Florida State University Libraries

Transcription

1 Florida State University Libraries Electronic Theses, Treatises and Dissertations The Graduate School 212 A Distributed Control Approach for Power and Energy Management in a Notional Shipboard Power System Qunying Shen Follow this and additional works at the FSU Digital Library. For more information, please contact lib-ir@fsu.edu

2 THE FLORIDA STATE UNIVERSITY COLLEGE OF ENGINEERING A DISTRIBUTED CONTROL APPROACH FOR POWER AND ENERGY MANAGEMENT IN A NOTIONAL SHIPBOARD POWER SYSTEM By QUNYING SHEN A Thesis submitted to the Department of Mechanical Engineering in partial fulfillment of the requirements for the degree of Master of Science Degree Awarded: Spring Semester, 212

3 Qunying Shen defended this thesis on January 13 th 212. The members of the supervisory committee were: David A. Cartes Professor Directing Thesis Emmanuel G. Collins Committee Member Leon Van Dommelen Committee Member Sanjeev K. Srivastava Committee Member The Graduate School has verified and approved the above-named committee members, and certifies that the thesis has been approved in accordance with university requirements. ii

4 This work is dedicated to: My parents and my husband, Jiabo Zhang iii

5 ACKNOWLEDGEMENTS I would like to thank my advisor, Dr. David A. Cartes. I would not have been able to start this work without your confidence on me. Your support and guidance were always a great accompany through my journey in completing this thesis work. I would also like to thank my supervisor, Dr. Sanjeev K. Srivastava, and Dr. Bhuvaneswari Ramachandran for their guidance and help on this work. My thanks also go to my other committee members, Dr. Emmanuel G. Collins and Dr. Leon Van Dommelen for their support and help. This work is based on the MVDC system which has been built by Michael Andrus using the Real Time Digital Simulator software RSCAD. I especially want to express my gratitude to Michael for his patience in answering my questions and his valuable advice for solving problems. I would also like to thank Troy Bevis, who has helped me to test the developed power controller algorithm on the hardware testing bed. I also extend my appreciation and thanks to all of the other people who have helped me in this work at the Center for Advanced Power Systems. This work is sponsored by the Office of Naval Research under grant # N iv

6 TABLE OF CONTENTS List of Tables... viii List of Figures... ix Abstract... xii INTRODUCTION Background Motivation Summary of Thesis Contributions Thesis Outline... 3 LITERATURE REVIEW Electric Warship History of Electric Ship Integrated Power System Next Generation Integrated Power System Power and Energy Management Objective Application of Power and Energy Management Quality of Service Multi-agent System Multi-agent System Theory Multi-agent System Application PRELIMINARY DESIGN Design Objectives Notional MVDC NGIPS Model NGIPS Module Types MVDC Topology v

7 3.2.3 RTDS Ship Model Power and Energy Management System System Definition and Architecture PCON Based Control Scheme PEMS Functionalities Energy Storage Module Energy Storage Technology Ultra-capacitors DC-DC Converter System and Zonal Energy Storage Module Control Design Load Shedding Two Types of Strategy Implementation in MVDC IPS SIMULATION System Initiate-Mode Selection Normal System Status Status under Launch Schedule Energy Storage Module Charging and Discharging SESM for Pulse Load ZESM for Zonal Load Load Shedding Fault Scenarios Loss of Main Generator vi

8 HARDWARE TESTING Multi-agent System What an Agent is Multi-agent System Advantages Multi-agent System Layout and Agent Communication Hardware Test Bed Agent Control Cases Loss of Main Generator SESM for Pulse Load CONCLUSION AND FUTURE WORK... 9 Appendix A Appendix B Appendix C References Biographical Sketch... 1 vii

9 LIST OF TABLES Table 2. 1 Comparison of general characteristics between non-ips and IPS... 1 Table 3. 1 Zone 1 loads category Table 3. 2 Power output for different modes when four generators online Table 3. 3 Generator launch schedule for different modes Table 3. 4 Comparison of energy storage technologies Table 3. 5 Converter parameters for SESM and ZESM Table 3. 6 Quality of service load category [19] Table 3. 7 Mission priority load category Table 5. 1 Agent properties and their definitions... 7 viii

10 LIST OF FIGURES Fig USS Langley off San Diego, CA [8]... 5 Fig DDG 1 renderings [9]... 5 Fig Power and propulsion systems integrated on steam side [8]... 7 Fig Power and propulsion systems separated [11]... 7 Fig Power and propulsion systems integrated on electrical side by IPS [11]... 7 Fig Anticipated power demands for sensor and weaponry in future [8]... 8 Fig Comparison of non-ips and IPS to fulfill projected power demand [4]... 9 Fig NGIPS technology development roadmap [4] Fig Classification of areas to use energy management [25] Fig Notional design example of high performance MVDC ship system design [3] Fig Zonal load center configuration [31]... 2 Fig Zonal loads deployment in RSCAD Fig Two-layer hierarchical architecture Fig Control scheme for power and energy management system Fig Schematic of a conventional capacitor [43]... 3 Fig Taxonomy of supercapacitors [43] Fig Schematic of an electrochemical double-layer capacitor [43] Fig Equivalent circuit of ultra-capacitor Fig Location of energy storage module Fig Circuit diagram of bi-directional DC-DC Converter [45] Fig Simulation module- dc/dc convertor with capacitor Fig SESM efficiency in constant power control Fig SESM efficiency in voltage control Fig SESM voltage and current variations in constant power control Fig SESM voltage and current variations in voltage control... 4 Fig ZESM efficiency in constant power control Fig ZESM efficiency in constant voltage control Fig ZESM voltage and current variations in constant power control Fig ZESM voltage and current variations in voltage control Fig Control configuration for bi-directional dc/dc converter ix

11 Fig Normal system status under battle mode Fig Normal system status under cruise mode Fig Normal system status under anchor mode... 5 Fig Cruise mode under generator launch schedule Fig Battle mode under generator launch schedule Fig From battle to anchor mode under generator launch schedule Fig Zonal energy storage module during charging Fig Zonal energy storage module during discharging Fig System energy storage module during charging Fig System energy storage module during discharging Fig System energy storage module for 7.41s-1MW pulse load in battle mode Fig System energy storage module for 3.33s-5MW 3-pulse burst load in cruise mode... 6 Fig Zonal energy storage in discharging-pcm 1 blocked in buck mode Fig Response in cruise mode-loss of main gen 1 w/ power&energy management Fig Response in cruise mode-main gen 1 restoration w/ power&energy management Fig Response in battle mode-loss of main gen 1 w/ power&energy management Fig Response in battle mode-loss of main gen 1 w/o power&energy management Fig Response in battle mode-main gen 1 restoration w/ power&energy management Fig Response in battle mode-main gen 1 restoration w/o power&energy management Fig Multi-agent system layout Fig MAMBA board test bed Fig Connection architecture of multi-agent hardware test bed Fig Loss of generator-agent communication Fig Agent control signal initial values Fig Main generator offline-agent signals and ESM modes Fig System status in cruise mode-main generator 1 offline-agent control... 8 Fig System status in cruise mode-main generator 1 restored-agent control Fig SESM discharging in cruise mode-main generator 1 offline-agent control Fig SESM back to charging in cruise mode-main generator 1 restored-agent control Fig System status in battle mode-main generator 1 offline-agent control Fig System status in battle mode-main generator 1 restored-agent control x

12 Fig ZESM in discharging-main generator 1 offline-agent control Fig SESM supply for pulse load-agent communication Fig SESM supply for pulse load in battle mode-agent control Fig SESM status in battle mode during pulse load-agent control Fig. B. 1 Closed-loop control block diagram [45] xi

13 ABSTRACT The main goal of this thesis is to present a power control module (PCON) based approach for power and energy management and to examine its control capability in shipboard power system (SPS). The proposed control scheme is implemented in a notional medium voltage direct current (MVDC) integrated power system (IPS) for electric ship. To realize the control functions such as ship mode selection, generator launch schedule, blackout monitoring, and fault ridethrough, a PCON based distributed power and energy management system (PEMS) is developed. The control scheme is proposed as two-layer hierarchical architecture with system level on the top as the supervisory control and zonal level on the bottom as the decentralized control, which is based on the zonal distribution characteristic of the notional MVDC IPS that was proposed as one of the approaches for Next Generation Integrated Power System (NGIPS) by Norbert Doerry. Several types of modules with different functionalities are used to derive the control scheme in detail for the notional MVDC IPS. Those modules include the power generation module (PGM) that controls the function of generators, the power conversion module (PCM) that controls the functions of DC/DC or DC/AC converters, etc. Among them, the power control module (PCON) plays a critical role in the PEMS. It is the core of the control process. PCONs in the PEMS interact with all the other modules, such as power propulsion module (PPM), energy storage module (ESM), load shedding module (LSHED), and human machine interface (HMI) to realize the control algorithm in PEMS. The proposed control scheme is implemented in real time using the real time digital simulator (RTDS) to verify its validity. To achieve this, a system level energy storage module (SESM) and a zonal level energy storage module (ZESM) are developed in RTDS to cooperate with PCONs to realize the control functionalities. In addition, a load shedding module which takes into account the reliability of power supply (in terms of quality of service) is developed. This module can supply uninterruptible power to the mission critical loads. In addition, a multi-agent system (MAS) based framework is proposed to implement the PCON based PEMS through a hardware setup that is composed of MAMBA boards and FPGA interface. Agents are implemented using Java Agent DEvelopment Framework (JADE). Various test scenarios were tested to validate the approach. xii

