurate casing treatment simulations. The numerical developments focus on the extension of an existing coupling algorithm in order to carry out unsteady calculations with any exterior geometry coupled to the main flow passage (in this case a single slot), having an arbitrary pitch. This extension is done by incorporating frequency domain, phase-lagged boundary conditions into this coupling procedure. Whereas the phase lag approach itself is well established and validated for standard rotor-stator calculations, its application to casing treatment simulations is new. Its capabilities and validation will be demonstrated on the given compressor configuration, making extensive use of the detailed particle image velocimetry flow field measurements near the rotor tip. Instantaneous data at all measurement planes will be compared for different rotor positions with respect to the stationary slots in order to evaluate the time-dependent interaction between the rotor and the casing treatment.

This contribution is aimed at summarizing the effort taken to apply stereoscopic PIV (SPIV) measurements in the tip clearance of a transonic compressor rotor equipped with a casing treatment. A light sheet probe was placed downstream of the stator and aligned to pass the light sheet through a stator passage into the blade tip clearance of the rotor. A setup with three cameras has been used in order to record the entire 2C velocity field and the smaller area of 3C field of view at the same time instance for comparison with earlier 2C PIV results. A homogeneous seeding distribution was achieved by means of a smoke generator, working with evaporated and subsequently re-condensed mineral oil. The main emphasis of the SPIV measurement was to establish a data set with high spatial resolution close to the compressor casing, where the aerodynamic effects of a CT are known to be strong. Additionally the SPIV data was intended to be used for comparison with numerical results and related code validation, as detailed laser-based flow field investigations in the vicinity of a casing treatment in a transonic compressor environment are barely reported in the literature. The paper will discuss some major aspects of the utilized PIV data processing and point out a variety of frequently underestimated error sources that influence the overall quality of the recovered data in spite of the fact that the individual PIV recordings seemed to be of very good quality. Thus the authors will not focus on the PIV results and related interpretation of the flow field, but on the optimization and procedures applied during setup of the experiment and data processing respectively.