Dynamic Response
Y. H. Park
Abstract
A hydraulically driven mold oscillator challenges estimating the dynamic state variables precisely. Significantly, the additional stiffness effect of hydraulic oil is variable according to operating conditions, and it is hard to formulate it as a mathematical expression. This study investigates the dynamic ...
Read More
A hydraulically driven mold oscillator challenges estimating the dynamic state variables precisely. Significantly, the additional stiffness effect of hydraulic oil is variable according to operating conditions, and it is hard to formulate it as a mathematical expression. This study investigates the dynamic characteristics of a mold oscillator operated by two hydraulic cylinders with other springs and dampers to determine the non-linear effect and estimate exact dynamic state variables to improve the accuracy control. The mold oscillator is excited in either step oscillation or sine-sweeping oscillation to measure its dynamic behaviors, including mold displacement and hydraulic cylinder pressure. Due to non-linear properties, the dynamic behavior change according to excitation conditions during sine-sweeping oscillation. Primarily, peak frequencies around 50 Hz are found from experimental pressure-displacement data in the frequency domain. To identify the oscillating mechanisms, equivalent 1-DOF and 2-DOF mass-damper-spring models for the mold oscillator are established. The fundamental system property is derived by experiment and a finite element multi-body dynamics model. In addition, inverse dynamics and numerical analysis are applied to derive the unknown force from the hydraulic servo system and structural characteristics. The unknown force is related to a friction problem and an elastic deflection by relative components near the mold. For high accuracy control, the unknown force model by an additional mass-spring model that causes high-frequency vibrations at 49, 48, 47, 46, or 45 Hz is suggested to formulate the equation of motion with the additional vibrations without any arbitrary modeling process.