14 CHAPTER ONE INTRODUCTION 1.1 Background Various components such as generators, cables, switchboards, load centers circuit breakers, bus transfer switches and loads constitute a typical shipboard electrical power system [1]. The integrated power system (IPS) is a design where the ship s service electrical system and the electric drive propulsion system are powered from a common electric bus. The advantage of the IPS is that it makes more intensively use of power electronics and control than the traditional shipboard power systems. It also increases the combat survivability by improved reconfiguration ability, where survivability relates to the ability of the power system to support ship s ability to continue its missions when it is damaged by a threat [2]. Electrical power can be more easily redirected to undamaged propulsion systems or mission combat systems during damage to the ship than mechanical power [3]. A program called Next Generation Integrated Power System (NGIPS) is under development by the Office of Naval Research (ONR) which will help the U.S. Navy to produce affordable power solutions for future all-electric ships. For the all-electric ship, all the steam powered, hydraulically powered, or pneumatically powered auxiliary systems will be converted to electrical powered systems. NGIPS is both a business and technical approach to define standards and interfaces, to increase commonality among combatants (surface and subsurface), and to efficiently use installed power [4]. Incorporating a Medium Voltage DC (MVDC) IPS for future surface combatants and submarines was decided as the goal of Navy in NGIPS technology development roadmap. It can offer smaller, simpler, more affordable and more capable ship s power system for future Navy platform [4]. 1.2 Motivation There are increasing power demands for future combatants armed with advanced sensors and weaponry loads such as lasers and rail guns [4]. These projected demands usually require high amplitude power within short period, which cause a step change to the system and will cause system collapses. On the other hand, increasing fuel cost forms a big portion of operating 1

15 cost, thus reduction of fuel consumption is also highly needed. A high level power and energy management system (PEMS) is necessary to satisfy these demands. It can assure that sufficient power be offered to all loads with sufficient rolling reserve to address possible step load changes during normal conditions. Energy balance between generation and loads during transients by large step load changes can be ensured too. Meanwhile, system survivability is also considered to achieve a more reliable and efficient shipboard power system. The Quality of Service (QoS) is measured as a Mean Time Between Service Interruption (MTBS) to verify the ability of the power system to supply power to loads and fulfill their power continuity requirements [5]. The PEMS should help to keep the continuity of power supply to the loads under normal conditions, and assure the ship s ability to continue its mission under threats, resulted in enhanced survivability. To realize these objectives, it is important to have the system provide the optimal allocation and distribution of energy onboard with the minimum operational cost, and should improve the controller and system robustness to faults and disturbances as well. For a shipboard integrated power system (SIPS), power can be supplied by several power sources such as generators, energy storage elements, and fuel cells. The design of these sources must consider that some loads in the system can be sensitive to an interruption of power supply caused by a system failure even for a very short time, which will be counted as a system failure in the QoS calculation. Overall, the PEMS objective is to deliver the right power to the right load at the right time to ensure that the SIPS satisfies the conditions of QoS and survivability. Furthermore, a power control module (PCON) is needed for the PEMS to realize all the control functionalities that are needed to accomplish the above mentioned objective. 1.3 Summary of Thesis Contributions The main contributions of this thesis are summarized as following: Development of a new PCON based power and energy management system for the shipboard integrated power system, Development of ultra-capacitor based energy storage module for shipboard integrated power system that is crucial to the basic functionalities of the PEMS, Development of load shedding module in a notional MVDC IPS, as primary functions of PEMS, 2

16 Implementation of the proposed distributed control scheme in a notional medium voltage direct current integrated power system, and Implementation of a multi-agent based framework in hardware testing to validate the recommendation of the researched PEMS for control of MVDC IPS. 1.4 Thesis Outline The rest of the thesis is organized as following: Chapter 2: The background information related to this research topic is discussed. Chapter 3: A high level PEMS is proposed and concerns of the control scheme design are discussed. Chapter 4: Implementation of proposed method in a notional MVDC IPS developed in Real Time Digital Simulator (RTDS) software RSCAD. The simulation results are shown to validate the function of the design. Chapter 5: A multi-agent based framework for hardware testing is proposed and implemented in real time using hardware setup (the MAMBA boards and FPGA interface) and RTDS. Chapter 6: Overall conclusion for this research is drawn and some possible fields of future work are discussed. 3

17 CHAPTER TWO LITERATURE REVIEW This chapter is intended to introduce the past literature work that dealt with related topics. In section 2.1, a basic electric ship history is introduced. The development of power and propulsion systems is discussed. The benefits and necessities for integrated power system are elaborated. An NGIPS approach is introduced for future IPS development. In section 2.2, research work that has been carried out so far on power and energy management is discussed. In section 2.3, quality of service as the objective of power and energy management is introduced. The multi-agent system, which is used for power and energy management is discussed in section Electric Warship History of Electric Ship The first U.S. naval ship powered by electric motors was the Navy s USS Langley (formerly Jupiter), launched in At that time, the electrically powered naval vessels employed two electrical systems: one for propulsion and the other for service loads [6]. Having two separate electric systems resulted in low efficiency, which made the most electric-drive ships fall out of favor after the 194s. Much after, in the 198s, shift occurred for commercial ship builders, when the development of motor drives allowed for the seamless connection of the two electrical systems. As a result, for the past 2 years, the cruise ships and military ships have been evolving towards electrically powered systems [6]. A shift from mechanical propulsion to electric propulsion produces a significant increase of electric power demand onboard in ships. The situation is more complex for all electric warships, where, besides electric propulsion, we may have those intermittent loads such as electronic weaponry or sensors that demand power with high amplitude and short duration [7], in the tens of MW range. Fig. 2.1 and Fig. 2.2 are two electric ships in U.S. Navy s electric ship history. The USS Langley (CV-1) in Fig. 2.1 [8] was the U.S. Navy s first aircraft carrier, and also the U.S. Navy s first electrically propelled ship. Fig. 2.2 [9] is the rendering of the first Zumwalt class destroyer 4

18 DDG 1, which features an integrated power system. These two ships represent over 8 years of electrical evolution onboard U.S. Navy vessels. Fig USS Langley off San Diego, CA [8] Fig DDG 1 renderings [9] 5

19 The electric ship has a lot of benefits because of the application of electric components and systems. They have lower weight, improved survivability, high fuel efficiency, and more compartment arrangement capability. For the shipboard power system design, survivability and continuity of power supply were regarded as the two primary objectives. Survivability is the ship s ability to stay working under its mission even under adverse conditions. Quality of Service (QOS) serves as a metric of the continuity of the electrical power supply for the normal, undamaged operation of its loads [1] Integrated Power System In the early days of electrification of ships, the ship s power and propulsion systems were supported by different stream turbines; however, multiple stream turbines were often powered by one or two boilers, which mean that the power and propulsion systems were highly coupled and vulnerable on the thermal energy side of the system, see Fig. 2.3 [8]. When evolution led to internal combustion engines, power and propulsion systems were separated completely adding considerable reliability to naval ships. They have individual prime movers as their energy sources. In Fig. 2.4 [11], the power conversion and distribution system on the right is powered by three prime movers and the propulsion system with the reduction gears is powered by another four prime movers. The IPS returns to share the source on the electrical side, which is enabled by solid state power electronics, multi-megawatt motor drives and automated controls. From Fig. 2.5 [11], the generators supply power to a common electric bus for propulsion motor drivers and ship service loads. Fewer prime movers are used for the system than conventional electric ship, which has a separated power source. In this way, the operation cost is saved not to mention more space to equip some other systems on the ship. It also offers the possibility to have more efficient control. For future combatants armed with advanced sensors and weapons such as rail guns and lasers, a high power is demanded. The U.S. Navy has been rapidly migrating back ship designs with electric propulsion, as a way to this high power demand. The anticipated power demands for future weapons are shown in Fig. 2.6 [8]. The power demand ranges from.4~3mw for the year from 214 to 22. To address these increasing power demands, ship designs are using IPS that provides electric power to propulsion and other electrical loads from a common set of sources, which supports significantly more electrical power than in the past and achieve this in a more efficient way. 6

20 Fig Power and propulsion systems integrated on steam side [8] Fig Power and propulsion systems separated [11] Fig Power and propulsion systems integrated on electrical side by IPS [11] 7

21 A comparison of a non-ips or IPS to fulfill projected power demands is shown in Figure 2.7 [4]. The DD-963 was the Spruance-class destroyers and DDG-51 FLT IIA was the Arleigh Bruke class of guided missile destroyer, both of them were with non-ips systems. The DDG 1 is the Zumwalt class destroyer equipped with IPS. Obviously, the IPS can satisfy much more power demand for ship service and mission systems, which include the service load such as the lights, pumps, and also the weaponry loads. Fig Anticipated power demands for sensor and weaponry in future [8] 8

22 Fig Comparison of non-ips and IPS to fulfill projected power demand [4] Table 2.1 shows the comparison of general characteristics between a non-ips Arleigh burke class destroyer DDG-64 and an IPS Zumwalt class destroyer DDG-1. From the characteristics, two items need to be noticed are the displacement and crew size. With automation system, the crew size is sharply decreased, which promises a great saving on operation cost. 9

23 Table 2. 1 Comparison of general characteristics between non-ips and IPS USS DDG-64 (non-ips) USS DDG-1 (IPS) Characteristics Length (ft) 55 6 Beam (ft) Draft* (ft) Speed (knots) 3 3 Displacement 8,9 14,564 (long tons) Installed Power(MW) Crew Size *Vertical distance between the waterline and the bottom of the hull Next Generation Integrated Power System To provide direction for future IPS development, the Navy initiated the Next Generation Integrated Power Systems (NGIPS) effort [1]. The NGIPS was described initially by Doerry [4]. Progress towards achieving these goals is described by Hoffman et al in [8]. The NGIPS enterprise approach seeks to improve the power density and affordability of Navy power systems and deploy appropriate architectures, systems, and components as they are ready into ship acquisition programs. The NGIPS technical approach utilizes common elements such as Zonal Electrical Distribution Systems (ZEDS), power conversion modules, and electric power control modules as enabler along an evolutionary development path [4]. A roadmap that defines path for NGIPS development is shown in Fig. 2.8 [4] which provides guidance to Navy and industry developing organizations and forms the basis for coordinated planning and future Navy investments [4]. The work of this thesis is focused on the power and energy control, which is part of the technologies for future electric warships in NGIPS roadmap. 1

24 A number of technical challenges need to be addressed to implement MVDC on surface combatants and submarines since naval warships require high power density. For amphibious warfare and auxiliary ships not requiring high power density, the MVAC systems will for the foreseeable future provide the most affordable solution for power generation and ship-wide distribution. Fig NGIPS technology development roadmap [4] 2.2 Power and Energy Management Objective With the widespread use of power electronics in shipboard power system (SPS), control of the power system implies a more complex interaction of switches and power converter control that allocates the actual flow of power [12]. An effective control approach is critical to achieve the power and energy management for SPS with the consideration of power balance and survivability. 11

