UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Design and implementation of microcontroller-based direct methanol fuel cell/lithium polymer battery… Chen, Di 2009

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2009_fall_chen_di.pdf [ 1.82MB ]
Metadata
JSON: 24-1.0067644.json
JSON-LD: 24-1.0067644-ld.json
RDF/XML (Pretty): 24-1.0067644-rdf.xml
RDF/JSON: 24-1.0067644-rdf.json
Turtle: 24-1.0067644-turtle.txt
N-Triples: 24-1.0067644-rdf-ntriples.txt
Original Record: 24-1.0067644-source.json
Full Text
24-1.0067644-fulltext.txt
Citation
24-1.0067644.ris

Full Text

   Design and Implementation of Microcontroller-Based Direct Methanol Fuel Cell/Lithium Polymer Battery Hybrid Energy Management System by Di Chen B.A.Sc., Nanjing University of Aeronautics and Astronautics, 2007 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE in The Faculty of Graduate Studies (Electrical and Computer Engineering) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) August 2009  ©Di Chen 2009   ii Abstract The Direct Methanol Fuel Cell (DMFC) has been considered as one of the competitive alternatives for battery technology as it has much higher energy density, faster recharging and does not require complicated control systems like a fuel reformer or compressed gas tank as needed by a hydrogen fuel cell. However, current DMFC technology suffers from the low power density caused by low reaction rate and undesired “methanol crossover” issues, which brings a big challenge for its application in practical systems. This thesis presents a practical design and prototype development of a DMFC/battery hybrid energy management system, which can be provided as one possible solution for the low power and cold start issues. First of all the existing fuel cell hybrid system schemes and design of the auxiliary units (BOP) are surveyed and compared. Based on the analysis above a microcontroller-based DMFC and Lithium Polymer Battery hybrid system is proposed. After that a novel “Battery-Current-Based Hybrid Control (BCBHC)” is proposed to provide active load sharing and proper battery charging and protection. The DMFC will follow the average battery current by neglecting the battery current transients and charge the battery by following the Constant-Current and Constant Voltage charging scheme when possible. A variety of battery protections, such as overcharging, overcurrent and charging current limitation, are implemented by the BCBHC and protection circuit. A detailed system design and modeling are then presented. The models are developed and simulated in PSIM. The simulation results are analyzed and showed the validity of proposed hybrid control. At the end a prototype hybrid EMS controller board has been implemented to further validate the hybrid system design. The dynamic behavior of DMFC/Battery hybrid system is examined and tested under a series of load experiments.  The measured results have proved the feasibility and stability of the designed hybrid control.   iii Table of Contents Abstract .........................................................................................................................ii Table of Contents ........................................................................................................ iii List of Tables ..............................................................................................................viii List of Figures ............................................................................................................... ix List of Abbreviations ................................................................................................... xii Acknowledgements .....................................................................................................xiii Dedication ................................................................................................................... xiv 1 Introduction ............................................................................................................ 1 1.1 Project Motivation ............................................................................................. 2 1.2 Design Objective ............................................................................................... 3 1.3 Thesis Outline ................................................................................................... 4 2 Background ............................................................................................................ 5 2.1 Basic Principle of Direct Methanol Fuel Cell ..................................................... 5 2.1.1 Performance of DMFC ............................................................................... 7 2.1.2 Cell and System Efficiencies ...................................................................... 8 2.2 Characterization of DMFC Stack ....................................................................... 9 2.2.1 Stack Specification ..................................................................................... 9 2.2.2 Polarization Curve ...................................................................................... 9 2.2.3 Cell Balancing .......................................................................................... 11   iv 2.3 Performance Evaluation of DMFC System ...................................................... 12 2.4 Balance of Plant (BOP) ................................................................................... 14 3 DMFC Hybrid System Architecture.................................................................... 18 3.1 Energy Storage Devices for Hybridization ....................................................... 18 3.2 Review and Analysis of Hybrid Setup Architectures........................................ 19 3.2.1 Passive Hybridization ............................................................................... 19 3.2.2 Active Hybridization ................................................................................ 21 3.3 Load Sharing Control Strategy ........................................................................ 26 3.3.1 Maximum Fuel Cell Current Limitation ................................................... 26 3.3.2 Constant Fuel Cell Current Output (No Load Following) ......................... 27 3.3.3 Limited Slow Load Following .................................................................. 28 3.4 Comparison ..................................................................................................... 28 3.5 Proposed Hybrid System Architecture ............................................................. 29 3.5.1 Microcontroller-Based Hybrid Energy Management System Board .......... 30 3.5.2 DMFC BOP ............................................................................................. 30 4 Proposed Battery-Current-Based Hybrid Control Strategy .............................. 32 1.1 Hybrid Control Considerations ........................................................................ 32 4.1.1 Proper Battery Charging ........................................................................... 32 4.1.2 DMFC Protection ..................................................................................... 33 4.1.3 Soft Start of DMFC .................................................................................. 34 4.1.4 Li-Po Battery Charging Current Limitations ............................................. 34 4.2 Proposed Battery-Current-Based Hybrid Control strategy ................................ 34 4.2.1 Battery Charging Management (BCM) ..................................................... 35 4.2.2 Battery Current Control Loop (BCCL) ..................................................... 36   v 4.2.3 Charging Current Protection (CCP) .......................................................... 37 4.2.4 Fuel Cell Current Control Loop (FCCL) ................................................... 37 4.3 Cold Start Procedures ...................................................................................... 38 4.4 DMFC BOP Control........................................................................................ 40 5 System Design and Simulation ............................................................................. 42 5.1 DC/DC Converter Design ................................................................................ 42 5.1.1 Converter Topologies ............................................................................... 42 5.1.2 Inductance Value Selection ...................................................................... 43 5.1.3 Input and Output Filter Capacitor ............................................................. 44 5.2 Modeling and Control Loop Design................................................................. 45 5.2.1 Design and Modeling of Current Sensing Feedback.................................. 45 5.2.2 Small-Signal Modeling of Average Input Current Mode Control .............. 47 5.2.3 FCCL PI Compensator Design ................................................................. 50 5.3 PSIM Simulation ............................................................................................. 53 5.3.1 Models for DMFC and Battery Source ..................................................... 54 5.3.2 Current Sensing Blocks ............................................................................ 55 5.3.3 Hybrid Controller Block in PSIM ............................................................. 56 5.4 Simulation Results .......................................................................................... 57 5.4.1 Dynamic Response of Fuel Cell Current Control Loop (FCCL) ................ 57 5.4.2 Dynamic Response of Battery Current Control Loop (BCCL) .................. 59 5.4.3 Charging Current Protection Mode ........................................................... 60 5.4.4 Pulse Load Test ........................................................................................ 62 6 Hardware Development and Testing ................................................................... 65 6.1 System Specification ....................................................................................... 65   vi 6.2 Hardware Components .................................................................................... 65 6.2.1 TPS3826-Based Digital-controlled Synchronous Buck Converter............. 66 6.2.2 +5VDC Auxiliary Power Supply (LDO)................................................... 66 6.2.3 DMFC/Battery Current Sensing Circuit .................................................... 67 6.2.4 Battery Protection Circuit ......................................................................... 68 6.3 Experimental Results and Analysis .................................................................. 69 6.3.1 Waveforms ............................................................................................... 70 6.3.2 Power Efficiencies ................................................................................... 72 6.3.3 Dynamic Responses ................................................................................. 73 6.3.4 Pulse Load Test ........................................................................................ 75 6.3.5 No Load Test (CCP) ................................................................................. 79 6.3.6 Battery CC-CV Charging ......................................................................... 80 6.3.7 Hybrid System Performance ..................................................................... 82 7 Conclusions and Future Work ............................................................................. 83 7.1 Conclusions ..................................................................................................... 83 7.2 Contributions .................................................................................................. 83 7.3 Future Work .................................................................................................... 84 7.3.1 Overall System Efficiency and Performance Improvement ....................... 84 7.3.2 Better Temperature Control of DMFC Stack by Self-heating .................... 85 7.3.3 Searching for New Hybrid System Scheme .............................................. 85 7.3.4 Hybrid Optimization for Long Term Lifetime Improvement ..................... 86 References .................................................................................................................... 87 Appendices ................................................................................................................... 89 Appendix A. Schematics of Prototype Hybrid EMS Board ..................................... 89   vii Appendix B. PCB Layout (Revision 1.1 02/17/2009) ............................................. 90 B.1 Top Layer and Silkscreen ................................................................................. 90 B.2 Bottom Layer ................................................................................................... 90   viii List of Tables Table 1.1 Energy density comparison for methanol and hydrogen storage technologies. .. 1 Table 2.1 Specification of Antig® 2W and 12W DMFC stacks ....................................... 9 Table 2.2 Methanol consumption rate ............................................................................ 12 Table 2.3 Summary of BOP for DMFC systems ............................................................ 17 Table 3.1 Comparison of different energy storage devices ............................................. 18 Table 3.2 Comparison of different hybrid configurations ............................................... 28 Table 5.1 Design parameters of synchronous buck converter ......................................... 43 Table 5.2 Design specification of current sensing feedback circuit................................. 55 Table 6.1 Prototype system specification ....................................................................... 65 Table 6.2 DMFC and battery operation status under different loads ............................... 