Sizing of a Propeller for Use in an Electric Boat, Simulation of Turbulent Flow Around the Propeller Using the Generalized k-omega (GEKO) Turbulence Model and Prediction of Propeller Performance

Project no. 31857 · Status: Kapanmış · 2023

Summary

This study presents comparative simulations of flow around marine propellers using the Generalized k-omega (GEKO) turbulence model, which is designed to unify different two-equation turbulence models commonly used in industrial flow scenarios. The adjustable parameters of GEKO allow for aligning RANS simulations with available experimental data while maintaining the core calibration of the base model. Computational flow simulations conducted on the INSEAN E779A propeller examine the influence of GEKO parameters CSEP, CMIX, and CCURV on the flow. The analysis focuses on operating conditions at very low (J = 0.05), low (J = 0.2), optimal (J = 0.88), and high (J = 1.012) advance ratios. The results indicate that CSEP significantly affects the 2D flow around blade sections but has minimal influence on blade boundary layers in 3D simulations due to rotational effects promoting attached flow conditions. However, this trend changes at low advance ratios, such as J = 0.05, where blade separation regions become prominent. Under these conditions, the SST model yields up to 18% error in thrust and torque coefficients compared to experimental data, whereas the GEKO model, with appropriately chosen CSEP parameters, reduces this error to 3%. Consequently, the GEKO model demonstrates consistency with experimental findings under separated propeller flow conditions in this project. Wake flow analyses reveal that CMIX increases the turbulent viscosity ratio, while CCURV suppresses turbulence levels at vortex cores, thereby extending the lifespan of tip and hub vortices downstream. Overall, GEKO parameters exhibit moderate effects on marine propeller flows characterized by mid-level turbulence. An exception arises under very low advance ratio conditions, where default GEKO parameters produce flow fields differing from those of the SST model. This highlights the necessity of case-specific model calibration procedures rather than relying on default parameter values for accurate predictions.

View details