25 The objective of a power management is to ensure continuous power supply for all loads under normal conditions with sufficient rolling reserve [13] and to supply continuous power to the mission critical loads under adverse conditions, which are defined by the mission priority, for the system survivability concern. By augmenting the survivability of the shipboard power system, the mission effectiveness of the warship is consequently enhanced [14]. At this point, by using PEMS can eliminate the preventable component failures, or some other life cycle costs, resulted in a lowered operation cost of the electric ships Application of Power and Energy Management Power and energy management has been used in several areas such as terrestrial power systems, microgrids, and space stations although not too much work has been done in naval vessel electric systems. In terrestrial power systems, Energy Management System (EMS) was proposed to perform Supervisory Control and Data Acquisition (SCADA) capabilities along with generation dispatch, scheduling and control capabilities [15]. Some researchers are also working on using energy management in microgrids, which is a localized grouping of electricity sources and loads that normally operates connected to the synchronous with the traditional centralized grid (microgrid) but can disconnect and function autonomously as physical and/or economic conditions dictate [16]. In microgrids, energy management system is used to make decisions regarding the best use of the generators for producing electric power and heat. Decisions will be made upon many considerations, such as the heat requirements of the local equipment, the weather, the price of electric power, and the cost of fuel [17]. Power and energy management is also used in the international space station for controlling the energy output from the energy storage module viz. the battery. The international space station survives on solar power and hence have several solar arrays stacked on the surface of the station. When sun is behind the earth, and when there is no solar power available to feed the loads, all the power is provided off the batteries, providing about a third of the station s power daily [18]. Reviewing the applications of the power and energy management in several areas reveals the benefits of using power and energy management. Since the all electric ship is expected to have high survivability especially in a battle, having an efficient PEMS that can control and 12

26 manage power flow between the generation and loads to maintain power balance, quality of service, and survivability is critical Quality of Service Quality of Service (QoS) is a metric of how reliable an electrical power system is, in providing electrical power to the user according to his requirements [19]. The MTBSI, mean time between service interruptions, is used to measure it. A service interruption is a power interruption longer in duration than a load can tolerate [5]. In the terrestrial sense, power interruption or departure from the required power quality that causes the load to be unable to perform its required function can be considered as a failure [2]. An energy storage device is helpful to balance generation and demand during transients for system reconfigurations or large step changes. Otherwise, a discontinuity of power service will occur in the system, which is counted as a system failure. The large step changes could be the pulse loads, which are war fighting loads such as electromagnetic guns, electromagnetic launch system, and free electron lasers [21], they draw high amplitude power within very short time. They re two potential problems for using energy storage to balance the power: overfilling and completely emptying, both of which will lead to a system failure [13]. A good control strategy for charging and discharging will solve this problem. This thesis suggests that using energy storage and load shedding will significantly mitigate the QoS failure. Energy storage unit work as the additional power source to distribute power to vital loads, which is defined by the duration the loads can tolerate an interruption of power supply. Based on the different tolerate capabilities, some loads called nonvital loads, are which can tolerate a long term interruption. These loads will be shed to assure that the uninterruptible loads, or vital loads in this work, will not experience a QoS failure during the limited power supply caused by component error or some other system failures. 2.3 Multi-agent System A multi-agent system (MAS) is a system composed of multiple interactions of intelligent agents [22]. MASs can be used to solve complex problems that are often difficult for an 13

27 individual controller or distributions of controllers to deal with. In a multi-agent system, agent has the properties such as autonomy, local views, and decentralization, etc Multi-agent System Theory Multi-agent system theory is evolved from distributed artificial intelligence (DAI) in the 7s and 8s. It is hard to find a unified definition about what an agent is. The MAS theory generally specifies that an agent is an entity owning some properties [23]: be capable of acting in an environment, be capable of perceiving its environment (to a limited extent), be able to communicate directly with other agents, be driven by a set of tendencies, in the form of individual objectives or of satisfaction/survival function which tries to optimize, can possesses its own resources, skills and can offer services, has only a partial representation of the environment, and perhaps none at all, satisfies its objectives, taking account of the resources and skills available to it and depending on its perception, its representations and communications Multi-agent System Application Currently, MAS technology is implemented in many subfields of computer science and artificial intelligence. Agents are being used in an increasingly wide variety of applications. The MAS has been proved to be an effective decentralized approach for power system control [24-26]. Increasingly studies on power and energy management by MAS have been published. The application fields of power or energy management using MAS can be distinguished as three fields which are production and storage, distribution and consumption symbolized in Fig. 2.9 [25]. For production and storage, mainly working on generator (P1, P2 and P3) and storage units (S1 and S2); for distribution, switches are the control objectives; and for consumption, the allocation of energy resources for the loads (C1, C2, and C3) is critical. 14

28 Fig Classification of areas to use energy management [25] For all the three fields, we can find application examples. A hybrid system connected to the grid is introduced in [25] to illustrate the management in production and storage. Much work has been done in the area of distribution management, such as the [27] for naval application and [28] for large-scale system. The management of consumption relates to solve the problem of resource sharing [29]. In the case of energy systems, the loads need to be controlled in order to preserve energy resources and improve the ability to properly allocate energy resources. In this thesis, a MAS will be applied for power and energy management for a notional MVDC IPS. 15

29 CHAPTER THREE PRELIMINARY DESIGN This chapter is intended to introduce the high level system design, and will not address specific design implementation issues of the hardware level of the integrated ship management level. In section 3.1, the design objectives are illustrated to identify how the proposed method satisfies the requirements. In section 3.2, a notional MVDC next generation integrated power system is introduced. The recommendations of this thesis will be implemented on this system. In section 3.3, a control scheme for power and energy management system is introduced. The functionalities of control module PCON are discussed. In section 3.4, the design of energy storage module is discussed. The load shedding strategy is introduced in section Design Objectives The US Navy intends to apply MVDC integrated power system to electric ships, in order to achieve its goal of producing affordable power solutions for future warships. During this development process, many technologies will be addressed. Among them power and energy control is prioritized for its ability to assure a high level of quality of service and survivability, since quality of service provides continuity of power to the loads during normal operation and survivability provides a power system the ability to assure the continuity of the ship s mission during a damage, these two concepts represent satisfy significant ship design requirements both in normal and adverse conditions. A complete process of preliminary design for power and energy management system can be achieved in Appendix A. 3.2 Notional MVDC NGIPS Model The design of the power and energy management system is based on the notional MVDC NGIPS model, which has been developed in the Center for Advanced Power Systems at the Florida State University. The design process starts from understanding the model NGIPS Module Types The standard NGIPS module types are [5]: Power Generation Modules (PGM) to convert fuel to electrical power 16

30 Power Distribution Modules (PDM) consist of switchgear and cabling necessary to distribute the power Power Conversion Modules (PCM) to convert power from one voltage /frequency to another Energy Storage Modules (ESM) to store energy provided by and provided to the electrical power system Power Loads use electrical power Propulsion Motor Modules (PMM) to convert electrical power into propulsion for the ship Power Control Modules (PCON) consists of software necessary to operate the power system. These standard module types are employed in all NGIPS power generation architectures and in the Zonal Electrical Distribution Architecture MVDC Topology MVDC is the final objective power generation architecture in NGIPS roadmap, the architecture of this notional MVDC system has been adopted in [3] as a notional design example of high performance MVDC ship system design and will be adhered to this thesis. The topology of the notional MVDC NGIPS model is shown in Fig A total installed electrical power of 8 MW is provided by two 36 MW (45MVA) main gas turbine generator sets and two 4 MW (5MVA) auxiliary gas turbine generator sets. One main and one auxiliary generator set are connected to each longitudinal bus, along with the drive inverter of one propulsion motor. It is a multi-zonal distributed system with ship service loads in four zones from bow to stern along the ship, is supplied from the zones with MVDC power from both of these port and starboard DC buses. Zone 5 represents superstructure of the ship and its largest electrical load high-power radar system. This architecture is developed to maximize the operational capability of the MVDC IPS even under adverse conditions, e.g., unexpected damage. Bow and stern cross-hull links are provided between the port and starboard dc buses to provide the capability of configuring a ringbus, from which AC power generation and load subsystems operate. Power system survivability can be enhanced by opening the bow and stern cross-hull disconnecting switches. This creates a split-plant configuration. All sources of electrical power and loads are connected to the MVDC 17

31 bus through disconnect switches. There are also disconnecting switches at each of the zonal boundaries. This enables reconfiguration operations to isolate a portion of the MVDC bus that may be damaged or experienced a fault. A capacitor bank is shown in Fig. 3.1 as the energy storage for the system. It is connected to both the port bus and the starboard bus when the cross-hull disconnects are connected. A pulsed load device is shown connected to the port and starboard MVDC bus when the cross-hull disconnect switches are connected. MVDC power from the port and starboard longitudinal buses is stepped down in the zonal load centers to low-voltage dc (LVDC, e.g., 8 V) by DC/DC converters. The load center topology is illustrated in Fig. 3.2 [31]. At the load center level, the bus transfer switches are shown represented by auctioneering diodes. Some loads operate directly off of the LVDC bus. Other ac loads are supplied with low-voltage, three-phase AC power (i.e., at 45 V) by a DC/AC converter operating off of the LVDC bus. Fig. 3.2 shows zonal load center configuration, ship service loads are categorized as either resistive (i.e., constant impedance), electronic equipment (i.e., constant power), or uncontrolled motor (e.g., pump and fan motors). Based on the topology we can see that some protection concepts were developed during the initial preliminary power system design, however the preliminary design still needs significant preliminary design of a high level control to coordinate and realize them. 18

32 Fig Notional design example of high performance MVDC ship system design [3] 19

33 Fig Zonal load center configuration [31] RTDS Ship Model We propose a distributed approach for power and energy management to achieve quality of service and survivability for the notional MVDC NGIPS. The MVDC model was developed in a Real Time Digital Simulator (RTDS) by its simulation software RSCAD. For the model in the RTDS, the energy storage and pulse load are connected to port and starboard no matter how the cross-hull disconnect switches work. This update is meaningful for the survivability concern. Fig. 3.3 shows the zonal circuit of the ship model that has been developed in the RTDS simulation software RSCAD. There are five service loads within the zone, two DC loads and three AC loads. Detailed information about the loads is listed in Table 3.1. Some of them are defined as vital loads, which cannot tolerate an interruption or can only tolerate a short-term interruption of power supply. Demanded power for each of them under battle and cruise are 2

