Abstract

The growing environmental concerns have prompted the use of alternate fuels and a strong demand for electronically controlled gas injectors. This thesis investigates a solenoid operated fuel injector discharging gaseous fuel directly into the combustion chamber of an engine. A switching circuits is designed to boost the solenoid coil current for quick electromagnetic force generation to perform a fast injector opening. A capacitor incorporated in the switching circuit is used to oppose and to quickly diminish the coil current for fast injector closing. To accurately control the gas injection transient process, within fractions of a millisecond, a model of the solenoid magnetic circuit, including effects from eddy current and flux leakage losses in order to account for the hysteresis of the magnetization characteristics for force generation, is developed and matched with experimental results. The performance studies showed that the stringent opening and closing timing requirements for the injector can be improved with proper selection of the design variables for the best performance. A systematic multivariable multiobjective constrained optimization procedure is developed to establish an optimal design. An optimized injection system could minimize the time delays and shape the profile of the injector needle motion in order to reduce the deviation between the expected and actual amount of the injected gas dose. A modified version of the Hooke and Jeeves unconstrained optimization method is used. Boundary constraints are placed on the design variables to ensure that the design is confined within realizable physical limits. To account for these constraints, a procedure is introduced that tracks three successive objective functions as the optimization iteration progresses, and when a boundary is violated. The penalty function method accounts for all other constraints. There are no known optimization methods that will result in a global minimum. Even if a global minimum is known to exist, then to determine the global minimum is to perform an exhaustive minimization search. The optimization method presented in this thesis results in a global minimization without the need to perform an exhaustive search on the design variables. This is because there is some "a priori" knowledge from the performance study about variable interactions, and about the trend of system performance for changes in the design variables. The research shows that a solenoid operated injector with small size and weight, can operate fast enough and accurately in direct injection applications, after its design is optimized with the use of a comprehensive solenoid model for fast transient operations.