Due to increase in the power generation from renewable sources, steam and gas turbines will be required to adapt for more flexible operations with frequent start-ups and shut-downs to provide load levelling capacity. During shut-down regimes, mixed convection takes place with natural convection dominance depending on the operating conditions in turbine cavities. Buoyant flows inside the turbine that are responsible for non-uniform cooling leading to thermal stresses and compromise clearances directly limits the operational flexibility. Computational fluid dynamics (CFD) tools are required to predict the flow field during these regimes since direct measurements are extremely difficult to conduct due to the harsh operating conditions. Natural convection with the presence of cross-flow -mixed convection has not been extensively studied to provide detailed measurements. Since the literature lacks of research on such flows with real engine representative operating conditions for CFD validation, the confidence in numerical predictions is rather inadequate. This paper presents a novel experimental facility that has been designed and commissioned to perform very accurate unsteady temperature and flow field measurements in a simplified turbine casing geometry. The facility is capable of reproducing a wide range of Richardson, Grashof and Reynolds numbers which are representative of engine realistic operating conditions. In addition, high fidelity, wall resolved LES with dynamic Smagorinsky subgrid scale model has been performed. The flow field as well as heat transfer characteristics have been accurately captured with LES. Lastly, inadequacy of RANS for mixed type of flows has been highlighted.

Since the renewable sources, which have gained great attention due to the low-carbon policies, are inherently intermittent, the conventional power generation systems will be in use to meet the power demand. These systems, however, must be capable of operating along with renewables, which will lead to a need for more operational flexibility with frequent system ramps. Therefore, understanding and control of thermal stresses and clearances are essential for improving flexibility of conventional power plants. Computational fluid dynamics tools are of great importance in predicting the turbomachinery flows design since the direct measurements of detailed and spatial flow and temperature distribution are often not trivial in the real engines. During shut-down regimes of steam turbines, natural convection takes place along with relatively weak forced convection which is not strong enough to prevent a rising thermal plume leading to a non-uniform cooling in the turbine cavities. Although natural and forced convection have been studied separately in the literature, mixed type of flows in turbine cavities have not been investigated extensively. This paper provides unique experimental data set for validation and development of the predictive tools, which is generated from the detailed flow field measurements in a test facility designed for mixed type of flows in the turbine casing cavities with engine representative conditions. Additionally, large eddy simulations have been performed and validated against the generated experimental data, to gain deeper insight into the flow field. Thus, this paper offers a great insight in these complex flow interactions and unique experimental data for enabling the flexible operations and the development of advanced turbulence modelling.

Due to increase in the power generation from renewable sources, steam and gas turbines will be required to adapt for more flexible operations with frequent start-ups and shut-downs to provide load levelling capacity. During shut-down regimes, mixed convection takes place with natural convection dominance depending on the operating conditions in turbine cavities. Buoyant flows inside the turbine that are responsible for non-uniform cooling leading to thermal stresses and compromise clearances directly limits the operational flexibility. Computational fluid dynamics (CFD) tools are required to predict the flow field during these regimes since direct measurements are extremely difficult to conduct due to the harsh operating conditions. Natural convection with the presence of cross-flow -mixed convection has not been extensively studied to provide detailed measurements. Since the literature lacks of research on such flows with real engine representative operating conditions for CFD validation, the confidence in numerical predictions is rather inadequate. This paper presents a novel experimental facility that has been designed and commissioned to perform very accurate unsteady temperature and flow field measurements in a simplified turbine casing geometry. The facility is capable of reproducing a wide range of Richardson, Grashof and Reynolds numbers which are representative of engine realistic operating conditions. In addition, high fidelity, wall resolved LES with dynamic Smagorinsky subgrid scale model has been performed. The flow field as well as heat transfer characteristics have been accurately captured with LES. Lastly, inadequacy of RANS for mixed type of flows has been highlighted.