Empirical modeling of air release in hydraulic reservoirs

Undissolved air is one of the most important contaminants in hydraulic oils, due to the negative effects on the system (damage to the system components, reduction of the oil stiffness), environment (acceleration of oil aging) and human health and well-being (noise pollution). In terms of air release from the system, the hydraulic reservoir is the most significant component. Determining the air release capability of a hydraulic reservoir is important for making profound reservoir selection for a hydraulic system. The air release characteristic of a hydraulic reservoir has been investigated in recent years both experimentally and with three-dimensional numerical simulations. However, both methods are time-consuming and require extensive expert knowledge so that they can hardly be used in an industrial environment. Therefore, so far, there are no efficient calculation methods to back up the design process. Within the scope of this dissertation, the empirical modeling of the air release function of hydraulic reservoirs based on a mathematical approach is presented. This study applies optimal design of experiments and metamodeling techniques for the modeling of the design process. The hydraulic reservoirs considered correspond to the relevant standards in terms of their design and construction. The developed model in this thesis evaluates the air release of the hydraulic reservoirs and represents air content in suction pipe flow depending on relevant variables (input parameters). The database for generating mathematical metamodels is determined using a CFD tool for simulating air bubbles in mineral oil as a hydraulic fluid. The metamodels are created for all standardized sizes according to DIN 24339 and for their configurations with and without baffle plate(s). Baffle plate is an important built-in accessory regarding air release. The statistical evaluation of the model rating shows very good values for its prediction quality. Furthermore, the developed mathematical model can also be applied in one-dimensional simulation, which enables efficient simulation for industrial usage.