77 Table 6.3 Energy and power density comparison of DFMC, Li-Po battery and hybrid system ........................................................................................................................... 82   ix List of Figures Figure 2.1 The schematics of PEM Direct Methanol Fuel Cell ......................................... 5 Figure 2.2 Typical polarization curve for DMFC with kinetic, ohmic, concentration and crossover potential losses. ............................................................................................... 7 Figure 2.3 Polarization curve of 2W Stack under different temperatures and methanol concentrations ............................................................................................................... 10 Figure 2.4 Polarization Curves of 12W Stack under different temperatures .................... 11 Figure 2.5 Single cell voltage balancing ........................................................................ 12 Figure 2.6 DMFC system performance evaluation ......................................................... 13 Figure 2.7 Full active DMFC BOP system ..................................................................... 14 Figure 2.8 An example of full active DMFC system design[11] ..................................... 15 Figure 2.9 A simple version of active DMFC BOP system ............................................ 16 Figure 2.10 Compact passive DMFC System ................................................................. 17 Figure 3.1 Passive hybrid: connected by two diodes ...................................................... 19 Figure 3.2 V-I characteristics of fuel cell/ battery passive hybrid system ....................... 21 Figure 3.3 Active hybrid: battery voltage regulated ....................................................... 21 Figure 3.4 Fuel cell and battery V-I characteristics with battery voltage regulated ......... 22 Figure 3.5 Active hybrid configuration 2 with BOP load ............................................... 22 Figure 3.6 Active hybrid: battery output controlled by bidirectional converter ............... 23 Figure 3.7 Bidirectional two-quadrant chopper (Buck-Boost) for active hybrid.............. 23 Figure 3.8 Active hybrid: dynamic DMFC output control .............................................. 24 Figure 3.9 Active hybrid: both DMFC and battery output controlled ............................. 25 Figure 3.10 Schematic diagrams of two types of multi-input converters......................... 26 Figure 3.11 The current profile for hybrid system with constant fuel cell output ............ 27 Figure 3.12 The hybrid system with bleeder circuit protection ....................................... 27   x Figure 3.13 Proposed hybrid system configuration ........................................................ 28 Figure 3.14 The Diagram of proposed DMFC/battery hybrid system ............................. 29 Figure 4.1 Battery-Current-Based Hybrid Control (BCBHC) ......................................... 34 Figure 4.2 Battery charging control algorithm ............................................................... 36 Figure 4.3 Dynamic responses of DMFC OCV establishment. ....................................... 38 Figure 4.4 Cold start procedures with OCV protection ................................................... 39 Figure 4.5 BOP control algorithm .................................................................................. 40 Figure 5.1 V-I characteristics of 12W DMFC stack and Li-Po battery ........................... 42 Figure 5.2 The schematics of synchronous buck converter ............................................. 43 Figure 5.3 Functional diagram of current sensing amplifier MAX4173[13] ................... 45 Figure 5.4 Ideal synchronous buck converter circuit model ............................................ 47 Figure 5.5 Block diagram of fuel cell current control loop ............................................. 50 Figure 5.6 Bode plot for uncompensated closed loop ( )uT s  ............................................ 51 Figure 5.7 Bode plot for compensated closed loop ( )cT s  ................................................ 52 Figure 5.8 PSIM simulation model for DMFC/battery hybrid system ............................ 53 Figure 5.9 Linear curve fitting based on V-I test data..................................................... 54 Figure 5.10 Fuel cell current sensing block .................................................................... 55 Figure 5.11 Bidirectional battery current sensing block ................................................. 56 Figure 5.12 PSIM hybrid controller block (BCCL + FCCL) .......................................... 56 Figure 5.13 Simulated step response of FCCL (Step: 1.2A Reference, Kp = 0.2,Ki = 500)  ...................................................................................................................................... 58 Figure 5.14 Simulated dynamic response of fuel cell current under load disturbance ..... 58 Figure 5.15 Simulated step response of battery current loop (Ki = 0.2, 0.4 and 0.8) ....... 60 Figure 5.16 Simulation result of battery charging protection mode ................................ 61 Figure 5.17 Load profile for test 1 ................................................................................. 62 Figure 5.18 Simulation results for Pulse Load Test 1 ..................................................... 63 Figure 5.19 Load profile for Pulse Load Test 2 .............................................................. 63   xi Figure 5.20 Simulation results for Pulse Load Test 2 (CCP mode) ................................. 64 Figure 6.1 Diagram of DMFC/battery hybrid energy management system board ........... 66 Figure 6.2 Basic circuit connection and output resistance selection[15] ......................... 68 Figure 6.3 Prototype circuit board of DMFC/battery hybrid EMS (Rev 1.1) .................. 69 Figure 6.4 Testing platform for hybrid EMS board prototype ........................................ 70 Figure 6.5 Synchronous buck converter gate driving signal ........................................... 71 Figure 6.6 Output voltage AC ripple waveform ............................................................. 71 Figure 6.7 Measured power efficiency under different duty cycles and loads ................. 72 Figure 6.8 Voltage drop and power losses on diode ....................................................... 73 Figure 6.9 Measured step response of fuel cell current loop (Kp = 0.2, Ki = 500)............ 74 Figure 6.10 Step response of BCCL (Ki_BCCL = 0.2)........................................................ 74 Figure 6.11 Experimental results for Pulse Load Test 1 ................................................. 76 Figure 6.12 Experimental results for Pulse Load Test 2 (battery charging current limitation) ..................................................................................................................... 76 Figure 6.13 Constant current load test (CC 2A@1min, 4A@1min, 1A@1min) .............. 78 Figure 6.14 DMFC/battery output under random load test ............................................. 79 Figure 6.15 No load test (CCP Mode) ............................................................................ 80 Figure 6.16 Experimental test results of constant current and constant voltage charging scheme .......................................................................................................................... 80 Figure 6.17  Battery charging control algorithm can avoid false triggering of battery charging caused by load disturbance .............................................................................. 81 Figure 7.1 Dual battery hybrid system for better battery charging control ...................... 85   xii List of Abbreviations ADC Analog-to-Digital Converter BCBHC Battery Current Based Hybrid Control BCCL Battery Current Control Loop BCM Battery Charging Management BOP  Balance Of Plant CC Constant Current CCP Charging Current Protection CV Constant Voltage DC Direct Current DMFC Direct Methanol Fuel Cell EMS Energy Management System FC Fuel Cell FCCL Fuel cell Current Control Loop HEMS Hybrid Energy Management System LDO Low Drop Out MEA Membrane Electrode Assembly OCV Open Circuit Voltage PEM Polymer Exchange Membrane PWM Pulse Width Modulation SOC State of Charge    xiii Acknowledgements First and foremost special thanks must go to my supervisor Dr. William G. Dunford, who introduced me into the graduate study and research, for his consistent support and guidance throughout my graduate study at the University of British of Columbia. His knowledge of state of the art research in both academia and industry made him an invaluable resource for this thesis. I would like to express my grateful thanks to Dr. Jacek Chrostowski, chairman and CTO of Methusala Microcells, Inc., for initializing this project and providing facilities, financial assistance and technical advice through the project. Many thanks are due to the development and research manager Mr. Robert Schilack, whom I have been working with for last 6 months. My project will not be finished without your kind assistance and advice. I would like to thank my peers in the power system lab: Feng Wei, Mehmet Sucu, Pooya, Sina and Tom, for the help and support of my master study and thesis. Special thanks to Feng Wei, who gave me a lot of useful advices and helped me with the component selections, and Mehmet for his support for my study and TA job. Thanks all forks! I will never forget the last 2 years experience working with all of you. Finally I would like to thank my families, my dear parents, for their true love and strong support for my study. Especially I have to thank beloved Ce Wang, for her love, encouragement and support.    xiv Dedication    Dedicated to  My parents and Ce Wang    1 1  Introduction The Direct Methanol Fuel Cell (DMFC) has been considered as the ideal fuel cell system for small and medium power application since it generates electric power by the direct conversion of the liquid methanol fuel at the fuel cell anode, instead of complicated and bulky hydrogen reformer or compressed hydrogen gas storage.  Table 1.1 compares the energy density of methanol with several other energy storage technologies [1]. Methanol is liquid at ambient temperatures and has much higher energy density (5.26kWh/kg) compared to the traditional compressed hydrogen gas fuel and battery. It can be stored unpressurized, which offers simpler storage systems, and can be distributed via the available infrastructure. Furthermore methanol can be produced on a large scale from natural gas with efficiencies of around 65-70%. If produced from biomass, methanol could participate in a closed carbon dioxide cycle, and thus help preventing global warming. Table 1.1 Energy density comparison for methanol and hydrogen storage technologies. Fuel Energy density Storage method Energy density considering storage Diesel  13.8kWh/kg Liquid in plastic tank 13.1kWh/kg Gasoline 12.2kWh/kg Liquid in plastic tank 11.6kWh/kg Methanol 5.52kWh/kg Liquid in plastic tank  5.26kWh/kg Hydrogen 33.3kWh/kg 300bar pressure  0.2kWh/kg Metal hydride cylinders 0.22kWh/kg Indirect methanol from methanol 2.3kWh/kg  Lithium-ion Polymer battery 0.013kWh/kg - 0.013kWh/kg Lead-Acid battery 0.025kWh/kg - 0.025kWh/kg The simple system structure and very high energy density of DMFC system make it a perfect solution for small industrial off-grid power supply and portable power for   2 consumer electronic devices where the long operational time is desired. The traditional solution for these applications is to use a large pack of batteries or diesel engine. However, the current energy density of the latest battery technology is still low (100- 150Wh/kg for lithium battery vs. 5kWh/kg for methanol fuel) and direct methanol fuel cell is much more environmentally friendly than diesel engines, though the energy density of . The future potential applications and market of DMFC are numerous. For industrial application the DMFC can provide power for remote telecommunication equipment and remote scientific investigation equipment such as weather station and wireless monitoring devices, emergency AC power or UPS for hospitals, etc. Many consumer applications will also emerge including mobile portable power for battery charging of personal mobile electronic devices. However, one big challenge associated with the DMFC is the slower fuel anode reactions than the hydrogen fuel cell. The oxidation of hydrogen occurs readily – the oxidation of methanol is a much more complex reaction, and proceeds much more slowly. This results in a fuel cell that has a far lower power for a given size. Another big issue for DMFC is “methanol crossover” through the PEM type of electrode, which further reduces the cell open circuit voltage and leads to lower power output. 