Swirling flow in a conical diffuser generated with rotor-stator interaction

Application Challenge AC6-14 © copyright ERCOFTAC 2024

Best Practice Advice

Key Fluid Physics

The main features of the flow are the on-axis recirculation region,the vortex rope and the vortex breakdown, and wakes of the blades. The separation from the blades, flow in inter-blade passages, separation in the divergent part of the draft tube and rotor-stator interaction are among other physical mechanisms which make the flow fields complicated and difficult to model.

Application Uncertainties

The complexity of the geometry, curved and bladed regions, tip-clearance and rotor-stator interaction, oscillation of the runner rotational speed which is absent in numerical simulations are some sources of uncertainties which make a high fidelity CFD model difficult to assemble.

Computational Domain and Boundary Conditions

The boundary conditions also play a prominent role to reproduce the physical mechanism of the vortex breakdown. The swirl intensity determines the occurrence of the vortex breakdown. The swirl depends on the axial and tangential velocity components, which are two dominant parameters and specify the physical mechanism of the breakdown. For swirling flows in a pipe, the former determines the radius of the vortex core and the later specifies the character of the on-axis axial velocity (jet- or wake-like). The inlet boundary condition is usually unknown at the draft tube inlet of the hydraulic turbomachines. To prevail this problem, the rotor-stator interaction, which is the interaction between the guide vane and the runner blades, is considered to retain the upstream effects on the flow in the draft tube.

Discretisation and Grid Resolution

Most of the simulations that are available in the literature are done using licensed codes. The open source codes are normally more sensitive to the non-orthogonality of the mesh. The mesh quality is thus highly crucial in such simulations. The maximum aspect ratio of a cell is around 400 close to the outlet of the draft tube. The minimum angle is 18$^{\circ}$ for nine elements and occurs close to the hub in the runner. The computational domain contains several regions, where GGI is used at the interfaces between the regions. The resolution spacing in the normal directions to the interface should be similar at both sides of the interface. The total number of cells used in the computational domain for the high-Reynolds number models is $5.05\times10^6$ and for the hybrid URANS-LES models is $13.25\times10^6$.

Physical Modelling

The shear layer between the two flow regions in the draft tube produces negative turbulence production \cite{Javadi2015} which makes it difficult for linear eddy-viscosity models to capture the physics of the flow. Javadi et al. \cite{Javadi2015b} studied the curvature correction modeling in hydropower applications and showed that the models that are sensitive to streamline and surface curvatures perform better than the conventional eddy-viscosity models. To resolve more unsteadiness and capture the physics of the on-axis recirculation region and the vortex breakdown, higher order numerical modelling is necessary. Models which have the potential to significantly improve the flow predictions by resolving anisotropy and incorporating more sensitivity and receptivity of the underlying instabilities and unsteadiness such as second-moment closure, hybrid URANS-LES and LES are able to capture the physical mechanisms in the flow fields very well. The second-moment closure models in swirling flows are studied by the authors. These models capture the unsteadiness in the flow fields very well, although they slightly overestimate the on-axis recirculation region.