On the Use of Unsteady Methods in Predicting Stage Aerodynamic Performance

In the last fifteen years aerodynamic designers have relied increasingly on a systematic use of Computational Fluid Dynamics. Previous evolution in computer performance as well as in computational methods made the development of a number of codes to integrate the Reynolds–averaged Navier–Stokes equations (RANS) possible. This opened new frontiers to turbomachinery designers, who added specific tools to investigate blade aerodynamic details to the classical throughflow models. In the beginning a lot of research was dedicated to the development and validation of 3D Navier–Stokes solvers to accurately predict the flow field in turbines and compressor blade passages. Nowadays, aspects like secondary flows, radial mixing, leakage flow, local separation, and heat transfer can be well estimated by means of a 3D viscous analysis. Viscous calculations of isolated blade rows have been well incorporated in advanced gas turbine design and have given rise to an impressive step up in component aerodynamic performance especially in the transonic regime. Nevertheless, for practical applications, it is common experience that these methods are not easy to implement in a multistage environment and they are often used by designers in conjunction with empirical through-flow methods to adjust inlet/outlet blade row flow fields. The use of “mixing planes” to pass information between blade rows can help to relax the problem but is not accurate enough to be extended to a large number of stages. In addition, for realistic configurations with narrow inter–blade axial gaps, the proper conservation of mass, momentum and energy may require some alteration in axial row distance. The averaged–passage flow model of Adamczyk [2] in which a time–averaged flow field is solved and the unsteady deterministic flow is accounted for in the momentum and energy equations by correlation seems to be very promising. Up to now, it has been commonly recognized that only an accurate control of the multistage environment can allow designers to improve both component and machine aerodynamic performance at the same time. The real flow in a multistage turbine or compressor is inherently unsteady because of the relative blade row motion. This causes unsteady interaction of pressure fields, shock waves, and wakes between stators and rotors. In addition to this forced unsteadiness related to the shaft rotational speed, natural unsteadiness like shock buffeting, wake instability and vortex shedding may also occur. These unsteady flows cause mixing and losses which need to be accounted for in order to compute a large number of stages simultaneously. In the last few years a lot of attention has been dedicated to the development of unsteady solvers to be used to investigate the physics of the rotor–stator interaction. With current computer power most of the practical applications are carried out in quasi–3D with relatively few examples of industrial fully 3D unsteady analyses, however the new generation of parallel computers will shortly make 3D unsteady simulation feasible for design application. From the design point of view, there is at least one other reason to look for unsteadiness. The current trend in gas turbine design is to strongly reduce the number of stages and blades per row. This will increase the stage Mach number and enhance the unsteady blade load with risks of possible structural failure. Unsteady load estimates are now needed by the designers to address the durability issue of the next generation of turbomachinery. The present overview is focused on unsteady rotor–stator interaction in turbine and compressors. The “dual time stepping” technique implemented by the authors in the last five years will be briefly presented. Afterwards, some applications will be discussed to highlight the capability, usefulness and the potential of the unsteady approach.