Analysis and Optimization of a Two-Way Valve using Response Surface Methodology

This paper describes the use of a numerical procedure developed by the authors for the analysis and optimization of hydraulic components. The element taken as reference is a two-way priority spool valve, typically utilized in steering systems with a load sensing control strategy in the presence of other actuators. The valve's purpose is to control the primary port flow rate, the exceeding flow being discharged to the secondary output port.The optimization algorithm is based on Response Surface Methodology techniques, adopting the path search method known as Steepest Descent. For this purpose, the component's behaviour is analytically described by means of a properly defined objective function. The procedure approximates this objective function with a simple model whose coefficients are evaluated using an AMESim® model of the valve, previously verified using test results. The simulations required to find the fitting model are planned using Design Of Experiments (DOE) methods.Because of the large number of factors characterizing valve design a preliminary analysis (screening) based on DOE algorithms was performed in order to identify the parameters which significantly influence valve behaviour. This allows the important factors to be considered for the optimization phase.The entire numerical procedure was implemented through MATLAB® scripts which automatically execute the AMESIM® simulations to perform the screening analysis or optimization.Considering a configuration pertinent to a stock version of the valve as starting point of the procedure, the paper proposes an optimal configuration. Experimental investigations performed on a prototype reveal the improved performance achieved with the proposed design in comparison with the behaviour observed in different stock versions of the valve, highlighting the potential of the optimization procedure developed.Moreover, the results presented in the paper illustrate how the procedure can also be utilized to perform other analyses of component behaviour, for example, proving, useful guidelines for the definition of dimensional tolerances.