1.1 Project Motivation The initial motivation of this project is to accommodate an existing DMFC system into a product that was powered by lithium battery pack. However later on we found that the power output of DMFC system with the same weight and size can never beat the old battery pack, though its energy density can be superior then battery with proper system design. Thus the low power problem must be solved before the use of DMFC for this application. There are many solutions for the low power problem of DMFC. For example, the membrane and catalyst can be improved to reduce the crossover and increase the current density but one of the fastest ways is to build a hybrid power source which consists of another high power source such as lithium battery to meet the high peak load. Moreover before the start up of DMFC stack, a back-up power source that can support the operation   3 of auxiliary units is required. Therefore a hybrid power system utilizing a secondary energy system as a supplementary source is required for a functional DMFC system. Then hybrid system control methodology of two sources must be determined. There is plenty of literature discussing the system architecture and control design of fuel cell and battery (or ultracapacitor) hybrid system. However, the papers that focus on the DMFC hybrid system are few. Reference [2] and [3] analyzed the performance of hybrid DMFC and battery hybrid system but its hybrid control is passive and simple without the consideration of DMFC and battery protection and no active load sharing. For general fuel cell and battery hybrid systems, paper from Lee and Chang [4] has given a good overview of fuel cell/battery hybrid system design and proposed a “Dynamic Duty Cycle Hybrid Control”, which is able to dynamically change the load sharing between FC and Battery. However, the proposed hybrid control results in a significant discontinuous current flow from battery and fuel cell, which is not desirable. Also their hybrid system requires additional battery charging circuit and increases the system complexity. Zhenhua and Thounthong present their control strategy and hybrid design of fuel cell/battery hybrid system in papers [5-9]. Their control strategies are either based on several states/modes or based on the SOC (state-of-charge) of battery and the hybrid system must be switched among different modes, which results in a complicated control. Also the normal battery charging scheme (constant current followed by constant voltage) is ignored and some basic battery protections are not fully covered in their design. Therefore there is a great demand for a novel design of DMFC hybrid EMS with a simple but effective hybrid control and complete DMFC and battery protection. The target system must be able to avoid any accidental damage to the DMFC, provide fast dynamic load following and have an improved performance. 1.2 Design Objective The main objectives of this project are the analysis of DMFC hybrid system architecture, proposal of a novel hybrid control and prototype implementation for DMFC-powered Hybrid Energy Management System (HEMS) including following features:   4  Active load sharing control. The load sharing between DMFC and battery must be actively controlled by a DC/DC converter. The hybrid control strategy must let the DMFC follow the average load and all peak transient load current should be supplied by battery.  Full DMFC overload protection. The DMFC voltage and current must be limited to provide protection for the cell.  Battery Protection. The HEMS must be able to monitor the battery operation status and adjust DMFC output or cut it off from load if battery undercharged or overcurrent is detected.  Proper Battery Charging. If the load is constant or not connected, DMFC must start up when the battery capacity is too low so that battery can be maintained within acceptable SOC range for normal system operation. The HEMS must actively control the DMFC to provide charging current for battery and average current for load. 1.3 Thesis Outline The remainder of this thesis is organized as follows: Chapter 2 provides a concise overview of DMFC technology, characteristics of DMFC stack which will be used for prototype testing and design of DMFC BOP system. Chapter 3 surveys the available DMFC hybrid system architectures and proposes system-level design of a microcontroller-based DMFC/Battery hybrid system. Then starting from Chapter 4 a novel Battery-Current-Based Hybrid Control (BCBHC) is proposed and detailed control algorithm is presented. Chapter 5 describes the detailed system design including DC/DC converter and control loop design. Also a PSIM models for the hybrid system is built and its simulation is presented and discussed. The prototype hardware testing results are presented and analyzed in chapter 6. The last chapter presents conclusions drawn from this research project, as well as contributions made and possible future work.   5 2  Background The fuel cell is an electrochemical device that converts chemical energy into electricity and heat without combustion. In contrast to batteries, the chemical energy is not stored in the cell. Instead, the chemical energy in form of the fuel and the oxidant is continuously fed to the fuel cell and converted into electricity. The process of the conversion continues as long as the fuel and the oxidant are fed to the system. 2.1 Basic Principle of Direct Methanol Fuel Cell  Figure 2.1 The schematics of PEM Direct Methanol Fuel Cell The Direct Methanol Fuel Cell (DMFC) has the similar structure as Proton Exchange Membrane (PEM) hydrogen fuel cell. As shown in Figure 2.1[10], the electrolyte material is sandwiched between two thin porous electrodes, the anode and the cathode.   6 The fuel is electrochemically oxidized in the anode and the oxidant is electrochemically reduced in the cathode. Since the cathode potential established by the oxidant reaction is higher than the anode potential, the electrons delivered by the fuel are flowing through an external circuit from the anode to the cathode. If a proton conducting electrolyte is used, the protons formed by the fuel oxidation migrate through the electrolyte from the anode to the cathode. At the cathode, the protons recombine with oxygen to form water as a reaction product. For PEM-based DMFC, the anode reaction is presented by the equation below[1]  3 2 2CH OH  H O 6H   6e  CO       (2.1) The H +  ions move through the electrolyte and the electrons move round the external circuit to form a load current flow. Water is needed at the anode for anode reaction, though it is produced more rapidly at the cathode via the cathode reaction:  +2 2 1 1 O  + 6H  + 6e  3H O 2    (2.2) Adding reaction equation (2.1)and (2.2) gives the total reaction equation for the DMFC:  3 2 2 2 1 CH OH + 1 O   2H O + CO 2   (2.3) The change in molar Gibbs energy for this reaction is 698.2 kJ/molG   , and the maximum theoretical voltage of single cell is  o G E zF    (2.4) Where z is the number of electrons transferred in the reaction per molecule methanol consumed, and F is the Faraday constant (96485.3 C/mol). For the reaction in (2.2) there are 6 electrons transferred thus at 25°C the reversible voltage is 1.21 V, which is very close to the reversible voltage for hydrogen-fed polymer electrolyte membrane fuel cells.     7 2.1.1 Performance of DMFC Current Density (A/cm2) V o lt ag e (V ) 1.21 0.9 Nerst Voltage Eo Actual OCV < Eo (A) Activation polarization (Kinetic Loss) (C) Concentration polarization (mass transport loss) iLim (B) Ohmic polarization (loss) (D) OCV loss  Figure 2.2 Typical polarization curve for DMFC with kinetic, ohmic, concentration and crossover potential losses. The polarization curve, which represents the cell voltage-current relationship, is the widely used standard figure of merit for evaluation of fuel cell performance. Figure 2.2 is an illustration of a typical polarization for a fuel cell. The losses in different regions are presented below:  Region (A) is dominated by Activation (Kinetics Loss) over potential of the electrodes.  In Region (B) the ohmic polarization of the fuel cell leads the most of losses. This losses are caused by the voltage drop on all the electrical and ionic conduction resistance through electrolyte, catalyst layers, interconnects of cell and wire contacts  The losses in Region (C) represent the concentration polarization of the fuel cell, caused by the mass transport limitations of the reactants to the electrodes.   8  The voltage losses in Region (D) between ideal maximum reversible voltage and actual OCV are mainly caused by undesired methanol crossover through the electrolyte, internal leakage current through the electrolyte, or some other contamination. 2.1.2 Cell and System Efficiencies Based on all the losses described above, the cell voltage when a load is applied is estimated by the equation:  cell rev ohm kin concV E V V V     (2.5) As shown in Figure 2.2, at low current densities, poor kinetics at both the anode and the cathode represent the main limiting factor. In the intermediate region, the internal resistance leads to a nearly linear drop of the voltage, before the mass transport limitation causes a rapid drop near the limiting current density. Usually, the maximum power point is reached before current reaches the maximum limitation. Cell efficiency can be estimated by using the equation below[1]:  100% 1.21  cellcell fu V    (2.6) in which fu  is the fuel utilisation coefficient with a typical value 0.8-0.95. From the equation above we can see a direct relationship between power/current output and cell efficiency of DMFC, since the cell voltage decreases with the rise of current output. Generally higher current density means higher power output but lower cell efficiency. However, the practical efficiency for DMFC is even lower than this, which is around 40% at maximum and 20% for operation. Please note that part of energy losses are transferred to heat so that with proper insulation the DMFC stack can be self-heated up to 40-50°C without external pre-heating, which is good for DMFC performance since the electrochemical reaction is sensitive to the temperature. Under higher temperature the power output of DMFC will increase dramatically.  Therefore the temperature stabilization is very important for the performance of DMFC stack. An additional heating system or a simple thermal insulation can be used to control and stabilize the stack temperature.   9 As for DMFC System efficiency, the power consumption of the auxiliary components that support the operation of DMFC (called “Balance of Plant”), such as circulating pump, air blower and/or electronic controllers, must be considered. The system efficiency is expressed as below:  100%loadsystem cell load BOP P P P      (2.7) 2.2 Characterization of DMFC Stack 2.2.1 Stack Specification Antig® 2W 4-cell passive air-breathing DMFC stack is used to for characterization testing to evaluate the performance of DMFC under different conditions. For this project the other Antig ®  12W DMFC stack with 24 cells is selected for system prototype testing. The detailed stack specification can be found in Table 2.1. Note that the methanol is diluted by water and all the methanol concentrations listed in this table and remaining thesis are expressed in percentage by volume. Table 2.1 Specification of Antig® 2W and 12W DMFC stacks Fuel Cell Stack Data 2W Stack 12W Stack Dimension 110(L)* 55(H) *7(W)mm 57 (L) * 61(W) * 161(H) mm Weight  64 ~68g 727 g Quantity of cells   4 cells 24 cells Cell MEA specification   Dupont MEA5 , Nafion 117 , 35 mm×35mm , Carbon Cloth Anode Reactant   10%(v/v) Methanol 10% (v/v) Methanol Cathode Reactant   Passive Air Breathing Active Air Blower Operating Temperature   0°C ~ 60°C 10°C ~70°C Fuel Concentration  6.0% ~ 10.0% (v/v) 6.0% ~ 10.0% (v/v) OCV  2.4~3.2V  16.8V ~19.2V Operating Voltage   1.2V ~ 1.6V 7.2V~10.8V Average Output Power   2W (±3~5%) @50°C,1.2V 12W(±6~10 %) @50°C, 7.2V 2.2.2 Polarization Curve The polarization curves and power output of 2W DMFC stack under different temperatures and methanol concentrations are measured using a fuel cell testing station   10 with built-in programmable electronic load and data acquisition system. By programming the load current from 0A to 2A with a step of 100mA and time interval 50 seconds, a series of stable cell voltage under different load current can be recorded. The measured results are presented in the figure below: 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.5 1 1.5 2 2.5 3 Current (A) V o la g e  ( V )   Polarization Curve (30 o C 6%) Polarization Curve (30 o C 10%) Polarization Curve (50 o C 6%) Polarization Curve (50 o C 10%)  Figure 2.3 Polarization curve of 2W Stack under different temperatures and methanol concentrations According the testing data in Figure 2.3, the dependence of the cell voltage as a function of the temperature and methanol concentration is validated. It can be found that increasing of operation temperature from 30°C to 50°C improves the stack performance dramatically. The reason for that is the improved catalyst activity. If we compare the polarization curves under methanol concentration 6v% and 10v% we conclude that the change of methanol concentration does not affect the performance significantly. In the high current range between 1.5 and 2 Amps it is apparent that the performance of both curves with 6v% (at 30°C and 50°C) drops more than the curves with 10v%. That is an indication that the fuel cell operating at lower concentration is converging towards the limit current. The performance of DMFC is even better with low methanol concentration within the low current range because of the more methanol crossover under higher concentration.   11 The polarization curves of 12W stack using 10v% methanol solution at different temperatures are shown in Figure 2.4.  At this time the constant voltage loads from 14.4V to 7.2V with a voltage step of 1.2V and time interval 50 seconds are used for polarization scan. The polarization tests under 30, 40 and 50°C are measured. Again the fuel cell voltage drops rapidly at low current range and then a linear voltage drop due to the internal resistance can be observed. It is obvious that the total performance of DMFC is very sensitive to the operating temperature. 0 0.5 1 1.5 2 2.5 3 6 8 10 12 14 16 18 20 Current (A) V o lt ag e (V )   Polarization Curve (30 o C) Polarization Curve (40 o C) Polarization Curve (50 o C)  Figure 2.4 Polarization Curves of 12W Stack under different temperatures 2.2.3 Cell Balancing To demonstrate the voltage balancing on each cell, the single cell voltage of 12W 24-cell DMFC stack are measured and presented in the Figure 2.5. The OCV of DMFC stack is 19.6V and single cell OCV is approximately 0.81V. The measured results show a very stable and balanced individual cell voltage distribution under different load conditions.   