34 listed. For our purposes only the ship is defined as battle mode when it s working at 3 knots, and defined as cruise mode when working at 2 knots. PCM1 in Fig. 3.3 is the bi-directional DC/DC converter, which converters the 5kV DC to 8V DC. PCM2 is the DC/AC converter to converter 8V DC to 45V AC. These two converters can be found in Fig. 3.2 as well, where the zonal load configuration is addressed. Fig Zonal loads deployment in RSCAD 21

35 Table 3. 1 Zone 1 loads category Loads Load Power Load Power Vital or Not Load Category Battle (kw) Cruise (kw) DC Loads Nonvital DC Constant Impedance Load (Resistive) DC Loads NA Nonvital DC Constant Power Load (Ave. value power, not switched) Type 1 AC Loads Nonvital Uncontrolled Induction Motor Load (speed-squared torque profile for fans/pumps) 12/28 VAC Loads Vital 45 VAC, 3 phase, Constant Power Load (Switched 6-pulse Bridge into resistive load) Type 1 AC Loads Vital 45 VAC, 3 phase, Constant Impedance Load (RL) 3.3 Power and Energy Management System Our work after understanding the model starts from the architecture of the power and energy management system we want to develop and then proceeds to control scheme and control responsibilities. The architecture is finally determined by a consideration of the power system topology sufficient to deliver and assess the functional requirements of these responsibilities. We considered the topology at two levels, which are the system level and zonal level. The system level relates to the 5kV DC buses, which are the main buses for system. The zonal level relates to the zones. Based on the zonal distribution, we propose a distributed approach for the system. PCON modules can be used to realize the functions such as monitoring and control, load shedding, system configuration and resource planning. We chose to use distributed PCONs as the power control modules for the power and energy management to realize its functionalities. The control scheme is demonstrated by using some modules that represent the components in the power system System Definition and Architecture PEMS, a high level control system, needs to deliver the right power to the right load at the right time, and it needs to assure the electrical systems contribution to QoS and survivability. For QoS concern, the power will be provided to the loads according to their degree of needs. The 22

36 power and energy management for survivability can be achieved by controlling the electronic components within the system to reconfigure and stabilize the system during a fault for continuity of mission effectiveness. The functionalities of the PEMS include: ship mode selection generator launch schedule blackout monitoring fault ride-through From the topology of the notional MVDC IPS, we have both service loads distributed in five zones and propulsion loads connected to system bus viz. port and starboard buses. To have complete and effective control of components connected to system bus and within the zone, a two-layer hierarchical architecture, viz. system level and zonal level is decided for this MVDC IPS ship model. The system level is supervisory control while the zonal level is decentralized control, which is based on the zonal distributed architecture of the notional MVDC IPS. The control architecture is general and applicable for most IPS. The architecture and the functions for each level are described in Fig

37 Fig Two-layer hierarchical architecture PCON Based Control Scheme A power control module called PCON is proposed. This computer software based module will coordinate with the behaviors of other modules within the system to execute the control. It can be interacted with the human operators by a Human-Machine Interface that will be part of the monitoring and control system. To realize the continuity of power supply to load and continuous ship mission for PCONs, we are focusing on monitoring, control, and load shedding. To satisfy the increasing power needs mentioned above, we develop an energy storage module on the system level which was shown in the topology but not well developed in the model. It is also used as an additional power source when it s necessary to assure the system survivability. To assure the quality of service, we extended the application of energy storage to each zone, which can provide power to some vital loads during a limited supply of power. Load shedding is also used to assure the power supply for vital loads. A detailed control scheme of PEMS is shown in Fig For simplicity, only one zone of the notional MVDC IPS is shown in the control scheme to depict the communication between system level and zonal level. Different types of modules are shown in the control scheme to represent the elements within the MVDC IPS. A detailed explanation of the modules can be reached in Fig. 3.5 or in Chapter Multiple PCONs are shown at the system level and they manage the system level elements including the PPMs, PCMs, ESMs, etc. These PCONs acting in a supervisory capacity 24

38 can send control signals to the zonal PCONs. Acting combined with those PCONs are the ESM and LSHED modules which are distributed in every zone and seek to assure the QOS and improve the survivability of the ship when a fault occurs in a zone. The size of ESM in each zone is decided by power needs of the vital loads in that zone. The communications between the various modules are elaborated in Fig. 3.5, while V is the voltage in kv, I is the current in ka, P is the power in MW, and f is the frequency in hertz. To simplify the scheme, only one zone was elaborated in Fig. 3.3 to show the system level and zonal level interaction. Zonal PCONs are provided at the zonal level for local control to formalize distributed control architecture. When a fault occurs in the DC/DC converter in one zone, the zonal PCON receives failure signal from the DC/DC converter and sends a control signal to activate the zonal ESM to power the vital loads and LSHED will be activated to shed the nonvital loads. Meanwhile, the DC/DC converter can be disconnected from the port and starboard buses to isolate the fault from the main buses. After the fault is cleared, DC/DC converter will be reconnected to reroute power to zonal loads, zonal ESM will be in charging mode and load shedding will be disabled. System PCONs work in system level for supervisory control to control the components in the system level and send control signals to the zonal PCONs. Normally, operators set the ship mode based on the ship mission through the Human Machine Interface (HMI), which is usually a graphics-based visualization of a control and monitoring system. System PCON initiates the sequence programs and delivers user request to related modules to satisfy the setting. System PCON executes the control for the whole system based on the request and the system status. It monitors the whole system, including the status of four generators, the converters, propulsion motors, and energy storage units. The modules represent the components in the power systems, thus, they continuously send components status to PCONs and system PCON does the top level control while zonal PCON executes local control PEMS Functionalities The first function of PEMS is to realize the ship mode selection. At the beginning of the operation of an electric ship, a ship mode can be selected by operators. Once the ship mode is selected, the system operation will be initiated to that requested mode. Three modes, including 25

39 battle, cruise and anchor are designed. Besides, an additional mode is offered for user to define specific request. Each mode offers a different running speed, which is defined in Table 3.2. Fig Control scheme for power and energy management system Second function of PEMS is generator launch schedule. Based on the system demand and supply, a detailed schedule of generators on or off under each mode is designed as shown in Table 3.3. By using this function, the fuel consumption can be lowered to achieve fuel economy. The simulation results will be discussed in Chapter 4. Third function of PEMS is blackout monitoring, which means the PCONs need to monitor the ship system over time and send status continuously to operator through HMI. In Fig. 3.5, we can see that all the modules connected to PCON are monitored and controlled. The PCON monitors the whole system through both system PCONs and zonal PCONs. The zonal PCON collects information from the zone and send that information to system PCON to let it 26

40 know the whole system status and send that information to operators through HMI. The faulty signals will be monitored and related control logic will be acted to avoid system blackout. By monitoring the components status in the power system, PCONs can report the conditions of power system equipments and loads after battle damage to the operator for replacement or maintenance. It also helps to determine what power system components and loads are safe to energize and also the priority in which they should be restored. The forth function of PEMS is fault-ride through, which is a key function to achieve a high level of survivability. If there s a problem in the main generator 1 and it cannot work under this situation, the abnormal signal will be sent to system PCON and system PCON will decide whether to shed some nonvital loads to have limited power supply the vital loads in the zone or to bring one more generator online if it s under a launch schedule function. Table 3. 2 Power output for different modes when four generators online Power Ship MTG1 MTG2 ATG1 ATG2 Propulsion Power to Speed Power Power Power Power Motor Zone 1 Mode (Knots) (MW) (MW) (MW) (MW) Power(MW) (MW) Battle Cruise Anchor Table 3. 3 Generator launch schedule for different modes Generator Main Main Auxiliary Auxiliary Total Mode Generator 1 Generator 2 Generator 1 Generator 2 Capacity (MTG1) (MTG2) (ATG1) (ATG2) (MW) Battle Online Online Online Online 8 Cruise Online Online Offline Offline 72 Anchor Offline Offline Online Online 8 27

41 3.4 Energy Storage Module We use energy storage module as part of the PEMS to coordinate with PCONs and LSHED modules to realize the quality of service and survivability. An energy storage module is developed and used both in system level and zonal level. A study of energy storage technology is executed and an ultra-capacitor based ESM is recommended. A bi-directional DC/DC converter is used to cooperate with ultra-capacitor to achieve charging and discharging for ultra-capacitor. Control logic is developed for ESM both in system level and zonal level Energy Storage Technology Energy storage elements have been widely used in a lot of electrical devices such as mobiles and calculators. Both industry and academics are working on advanced technologies to use them in electricity network or transport areas to save energy and achieve lower carbon energy system [32]. It is becoming a key technology aiming to improve the power quality in the power applications. Two factors characterize the application of an energy storage technology. One is the amount of energy that the device can take in and deliver, which is defined by specific energy (measured in Watt-hour per kilogram, or Wh/kg) in combination with the physical weight of the storage device. Another is the rate at which energy can be transferred into or out of the storage device. This depends mainly on the peak power rating of the power conversion unit, but is also impacted by the response rate of the storage device itself [33]. Batteries, ultra-capacitors, flywheels are possible choices for energy storage system using in the MVDC IPS. A comparison of their properties is shown in Table 3.4 [34-36]. 28

42 Table 3. 4 Comparison of energy storage technologies Property Technology Types Batteries Ultra-capacitors Flywheels Effiency 7-9% 9% 9% Specific Power (W/kg) Specific Energy (Wh/kg) Max 6 Max 3 Max 2 Charge Time Hours Seconds Minutes Capital Cost($/kW) Batteries own large specific energy, or energy density but low specific power, or power density compared to others, it is the first choice for long term or medium term power source supply. Ultra-capacitors and flywheels have significantly faster response in providing power to the load than batteries, while ultra-capacitors show better characteristics than flywheel in power density. In PEMS, ESM has two main functions, one is for the large step loads,or pulse loads, which requires high amplitude power with short duration, and the other is for the zonal vital loads when power supply is not enough for all the loads in the zone. Considering the fast response demand for system safety, we chose to utilize ultra-capacitor energy storage as the ESM element and implement it as part of the PEMS in the notional MVDC IPS Ultra-capacitors Ultra-capacitors, also known as supercapacitors or electrochemical capacitors, are made with high surface area electrode materials and thin electrolytic dielectrics, which help to achieve capacitance several orders of magnitude larger than conventional capacitors [37-41]. It stores energy in the electrical double layer at the electrode/electrolyte interface. Ultra-capacitors are increasingly becoming important in onboard energy storage applications in electric ship, electric and hybrid electric vehicles in improving power quality and reliability [42]. 29

