Dynamic Analysis of Axial Piston Pumps under Uncertainty using Karhunen– Loève Decomposition

Abstract

Hydraulic systems are commonly employed in industrial systems and machinery due to their advanced characteristics. Among these, axial piston pumps, as the primary component, play a key role in developments and refinements of modern hydraulic systems. In this study, the Karhunen-Loève (KL) expansion is employed to analyse the pump dynamic characteristics (i.e., outlet pressure and flow rate). A dynamic model of the axial piston pump is formulated to investigate the pump dynamic behaviors using the numerical Runge-Kutta (RK) scheme at a reasonable convergence level. Subsequently, surrogate models are reconstructed using a truncated KL expansion to analyse the pump dynamic responses while accounting uncertainties of system parameters in a high-dimensional space. The results demonstrate that, with a convergence level at a relative error of 1×10⁻⁷, the forth-order truncated surrogate model shows a strong predictive capability in characterizing the pump dynamic responses having transition regions. Additionally, in terms of uncertainty quantification, pump characteristics appear to follow a normal distribution function regardless of whether the input system parameters are normally or uniformly distributed