Abstract

Rotating instability (RI) in steam turbines is a problem that is posed by the highly flexible operation of modern powerplants due to the addition of renewable energy sources to the power grid.

RI is a phenomenon well-known from compressors but may also be observed in mainly the last stage of low-pressure steam turbines operating under low volume flow conditions.   

In this paper, the last two stages of a low-pressure turbine including the downstream situated exhaust with rotational symmetry are modelled. Tip gaps are included in the rotating blade rows.

Various CFD investigations into RI have been undertaken. However, advanced Scale-Resolving Simulation (SRS) models such as Large Eddy Simulation (LES) or Improved Delayed Detached Eddy Simulation (iDDES) have not been used to analyse RI yet.

Currently iDDES is not widely spread in turbomachinery applications. Especially the high computational effort, as well as the need for a specific mesh density and aspect ratios in the boundary layer pose a problem to the traditional simulation of turbomachinery flows. This paper demonstrates a process-chain that can be used to implement iDDES for the analysis of the flow field in steam turbines.

Multiple meshing software used in turbomachinery applications define the geometry using point-file geometric data. In order to create a CAD model, a method for processing point-cloud data in NX using reverse engineering tools is presented. This was a mandatory step for the creation of a closed and watertight CAD model for the import and CFD analysis in Star-CCM+.

Commonly, structured, hexahedral grids are used for the meshing of turbine blades. This paper is presenting an alternative approach using the polyhedral mesh structure offered by Star-CCM+. In order to capture the flow in the boundary layer accurately, prism layers were generated on the walls of the domain.

Finally, a strategy for obtaining a physically reasonable iDDES flow solution efficiently is proposed.

When using the iDDES flow model, achieving a converged solution can be problematic, as this flow model is heavily dependent on the initial conditions used. Therefore, a RANS solution is used for the initialisation of the iDDES case. To obtain the RANS solution efficiently, advanced initialisation methods offered by STAR-CCM+, such as CFL ramps and Grid Sequencing, are used.

In order to identify turbulent effects caused by the existence of RI, the flow solution is compared to a baseline test case operating under design conditions.

The results are presented using 2D and 3D visualisation methods.

The obtained results are going to be taken forward for the future analysis of the flow physics behind RI.