43 For conventional capacitors, the electric charges are stored on the electrode performing absorption and desorption of the ions in electrolysis liquid (see Fig. 3.6 [43]). Capacitance is proportional to the area of overlap and inversely proportional to the separation between conduction sheets as following: (3.1) where C is the capacitance, A is the area of overlap of the two plates, ε r is the relative static permittivity (sometimes called the dielectric constant) of the material between the electrodes, ε is the electric constant (ε Fm -1 ), and d is the separation between the electrodes. The energy stored on the plates is: (3.2) where E is the energy, in joules; C is the capacitance, in farads; and V is the voltage, in volts. Based on current trends, ultra-capacitors can be divided into three general classes: electrochemical double-layer capacitors, pseudocapacitors, and hybrid capacitors. Taxonomy of the ultra-capacitors is shown in Fig [43]. Fig Schematic of a conventional capacitor [43] 3

44 Fig Taxonomy of supercapacitors [43] Compared with conventional capacitor, ultra-capacitors don t have conventional dielectric. Rather than two separate plates separated by an intervening substance, a vanishingly thin physical separator is used to separate charge within the two electrodes. Fig.3.8 [43] shows the schematic of an Electrochemical Double Layer Capacitor (EDLC). As voltage is applied, charge accumulates on the electrode surfaces. Because of the natural attraction of unlike charges, ions in the electrolyte solution diffuse across the separator into the pores of the electrode of opposite charge. Thus, a double-layer of charge is produced at each electrode. These doublelayers, coupled with an increase in surface area and a decrease in the distance between electrodes, allow EDLCs to achieve higher energy densities than conventional capacitors [37-39]. 31

45 Fig Schematic of an electrochemical double-layer capacitor [43] An equivalent circuit of ultra-capacitor shown in Fig. 3.9 is used in simulation. The resistance of the resistor represents the equivalent series resistance (ESR), and the capacitance of the capacitor represents the total capacitance of the ultra-capacitor bank. This circuit is chosen to simplify the problem. A further study on this model can be followed up for future work. Resistor Capacitor Fig Equivalent circuit of ultra-capacitor DC-DC Converter Two types of ESM are designed for the system, one is SESM and the other is ZESM. The architecture of the model is identical for these two except the parameters of the components inside the modules since the system ESM is connected to the 5 VDC port bus while zonal ESM is in the zone so that it is connected to 8 VDC buses as shown in Fig The system 32

46 ESM and pulse load follow the design in the system model developed in RTDS that is connected to both port and starboard buses, which is different from that shown in the topology. Fig Location of energy storage module Based on the decreasing output characteristic of the ultra-capacitor, to get steady output voltage or current during discharge, a bidirectional DC/DC converter is coupled with the ultracapacitor equivalent circuit as the ESM. The DC/DC converter used in this thesis is originally proposed by [44] and developed for MVDC IPS by [45]. The circuit diagram of bi-directional dc/dc converter introducted in [45] is shown in Fig [45]. The bi-directional DC/DC converter works in two modes, which are buck and boost. The electrical power flows from the high voltage side to the low voltage side in buck mode and reverse in boost mode. In buck mode, the converter acts as a voltage fed full bridge converter, 33

47 and in boost mode, it acts as current fed full bridge converter. A brief theoretical introduction of this DC/DC converter can be reached from Appendix B. Current-fed Full-bridge Converter Voltage-fed Full-bridge Converter Local Load Side (8Vdc) MVDC Bus Side (5kVdc) Bidirectional Power Flows: Boost Mode (Discharging LV Storage) Buck Mode (Charging LV Storage) Fig Circuit diagram of bi-directional DC-DC Converter [45] System and Zonal Energy Storage Module The configuration of the ESM in simulation model is shown in Fig The high voltage side connects to the port and starboard buses for system ESM and to the output of the DC/DC converter (5kV/8V DC) as shown in Fig. 3.1 for zonal ESM. Based on the bidirectional DC/DC converter derived in [45] and the actual application in the simulation for MVDC IPS, the parameters of the DC/DC converters used in SESM and ZESM are listed in Table

48 Fig Simulation module- dc/dc convertor with capacitor Table 3. 5 Converter parameters for SESM and ZESM Parameters SESM value ZESM value MVDC voltage (V HV ) 5 kv 8V Energy storage voltage (V LV ) 8V 5V Choke inductance (L) Leakage inductance at the low voltage side (L lk ) Switching frequency(f s ) Clamp capacitance (C c ) Filtering capacitance (C 1 ) Filtering capacitance (C 2 ) 1mH 27uH 7Hertz 47uf 2mF 2.7mF Source and line resistance(r s ).1 Boost-mode output resistance (R o ).1 35

49 From Table 3.5, SESM contains a 5kV/8V converter while ZESM contains an 8V/5V converter. During the buck mode, the convert delivers the power from high voltage side to the low voltage side to store energy on the capacitor; when in the boost mode, the capacitor is distributing stored power to the high voltage side. SESM is designed to power the pulse load which is connected to the 5kV DC bus. The pulse load requires large amplitude of power within very short time. The energy stored on the capacitor depends on the capacitance and the bus voltage through Eq. (3.2). Given the power and time duration of the pulse load, we can get the energy through: (3.3) where E is the energy, P is the power, and t is the time duration. When a specific pulse load is given and the efficiency is known, Eq. (3.2) can be used to calculate the capacitance of ultra-capacitor. In order to get the efficiency of the bi-directional dc/dc converter, we set a random capacitance on the capacitor and run the ESM in the RTDS. The efficiency is calculated through: (3.4) where P out means the power output from the dc/dc converter and P in means the power input to the dc/dc converter. ZESM is targeted to supply power for vital load in the zone. The power requested by each zone in battle or cruise mode is shown in Table 3.6. In this thesis work, only Zone 1 is included in the model. The power output of ZESM is designed as 1 MW for power supply. For Quality of Service consideration, to satisfy the ESM-F1 proposed in [5], ESM should assure a 1s run time. Use Eq. (3.3), energy output from the ZESM should be 1 MJ Control Design Usually the controller in buck mode controls the voltage of the LV side to get steady output voltage. We make use of this bi-directional converter to realize the charging and discharging for the ultra-capacitor. Thus, the voltage on the low side, or capacitor side, is not necessary to be controlled. Through simulation, we found, if the LV voltage is set to be controlled for the buck mode, a current limit control should be coupled to assure the system stability because at the beginning of charging, the current at the high voltage side could be very large, which causes a huge power 36

50 input to the DC/DC converter and cause the power generators to exceed the maximum power outputs. Another option for buck mode control is the power control. Connecting the ESM to the buses makes the ESMs as loads for the buses during charging. A good way to less the fluctuations on the buses is to have the ESMs working as constant power load in the buck mode. A comparison between these two controls is conducted in this thesis work. The converter efficiency and the transition of voltage and current both on high side and low side are shown in Fig In the simulation, the constant power control shows a better efficiency results. The voltage control shows a less variations on the high side current since coupled with high side current control. In our testing, we will use the constant power control based on the efficiency concerns and the ability of the ESM to provide a constant power to the power system, which is important to eliminate the variations in the power system performance. 37

51 1 EffES Efficiency (%) Time (s) Fig SESM efficiency in constant power control 1 EffES Efficiency (%) Time (s) Fig SESM efficiency in voltage control 38

52 Voltage (kv) Voltage (kv) Current (ka) Current (ka) High Side Voltage Low Side Voltage High Side Current Low Side Current Time (s) Fig SESM voltage and current variations in constant power control 39

53 Voltage (kv) Voltage (kv) Current (ka) Current (ka) High Side Voltage Low Side Voltage High Side Current Low Side Current Time (s) Fig SESM voltage and current variations in voltage control 4

54 1 EffES Efficiency (%) Time (s) Fig ZESM efficiency in constant power control 1 EffES Efficiency (%) Time (s) Fig ZESM efficiency in constant voltage control 41

55 Current (ka) Current (ka) Voltage (kv) Voltage (kv) High Side Voltage Low side Voltage High Side Current Low Side Current Time (s) Fig ZESM voltage and current variations in constant power control 42

56 Current (ka) Current (ka) Voltage (kv) Voltage (kv) High Side Voltage Low side Voltage High Side Current Low Side Current Time (s) Fig ZESM voltage and current variations in voltage control In boost mode, the voltage will be controlled. However, here, the SESM is connected to the port and starboard buses, on which the voltage is regulated by the rectifiers. Thus, the high voltage side current is to be controlled in boost mode. With the regulated voltage on the high side, the output of ESM is close to a constant value, which can be used to power the pulse load within a period of time or for some loads which need steady power supply for a while. All these control objectives are achieved by manipulating the duty cycles of the gating signals for switches in DC/DC converter. Feedback control is used for this bi-direction DC/DC 43

57 converter controller, whose basic configuration is shown in Fig In buck mode, the input side is the high voltage side and the output side is the low voltage side; in boost mode, it reverses. Fig Control configuration for bi-directional dc/dc converter 3.5 Load Shedding Two Types of Strategy Load shedding is a method being used to reroute the power to assure continuously power supply to vital loads in the case of loss of generations by system failures or damages. At this point of view, load shedding plays a critical role in assuring the system survivability, or in other words, in keeping the continuity of power service to vital loads. There re two strategies for loading shedding which were defined in [19], the QoS based load shedding and the mission priority based load shedding. Before introducing these two strategies, two ways to categorize the loads in the ship power systems should be elaborated. In [19], loads are categorized as Long Term Interruptible Load (LTI), Short Term Interruptible Load (STI), and Uninterruptible Load (UI) according to their abilities to tolerate an interruption 44

