Mapeamento dos coeficientes dinâmicos de selos planos de máquinas rotativas

Abstract

Mechanical seals are widely used components in rotating machinery, designed to prevent fluid leakage. The study of their dynamic coefficients is essential in the context of Rotating Machinery Dynamics. Typically, seals are incorporated into the equations of motion of rotors through stiffness and damping coefficients, directly influencing the vibration response and operational stability of these machines. The dynamic coefficients of mechanical seals with simple geometries, such as plain seals, are determined using simplified mathematical/physical models for the seal's lubricant fluid flow, known as bulk-flow models. A common simplification in such models is assuming the shaft center position is concentric with the seal, thereby neglecting significant variations in the dynamic coefficients calculated for eccentric positions. Consequently, when relating the rotor center's displacements and velocities with stiffness and damping coefficients calculated for a concentric position, the effects of these variations may not be accurately represented in the vibration response to unbalance. In this context, the present work proposes a bulk-flow model to calculate the dynamic coefficients of plain seals using the Finite Volume Method (FVM) without simplifications regarding the rotor center's position. The model is developed in a Python environment and integrated with the Rotordynamic Open-Source Software (ROSS) library. Furthermore, this work evaluates the impact of variations in these dynamic coefficients with the rotor center's position on the vibration response to unbalance of a water injection pump. This is achieved through a comparative analysis between the vibration response to unbalance determined using dynamic coefficients calculated for a position concentric with the seal and the vibration response to unbalance obtained considering dynamic coefficients varying with the rotor center's position, using for that the mapping of the dynamic coefficients and Kriging metamodels.