Numerical Analyses for Design and Performance Prediction of Steam Turbine Axial Exhaust Hood

Abstract Over the years the use of renewable energy as a source in Rankine cycles has led to a paradigm shift in steam turbine design. Given the inherently random nature of these energy sources, steam turbines are often forced to operate under offdesign conditions, making the ability to function flexibly across a wide range of discharge pressure a key design objective. In this context, exhaust hoods play a crucial role in the overall performance of the machine, as they facilitate the conversion of kinetic energy into pressure energy, reducing the pressure level at the exit of the last stage and thereby increasing the extractable work. In this context, the highly three-dimensional flow within the exhaust casings and the interaction with the structural components involved lead to an increase in pressure losses that can undermine the recovery capability of the casing itself. For this reason, nowadays, the design phase of the exhaust system can hardly be accomplished without the use of Computational Fluid Dynamics (CFD) as a tool for investigating and quantifying losses. Moreover, it has been widely demonstrated in literature that the flow field of the last stage strongly depends on the conditions inside the exhaust system. Therefore, a reliable simulation of the exhaust system must be coupled with the domain of the last stage of the low-pressure turbine. In this work, several CFD approaches with different levels of precision are presented for estimating the performance of an axial exhaust system currently under development at Baker Hughes in Florence. The objectives of these approaches are to calculate the performance around the nominal conditions and to estimate the low-frequency forces induced by the non-axisymmetry of the exhaust, primarily due to the presence of pipes and struts. In steady simulations context, this can be done only with appropriate interface managements between stationary and rotating domains. Finally, a full-unsteady analysis was performed on three low-load conditions to detect flow instabilities and their effect on the spectrum of the dynamic load on the rotor blades. Furthermore, a method based on the Fourier transform of the pressure signal at specific positions in the casing is also proposed to estimate the dynamic load frequency range during the operation of a real machine under low-load conditions.