Second Stage Optimization of Helical Groove Seals Using Computational Fluid Dynamics to Evaluate the Dependency of Optimized Design on Preswirl
Helical groove seals are non-contacting annular seals used in pumping machinery to increase the efficiency and, in the case of the balance drum, to manage the axial force on the thrust bearing. Prior work has shown that optimization of helical groove seals can reduce the leakage by two thirds given a desired pressure differential or, conversely, can significantly increase the pressure differential across the helical groove seal given a flow rate. This study evaluates the dependency of the optimal helical groove seal design on the inlet preswirl, which is the ratio of the inlet circumferential velocity to the rotor surface speed. To accomplish this goal, second stage optimization from the previously optimized helical groove seal with grooves on the stator and water as the working fluid were conducted at a series of preswirls ranging from −1 to 1. Optimization is performed using ANSYS CFX, a commercial computational fluid dynamics software and mesh independence is confirmed for the baseline case. For each preswirl case, design of experiments for the design parameters of groove width, groove depth, groove spacing, and number of grooves was performed using a Kennard-Stone Algorithm. The optimized solution is interpolated from the simulations run by using multi-factor quadratic regression from the 30 simulations in each optimization and the interpolated solution is simulated for comparison. In addition to evaluating the optimized solution’s dependency on preswirl, the viability of using swirl breaks or swirl promoting inlet passages to improve the overall efficiency of the seal is discussed. Finally, the power loss performance is evaluated for each of the seal designs simulated so that potential trade-offs can be evaluated. Overall, the results show that increasing preswirl can increase the efficiency of the helical groove seal both by improving power loss and by improving leakage.