Real-time Power Management of Hybrid Power Systems in All Electric Ship Applications.

Abstract

Motivated by the need for achieving flexible shipboard arrangement and meeting future on-board power demand, the concept of all-electric ships (AES) has been pursued. The integrated power systems enable this initiative by providing a common electrical platform for the propulsion and ship-service loads and are a classic example of hybrid power systems (HPS). In order to leverage the complementary dynamic characteristics of the diverse sources, effective power management (PM) is essential to coordinate the sources and energy storage to achieve efficient power generation and fast load following. Although extensive research has been done on the PM of hybrid land vehicles for commercial applications, this problem for shipboard military applications remains largely unaddressed, leading to its exclusive focus in this dissertation.

While HPS brings in many opportunities for power management, there are many associated challenges for systems used in military applications since both performance as well as survivability criteria have to be satisfied. While the on-demand goal for the power management problem makes real-time control a key requirement, leveraging the look-ahead opportunities for the shipboard missions makes it difficult to attain this goal. Furthermore, the nonlinearity and the complexity of hybrid power systems, make the optimal control of HPS challenging. In this dissertation, we address real-time power management for the AES and general hybrid power systems targeting military applications. The central theme of this work is the development of power management schemes with real-time computational efficiency by exploring HPS dynamic properties, for improved performance (namely fuel economy and fast load following) during normal mode conditions as well as increased survivability during component failure.

A reduced order dynamic HPS model and a scaled test bed is developed as a numerical tool for controller design and validation. The power management (PM) schemes for both normal as well as failure mode conditions are proposed and implemented on a real-time simulator which demonstrated the real-time performance of the proposed method. While the normal mode PM leverages the complementary dynamic characteristics of the HPS for real-time look-ahead control and performance, the failure mode PM uses a reference governor approach for real-time constraint enforcement.