An idealized transonic blade configuration is studied using Delayed Detached Eddy Simulations (DDES), with focus on the loss mechanisms associated with the base pressure and turbulent wake development. The working fluid is Novec649, which is commonly used in low temperature Organic Rankine Cycle (ORC) power plants. Due to the molecular complexity of the fluid, a lower pressure ratio and exit Mach number are achieved compared to an air flow through the same geometry, a phenomenon that becomes more pronounced when considering thermodynamic operating conditions that lead to stronger non-ideal gas behavior, and which is well captured by models of standard use in ORC design and analysis, such as Reynolds-Averaged Navier–Stokes (RANS) models. Overall, as the fluid isentropic exponent and the exit Mach number decrease, the losses tend to increase. When performing a breakdown of the losses, the wake region provides the dominant contribution. Unlike RANS models, the DDES resolves the largest turbulent scales in the wake region, allowing fine-detail analysis of the unsteady wake dynamics, namely vortex shedding, and its influence on the loss coefficients. In the low supersonic Mach number range considered (M2∼1–2), the vortex shedding exhibits a flapping motion in the reattachment region behind the trailing edge. The Strouhal number, classically formed with the diameter of the trailing edge, is found to increase with increasing outlet Mach number. However, when introducing a new scaling based on the neck width of the reattachment region, directly related to the shape of the recirculation bubble and the associated reattachment shocks, the Strouhal number collapses to a relatively constant value of about 0.3 in all cases. DDES also shows that, due to the large heat capacity of Novec649, temperature fluctuations in the wake are greatly reduced, leading to the suppression of the so-called energy separation phenomenon occurring in air flows. While RANS simulations capture satisfactorily the main features and global quantities of the flow, they fail in describing the flow around the trailing edge. This results in inaccurate estimates of the back pressure, which ultimately leads to an underestimation of the losses by about 20%. Unsteady RANS simulations can predict the wake unsteadiness, but do not capture the couplings between large coherent structures, the recirculation, and the shocks, also resulting in inaccurate performance predictions. Thus, the present result showcase the importance of using advanced scale-resolving simulations for improved analysis and design of ORC turbines.
Read full abstract