Design and control of unstable valves for high performance applications

In single-stage electrohydraulic proportional valves, the spools are mainly stroked directly by electromagnetic actuators. They are cheaper and more reliable than multistage servo valves. Their use, however, is restricted to low bandwidth and low flow rate applications due to the force and power limitation of the actuators. The research in this dissertation focuses on improving the performance of single-stage valves in high flow-rate and high frequency applications. The approach adopted is to take advantage of unstable flow induced forces via valve design, thereby alleviating the large solenoid need. The flow induced force models are extensively investigated and verified using Computational Fluid Dynamics (CFD) and experimentally. It has been found that both transient and steady flow forces can be utilized to improve the spool agility by simply changing the valve geometry. Next, a robust valve design methodology is presented to minimize the steady flow forces under variable operating conditions. The optimization problem is formulated as a linear fractional transformation (LFT) so that robust control theories can be applied. The robust optimal solutions show the improved performance compared to the nominal one. Based on the robust design solution, a prototype of the unstable valve is designed and manufactured with some necessary modification. The experimental study of the prototype is extensively conducted. The unstable valve prototype, with the smaller size-3 solenoids, has the faster step response and larger bandwidth than the commercial counterpart with the same flow rating, in which the larger size-5 solenoids are used. The research objectives, using unstable flow forces to improve performance and to alleviate large solenoid requirement, have been verified. In a typical hydraulic application where a cylinder is controlled by the prototype, two different controllers, the velocity feedforward controller and the Internal Model Controller (IMC), are successfully developed. Finally, this dissertation investigates self-sensing solenoid actuators. Based on the relationship between solenoid inductance and spool displacement, two types of model based observers, the Boxcar observer and the Kalman filter, are developed. A self-calibration method is also developed to estimate the parameters of the two-parameter solenoid model. The concepts are verified both in simulation and experimentally.