12 0 5 10 15 20 25 0 0.2 0.4 0.6 0.8 1 Number of Cell V o lt a g e  ( V )   Open Circuit Voltage 10V load 9.6V load  Figure 2.5 Single cell voltage balancing 2.3 Performance Evaluation of DMFC System The methanol consumption rate for the 12W stack is measured as shown below: Table 2.2 Methanol consumption rate DMFC Methanol Consumption Rate Testing Condition 24 cells 12W Stack 0.75~0.85 Wh/cc 1. Methanol Concentration: 8V% ~ 10V% 2. Average Power output: 13 W 3. Operation Temperature: 50C~55C The performance of DMFC system, such as energy density, strongly depends on the system configurations. The pure methanol has extremely high potential energy density (5.26kW/kg) but the real DMFC system includes the weight of DMFC stack and auxiliary units thus the actual energy density will be much lower. More importantly, none of current DMFC can have a good performance under high methanol solution so the total energy density of DMFC system will be very low if 10v% methanol solution tank is used. The Figure 2.6 presents the performance comparison of different DMFC system configuration as well as lithium ion battery. If a fuel tank with 200cc pure methanol is used then the total energy density is approximately 192Wh/kg, which is higher than   13 energy density of current typical Lithium ion battery technology. A larger tank with 500cc methanol is able to increase the energy density up to almost 400Wh/kg, which is 2.5 times higher than Li-ion battery. A lithium ion battery with same stored energy is selected here for comparison. It can be seen that a heavier (2.83kg) battery pack is required to provide the same energy.  Figure 2.6 DMFC system performance evaluation However if 500cc 10v% methanol instead of neat methanol is used, which is common for some current passive DMFC systems, the energy density will be only 35Wh/kg. Since the current MEA only tolerates low methanol concentration, one solution is to contain a small 10v% methanol solution tank along with the other big neat methanol. The system controller will monitor the methanol concentration in the small tank and open the valve (or pump) to pump in a small amount of neat methanol from big tank if methanol concentration is lower than the limit. Nevertheless this increases system complexity and may not be suitable for some small passive DMFC applications.   14 2.4 Balance of Plant (BOP) Water Tank 100% MeOH Fuel Tank DMFC Stack CO2 Removal Air intake Fuel Supply Pump Methanol Circulating Pump Air Out Mixing Chamber Water supply Pump Liquid Level Sensor Condensed Water Water Recycling Pump Methanol Conc. Sensor  Figure 2.7 Full active DMFC BOP system The auxiliary units in DMFC system that support the operation of stack are called “Balance of Plant” (BOP). An ideal full active DMFC system (Figure 2.7) must be able to dynamically control the methanol concentration, stack temperature and air flow in order to optimize the DMFC performance. One example active system is shown in Figure 2.8. The anode side fuel supply is implemented by neat methanol tank, small water tank, mixing chamber and two pumps. On the cathode side the air blower is used to flow fresh air into the air chamber for reaction. The pure methanol is diluted in mixing chamber and its concentration can be dynamically adjusted by controlling the Methanol supply and water supply pumps. The water supply tank can be small compared to fuel tank for the minimization of system space and weight. Meanwhile the water produced on the cathode can be condensed and reused for fuel mixing. If possible the stack temperature can also be actively controlled by a heater, though in most cases the heat generated by reaction and crossover can be used to heat up the stack with good thermal insulation.   15   Figure 2.8 An example of full active DMFC system design[11] The drawback of this active system is the complicated system structure and high power losses on BOP. Thus for small portable DMFC system the BOP can be simplified, as shown as Figure 2.9. This type of “Half Active System” provides a simpler solution for methanol supply: The pre-mixed 10v% methanol solution is pumped to anode side by a circulating pump and fuel supply pump (or active controlled valve) will only turn on a short time to pump neat methanol into the methanol tank if its concentration drops too much. The methanol concentration sensor can also be removed by simply estimating the methanol concentration and turn on the fuel supply pump periodically. The water recycling can be neglected as well.   16 8v%-10v% MeOH Tank 100% MeOH Fuel Tank DMFC Stack CO2 Removal Air intake Fuel Supply Pump Methanol Circulating Pump Air Out  Figure 2.9 A simple version of active DMFC BOP system For some applications where the system size is very sensitive, such as low power supply for a wireless device, the BOP can be minimized by pure passive control. A small piezoelectric pump can be used to provide slow flow for the anode methanol supply and Carbon Dioxide flush while on the cathode side the air is supplied only passive air breathing. The passive DMFC system is shown in Figure 2.10. According to the analysis above, three types of DMFC BOP systems, “compact passive”, “Half Active System” and “Full Active System” can be defined for different applications, as summarized in Table 2.3.    17 8v%-10v% MeOH Tank DMFC Stack CO2 Removal Passive Air Breathing Methanol Circulating Pump Figure 2.10 Compact passive DMFC System Table 2.3 Summary of BOP for DMFC systems  Compact Passive  Half Active System  Full Active System Circulation Pump Needed for CO2 Flush For methanol supply and CO2 Flush For methanol supply and CO2 Flush Active methanol concentration control No  Yes  Yes Air blower fan  No. Air breathing  Yes  Yes. Heating system  No. Self-heating  No. Self-Heating  Maybe needed. Cathode Water Condenser No  No Yes. Temperature Sensor  No  No Yes if heating included Concentration Sensor  No.  No.  Yes. Liquid Level Sensor  No  No  Yes Cartridge 8 -10v% MeOH tank (maybe higher in future) 1) 8-10v% MeOH chamber for operation 2) Neat MeOH tank 1) Mixing Chamber  (2) Water tank 3) Neat MeOH tank Application Low power wireless device Small power DMFC system Medium power DMFC system   18 3  DMFC Hybrid System Architecture 3.1 Energy Storage Devices for Hybridization Table 3.1 Comparison of different energy storage devices  ANTIG 12W DMFC Stack1 KOKAM 640mAh 2Cell Li-Po Battery MAXWELL BCAP0120 P250 Ultracapacitor Total Weight (g) 1123 30.6 29 Energy Density (Wh/kg) 378.5 154 3.59 Power Density (W/kg) 9.9 967.7 21,500 Voltage Range (V) 7.2-20 6-8.4 0-2.5 Charging N/A Slow (1C) Very Fast Life Cycle Low 500-1200 Over 500,000 The ideal secondary energy storage device for this DMFC hybrid system must have high power density and acceptable energy density for short term load supply. The Table 3.1 lists some basic parameters of two types of positional energy storage devices for DMFC hybrid system: lithium polymer battery and ultracapacitor with comparison with DMFC Stack. The advantages of the ultracapacitor are its virtually unlimited life cycle and extremely high power density. Also since the charging and discharging of ultracapacitor does not involve in any chemical reaction it can be charged at same rate as discharging thus a faster and easier charging can be implemented. Like traditional capacitor, the energy stored in ultracapacitor is proportional to its voltage. Hence the ultracapacitor has a poor voltage discharge curve which can be down to zero if it is fully discharged. Thus the  1 Note that the total weight, energy and power density of DMFC stack are calculated based on the example stated in section 2.3.   19 energy stored in ultracapacitor could not be fully used. Also another DC/DC converter used as the interface between the ultracapacitor and load must be needed for voltage regulation. Compared to the ultracapacitor, lithium polymer battery has lower power density but much better energy density.  One big disadvantage of battery is its very slow and complicated charging process: the battery can only be low charging current up to 1C (for example, if a lithium battery has a capacity of 500mAh then “1C” stands for 500mA current) and CC-CV charging scheme required. However, its high energy density, large amount of mature commercial products and existing charging ICs make it as our first choice for the secondary energy source of DMFC hybrid system. 3.2 Review and Analysis of Hybrid Setup Architectures There are many existing hybrid configurations in the literature. Generally there are two types of hybrid setups: Passive and Active hybridization. Passive hybridization means that two energy sources are connected to the load directly (or through a diode) without any active and dynamic power conditioning. The active hybridization is more advanced compared to passive hybridization since either fuel cell, battery or both outputs are controlled to adjust the load sharing between two sources. 3.2.1 Passive Hybridization Config.1: DMFC/Battery Connected by Diodes: Max DMFC Current Limited Fuel Cell Stack D1 Load D2 Battery  Figure 3.1 Passive hybrid: connected by two diodes The simplest passive hybrid system is the direct parallel connection of two power sources. However this hybrid configuration is not within our consideration since the fuel cell   20 cannot tolerate reverse charging current. Figure 3.1 is a very simple passive hybrid architecture proposed in [4], which can prevent the fuel cell reverse charging. Two diodes (or MOSFET diode for low power application) are placed in series with fuel cell and battery in order to prevent the reverse current for fuel cell and excess charging for the battery.  The typical voltage-current characteristics of fuel cell and battery are shown in Figure 3.2 a). The maximum fuel cell current IFC_MAX is determined by V-I characteristics of fuel cell and battery. As can be seen from Figure 3.2 b), when the load current is larger than IFC_MAX, the fuel cell voltage will drop to lower than battery voltage. So diode D2 will be turned on. The fuel cell will continue to output IFC_MAX and the battery will provide the remaining current.  The diodes work as a logic OR that the energy source with higher voltage will be selected for load supply. The advantages of this hybrid configuration are:  Simple and cost-effective. No complicated control is needed here and all the load sharing is accomplished passively.  Fuel Cell reverse current blocked. The problems with this architecture are:  Maximum fuel cell current varied. With the discharge of battery the battery voltage will vary and thus the maximum current that the fuel cell can provide may changes as well. Also  No Battery protection.  Battery overcharging/discharging and overcurrent protection cannot be implemented with this hybrid configuration.   A separate charging circuit is needed for the battery charging since the battery current can only be unidirectional.   21 Fuel cell Battery I FC_MAX i v i v D1 is ON D2 is OFF D1 is ON D2 is ON V FC I FC_MAX a) The V-I Characteristic of Fuel cell and Battery b) The V-I Characteristic of Passive hybrid system Figure 3.2 V-I characteristics of fuel cell/ battery passive hybrid system 3.2.2 Active Hybridization Config.2: Battery Tied to DC/DC Converter: Max DMFC Current Adjusted Then quickly one can think of inserting a DC/DC converter in series of battery and the battery output voltage will be controlled (Figure 3.3).Hence the fuel cell maximum output power can be fixed or adjustable, as shown in Figure 3.4. The merit of this architecture is the controllable maximum fuel cell current. Also since the battery output voltage is regulated, its output can be used as the auxiliary power supply for some BOP components such as circulating pump or air blower fan, as shown in Figure 3.5. Again for Config.2 the battery charging is not available so battery charging must be implemented using a separate battery charging circuit. Fuel Cell Stack D1 Load Battery DC/DC Conv. IBAT IFC D2  Figure 3.3 Active hybrid: battery voltage regulated   22 Current Voltage VBAT_DC VFC IFC_MAX Figure 3.4 Fuel cell and battery V-I characteristics with battery voltage regulated Fuel Cell Stack D1 Load Battery DC/DC Conv. BOP D2  Figure 3.5 Active hybrid configuration 2 with BOP load Config.3: Battery Tied to Bidirectional Converter: Dynamic Battery Output Control By replacing the DC/DC converter and D2 in Config.2 with a bidirectional Converter (such as 2 quadrant chopper), the battery discharging/charging can be controlled dynamically, as shown in Figure 3.6. In this hybrid configuration the battery current can be in both direction and adjusted by controlling the duty cycle of bidirectional DC/DC converter. One diode is needed between fuel cell and load to block reverse current and FCI is limited by dynamic control of BATI . Buck-Boost Converter, or 2-Quadrant Chopper, is widely used for this application as shown in Figure 3.7. Many references can be found for this hybrid architecture and most of them are designed for fuel cell/Ultracapacitor hybrid system.   23 Fuel Cell Stack Load Battery Bidirectional DC/DC IFC IBAT D1  Figure 3.6 Active hybrid: battery output controlled by bidirectional converter DMFC Stack Q1 D1 Q2 Cbus L Battery D2 D3 Cin LOAD Bidirectional Two-Quadrant Chopper (Buck-Boost)  Figure 3.7 Bidirectional two-quadrant chopper (Buck-Boost) for active hybrid The advantages of this configuration are:  Good management of battery discharging and charging management. The battery current is directly controlled by the bidirectional converter so that the battery overdischarge/charge and overcurrent protection can be implemented.  Fuel cell current can be limited. Although FCI is not controlled directly, it can be limited by adjusting battery current. Also the diode can prevent reverse current flow.  No separate battery charging circuit needed. The battery current now can be controlled to flow into the battery and no additional battery charging circuit is required.   24 The drawbacks of this hybrid configuration are:  High speed and complicated control required. The main objective of the hybrid system is to use a battery for transient peak load power supply and the DMFC only provides average load current. Therefore this bidirectional converter requires very high speed control to ensure fast dynamic response and follow transient load changes. Furthermore compared to the traditional DC/DC converter it is more complicated for bidirectional current control.  