Effects of Prandtl’s Secondary Flows of the Second Kind on sCO2 Heat Transfer
Abstract
Numerical simulations of supercritical carbon dioxide (sCO2) in a square duct were conducted to evaluate Prandtl's secondary flows of the second kind and their impact on supercritical heat transfer. A Reynolds stress model (RSM) was implemented in ANSYS Fluent to capture Reynolds stress anisotropy, the mechanism which induces such secondary flows in the corners of non-circular ducts. An RSM was selected as a compromise between computationally expensive high-fidelity models and traditional RANS turbulence models which implement the Boussinesq eddy viscosity hypothesis and therefore, by nature, cannot model the desired anisotropy. The RSM settings were validated against sCO2 DNS data in a circular duct. A square duct was then used to observe the secondary flows. Additional simulations were conducted with the commonly employed k-omega shear stress transport (SST) model to highlight the importance of capturing such nuanced flow phenomena. Results indicate large deviations in local temperature and velocity measurements between the RSM and SST models, especially in local bulk temperature.
Effects of Prandtl’s Secondary Flows of the Second Kind on sCO2 Heat Transfer
Numerical simulations of supercritical carbon dioxide (sCO2) in a square duct were conducted to evaluate Prandtl's secondary flows of the second kind and their impact on supercritical heat transfer. A Reynolds stress model (RSM) was implemented in ANSYS Fluent to capture Reynolds stress anisotropy, the mechanism which induces such secondary flows in the corners of non-circular ducts. An RSM was selected as a compromise between computationally expensive high-fidelity models and traditional RANS turbulence models which implement the Boussinesq eddy viscosity hypothesis and therefore, by nature, cannot model the desired anisotropy. The RSM settings were validated against sCO2 DNS data in a circular duct. A square duct was then used to observe the secondary flows. Additional simulations were conducted with the commonly employed k-omega shear stress transport (SST) model to highlight the importance of capturing such nuanced flow phenomena. Results indicate large deviations in local temperature and velocity measurements between the RSM and SST models, especially in local bulk temperature.