15th European Conference on Turbomachinery Fluid dynamics & Thermodynamics

Paper ID:


Main Topic:

Turbulence modelling



Julien Marty - ONERA, Université Paris Saclay, France
Benjamin François - ONERA, Université Paris Saclay, France
Cédric Content - ONERA, Université Paris Saclay, France


Turbine flow is quite specific as the Reynolds number (100,000) is quite small while adverse pressure gradient can be high. A significant part of the flow at the turbine blade walls can be laminar which influences the heat transfer, separation. Thus, the assumption of fully turbulent flow is not well suited for turbine applications. The present study aims at validating the Menter-Langtry transition model for unsteady transition. The configuration supporting this study is the CT3 turbine stage, experimentally tested at the van Karman Institute. To carry out unsteady simulations without simulation the full annulus configuration two methods can be used, among other things: the phase-lag – also called chorochronic – approach and, the sliding join technique based on a reduced blade count leading to a contraction or a dilatation of the rotor-stator interface (the blade geometry is not modified). Thus, the model validation is completed by a comparison of these methods. Moreover, as the elsA software enables the use of structured and unstructured meshes in the same computational domain, this validation is performed on both mesh types. The comparison to experimental data highlights a good agreement and validate the model to predict the time-averaged performances. For this configuration, only the unsteady simulations based on a transition model are able to capture both isentropic Mach and Nusselt numbers over rotor and stator walls. Other simulations (steady one with or without transition model, unsteady one with a full turbulent flow assumption) fail at matching measurements. The mesh type and numerical approach effects are also reported.


Download it! Paper is available for download