Despite LES and DNS being widely used in academia, the rapid increase in computational
power in recent years has not led to wide use of computation-intensive models for the design
of turbomachinery components in commercial companies. Instead, simulations with reduced
order turbulence and transition models are used much more frequently, a situation which is to
stay for several years. In order to ensure quick turnaround times in optimization processes,
steady state RANS methods are often preferred to unsteady time-resolved computations.
When utilized in multistage turbomachinery applications, steady state computations involve
the use of mixing plane interfaces that perform a circumferential averaging of all flow
variables. While this is the preferred way in steady state simulations as the circumferential
averaging of periodically unsteady flow results in a steady boundary condition for the
downstream blade rows, the use of mixing planes also leads to unwanted side effects that
ultimately lead to altered production and destruction mechanics of turbulent kinetic energy
k. As k is an important variable that largely impacts the turbulent stress computation and is
also the input for widely used correlation-based transition models such as the γ − Reθ model,
it is important that the distribution of turbulent kinetic energy in the flow field is accurately
predicted. This study presents an extension for k − ω based turbulence models using one
additional transport equation in order to improve the modeling of turbulent kinetic energy
decay while correcting the eddy viscosity μt
that also greatly impacts the flow solution. In
test cases of gradually increasing complexity, it is shown that the Multistage Turbulence
Extension (MTE) is able to produce equal or better results than the status quo at only minimal
additional computational cost.
«Despite LES and DNS being widely used in academia, the rapid increase in computational
power in recent years has not led to wide use of computation-intensive models for the design
of turbomachinery components in commercial companies. Instead, simulations with reduced
order turbulence and transition models are used much more frequently, a situation which is to
stay for several years. In order to ensure quick turnaround times in optimization processes,
steady state RANS methods are often...
»