Higher stresses for converter components. Since battery is used to provide peak load power, the transient power demand from battery is larger than DMFC output. Thus the converter needs to tolerate higher current and power stresses. Config.4: DMFC Tied to DC/DC Converter: Dynamic DMFC Output Control Fuel Cell Stack Load Battery DC/DC Conv. IFC IBAT  Figure 3.8 Active hybrid: dynamic DMFC output control As an alternative version to the previous configuration, a unidirectional DC/DC converter can be placed between DMFC and load and battery is directly attached to the DC bus, as shown in Figure 3.8. The converter can be buck, boost or buck-boost depending on the fuel cell stack voltage range. The main advantages of this setup are:  Direct fuel cell output control. Now the fuel cell current can be controlled directly and accurately.  DC/DC Converter power rating reduced.  Fuel cell only output average load power so the maximum power through converter is lower than the converter on battery side and inductor and capacitor size and weight can be reduced.   25  Battery charging/discharging is controlled. Battery current is able to flow in both directions and can be controlled indirectly by adjusting fuel cell output. However the disadvantage is:  Battery protection is not available. Although the battery current can be adjusted by changing fuel cell output, the hybrid system is not capable of dealing with battery overcurrent caused by heavy load or even a short circuit. Config.5: Both Battery and DMFC Tied to DC/DC Converter: Advanced Load Sharing More advanced hybrid system may be capable of doing dynamic fuel cell and battery current sharing control, considering the SOC of battery, DMFC/ battery protection and DC Bus voltage regulation. This may need two DC/DC converters or one multi-input converter connected with both fuel cell and battery, as shown in Figure 3.9: Fuel Cell Stack Load Battery IFC IBAT ILoad DC DC DC DC Bidirectional Load Battery Multi-Input DC/DC Conv. IFC IBAT ILOAD Fuel Cell Stack a) Two converters are used to provide active load sharing between two energy sources b) Hybrid Load Sharing by one multi-input converter Figure 3.9 Active hybrid: both DMFC and battery output controlled   26 Fuel Cell Stack Load Battery Load Battery IFC IBAT ILOAD Fuel Cell Stack a) Diagram of Multi-Phase Synchronous Buck Converter b) Diagram of Isolated Multi-Input Converter Q1 Cbus L D1 D2 Q3 Q4 D3 D4 DC AC DC AC AC DCQ2 D5  Figure 3.10 Schematic diagrams of two types of multi-input converters The multi-input converter can be a simple Multi-Phase Buck Converter (Figure 3.10 a)) if the power is low and galvanic isolation is not needed. For high power and large scale application where isolation and large voltage needed, a multi-input isolated converter as shown in Figure 3.10 b) can be selected. This multi-input converter consists of two DC/AC inverters at input-stage, one AC/DC rectifier at output-stage and three-winding coupled transformer. The advantage of this hybrid configuration is obvious: both fuel cell and battery output can be controlled precisely and DC bus voltage can be regulated. Nevertheless, it requires more converter components and more complicated controls, which dramatically increases system weight, cost and design difficulty. 3.3 Load Sharing Control Strategy Typically there are three types of fuel cell and battery load sharing strategies: 3.3.1 Maximum Fuel Cell Current Limitation A simple way of load sharing is to just limit the maximum fuel cell current. Once the load power is too high, the extra load current will be supplied by battery. This current limitation can be achieved by a simple two-diode connection as shown in Config.1 and 2.   27 3.3.2 Constant Fuel Cell Current Output (No Load Following) For Config.2 to 5 where the fuel cell output current can be controlled, keeping the fuel cell at a constant and optimum operating point could make the system efficiency higher and expand the lifetime of fuel cell module. The battery will run in “Free mode”, which means that the charging and discharging of battery is not actively controlled. Instead, the battery is discharged or charged depending on the load profile, as shown in Figure 3.11. IFC_const ILoad t C u rr en t Battery Discharge Battery Charge  Figure 3.11 The current profile for hybrid system with constant fuel cell output This system is suitable for application when the load profile does not change that much. Otherwise in case the load current keeps low and the battery has already been fully charged, the extra energy generated by fuel cell should be dissipated by a “bleeder circuit” to prevent the over-charging of battery. The bleeder circuit will protect the battery but obviously decrease the system efficiency. The hybrid system with bleeder protection circuit can be seen in Figure 3.12. Fuel Cell Stack Load Battery DC/DC Conv. IFC IBAT Bleeder Circuit Iw ILoad  Figure 3.12 The hybrid system with bleeder circuit protection   28 3.3.3 Limited Slow Load Following A more advanced solution can be current-adjustable converter for the fuel cell. The hybrid controller will dynamically monitor the load current and adjust the fuel cell to follow the load current with a slow response and maximum current limitation. By doing this the fuel cell only supplies the average load power and the peak transient load is provided by battery. 3.4 Comparison Table 3.2 Comparison of different hybrid configurations  DMFC Current Limit DMFC Active Control Battery  Protection Battery Active Control Battery Charging System Complexity Config.1 Not fixed No No No No Very simple Config.2 Yes  No Yes No No Low Config.3 Yes Indirect Yes Yes Yes Medium Config.4 Yes Yes No Indirect Yes High Config.5 Yes Yes Yes Yes Yes Most Complicated Fuel Cell Stack Load Battery DC/DC Conv. IFC IBAT Battery Protection Module D1  Figure 3.13 Proposed hybrid system configuration The comparison of all five hybrid configurations is listed in the Table 3.2. For Config.1 and 2 the battery can only be discharged so another separate charging circuit is needed. Generally Config.4 provides almost all the functions that we desire for the hybrid system   29 with acceptable system complexity except battery protection. Considering the functionality, system complexity, design difficulty and total cost, hybrid configuration 4 with an additional battery protection circuit is selected for this project, as shown in Figure 3.13. To provide better battery and DMFC protection, the limited slow load following is selected for our hybrid system. 3.5 Proposed Hybrid System Architecture According to the analysis in section 2.2 and 3.2, a proposed Microcontroller-based DMFC/Li-Po battery hybrid system architecture is shown in Figure 3.14. Air intake Air Out 8v% MeOH Antig 12W (24 cells) DMFC Stack Li-Po Battery  + - (2 cells 7.4V) Digital-controled DC/DC Converter Load Anode fuel supply Fuel Tank Circulating Pump Battery Protection Module Hybird EMS Board Voltage Regulator MSP430F2274 Microcontroller DMFC Input Ibat IfcVDC +5/3.3V  Figure 3.14 The Diagram of proposed DMFC/battery hybrid system   30 The proposed Microcontroller-based DMFC/Battery Hybrid System consists of the components listed as follow: 3.5.1 Microcontroller-Based Hybrid Energy Management System Board  MSP430 microcontroller. This is the brain of entire hybrid energy management system. It measures and collects all the system information such as voltage and current to monitor the system performance. Based on the hybrid algorithm that is programmed in, the MSP430 will send signal to the digital-controlled DC/DC power converter and battery protection circuit to control and limit the fuel cell and battery output current.  The BOP of DMFC system can also be controlled by MSP430.  Digital-controlled DC/DC converter. A power converter controlled by microcontroller will be used as an interface between fuel cell and battery to actively control the power output of DMFC.  Li-Po battery pack.  This supports the power for the EMS board itself as well as the BOP components before the DMFC has fully functional. Meanwhile it supplies the transient and peak load so that the DMFC output will be smooth.  Battery protection module. Two back-to-back P-channel MOSFETS with digital-compatible driving circuit can be placed on the high side of battery pack to provide battery protection. Once the MSP430 detects that the battery voltage is too low (overdischarged) or too high (overcharged), it will switch off the MOSFET to disable battery discharge or charge, in order to avoid further damage.  DC +5V system auxiliary power supply. The battery pack is also responsible for the power supply for MSP430, pump, fan and all other components before DMFC starts up. A simple low drop out voltage regulator can be used here to generate either +5V or +3.3V supply. 3.5.2 DMFC BOP The main objective of this project is to develop a prototype hybrid EMS to control the DMFC and battery output. Therefore the advanced design of BOP is not considered here   31 and initially the BOP will be supported by DMFC testing station. To put it simply, only one circulating pump and air blower is included in the DMFC system. Once the DMFC starts up, the circulating pump is turned on by EMS to produces liquid flow into and out of the DMFC stack to supply the methanol fuel for anode reaction. Meanwhile on the cathode side the air blower provides enough oxygen for cathode reaction. The temperature of DMFC stack is not actively controlled here. Instead a good insulation of stack and tank can help the self-heating of DMFC stack itself.   32 4  Proposed Battery-Current-Based Hybrid Control Strategy After the hybrid system architecture is determined, the next key task is to design the hybrid control strategy for the system. Both the DMFC and lithium polymer battery are very sensitive to overvoltage and overcurrent so particular protection must be implemented to avoid any damage under different load condition. In this chapter some control considerations are discussed and a novel “Battery-Current-Based” hybrid control strategy is proposed based on the idea that DMFC will be the “Battery Maintainer” trying to let the battery current follow the normal constant voltage – constant current charging scheme. 1.1 Hybrid Control Considerations 4.1.1 Proper Battery Charging In order to keep the battery in a good condition, a Li-ion battery charging process should consist of three stages:  Pre-charging: Slow charging using current of 0.1C.  Fast charging (constant current): Constant current charging stage using current of 1C  Constant voltage charging stage: A full charged condition is reached while the current drops down to 0.1C. Most Li-ion battery chargers are based on this Constant Current and Constant Voltage (CC-CV) mode. As we can see, the termination of charging is not only based on the terminal voltage but also the amount of charging current under constant voltage charging stages. Nowadays lithium battery chargers designed solely for battery charging make charging and termination decisions based on the current and voltage out of the charger.   33 Thus if the battery is directly connected to the system load, the load current disturbance will cause several common problems[12]:  Endless charging. If the load current is high the charger cannot terminate charging during the constant voltage charging stage. Also the frequent high load pulse may continually reset the timer and prevents termination.  False fault mode. Under the pre-charging mode (<3V, 0.1C), typically after 30 minutes, if the battery voltage is not above 3 V, some types of charger declares a dead battery and enters a fault mode. Applying a system load, while the charger is in pre-charge conditioning mode, reduces or eliminates the pre-charge current potentially keeping the cell below 3 V, causing the charger to enter fault mode. Therefore it is not recommended to attach the system load directly to the Li-ion batteries when using a stand-alone Li-ion battery charge management controller IC with automatic termination feature. For this hybrid system a customized battery charging algorithm must be developed to follow CC-CV charging with tolerance of system load disturbance. 4.1.2 DMFC Protection 1) Low OCV Protection. When the system is starting up, the air blower pumps air into the cathode side of DMFC and methanol solution in the tank will be pumped into the anode by circulation pump. After several seconds an OCV will be established on the output of DMFC. The OCV can be used to check if the BOP of DMFC is working properly or if the stack is broken. The system will terminate if an unexpected low OCV is measured during the start-up, which indicates the potential fault of BOP and stack. 2) Voltage and Current Limitation.  The minimum operating voltage and maximum output current must be limited to protect the DMFC stack.  A too low voltage may cause reverse voltage added to some cells in the stack due to the inherent imbalance on each cell. Based on the experimental data, 4.8V (0.2V per cell) is the lowest allowed voltage. The fuel cell cannot tolerate any reverse charging so a diode (or active-controlled MOSFET) is needed to prevent any reverse current into the fuel cell. Also the fuel cell maximum current must be limited.   34 4.1.3 Soft Start of DMFC As illustrated before the start-up and dynamic response of DMFC is very slow (in range of seconds). Thus the soft start must be implemented and the frequent switching off of DMFC system is not desired. 4.1.4 Li-Po Battery Charging Current Limitations Although the maximum discharge current of lithium polymer battery can be up to 5-8C, the maximum charging current must be strictly limited to 1C. Therefore the fuel cell output needs to be adjusted when the system load is disconnected or a sudden load current decrease occurs. 