58 before a QoS failure is encountered. A detailed time slot for these different types of loads is listed in Tabel 3.6 based on the definition in [19]. QoS Load Shedding is applicable only for the LTI, which means under QoS load shedding strategy, when load shedding is necessary, only the LTI load will be shed. Then if after five minutes power has not been restored, the Mission Priority Load Shedding is initiated [19]. Mission Priority load category is defined based on severity of power demand, which means, the loads are shed only based on the level of the necessity for that specific mission Implementation in MVDC IPS Our aim is to first categorize the service loads to vital and nonvital loads (see Fig. 3.1) for the first level shedding. And then define the loads to different priority level for necessity under specific ship mode. Three types of ship modes are introduced, which are anchor, cruise, and battle. Table 3.6 shows the details about the category for the mission priority loads under different ship modes. For the first level shedding, the vital loads can be regarded as the UI and STI loads, while the nonvital loads can be regarded as the LTI loads defined in the QoS load shedding strategy. The second level shedding is based on the Mission Priority Load Shedding in [19]. High priority loads are with high necessity for that specific mission and will be the last to be shed. Low priority will be shed first and medium priority followed. In this thesis work, load shedding under cruise and battle are both studied under fault. Since continuous loads in the model are service loads and propulsion loads, the simulation for load shedding for battle mode will be on service loads and propulsion loads, and only service loads for cruise mode. In cruise mode, the total power capability is about 8 MW (four generators ON) or 72 (2 main generators ON), which is dependent on whether it s in normal or generator launch schedule as shown in Table 3.3. In battle mode, when system cannot sustain under some faults or damages, both the nonvital service loads in the zone and the partial propulsion loads will be shed. The system discontinuity is avoided and system survivability is improved by using this load shedding method. The simulation results will be discussed in Chapter 4. 45

59 Table 3. 6 Quality of service load category [19] Categories Uninterruptable Load (UI) < 2 seconds Time the load can tolerate an interruption before a QOS failure occurred Short Term Interrupt Load (STI) >=2 seconds and <5 minutes Long Term Interrupt Load (LTI) >=5 minutes Table 3. 7 Mission priority load category Load Priority Ship Mode Anchor Cruise Battle High Priority Service Loads Propulsion Loads Weapon Loads Medium Priority Propulsion Loads Service Loads Propulsion Loads Low Priority Weapon Loads Weapon Loads Service Loads 46

60 CHAPTER FOUR SIMULATION This chapter is for simulation results. A 2 rack MVDC IPS model is used to implement the proposed PEMS, which means there re two subsystems in the model, one for zone 1 and the other for power generation. In section 4.1, the mode selection function will be demonstrated. Section 4.2 will show the working status of ESM, including SESM for pulse load and ZESM for zonal uninterruptible and short term interruptible loads. The load shedding function is shown in section 4.3. In the section 4.4, several scenarios are elaborated and the system responses based on PEMS are discussed. 4.1 System Initiate-Mode Selection Normal System Status There re three modes created in the integrated power system. Fig. 4.1 shows the system status under battle mode. The output power from MTG1 is about MW, the MTG1 frequency is 377 rad/s, the MTG1 output voltage is about 5 kv, the propulsion motor power varies between MW, the propulsion motor speed is around rad/s. In cruise mode, the power output from MTG1 is about 8.4MW, the MTG1 frequency is 377 rad/s, the MTG1 output voltage is about 5 kv, the propulsion motor power mainly varies between MW, the propulsion motor speed is around rad/s. The status of this mode is shown in Fig In anchor mode, the power output from MTG1 is about 3.5 MW, the MTG1 frequency is 377 rad/s, the MTG1 output voltage is about 5 kv, the propulsion motor power mainly varies between MW, the propulsion motor speed is around rad/s. The status of this mode is shown in Fig

61 Frequency(rad/s) Power (MW) Power (MW) Voltage (kv) MTG1 Power MTG1 Frequency MTG1 Bus Voltage Propulsion Motor Power Propulsion Motor Speed 59 Speed (rad/s) Fig Normal system status under battle mode 48

62 16 MTG1 Power Power (MW) MTG1 Frequency Frequency(rad/s) MTG1 Bus Voltage 5.7 Voltage (kv) Propulsion Motor Power 8 Power (MW) Propulsion Motor Speed 38.5 Speed (rad/s) Fig Normal system status under cruise mode 49

63 Frequency(rad/s) Power (MW) Power (MW) Voltage (kv) MTG1 Power MTG1 Frequency MTG1 Bus Voltage Propulsion Motor Power Propulsion Motor Speed 19 Speed (rad/s) Fig Normal system status under anchor mode Status under Launch Schedule Under launch schedule, a different combination of generators as power source under the three modes is presented in Table 3.3. The status of the system under launch schedule for different modes is shown in Fig. 4.4, Fig. 4.5, and Fig.4.6 individually. 5

64 In Fig. 4.4, we can see the cruise mode under generator launch schedule. The main generator 1 and 2 were online, and the other two auxiliary generators were offline. In Fig. 4.5, the two main generators were online while the two auxiliary generators were offline. In Fig. 4.6, it is switched from battle to anchor mode under generator launch schedule. Two auxiliary generators are online when the two main generators are offline. ON/OFF ON/OFF ON/OFF ON/OFF BRKMTG BRKMTG BRKATG BRKATG Time (s) Fig Cruise mode under generator launch schedule 51

65 ON/OFF ON/OFF ON/OFF ON/OFF BRKMTG BRKMTG BRKATG BRKATG Time (s) Fig Battle mode under generator launch schedule 52

66 ON/OFF ON/OFF ON/OFF ON/OFF BRKMTG1 BRKMTG2 BRKATG1 BRKATG Time (s) Fig From battle to anchor mode under generator launch schedule 4.2 Energy Storage Module Charging and Discharging Fig show the charging and discharging of the SESM and ZESM. The controller executes input power control during charging and high side current control during discharging. High side and low side mean the high and low voltage sides of the DC/DC converter. In Fig. 4.7, the power input goes to 2 MW finally during charging. In discharging, both the SESM and ZESM can satisfy a 1s constant power supply. Some fluctuations are found during the buck mode or boost, however the system will sustain, and the power outputs from the ESMs can satisfy the system demands. 53

67 About a 1s power supply is achievable for both system and zonal energy storage unit. Voltage (kv) Voltage (kv) Current (ka) Current (ka) Power (MW) Power Input-ZESM High Side Voltage Low side Voltage High Side Current Low Side Current Power Output-ZESM Time (s) Fig Zonal energy storage module during charging 54

68 Voltage (kv) Voltage (kv) Current (ka) Current (ka) Power (MW) Power Input-ZESM High Side Voltage Low side Voltage High Side Current Low Side Current Power Output-ZESM Time (s) Fig Zonal energy storage module during discharging 55

69 Voltage (kv) Voltage (kv) Current (ka) Current (ka) Power (MW) High Side Voltage Low Side Voltage High Side Current Low Side Current Power Input-SESM Power Output-SESM Time (s) Fig System energy storage module during charging 56

70 Voltage (kv) Voltage (kv) Current (ka) Current (ka) Power (MW) High Side Voltage Low Side Voltage High Side Current Low Side Current Power Input-SESM Power Output-SESM Time (s) Fig System energy storage module during discharging SESM for Pulse Load The ship was working in battle mode under 3.2 knots. Propulsion motors consumes about 27 MW each at this speed and four generators are all online. Fig shows how the SESM compensates the pulse load. A 1 MW pulse load is triggered with a 7.41 s run time. MTG1 and MTG2 are main generators, ATG1 and AGT2 are auxiliary generators, and two pictures at the bottom show the power outputs from the four generators. The first picture shows the pulse load power and the second picture shows whether SESM is in charging or discharging, where 1 means charging and means discharging. The third picture is the power input and output from the SESM, where positive power means consuming power from generator and negative power mean supplying power to buses. For input 57

71 power, it means the source of power; for output power, it means the user of the power. The power user in charging mode is the ultra-capacitor, while also the source of power in discharging mode. On the 5kV bus side, it represents as the power source during charging and the power user during discharging. The waveforms can be divided into several stages. During -2.4s, ESM was working in charging mode. During s, ESM was working in discharging mode, and pulse load was triggered during this period. From the output power of MTG1 and MTG2 during this period, we can see that at the beginning the output power went to 36MW, which is the maxima power from the main generators. After that, SESM was in discharging, the output power from the generators was lowered to a safe range. During s, SESM returned to work in charging mode. 58

72 Power (MW) 1-charge/-discharge Power (MW) Pulse Load Power SESM charge/discharge Power Input -SESM Power Output-SESM Power (MW) Power (MW) MTG1 Power MTG2 Power ATG1 Power ATG2 Power Time (s) Fig System energy storage module for 7.41s-1MW pulse load in battle mode Another case that ran in simulation for a 3-pulse bust load is elaborated in Fig The power value for the pulse load was 5MW, each pulse lasted 3.33s. The SESM supplied power when a pulse load was fired, and went back to charging when it was off. It proves that the SESM can supply constant power to the bus even for consequent firing of pulse loads. 59

73 Power (MW) Power (MW) Power (MW) 1-charge/-discharge Pulse Load Power SESM charge/discharge Power Input -SESM Power Output-SESM MTG1 Power MTG2 Power Power (MW) 6 ATG1 Power ATG2 Power Time (s) Fig System energy storage module for 3.33s-5MW 3-pulse burst load in cruise mode 6

74 4.2.3 ZESM for Zonal Load ZESM is designed to supply power to zonal vital loads under the situation when there s no power supply from the 5 VDC buses to the PCM1, the DC/DC converter, which connects zonal loads to port and starboard buses. Once scenario was tested in RTDS when the PCM1s were blocked in buck mode, which means no power input to PCM1s. This happens when the main buses are encountered with a large disturbance and need to disconnect the loads to help the main buses to recover. ZESM was working in discharging mode and a constant power was supplied to the 8V DC bus. The duration of this power can be more than 1s. Simulation results for this scenario are shown in Fig PinES1 and PoutES1 mean the power input and power output of the zonal energy storage. At about 2s, the ZESM changed from charging to discharging. Positive power means ZESM is consuming power from the bus, while negative power means ZESM is supplying power to the bus. In this case, the last picture shows the power supplied to the bus, which is about.5mw. 61

