Control and Optimization of Fuel Cell Based Powertrain for Automotive Applications

Abstract: Fuel cell powered electric vehicles, with fast-refueling time, high energy density, and zero CO2 emissions, are becoming a promising solution for future fossil-free transportation. However, the relatively slow dynamic response and the inability of recovering the regenerative energy make vehicles solely powered by fuel cells not an immediately attractive solution. Instead, hybrid vehicles powered by fuel cells combined with energy buffers such as batteries and supercapacitors could be of more interest. Due to the unique characteristics of each energy buffer, the vehicle performance may vary with the hybrid energy storage system configuration. This thesis performs a comprehensive study on various energy storage configurations for applications in fuel cell hybrid electric vehicles. This thesis first examines the fuel cell/supercapacitor passive hybrid configuration where the fuel cell and supercapacitor share the same DC-link voltage. The power distribution between them is inherently determined by their internal resistances. Therefore, the DC-link voltage varies and depends on the vehicle power demand. In this work, a fuel cell/supercapacitor passive hybrid powertrain is first modeled and evaluated. Simulation results show that the energy efficiency is 53%–71% during propulsion and 84%–94% during braking, respectively. Moreover, a 3 kW lab-scale fuel cell/supercapacitor passive hybrid system is designed and investigated. Experimental results show that the fuel cell takes time to respond to a load change, while the supercapacitor provides the transient power, which makes it possible to downsize the fuel cell. Since the passive configuration loses the active controllability, this thesis further considers a fully-active fuel cell/supercapacitor system to improve the controllability of the power distribution. This configuration requires a boost converter for the fuel cell and a buck-boost converter for the supercapacitor. In this work, an adaptive power split method is used to smooth the fuel cell current and prevent the supercapacitor from exceeding its lower and upper charge limits. The cut-off frequency of the low-pass filter is adaptively controlled by the spectrum area ratio. Experimental results show that the supercapacitor state-of-charge is effectively controlled within the desired range. Moreover, a load disturbance compensator is proposed and demonstrated to improve the control performance such that the DC-link voltage fluctuation caused by the load current variation is significantly reduced. This thesis also investigates the cost-effectiveness of different energy buffers hybridized with fuel cells in various trucking applications. First, a chance-constraint co-design optimization problem is formulated. Convex modeling steps are presented to show that the problem can be decomposed and solved using convex programming. Results show that the power rating of the electric machine can be dramatically reduced when the delivered power is satisfied in a probabilistic sense. Moreover, the hybridization of fuel cells with lithium-ion batteries results in the lowest cost while the vehicle using lithium-ion capacitors as the energy buffer can carry the heaviest payload. This work also develops a robust co-design optimization framework considering the uncertainties in parameters (e.g., vehicle movement) and design decision variables (e.g., scaling factors of fuel cells and batteries). Results show that these uncertainties might propagate to uncertainties in state variables (e.g., battery energy) and optimization variables (e.g., battery power), leading to a larger battery capacity and therefore a higher total cost in robust optimal solutions. In summary, this thesis performs a comprehensive study on control and optimization of fuel cell based powertrains for automotive applications. This will provide a guidance on component selection and sizing, as well as powertrain system configuration and optimization for design of fuel cell powered electric vehicles.