Abstract
The present paper aims at investigating the impact of part-span snubber applied on a last rotating blade of an industrial low pressure steam turbine. The large increase in steam specific volume during the expansion forces low pressure turbine designers to develop high aspect ratio blades for the last stage. Considering also the relatively low thickness of the airfoil required to keep a high efficiency, the rotor becomes prone to aerome-chanical instabilities. Therefore, the design process must search for a tradeoff between aerodynamical and mechanical requirements that are usually conflicting. Part-span snubbers can be employed to grant additional mechanical stability to the blade thus avoiding the onset of vibrations which could lead to high cycle fatigue blade failure. In this work, a part-span snubber was introduced in the numerical model of an industrial rotor to assess its impact on the aerodynamic performance. The impact on the flow solution and on the computational cost was evaluated by using two different simplified approaches: firstly, the presence of the damping wire was introduced by means of a body forces model without any change in the aerodynamic geometry (that is blade without snubber). Then, the airfoil shape was modified to include the connections between the blade and the damping wire without a physical description of the damping wire itself which was once again modelled by the body forces. The CFD simulations were performed on hexaedrical structured grids, and a sensitivity with respect to the grid clustering near the damper devices was carried out to select a suitable tradeoff between computational cost and discretization accuracy. The aerodynamic analyses were performed with RANS computations and the flow distortions induced by the snubber are described in detail. In addition, uncoupled time domain flutter simulations were also performed to test the capability of the mesh deformation strategy to deal with snubber elements ensuring a smooth grid deformation without any cell interlacing. The simplified body forces model showed a good balance between results accuracy and computational cost, while the more complex model turned out to be more time consuming because of both the higher grid density required to correctly capture all the geometrical features of the connections, and for the higher efforts in generating a good quality mesh.