75 Fig Zonal energy storage in discharging-pcm 1 blocked in buck mode 62

76 4.3 Load Shedding Load Shedding is executed during the faults or special scenarios. We tested this function in two modes which are cruise and battle. The detailed results can be seen in the fault scenarios simulation next. The loads within the zone are listed in Table 3.2 discussed in Chapter 3. Among these loads, the DC1 and DC2 loads are regarded as non-vital loads during cruise and battle mode, and the Type 1 AC load (Uncontrolled Induction Motor Load ) is regarded as the non-vital load during battle, all other loads are vital loads. This definition is also listed in Table Fault Scenarios The fault scenarios were running at the normal status, which means when four generators are online at the beginning Loss of Main Generator In this scenario, we want to show how the PCON works when the loss of one main generator occurs during the cruise and battle mode, and also the responses for its restoration. We will show several critical parameters in the system, including the power outputs from the two main generators, the voltage of the 5kV DC bus, the loads power consumption, propulsion propeller speed, and the power outputs from the SESM. Fig shows the simulation results for the loss of one main generator in cruise mode with PEMS. The ship was running at 2 knots. By design, when the MTG1 or MTG2 is offline, the SESM will work in discharging mode to supply power to the 5kV DC bus, the non-vital loads in the zone, which are DC Load 1, DC Load 2 Type 1 AC Load will be shed. According to the load shedding strategy for cruise mode, since the propulsion mode is defined as high priority load, the propulsion motor speed will not change. When the generator is restored, the SESM will be in charging mode and the zonal load that had been shed will be restored, the detailed response is shown in Fig From Fig.4.14 we can see that all the nonvital loads were shed and SESM worked in discharging mode to supply power to buses, which makes the MTG2 keeps almost the same power output as before the loss of MTG1. The propulsion speed has a very small change, which would be caused by the system transient and the system bus voltage variation for the loss of the main generator. 63

77 Voltage (kv) Speed (rad/s) Power (MW) Power (MW) Power (MW) MTG1 Power MTG2 Power MTG1 Voltage MTG2 Voltage DC Load 1 DC Load 2 Type 1 AC Load Power Input-SESM Propulsion Motor Power Output-SESM Time (s) Fig Response in cruise mode-loss of main gen 1 w/ power&energy management 64

78 Fig shows the restoration of MTG1with PEMS and we can see that the nonvital loads were restored. The SESM went to charging mode. The propulsion motor speed went back to the initial value. Speed (rad/s) Power (MW) Voltage (kv) Power (MW) Power (MW) MTG1 Power MTG2 Power MTG1 Voltage MTG2 Voltage DC Load 1 DC Load 2 Type 1 AC Load Power Input-SESM Propulsion Motor Power Output-SESM Time (s) Fig Response in cruise mode-main gen 1 restoration w/ power&energy management For the battle mode, the ship is running at 3 knots. By design, when the MTG1 or MTG2 is offline, the SESM will work in discharging mode to supply power to the system. According to the load shedding strategy introduced in Chapter 3, the non-vital loads will be shed, and with partial propulsion loads (by lowering the propulsion motor speed). The propulsion motor speed that needs to be lowered is calculated using the power loss and the power from the ESM. We can calculate how much power the propulsion motor needs to be lowered by each, 65

79 which can then be converted to motor speed in the speed control module. For this scenario, under battle mode, the load shedding is working for both the zonal loads and propulsion loads. When the generator is restored, the SESM will work in charging mode, the shed loads will be restored, and the ship speed as well. In Fig Fig. 4.19, this scenario is well presented. In this scenario, we also present the response without power and energy management to further understand the function of PEMS. For the scenario with power and energy management, the bus voltage will recover to 5kV DC when the main generator restored while the scenario without PEMS cannot recover. 6 Power-MTG1 Power-MTG2 Power (MW) 4 2 Voltage (kv) Power (MW) Voltage-MTG1 Voltage-MTG2 DC Load1 DC Load2 Type 1AC Load Propulsion Motor Speed (rad/s) PinSESM PoutSESM Power (MW) Time (s) Fig Response in battle mode-loss of main gen 1 w/ power&energy management 66

80 1 Power-MTG1 Power-MTG2 Power (MW) Voltage-MTG1 Voltage-MTG2 Voltage (kv) DC Load1 DC Load2 Type 1AC Load Power (MW) Propulsion Motor Speed (rad/s) E-16 PinSESM PoutSESM 2E-16 Power (MW) 1E-16-1E Time (s) Fig Response in battle mode-loss of main gen 1 w/o power&energy management 67

81 Power (MW) Voltage (kv) Power-MTG1 Power-MTG2 Voltage-MTG1 Voltage-MTG2 DC Load1 DC Load2 Type 1AC Load Power (MW) Propulsion Motor Speed (rad/s) Power (MW) PinSESM PoutSESM Time (s) Fig Response in battle mode-main gen 1 restoration w/ power&energy management 68

82 Power (MW) Power-MTG1 Power-MTG2 Voltage-MTG1 Voltage-MTG2 Voltage (kv) DC Load1 DC Load2 Type 1AC Load Power (MW) Propulsion Motor Speed (rad/s) E-16 PinSESM PoutSESM 1.5E-16 1E-16 Power (MW) 5E-17-5E Time (s) Fig Response in battle mode-main gen 1 restoration w/o power&energy management 69

83 CHAPTER FIVE HARDWARE TESTING 5.1 Multi-agent System What an Agent is The most general way in which the term agent is used is to denote a hardware or (more usually) software-based computer system that enjoys the following properties [46-48]: Table 5. 1 Agent properties and their definitions Properties Autonomy Social ability Reactivity Pro-activeness What agents do Operate with no direct intervention of humans or others; Have control over agents actions and internal state; Interact with other agents (possibly humans) through agent communication language; Perceive environment, which may be physical world, a graphical user interface, etc; Respond in a timely fashion to changes that occur in it; Not simply act in response to their environment; Be able to exhibit goal directed behavior by taking the initiative Although this notion of agent is widely recognized by some researchers in a variety of areas [48-51], others may regard it as a weak notion. For some researchers, particularly those working in AI, an agent is regarded as a computer system that, in addition to having the properties identified above, is either conceptualized or implemented using concepts that are more usually applied to humans [52-55]. 7

84 5.1.2 Multi-agent System Advantages MAS can be defined as a loosely coupled network of agents that interact to solve problems that are beyond the individual capabilities or knowledge of each agent. Compared to conventional approach for energy management system, a MAS presents several advantages, they are: having more reliability which can tolerant a fault for one or more elements; avoiding a blackout by a distributed control in the case of bug from one program; owning an easier system design; and an openness of the system to integrate new elements [25]. Multi-agent systems in many ways mimic human organizations, which make the people tend to feel fairly comfortable when working with them. These characteristics greatly facilitate the introduction of multi-agent design systems (MADS) into the workplace and will support the rapid development and use of the technology [56]. In our work, based on system architecture and distributed PCONs in PEMS which are designed to realize the control work for power and energy management, an effective framework is necessary. We chose to use MAS based on its character that it s a combination of agents which work together to solve complex problems. By using MAS, we can use different agents to represent both the components in the power system and PCONs. The agents who represent components in the system are responsible for the components control, the PCON agents are responsible for system and zonal level control which include the communication between component agents and PCON agents in different level, which are system or zonal level Multi-agent System Layout and Agent Communication Based on the topology of MVDC ship model which is shown in Fig. 3.1, by using agents to represent components in the ship model and PCONs in the proposed control scheme in Fig. 3.5, we got the layout of the multi-agent system for control realization. Fig. 5.1 shows us this layout. The HMI is used to show how the PCONs communicate with operator. It s the interface between the PCONs and operator. In the system level, we can see that we designed more than one system PCON (in the Fig. 5.1, it is shown as n ), which is for system survivability concern. Thus, at least one more system PCON is needed, which is a duplication of the system PCON. These two system PCONs are monitoring the same information from the components agents and zonal level PCONs in case of the corruption for one of them during the work. From Fig. 5.1 we can see that we have pulse load agent for pulse loads, system ESM agent for energy storage 71

85 element, main gen agent for main generators, auxiliary gen agent for auxiliary generators, system converter agent for system level converters, propulsion agent for propulsion loads and radar agent for radar which are connected to system PCONs. Since the ship model is a zonal distributed architecture, several zonal PCONs are designed to represent different zones and under the zonal agents are the load agents, converter agents and zonal ESM agents. Fig Multi-agent system layout From an overall system level control point of view, the system PCON agent is responsible for supervisory control while zonal PCON agent is responsible for local control and it reports to system PCON agent. Since at least two system PCON agents are distributed in the ship to monitor other agents and to execute control for system QoS and survivability, the coordination between them needs some attention. Although they can be physically distributed within the ship, they communicate with each other to compare the information they receive from other agents to make sure they are identical. Among them, one will be the main system PCON which has a TOKEN that allows it to take the responsibility of control. Once the main system 72

86 PCON cannot continue its functions due to damage or error, it will ask another system PCON to take the TOKEN and take over its responsibility, in some cases the subordinate can override its lack of TOKEN and assume power, if it finds that it is divorced from the supervisor PCON. Once the supervisor PCON recovers, also it can reclaim the TOKEN from other system PCON as needed. A detailed list of what each agent does and their relationship with System PCON Agent is shown in Appendix C. It is illustrated by what signals the System PCON Agent receives from or sends to each agent listed in agent s column. 5.2 Hardware Test Bed The multi-agent system test bed is a distributed control test bed designed for the implementation of a flexible multi-agent architecture. The goal is to provide a test bed architecture that allows an agent programmer to focus solely on agent implementation and not the underlying communication protocol or hardware implementation, thus leading to shorter development time. The multi-agent systems are capable of interfacing with numerous systems, both in simulations and actual hardware. The capability of controlling a real, physical system via the test bed is desired to convey the usefulness of multi-agent systems for various projects. The choice of controller is the MAMBA by Versalogic Corporation. The MAMBA is an EBX single board computer with a 2.6 GHz Intel Core 2 Duo processor, 2 GB of RAM, dual gigabit Ethernet ports, and both analog and digital I/O. The operating system is Debian, using a custom Linux kernel provided by Versalogic Corp. There are six MAMBA boards in the test bed, shown in Fig 5.2, and each board can run several agents concurrently. 73