4.2 Proposed Battery-Current-Based Hybrid Control strategy Considering all the aspects stated above, a Battery-Current-Based Hybrid Control (BCBHC) strategy is proposed as shown in Figure 4.1. The philosophy of BCBHC is that the DMFC, or “battery maintainer”, is always trying to compensate the battery current caused by load in a slow manner to fulfill a pre-set battery CC-CV charging in a long term rather than transients. The merit of this BCBHC is that the hybrid controller only cares and senses the battery current without additional sensing and control loop for load current. ADC IBAT + - Battery Charge Control ADC VBAT IBAT_REF I Controller Charging Current Protection + - IFC_REF ADC IFC PI Controller PWM Duty Cyle To DC/DC Converter Battery Charging Management VBAT → IBAT_REF Battery Current Control Loop IBAT → IFC_REF Fuel Cell Current Control Loop IFC_REF → Duty Cycle Ki_fast/slow  Figure 4.1 Battery-Current-Based Hybrid Control (BCBHC)   35 4.2.1 Battery Charging Management (BCM) The Battery Charging Management (BCM) checks the status of battery by monitoring its voltage and adjusts battery current reference value trying to let it follow CC-CV charging scheme. Since the battery state-of-charge (SOC) changes very slowly, this loop can be called by an interrupt from Timer in the microcontroller every several seconds. The detailed charging control algorithm is shown in Figure 4.2. As stated before that the standard CC-CV charging scheme may not work well when the battery is connected to the system load, some important changes are made here to avoid this problem. In the practical system the battery is being used all the time and thus the charging can never stop due to the interrupt by the load and the frequent switching between CC and CV. Normally we expect the battery running at the range between 60- 80% of total capacity so pre-charge stage and charging timeout error is neglected. The Battery Charging will stop ( _ 0FC REFI  ) if the battery current reference calculated from constant voltage charging control is less than 0.1C for at least 30 seconds, which can avoid fake charging termination caused by load current disturbance.    36 Interrupt START Battery Charging Enable CCM  Set IBAT_REF = 1C VBAT < 8.4V? VBAT < 5V? Battery Charging Disable END IBAT_REF = 0 IBAT_REF<0.1C? CVM: VBAT → 8.4V Update IBAT_REF CCM: Constant Current Mode CVM: Constant Voltage Mode Battery Charging Disable IBAT_REF = 0 System Fault and Power Off No Yes Yes No No Yes VBAT < 8V? Charging Enable? Yes No Battery Charging Enable No Yes  Figure 4.2 Battery charging control algorithm 4.2.2 Battery Current Control Loop (BCCL) Once the battery current reference _BAT REFI  is obtained from battery charging control function, the error between reference and measured battery current will be calculated. The control goal of BCCL is to adjust fuel cell input reference smoothly to follow the average battery current. Thus a pure integration controller with a large time constant (very small iK ) can be used here to eliminate steady state error and act as a low pass filter to smooth out fast dynamic response. The discrete integration controller is implemented in microcontroller by using the equation below:  _ ( ) integral( ) ( ) 0       FC REF iI n n K rr i i n e  (4.1)   37 Then fuel cell current reference is given by the integral of errors with upper and lower limit and passed to inner fuel cell current regulation control loop. 4.2.3 Charging Current Protection (CCP) Normally we hope that the coefficient of integral Ki to be small (large time constant) so that the fuel cell current will only follow the average value of battery current ignoring the dynamic and transient load disturbance. However in case of emergency when the load is switched off, a large fuel cell current that was supposed to supply the load will rush into battery producing a high charging current that is many times larger than rated maximum limitation (1C), which will damage the battery if happens frequently. In this case iK needs to be adjusted so that the fuel cell current reference can respond quickly to protect the battery. To make it simple and reliable, two sets of iK , _i slowK  and _i fastK , will be chosen depending on the battery charging current every time when the fuel cell reference is generated:  _ _ _ max_ _ _ max_ ( ) ( 1),  if ( ) ( -1), if  ( )          FC REF FC REF BAT BATCHi slow FC REF BAT BATCHi fast err n I n err n I n k I I I n k I I  (4.2) The fuel cell reference is generated by the previous reference value plus the error times the integration coefficient (large coefficient when the battery charging current is too large or small coefficient during normal mode).The values of integral coefficients must be selected carefully so that the system can be stable with both coefficients and the dynamic response of fuel cell is within acceptable range. The advantage of this dynamic integral time constant control is that the fuel cell current reference and system will change smoothly between the boundaries of two modes. 4.2.4 Fuel Cell Current Control Loop (FCCL) The inner fast fuel cell current loop will dynamically adjust the duty cycle allowing the fuel cell output exactly follow its reference given by BCCL. The PI parameters will be estimated by the system small signal modeling (In section 5 ) and adjusted in more detailed PSIM simulation and experimental tests.   38 4.3 Cold Start Procedures The Open Circuit Voltage (OCV) of DMFC must be checked after the system has started up. A low OCV indicates potential faults of circulation pump, air blower or stack. The Figure 4.3 presents the dynamic response of DMFC voltage establishment. Note that when the fan turned on at 16t s , the OCV started to rise and stabilized at about 32s. Hence this DMFC stack needs approximate 15 seconds until the final OCV voltage is established. Therefore once the system is started, the circulation pump and air blower will operate for 30 seconds before connecting to load in order to make sure that the voltage of the DMFC is stable. The detailed cold start procedures can be found in Figure 4.4. 0 5 10 15 20 25 30 35 40 45 50 2 4 6 8 10 12 14 16 18 20 Time in Second V o lt a g e  ( V ) OCV stable Fan was turned on About 15 seconds  Figure 4.3 Dynamic responses of DMFC OCV establishment.   39 SYSTEM START Turn Off Pump, Air Blower Wait for 30 sec? OCV > 16.8V? Turn On Pump, Air Blower END SET System Fault Alarm SYSTEM STOP Yes Yes No No  Figure 4.4 Cold start procedures with OCV protection   40 4.4 DMFC BOP Control INTERRUPT A/D Sample DMFC_STATUS = ON BOP ON END Yes No IFC_REF < IFC_REFMIN? Timer Started? DMFC_STATUS = OFF BOP = OFF BOP = OFF? In Timer Waiting? IFC_REF = 0 BOP = OFF? SET TIMER 30sec CALL COLD_START No Yes Yes No Yes No Yes No  Figure 4.5 BOP control algorithm Normally Balance of Plant (BOP) including the pump and air blowing fan must be turned on all the time when DMFC stack is operating. However in some cases the BOP can be switched off if the DMFC stack is not being used for a certain long time to save power and increase total efficiency. If there are no load on the DC bus and Li-Po battery is fully charged (Battery Charge Enable = False), the _FC REFI must be zero. Since frequent   41 switching on and off of BOP is not desired due to the slow responses of the mechanical components, a delay (for example 30 seconds) can be used before re-open. Since the DC +3.3/5V secondary power supply which supplies hybrid energy management system board always requires a small amount of current to support the normal operation of control board (about 30mA based on the experimental measurement), the fuel cell current reference may not be exactly zero even if no load is applied to the system and the battery is fully charged. A minimum value for fuel cell current reference _ _FC REF MINI  is checked (In this case it is set to 50mA) so that if the _FC REFI has been constantly lower than _ _FC REF MINI  for 30 seconds, which indicates a long term no load condition, the BOP will be switched off for at least 30 seconds (see Figure 4.5). If a large enough battery discharging current is detected then cold start procedure (Figure 4.4) must be applied to ensure secure start-up.   42 5  System Design and Simulation Starting from this chapter the detailed system design including power circuit topologies, control system modeling and parameter calculation will be discussed. After that a hybrid system model will be implemented in PSIM and its simulation results will be presented and analyzed. 5.1 DC/DC Converter Design 5.1.1 Converter Topologies  Figure 5.1 V-I characteristics of 12W DMFC stack and Li-Po battery   43 Vin Cin Q1 Q2 L RLoadCout  Figure 5.2 The schematics of synchronous buck converter The first priority of converter design is to determine which topologies can be used for this hybrid system. The comparison of voltage-current characteristics between 12W DMFC Stack (24 cells) and KOKAM Lithium Polymer Battery Pack (2 cells) is shown in Figure 5.1. It can be observed that although the voltage drop rate of DMFC is larger (20V to 7.2V) compared to Li-Po battery (7.8V to 6.3V), the voltage of DMFC is always higher than battery voltage in the entire normal operation range. Therefore a Buck (step-down) converter can be chosen for this design as the interface between DMFC and battery/load. To further preserve the energy and increase the total power efficiency, Synchronous Buck Converter which replaces the diode with a MOSFET (Figure 5.2) will be used for the further design. 5.1.2 Inductance Value Selection Table 5.1 Design parameters of synchronous buck converter Design Parameters Max. Rated Min. Input Voltage ( inV ) 20V 9.6V 7.2V Input Current ( inI ) 2A 1.2A - Output/Battery Voltage ( Vbat ) 8.4V 7.3V 6V Inductor Ripple Current ( Li ) 10% - - Voltage Ripple ( outv ) 1% - - Switching Frequency ( swf ) - 125KHz - The specification and design parameters of synchronous buck converter are shown in Table 5.1. The inductor current ripple in Buck converter can be expressed as:   44  in outL sw V V i Lf     (5.1) The output voltage inV  depends on the State-of-Charge of Battery and inductor current ripple reaches its maximum value if _ minV V 6Vout bat  when DMFC is running at its rated operating point. In this case duty cycle _ 6 0.625 9.6 in bat min V D V    Also the maximum inductor current is  _ _ 12 2 6 max out max bat min P I A V    if the power losses on the converter are neglected. Thus the maximum ripple occurs when V 6Vout  and D 0.625 and must meet the requirement of maximum current ripple:  _ _ ( ) (1 ) 10% 0.2 in out min L max out sw V V D D i I A Lf        (5.2) The minimum inductor that can meet this current ripple limitation can be estimated by using (5.2):  _ _ (1 )( ) 33.75 in out min min L max sw D D V V L H i f        (5.3) Practically a 47.3 H inductor is selected for hardware implementation. 5.1.3 Input and Output Filter Capacitor The minimum output filter capacitor inC can be selected based on the input voltage and RMS ripple current requirement. The capacitor RMS value of input current ripple can be estimated by using the equation below:  _ (1 )in rms inI I D D   (5.4) The rated input fuel cell current 1.2inI A and duty cycle is 62.5%. By using Eq. (5.4) the RMS ripple can be calculated  _ 1.2 0.625 (1 0.625) 0.5in rmsI A      (5.5)   45 For an input voltage of 20V, two 100μF Panasonic Capacitors ECA1HM101 (50V/RMS 0.25A) can be put in parallel. The output capacitance inC is calculated according to the output voltage ripple requirement:  _1% 0.06ripple out minv V V    (5.6)  2 0.625 167 0.06*125000 out in ripple sw I D C F v f         (5.7) Thus two 100μF Capacitors can be placed in parallel as output filter. 5.2 Modeling and Control Loop Design 5.2.1 Design and Modeling of Current Sensing Feedback  Figure 5.3 Functional diagram of current sensing amplifier MAX4173[13] Since the fuel cell output current and battery charging/discharging current must be precisely controlled to provide active hybrid and current limitation, two current sensing feedback loops have to be designed properly. In this project the High Side Current Sensing technique (place a sensing resistor on the positive side, not ground compared to   46 traditional low side sensing) is used to sense both current signals. The advantage of high side current is that it will not cause any ground disturbance which exists in low side sensing due to the voltage drop on the sensing resistor. However, one big challenge of high side current sensing is the high common mode voltage on the positive terminal of sensing resistor because it is directly connected to DCV  . Therefore some specific current sensing amplifier ICs must be used for the high side current sensing. Figure 5.3 shows an example of typical high side current sensing circuit using MAX4173.  The basic operating principle for high side current sensing is: The voltages on the non-inverting and inverting ports of internal amplifier are the same: V V V I RSOURCE LOAD SENSE    . Thus we have:  1 1 LOAD SENSE RG G I R I R   (5.8) Then a current mirror is a current amplifier with gain β. The current IRGD  can be expressed as:  1 1 LOAD SENSE RGD RG G I R I I R     (5.9)  Voltage across the Resistor RGDR is:  1 1 LOAD SENSE OUT RGD RG RGD RGD CS LOAD G I R V R I R R A I R     (5.10) Therefore the current sensing circuit can be modeled as a Current-to-Current amplifier with a gain CSA and a Current-to-Voltage amplifier with a gain CS OUTG A R  . In practical design a small ceramic capacitor can be placed at the voltage output terminal in parallel with internal 1GDR to smooth the sensing signal and eliminate current ripple caused by high frequency switching. For example if a capacitor 0.1fC F is used for filtering and the transfer function of current sensing feedback can be expressed as:  1 1 1 1 1 0.0012 GD CS sc CS GD f R G R A sR C s      (5.11) The cutoff frequency of RC low pass filter is shown as below:   47  1 1 132 2 cutoff GD f f Hz R C    (5.12) which is significantly lower than switching frequency. Since the fuel cell current is not supposed to be adjusted very fast, a low pass filter in the current feedback which causes the slow response of total system is acceptable for this design. 