Highly resolved simulations of turbulence and fluid-structure interaction in blade cascades during transients
||Highly resolved simulations of turbulence and fluid-structure interaction in blade cascades during transients|
||SNIC Large Compute|
||Håkan Nilsson <firstname.lastname@example.org>|
||Chalmers tekniska högskola|
||2020-07-01 – 2021-01-01|
Renewable electric energy production often involves unsteady flow in blade cascades. The unsteadiness can be caused by:
* Turbulence: The high Reynolds numbers gives a wide range of temporal and spatial scales. The numerical modeling techniques must be capable of resolving both the small and large scales, which yields short time steps during long events.
* Rotor-stator interaction: The passing of moving blades through the wakes of steady blades require numerical techniques for the relative motion of blades, and the coupling between different mesh regions. Simulations must be long enough to fully develop the interaction before the results can be used for analysis. Flow instabilities, such as rotating stall, requires even longer time to be fully established.
* Varying operating conditions: The deregulated energy market forces the production units to regulate at a wide range of time-scales. Highly resolved flow simulations usually assume a periodic behavior in both time and space, which is not applicable in energy production units operating under varying operating conditions. The entire process of the variations must be included in the simulations, at the same time that the short time scales must be resolved. This means that highly detailed turbulent flow simulations must be done for a much longer time than is usually the case. Flow instabilities during such operation, such as large-scale vortex breakdown, are highly dependent on accurate representation of the smaller turbulent scales.
We will apply highly resolved LES, DES and hybrid techniques to do detailed studies of the turbulent flow in blade cascades, with rotor-stator interaction and variation in operating conditions. Our present focus is on:
1: Transients between different operating conditions in hydraulic turbines: Very long time scales need to be included, related to those of the transient processes. Very fine meshes and short time steps need to be used to resolve the small scales of turbulence. There is therefore a need for a substantial amount of computational resources. These studies will be done in two separate projects, looking at already established hydraulic turbines v.s. a novel counter-rotating design for pumped-hydro.
2: Fan blades in electric generators: The flow unsteadiness makes the blades vibrate and eventually break, causing great damages to the machines. We will use LES and Fluid-Structure-Interaction methodologies to study the physical processes of the fan blade vibrations. That requires excessive computational meshes, and thus a substantial amount of computational resources.
A single shut-down simulation of a water turbine is estimated to need 480kh, and we need several similar simulations. Just an example - many other large simulations are also needed.