87 Fig MAMBA board test bed Each board will control its environment using the analog/digital I/O, similar to a classic centralized controller. A special case for environment control will be when simulating systems using the RTDS. When simulating with the RTDS, an FPGA will be utilized to interface the MAMBA boards and an RTDS module. The FPGA will translate the fiber optic protocol used by the RTDS to TCP/IP protocol sent to the MAMBA boards over Ethernet, and vice-versa. Preliminary tests show that the RTDS-FPGA-MAMBA connection is capable of sending and receiving data sets nearly 9 times per second. Each board (agent) will communicate with other boards (agents) using Ethernet via a Local Area Network (LAN). The agent communication framework used during the control implementation is the Java Agent DEvelopment (JADE) framework, developed by Telecom Italia. JADE is developed in Java and provides a communication protocol implementation for MAS. The MAMBA board connection architecture is shown in Fig

88 Ethernet Switch FPGA Fig Connection architecture of multi-agent hardware test bed 5.3 Agent Control Cases Loss of Main Generator This case is about the loss of one main generator during the cruise and battle modes, to show how the system responds to this failure in the system. Also see how it responds when the lost main generator is restored. 75

89 The agent communications for this case are listed in Fig System PCON Agent receives Generator Power, which is the power output from Main Generator 1. It was received by Main Gen Agent from RTDS through FPGA interface. Main generator agent sends the generator power signal to System Agent PCON 1 to inform the system status and let it decide whether any control actions are needed. The logic to action here is that the generator power needs to be compared with a value α, which is dependent on system scenario and tuning result. For this scenario, we initially set α as, however during tuning, we finally chose α as.1. When control action is needed, the system PCON agent will send speed selector signal to Propulsion Agent to lower the motor speed if necessary. By calculating the difference between the lost power supply and the power supply from SESM, a value of how much the propulsion motor power should be lowered is achieved. Besides the speed selector, a control signal will be delivered to Zonal PCON Agent to implement zonal control, including shedding loads through Load Agent and discharging zonal ESM through Zonal ESM Agent. Meanwhile, the control signal for system ESM is also delivered to system ESM agent. Finally, those control signals, including the speed selector will be forwarded to RTDS through the FPGA interface, which assure the quality of service and survivability. 76

90 Fig Loss of generator-agent communication In the simulation with hardware test bed cooperated control, there re several signals from the agents through FPGA interface to send to RTDS to conduct the control. Fig. 5.5 shows the initial status of these signals, where Spdselnew is the ship speed selector signal, which works in speed control selector to choose control objective. The initial value given in the JADE code for this signal is 5, which is the ship mode based speed control. MTGOFF_A is the main generator offline signal to shed the nonvital loads and make the ESMs work in discharging. PUESMA is the signal to control the SESM to supply power for pulse loads. CTLModeE1 and CTLModeES are the working modes for the ZESM and SESM. 77

91 SESM Mode ZESM Mode SESM-A MTG off-a Selector-A Spdselnew MTGOFF_A PUESMA CTLModeE1 CTLModeES Time (s) Fig Agent control signal initial values The SESM requests 3 MW constant power in charging, and the ZESM requests 1.5 MW constant power for charging mode. When the MTG1 is offline in the run time, the Generator Power signal received by the Main Gen agent is less than α, or.1 in this case, thus the control signal MTGOFF_A will be, as well as the PUESMA. The control signals and mode status under this situation are updated in Fig

92 SESM Mode ZESM Mode SESM-A MTG off-a Selector-A Spdselnew MTGOFF_A PUESMA CTLModeE CTLModeES Time (s) Fig Main generator offline-agent signals and ESM modes The system status for the loss of main generator 1 and its restoration under cruise mode are shown in Fig. 5.7 and Fig The power of MTG1 went to and 5kV DC bus experienced some oscillations, the propulsion motor power oscillating within a small constant range, and the motor speed keeps unchanged. During the generator restoration, the propulsion motor power and speed are constant but the generator power, frequency and voltage experienced a transient. 79

93 15 MTG1 Power Power (MW) 1 5 Frequency(rad/s) Voltage (kv) Power (MW) MTG1 Frequency MTG1 Bus Voltage Propulsion Motor Power Propulsion Motor Speed 38.5 Speed (rad/s) Time (s) Fig System status in cruise mode-main generator 1 offline-agent control 8

94 2 MTG1 Power Power (MW) Frequency(rad/s) MTG1 Frequency Power (MW) Voltage (kv) MTG1 Bus Voltage Propulsion Motor Power Propulsion Motor Speed 38.5 Speed (rad/s) Time (s) Fig System status in cruise mode-main generator 1 restored-agent control 81

95 The SESM discharging status in cruise mode is shown in Fig The SESM was supplying 5MW constant power for about 1s to the 5kV DC bus. Voltage (kv) Voltage (kv) Current (ka) Current (ka) Power (MW) Power Input-SESM High Side Voltage Low Side Voltage High Side Current Low Side Current Power Output-SESM Time (s) Fig SESM discharging in cruise mode-main generator 1 offline-agent control 82

96 1 restored. Fig. 5.1 shows the SESM status when it went back to charging once the main generator Voltage (kv) Voltage (kv) Current (ka) Current (ka) Power (MW) High Side Voltage Low Side Voltage High Side Current Low Side Current Power Input-SESM Power Output-SESM Time (s) Fig SESM back to charging in cruise mode-main generator 1 restored-agent control 83

97 The system status for the loss of main generator 1 and its restoration under battle mode are shown in Fig and Fig The power of MTG1 went to and the 5kV DC bus experienced a transient and went back to 5kV, the propulsion motor power was lowered to decrease the propulsion load, which resulted in a lowed motor speed. During the generator restoration, the propulsion motor power and speed went back. The generator power, frequency and voltage experienced a transient and back to normal operation status. 4 MTG1 Power Power (MW) Frequency(rad/s) MTG1 Frequency Voltage (kv) Power (MW) MTG1 Bus Voltage Propulsion Motor Power Propulsion Motor Speed Speed (rad/s) Time (s) Fig System status in battle mode-main generator 1 offline-agent control 84

98 4 MTG1 Power Power (MW) Frequency(rad/s) Voltage (kv) MTG1 Frequency MTG1 Bus Voltage Power (MW) Propulsion Motor Power 1 6 Propulsion Motor Speed 55 Speed (rad/s) Time (s) Fig System status in battle mode-main generator 1 restored-agent control 85

99 Fig shows the ZESM in discharging status when the main generator 1 was offline. It was supplying 1MW constant power to the 8V DC zonal bus. It can provide over 1s power supply continuously. Fig ZESM in discharging-main generator 1 offline-agent control SESM for Pulse Load A 5MW pulse load lasts about 7.41s to the port and starboard buses. Pulse load agent receives the pulse load power from the RTDS through FPGA interface; it will send this signal to the system PCON agent, then system PCON agent evaluates the value and make a control decision, for the logic to act, we compare the load power with β, which is 86

100 related to system. In this case, β equals to 1. If a control action is needed, then a control signal will be sent to SESM agent and that agent will send the control signal to RTDS to let the SESM work in discharging mode. In this way, power from the SESM can be supplied to the pulse load, and system survivability is achieved. The agent communications are elaborated in Fig Fig SESM supply for pulse load-agent communication Fig shows how the system responds for agent control. When the pulse load occurred at about 2.4s, the SESM turned into discharging to supply power to the pulse load. Meanwhile, the power outputs of the generators were lowered compared with the time period of -2.4s.At about 9.8s, the pulse load ended so that the SESM went back to charging mode and the main generators and auxiliary generators continued to normal working status. From Fig we can see a power drop at 9.8s which was caused by the response time of receiving the end of pulse load signal and sending stop discharging signal to SESM. Fig shows the SESM discharging status, in the occurrence of the pulse load. 87

101 Power (MW) Power (MW) Power (MW) 1-charge/-discharge Pulse Load Power SESM charge/discharge Power Input -SESM Power Output-SESM MTG1 Power MTG2 Power Power (MW) 28 ATG1 Power ATG2 Power Time (s) Fig SESM supply for pulse load in battle mode-agent control 88

102 Voltage (kv) Voltage (kv) Current (ka) Current (ka) Power (MW) High Side Voltage Low Side Voltage High Side Current Low Side Current Power Input-SESM Power Output-SESM Time (s) Fig SESM status in battle mode during pulse load-agent control 89

103 CHAPTER SIX CONCLUSION AND FUTURE WORK The proposed control scheme by using of PCONs for the notional MVDC IPS in Real Time Digital Simulator has been implemented both in simulation and hardware testing. Based on the simulation results, the proposed distributed PCON based PEMS has been verified to be working under battle and cruise modes to assure the system survivability and the continuity of power to the loads under normal or adverse conditions. PEMS offers the operator designed ship modes to choose from and to initiate the system under generator launch schedule function. Ship can work under several conditions based on the ship mission and system demand. Energy storage and load shedding methods are used to coordinate the PCONs to realize the control functionalities. The ultra-capacitor based energy storage modules are used both for system level and zonal level. For system level ESM, it can supply power to the pulse load which has high power amplitude within short time and also can supply power to system loads when the loss of one main generator happens. Zonal level ESM, defined as the backup power source for the zonal vital loads, works under discharging status to supply power to the zone when the PCM 1 cannot receive power from the port or starboard bus. Both the simulation and hardware testing verify the function of this module. Through the testing of fault scenarios, we can see that using the PEMS in the notional MVDC IPS, the system can tolerate the loss of the main generator and recover after its restoration. This is a critical capability of the PEMS which can improve the system survivability. Meanwhile, the vital loads can receive uninterruptible power to avoid the QoS failure. The results of the cases tested on hardware test bed verified that the multi-agent control works well by using the proposed PCON based distributed control algorithm. Both cases presented effective agents communication and good control realization. For future work, the robustness of the system with PCONs could be studied. An algorithm of the coordination between multiple system PCONs can be studied further, mainly about how they will cooperate with each other to realize the supervisory control more smoothly and safely. 9

104 APPENDIX A PRELIMINARY DESIGN PROCESS 91