Control, Modeling, and Design Towards Higher Performance Energy Systems

Abstract

Fostered by the high controllability, speed, and power density of power electronics, future energy systems can be more heterogeneous in form, more agile in actuation, larger in scale, and more distributed in function. The ultimate energy system should deliver the exact energy at the exact time. However, the path of energy transfer is still restricted by the cost of power converters. Moreover, the bandwidth of energy transfer is limited by the speed of power converters.

Power converters must endeavor to deliver the exact energy at the exact time. Therefore, the dissertation follows two themes: (I) better power-processing architectures to move the right amount of energy to the right place; (II) better power-control frameworks to move the right amount of energy at the right time.

The first part of this dissertation investigates a new power-processing architecture that enables more effective energy transfer. Dispensing energy effectively means high utilization of energy storage, high power efficiency, cost-effectiveness, and sustainability. Reusing retired electric vehicle batteries for energy storage has a significant impact on both sustainability and the economy because these batteries are recycled when the capacity only drops by about 20%; this impact will grow with the inevitable dominance of the EV market. The technical challenge in this solution is that second-use batteries are heterogeneous in their capability which includes their capacities, voltages, and state of health. The conventional approach, in which every battery pack needs a power converter, is both expensive and inefficient. We provide a hierarchical partial-power-processing architecture that is sparse in power converters. Power converters are designed in multiple layers to handle the different heterogeneities of the second-use batteries. This architecture allows designers to push the multiple tradeoff boundaries which include performance, cost, and heterogeneity.

The second part of this dissertation studies a new control framework that enables faster energy delivery. Quicker energy delivery is demanded in many emerging applications such as LiDAR, microprocessors, communications circuits, and memory. We propose three new technologies that overcome the speed limitations from three perspectives. (1) Cycle-by-cycle digital control and variable-frequency power converters are better for fast energy delivery. We provide a switching-synchronized sampled state space (5S) framework which enables the simple-yet-effective controller design together with mathematically rigorous stability analysis. The transient response of the dc-dc converters is experimentally improved by more than a factor of two. (2) Interference in current measurement limits the highest switching frequency of current-mode power converters. We demonstrate a control conditioning approach that makes the interference inherently part of the controller design. The resulting current-mode power converter can safely switch as high as 5MHz while multi-megahertz current-mode converters are rarely reported in the literature. (3) The fastest transient response of typical dc-dc converters is fundamentally limited by the inductor current slew rate. We employ a saturating inductor with a systematic and rigorous control design to increase the slew rate without introducing extra hardware. The transient speed is boosted by 40% with a saturating inductor.

Supported by the aforementioned innovative modeling frameworks and advanced control techniques, the speed, and efficiency of energy systems are significantly improved as validated by extensive simulation and hardware experiments.