5.2.2 Small-Signal Modeling of Average Input Current Mode Control VFC L Cout IL ZloadIFC S2 S1 200μF 10V Cin 200μF + - Vout 47μH IoutIin Vin + - Icin Icout Rbat Vbat RLoad D  Figure 5.4 Ideal synchronous buck converter circuit model The ideal circuit model for synchronous buck converter without the consideration of sonR  , and ESRR  is shown in Figure 5.4. An ideal continues time small-signal ac model can be derived based on this circuit. The state-space averaged models can be obtained as follows[14]:  ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ( ) ( ) ) L out in in FC in out out out L L in oad d i t L d t v v dt d v t C i t i t dt d v t v C i t t t dt Z t        (5.13) And input current of buck converter can be written as:  ( )( ) ( )in Li d t it t  (5.14) All the averaged terms enclosed angle brackets can be divided into steady-state plus perturbed quantities as below:   48  ˆ( ) ( ) ˆ( ) ( ) ˆ( ) ( ) ˆ( ) ( ) ˆ( ) ( ) L L L in in in in in in out out out i t I i t i t I i t v t V v t v t V v t d t D d t            (5.15) From Eq. (5.13), (5.14) and (5.15), the buck converter continuous-time small-signal models can be obtained with small-signal approximation:  ˆˆ ˆ ˆ( ) ( ) ( ) ( ) ˆ ˆˆ ( ) ( ) ( ) ˆˆ ˆ( ) ( ) ( ) ˆ ( )ˆˆ ( ) ( ) L in in out in FC in in L L out out L Load sLi t Dv t V d t v t sCv t i t i t i t Di t I d t v t sCv t i t Z                   (5.16) The transfer functions ( ( )idG s ) from fuel cell output current ˆ ( )FCi t  to duty cycle ˆ( )d t  is of interest here. Hence the perturbation of buck converter input voltage can be ignored and we will have  ˆ ˆ( ) ( ) 0FC inv t v t   (5.17) Substituting (5.17) into (5.16) then ˆ ( )ini t  can be expressed by equations only containing ˆ( )d t :  ( || ) ˆˆ ˆ( ) ( ) ( ) ( ) LOAD FC FC in L L C Z DV i t i t I d t Z      (5.18) Where ( || )LOADL C ZZ  is the impedance of L in series with output capacitor outC  parallel with LoadZ , which can be expressed as:  2 ( || ) 1 1LOAD Load out Load Load L C Z out Load out Load Z s LC Z sL Z Z sL sC Z sC Z          (5.19)  By substituting (5.19) into (5.18) , the i-to-d transfer function ( )idG s  is obtained as follows:  2 2 ( ) ( ) ( ) L out Load FC out Load L FC L Loadid out Load Load I LC Z s DV C Z I L s DV I Z G s LC Z s Ls Z         (5.20)   49 The steady state rated operating point ( _ 1.2FC FC ratedI I A  ) for buck converter ( LI , D, FCV ) can be estimated as follows:  _ 9.6 (@ 1.2 ) 7.3 76% 9.6 1.58 FC FC Bat Rated FC FC L V V I A V D V I I A D         (5.21) Inductance and output capacitor can be found in section 5.1: 47L H and 200outC F . The load impedance LoadZ depends on the types of loads and battery conditions. To put it simply, the load can be modeled as a pure resistance 4.7LoadR    so that when the output current is 1.58A as calculated in Eq. (5.21) the DC bus (or load voltage) is close to battery voltage 7.3V, which is close to the real system load. Based on the V-I characteristics of Li-Po battery shown in Figure 5.1, a rough estimation of battery internal resistance can be done: 0.39batR   . Thus the load impedance with and without battery can be obtained:  4.7, Load without battery 0.36, Load with battery       Load Load Load bat Load bar R Z R R R R  (5.22)  The load impedance with battery is of interest here since in the hybrid system the output of buck converter will be connected to Li-Po battery. From (5.21), (5.22) and (5.20), the final result for transfer function of input current-to-duty cycle is shown as below:  9 2 9 2 5 5.274 10 0.0006055 7.953 ( ) 3.384 10 4.7 10 0.36 id s s G s s s             (5.23)   50 5.2.3 FCCL PI Compensator Design Gpi(s)+ - Gid(s) Gcs(s) Gpwm Gad IFC_REF IFC_AD err d iFC T(s)  Figure 5.5 Block diagram of fuel cell current control loop  Figure 5.5 shows the complete block diagram for Fuel Cell Current Control Loop (FCCL). The PWMG is the transfer function of PWM regulator, which can be simply modeled as:  1 PWM M G V   (5.24) In analog system the PWM signal is generated by comparing the PWM control signal with the modulation waveform, for example triangle. Thus the gain of PWM generator can be written simply as one over the peak maximum value of PWM modulation waveform. For microprocessor-based digital control system, the modulation waveform is replaced by a continuous up-counting variables generated by the timer. Thus MV  here should be the maximum resolution of the counter. For this project considering the microcontroller speed and switching frequency, a 7-bit (0 - 128) PWM resolution is quite enough for accurate current regulation thus we will have:  7 1 1 2 1 127    PWMG  (5.25) ( )CSG s  represents the transfer function and gain caused by current sensing feedback circuit, as illustrated in section 5.2.1. The effect of A/D sampling in the feedback loop is modeled as a gain ( )ADG s , which can be expressed as:   51  N-bit _ 1 2AD ad ref G V    (5.26) where _ad refV  is the reference voltage for AD conversion and N-bit is the number of bits of AD conversion. For example if the ADC module in microcontroller has a 10-bit resolution ADC with a 2.5V voltage reference then  102 409.6 2.5 ADG    (5.27) Therefore the closed loop transfer function ( )T s  can be written as:  ( ) ( ) ( ) ( )id cs ad pi pwmT s G s G s G G s G  (5.28) The uncompensated closed loop transfer function (set ( ) 1piG s  ) ( )uT s  can be obtained:  8 2 12 3 8 2 1.699 10 0.001951 25.63 ( ) 4.061 10 5.978 10 0.000479 0.36 u s s T s s s s              (5.29) The Bode plot of ( )uT s  using MATLAB is shown in Figure 5.6. -80 -60 -40 -20 0 20 40 M a g n it u d e  ( d B ) System: T Frequency (rad/sec): 2.28e+004 Magnitude (dB): -0.0767 10 1 10 2 10 3 10 4 10 5 10 6 10 7 -180 -135 -90 -45 0 System: T Frequency (rad/sec): 2.28e+004 Phase (deg): -161 P h a se  ( d e g ) Bode Diagram Frequency  (rad/sec)  Figure 5.6 Bode plot for uncompensated closed loop ( )uT s   52 Without the PI compensator the phase margin is about 180° - 161° = 19°. To accomplish acceptable dynamic response and stability, a phase margin at least 45° is desired. The transfer function of typical PI compensator is:  ( ) (1 )Ipi pG s K s     (5.30) The pK  can be determined by the desired phase margin and / 2I If    must be significantly smaller than closed loop crossover frequency so that the phase margin will not be changed. The desired 45° phase margin occurs at around 410.6 10 rad/sec with magnitude 14.1dB so the PI compensator must decrease the closed loop magnitude by - 14.1dB:  14.1 0.2pK dB    (5.31) -80 -60 -40 -20 0 20 40 60 M a g n it u d e  ( d B ) System: T Frequency (rad/sec): 1.06e+004 Magnitude (dB): 0.19 10 2 10 3 10 4 10 5 10 6 10 7 -180 -150 -120 -90 System: T Frequency (rad/sec): 1.06e+004 Phase (deg): -141 P h a se  ( d e g ) Bode Diagram Frequency  (rad/sec)  Figure 5.7 Bode plot for compensated closed loop ( )cT s The new crossover frequency is about 41 10 rad/sec. The zero in the PI compensator must be put at the frequency at least 10 times lower than crossover frequency. Thus   53 1000I  rad/sec is selected. The transfer function of PI compensator can be obtained as follows:  1000 ( ) 0.2(1 )piG s s    (5.32) The compensated closed loop transfer function ( )cT s  is calculated:  9 3 4 2 12 4 8 3 4 2 3.587 10 4.128 10 5145 ( ) 4.286 10 5.997 10 5.03 10 0.38 c s s T s s s s s                  (5.33) The Bode diagram of ( )cT s  is shown in Figure 5.7. From the figure above the final phase margin is about 40° while the crossover frequency is about 41.06 10 rad/sec. 5.3 PSIM Simulation  Figure 5.8 PSIM simulation model for DMFC/battery hybrid system The modeling and control loop design illustrated in the previous section are all based on continuous-time ideal model.  The digital loop delays including A/D sampling/hold and computation delay and practical parameters of real components such as LR , sonR and current sensing resistors were not considered. Hence the hybrid control system is modeled (except battery charging Loop) and simulated in PSIM to validate the FCCL   54 control loop parameters, design BCCL and observe dynamic response as well as other detailed hybrid system behavior. The battery charging control, cold start procedure and BOP control are not included in this simulation and will be verified in further hardware tests. The PSIM model for the entire DMFC/Battery hybrid power system (Figure 5.8) consists of simple models for DMFC/Battery sources, a synchronous buck converter, current sensing feedback circuits and hybrid controller. 5.3.1 Models for DMFC and Battery Source  (a) Linear curve fitting for DMFC stack     (b) Linear curve fitting for Li-Po battery Figure 5.9 Linear curve fitting based on V-I test data. Usually the DMFC is running at its ohmic losses range so it can be simply represented by a voltage source plus equivalent internal resistance. The Li-Po battery can also be modeled as voltage source that represents OCV and _bat inR .  Based on the curve fitting data in Figure 5.9, DMFC and Li-Po battery models can be obtained: 12.24FCV V , 1.9FCinR   , _ 8bat ocvV V  and _ 0.397bat inR   . Note that the battery charging behavior, dynamics of DMFC and battery are not considered in this model since the steady-state performance at rated operating point is of interest here.   55 5.3.2 Current Sensing Blocks According to the section 5.2.1 the fuel cell current sensing feedback loop can be modeled in PSIM as shown as follows:  Figure 5.10 Fuel cell current sensing block Table 5.2 Design specification of current sensing feedback circuit  Fuel Cell Current Sensing  Battery Current Sensing Current Sensing Type High Side Unidirectional High Side Bidirectional Voltage Output (mV) 0 – 2500 0-2500 Zero Offset (mV) 0 500 Sensing Range  0-2500mA -1000 – 4000mA Sensing Resistor senR  0.02mΩ 0.05mΩ Output Resistor soutR  12kΩ 10kΩ Current gain iA  1/240 1/500 Total i to v sen i soutG R A R      1 0.5 The CS+ and CS- are connected to positive and negative terminal of current sensing resistor and voltage drop sensed across the resistor will be amplified by gain iA  and converted to current signal modeled by a voltage-Controlled Current Source.  In order to obtain a suitable current sensing range and acceptable voltage output for A/D sampling (2.5V Reference), the proper values of gain iA  and soutR  for fuel cell and battery current sensing feedback loops are presented in Table 5.2. Since bidirectional current sensing is needed for battery current measurement, an offset voltage (500mV) is superimposed to the current sensing signal so that battery is being   56 charged when then current sensing output is lower than 500mV and being discharged if higher than offset voltage. The bidirectional battery current sensing circuit is modeled as shown in Figure 5.11.  Figure 5.11 Bidirectional battery current sensing block 5.3.3 Hybrid Controller Block in PSIM PWM Fuel Cell Current Loop (FCCL) Battery Current Control Loop Charging Current Protection  Figure 5.12 PSIM hybrid controller block (BCCL + FCCL) The hybrid controller block contains the PWM duty cycle generator, inner fuel cell current loop and battery current control loop as well as battery charging current protection, as shown in Figure 5.12.  As mentioned before that the BCC (Battery Charging Control) is not included in the PSIM simulation model since the battery model used here is simply an ideal voltage source plus internal resistance thus battery charging   57 behavior is not considered and simulated in this model. Alternatively a simple constant maximum charging current reference (600mA) is applied to hybrid controller in the place of battery charging control. The battery current reference signal _BAT REFI  is passed to the BCCL and the error between reference and measured value is calculated and accumulated in I controller with integral gain iK . A/D sampling is modeled by a Zero-Order Hold block with sampling time 1/12500secadcT   plus the gain of A/D conversion 409.6adG  . Output of I controller is limited within the range between 0 to _ _FC REF MAXI . Meanwhile a Charging Current Protection (CCP) is added to the BCCL to obtain faster dynamic response when the charging current is too high in order to suppress the inrush high charging current, which may damage the battery. The CCP has a faster sampling rate thus the equivalent iK will increase proportionally and the responses of fuel cell reference to the changes of battery current will be faster. The CCP will be activated and added to the I controller only if the battery charging current is higher than maximum allowed value _BAT CMAXI . The output of BCCL _FC REFI will be sent to FCCL and required duty cycle of buck converter can be generated by a PI controller. The duty cycle signal is compared with a saw tooth modulation signal from continuous up-counting timer output and the output of comparator will be used as a driving signal for the synchronous buck converter. 5.4 Simulation Results 5.4.1 Dynamic Response of Fuel Cell Current Control Loop (FCCL) The PI parameters of FCCL in the PSIM model are set firstly based on the PI compensator results (Eq. (5.32)). The step response of FCCL is shown in Figure 5.13. It can be seen that the overshoot and settling time of the fuel cell current are acceptable. The fuel cell current stabilized after 6ms. A load disturbance is added to test the dynamic response of fuel cell current regulation under sudden load disturbance. At 104ms load current increased and battery terminal voltage decreased from 7.8V to 7.3V due to voltage drop on internal resistance caused by higher discharge current.   58 0.0 0.20 0.40 0.60 0.80 1.00 1.20 1.40 Fuel Cell Step Reference (A) Fuel Cell Current (A) 85.00 90.00 95.00 100.00 105.00 110.00 Time (ms) 7.00 8.00 9.00 10.00 11.00 12.00 13.00 VFC VBUS  Fuel Cell Current Fuel Cell Current Reference Fuel Cell Voltage Battery/DC Bus Voltage C u rr en t (A ) V o lt a g e (V )  Figure 5.13 Simulated step response of FCCL (Step: 1.2A Reference, Kp = 0.2,Ki = 500) 0.60 0.80 1.00 1.20 Fuel Cell Step Reference (A) Fuel Cell Current (A) 100.00 102.00 104.00 106.00 108.00 Time (ms) 7.00 8.00 9.00 10.00 11.00 VFC VBUS  Fuel Cell Current ReferenceFuel Cell Current C u rr en t (A ) V o lt a g e (V ) Battery/DC Bus VoltageFuel Cell Voltage  Figure 5.14 Simulated dynamic response of fuel cell current under load disturbance   59 5.4.2 Dynamic Response of Battery Current Control Loop (BCCL) Unlike the FCCL, the fast response of fuel cell current reference is not desired and frequent switching on and off of the DMFC will make the system efficiency low due to the slow dynamic response of the DMFC and its BOP. Once the fuel cell started, it is desirable that DMFC stack can output a smooth and average load power by neglecting transient load disturbance. On the other hand the DMFC current also needs to be controlled precisely to compensate the average load and charge the battery when necessary. Thus BCCL is very important for the hybrid system and the control parameters must be designed properly. Proportional control is not needed and a pure integral controller is used for BCCL to allow the fuel cell current follow the required battery current with a slow pace. The gain of I controller iK  will affect the rising time and dynamic response of fuel cell output with respect to battery current changes. The Figure 5.15 shows the step response of fuel cell current to a step battery current reference with different values of integral gain iK .  The battery current reference is set to -600mA (maximum charging current) while a 4.7Ω load is connected to the battery, which caused a battery discharge current about 1.55A. Thus a large error signal was sent to the Integral controller and fuel cell current will increase slowly in order to eliminate the error until it reaches its maximum value. Obviously the rising time for the fuel cell current increases with lower iK . Also one can consider this I control as a type of low pass filter so that the fuel cell current will only follow the average battery current and small load disturbance can be neglected. Generally a settling time around 5-10 seconds will be enough for DMFC to react. According to the waveform data shown in Figure 5.15, a gain iK  from 0.2 to 0.4 would give a reasonable dynamic response.   60 0 1 2 3 4 5 6 7 8 9 10 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 Time (s) C u r r e n t (A )   I bat  K i  = 0.2 I FC  K i  = 0.2 I bat  K i  = 0.4 I FC  K i  = 0.4 I bat  K i  = 0.8 I FC  K i  = 0.8  Figure 5.15 Simulated step response of battery current loop (Ki = 0.2, 0.4 and 0.8) 5.4.3 Charging Current Protection Mode As shown in previous section that BCCL has a very slow response to the battery current so that the DMFC can output average and smooth power to the load and battery will provide any transient load disturbance. However, according to the section 4.1.4, the battery charging current must be limited if a sudden load current drop occurs. Thus the function of Battery Charging Protection (CCP) is to prevent long term high inrush charging current to the battery. Once the system is entered into CCP mode, the fuel cell current reference will follow the battery current with a much faster dynamic response then BCCL. This can be done by using a much smaller sampling time for Integral controller (more frequent calling of I controller calculation). A discrete integrator can be expressed in z-domain transfer function by using Backward Euler:  s to z 1 1 1 s i id T K K s z    (5.34) For a discrete integrator changing sampling frequency or time interval of calculation will change the equivalent iK  in continuous time as well. For example in this hybrid system the BCCL samples the battery current by averaging every 1024 sampling data so the sampling frequency of BCCL will be only 1/1024 of A/D sampling frequency   61 (12.5KHz/1024=12.2Hz). For CCP mode the BCCL can sample the battery current after each A/D conversion so that the equivalent integrator gain  _ _ _ _ _ 1024 s bccl i fast i slow i slow s ccp T K K K T      (5.35) When the system detects that the battery charging current _ minbat batI I , the BCCL will run in CCP mode and the error between measured battery current and its reference can be minimized quickly. System will quit CCP mode if the battery current is higher or equal to preset lower limit. A test in PSIM simulation model is done to validate this design. The results (Figure 5.16) show that the battery current has been successfully clamped to -600mA within 30ms after the load is disconnected. It can be concluded that the simulation results accord with our design that in spite of slow response of BCCL the potential high battery charging current was suppressed by charging current protection. 7.6 7.8 8 8.2 8.4 8.6 -2 -1.5 -1 -0.5 0 0.5 1 1.5 Time (s) C u r r e n t (A )   I FC  I bat Battery Charging Protection Mode Battery current has clamped at -600mA (Max Charging) within 30ms  Figure 5.16 Simulation result of battery charging protection mode    62 5.4.4 Pulse Load Test More comprehensive tests can be done by adding periodic and pulse load to the system to verify the hybrid behavior. A pulse load profile sample as shown in Figure 5.17  is selected for the first test. The load period is 2.5 seconds and it comprises a 2.3A peak load current lasting for 0.375seconds (15% duty cycle) coming with a normal load current about 0.8A. It is implemented in PSIM by switching a second load resistor on and off. By observing the simulated fuel cell, battery and load current waveforms (Figure 5.18), it can be found that the fuel cell output current follows the battery current slowly and only supplies average load current while the battery takes care of transient peak load current. A more extreme case (Pulse Load Test 2) is simulated in PSIM to observe the dynamic battery charging current protection.  The load profile for test 2 is shown in Figure 5.19, in which the high load current (2.3-2.4A) lasts about 2.125 seconds (85%) and then all the loads are switched off so that the load current drops to zero. According to the simulation results in Figure 5.20, the CCP (charging current protection) is enabled after load is disconnected and pulls down the fuel cell current immediately to prevent extreme high charging current into the battery. In this test the fuel cell current shows a saw-tooth style waveform with slow increasing rate but fast dropping rate during CCP mode. 0 0.5 1 1.5 2 2.5 3 1 1.5 2 2.5 Time (s) C u r r e n t (A ) Pulse Load Period Time 2.5s 15% 0.375s 85% 2.125s 0.8A 2.3A  Figure 5.17 Load profile for test 1   63  0 5 10 15 20 25 30 35 40 45 50 -1 0 1 2 3 C u r r e n t (A )   I FC I BAT I LOAD 35 36 37 38 39 40 41 0 1 2 Zoom In Time (s) C u r r e n t (A )  Figure 5.18 Simulation results for Pulse Load Test 1 0 0.5 1 1.5 2 2.5 3 -0.5 0 0.5 1 1.5 2 2.5 Pulse Load Period Time 2.5s 85% 2.125s 15% 0.375s Load is disconnected  Figure 5.19 Load profile for Pulse Load Test 2    64 0 5 10 15 20 25 30 35 40 45 50 -1 0 1 2 3 C u r r e n t (A ) Pulse Load Test 2 (Dynamic battery charging current limitation)   32 32.5 33 33.5 34 34.5 35 35.5 36 -1 0 1 2 Time (s) C u r r e n t (A ) Zoom In   I FC I BAT CCP Battery charging current is clamped at 600mA (~1C) for battery protection  Figure 5.20 Simulation results for Pulse Load Test 2 (CCP mode)   65 6  Hardware Development and Testing A prototype of hybrid Energy Management System board based on the previous design is built to validate the proposed battery-current-based DMFC/Battery hybrid control strategy. The testing results will be used to compare with the analytical and simulation results in the previous sections. 6.1 System Specification The detailed system specification is listed in the table below: Table 6.1 Prototype system specification  Min Rating Max DMFC Stack Output Current (A) - 1.2 2 Voltage (V) 7.2 9.6V - Output Power (W) - 12 20 Li-Po Battery Charging Current (A) - 0.6 (1C) 0.6 Discharge Current (A) - 1.28 (2C) 5.12 (8C) Voltage (V) 5 7.4 8.4 Hybrid Energy Management System Current from FC (A) - 1.2 3 Power from FC(W) - 12 20 Current from Battery (A) -1 (charge) - 4 (discharge) Power From Battery(W) - - 40 Total Power Output (W) - 12 60 6.2 Hardware Components The diagram that presents the basic hardware components of hybrid EMS board is shown in Figure 6.1. The detailed circuit schematic can be found in Appendix A.   66 DMFC 10μF 100μF Q1 Q2 100μF D1 Cbus 10μF Li-Po Battery LOAD Rs_bat 0.05 Q4Q3 MAX4173 Current Sensor TPS2836 MOSFET Driver INA170 Current Sensor LT1120 LDO Voltage Regulator MSP430F2274 Microcontroller Board DC +5/+3.3V BOP (pump) Battery Protection Driving Circuit Rs_fc 0.02 L1 47μH Battery Discharge Enable Battery Charge Enable IFCVFC PWM Vbat Ibat Vref 2.5V  Figure 6.1 Diagram of DMFC/battery hybrid energy management system board 6.2.1 TPS3826-Based Digital-controlled Synchronous Buck Converter The DC/DC synchronous buck converter is implemented by two IRF7809 MOSFETs driven by Texas Instruments ®  Synchronous Buck MOSFET Driving IC TPS2836. TPS2836 is capable of providing dead time between two driving signals and high driving voltage to turn on the high side MOSFET.  Since the synchronous buck converter allows bidirectional current/power flow, which is not desired for DMFC system, a diode IN5811 is added between the output of Buck converter and load to protect the DMFC. However this diode has a voltage drop around 0.3-0.6V, which is a big concern for overall system efficiency. 6.2.2 +5VDC Auxiliary Power Supply (LDO) For the cold start of DMFC system, the circulating pump, air blower and hybrid controller board must be powered by battery before the DMFC is fully started up.  A LDO (Low Drop Out) 5V Voltage Regulator LT1120 from Linear Technology is chosen to supply power for controller board and BOP. It can output maximum 125mA current   67 and comes with an accurate 2.5V voltage reference which can be used for battery current sensing circuit. 6.2.3 DMFC/Battery Current Sensing Circuit As analyzed in section 5.2.1, Maxim ®  MAX4173FESA Voltage-Output High Side Current Sensing Amplifier is selected for the implementation of DMFC output current monitoring. MAX4173 has a voltage-to-voltage gain 50v vG    thus a 0.02Ω current sensing resistor is used to provide a total current sensing gain 1csG  . The fuel cell current sensing range is 0- 2.5A limited by the A/D conversion voltage reference 2.5V. A 0.1μF ceramic capacitor is added to the output terminal to smooth out the current ripples and noise. Due to the high output impedance ( 1 12GDR k ), a voltage buffer made by OPA340 is designed to interface between the MAX4173 output and MSP430 A/D conversion. Battery current monitoring is more complicated since bidirectional current sensing is required for both battery discharge and charging current sensing. Here TI ®  INA170 High- Side Bidirectional current shunt monitor IC is used as the bidirectional current sensing solution.  The basic circuit connection reference and output resistance selection is shown in Figure 6.2. INA170 is a current-output current sensing amplifier so the total output voltage to current gain can be calculated using the equation below[15]:  LOUT S S m L REF OS R V I R g R V R    (6.1) The first term S S m LI R g R is the voltage output without offset, in which SI stands for the sensing current, SR  is the sensing resistance and LR  is the output resistor chosen by user. The transconductance, or voltage-current amplifying gain mg , of the INA170 is 1000µA/V. INA170 requires an external voltage reference REFV  to provide initial voltage offset for bidirectional signal output.  The final offset voltage can be selected by offset resistor OSR . The A/D conversion voltage range is 0-2.5V and current sensing range is from -600mA (1C Maximum charging Current) to 4A. Thus a total current sensing gain 0.5 with an offset voltage 500mV is selected for INA170 current sensing circuit so that the current monitoring range is from -1A (0V) to 4A (2.5V). A current sensing resistor 0.05SR    is used thus the voltage gain of INA170 can be selected as 10. According to   68 (6.1), the output resistor LR  must be 10KΩ to obtain voltage gain of 10. The LT1120 LDO voltage regulator can output 2.5V voltage reference and based on Eq. (6.1) the offset resistor OSR  can be determined:  1 5 10 50 0.5 REF OS L V R R K K V       (6.2)  Figure 6.2 Basic circuit connection and output resistance selection[15] Again a filter capacitor 560fC pF is inserted in parallel with LR to implement a RC low pass filter for current sensing signal smoothing.  Voltage buffer circuit is used as well to interface with the MSP430 A/D conversion. 6.2.4 Battery Protection Circuit IRF7329 contains Two P-channel MOSFETs, which can be connected in a Back-to-Back configuration to protect the battery from over-current and over-discharge.  The advantage of P - channel MOSFET is that it can be driven to turn on by pulling the gate terminal to ground without the need of high driving voltage. The gate of MOSFET can be connected zero by turning on N-Channel digital transistor FDV301N using a CMOS logic level voltage (3.3V) from General I/O of MSP430.   69 The discharge/ charge can also be enabled or disabled separately if needed. However if one MOSFET is turned off, the battery current has to pass through its body diode, which may be an issue since the body diode in MOSFET can tolerate less current. For example , the datasheet of IRF7329[16] states that its internal reverse p-n junction body diode can only allow continuous 2A, almost 4 times lower than maximum drain current. 6.3 Experimental Results and Analysis A prototype PCB is manufactured to test the hardware design, as shown below:  Figure 6.3 Prototype circuit board of DMFC/battery hybrid EMS (Rev 1.1) For this stage of prototype testing, the BOP components are included and controlled by DMFC test station. A programmable electronic load on the test station can be zused to simulate a variety of different load profiles for hybrid system testing. The Figure 6.4 is the photo for the entire hybrid system testing platform including the 12W DMFC stack, 200cc 8v% Methanol Solution Tank, Circulating Pump, Li-Po battery pack and Hybrid controller board.   70  Figure 6.4 Testing platform for hybrid EMS board prototype 6.3.1 Waveforms The driving signal waveforms for high side and low side MOSFETs can be found in Figure 6.5. The dead time that protects the system from short circuit is observed.   71  -10 -5 0 5 10 15 20 -5 0 5 10 15 Time (S) V o lt a g e  ( V ) High Side Gate Signal Low Side Gate Signal Dead Time to prevent short circuit  Figure 6.5 Synchronous buck converter gate driving signal The voltage ripple is shown in Figure 6.6. The maximum peak-to-peak voltage ripple (50% duty cycle) is only 60mV, which is lower than required 1% voltage ripple. However, voltage spikes are observed when the MOSFETs are being switched on or off. -250 -200 -150 -100 -50 0 50 100 150 200 -250 -200 -150 -100 -50 0 50 100 150 200 Time (S) V o lt a g e  ( m V ) V out  DC = 8V V p-p  = 60mV < 1%V out  Figure 6.6 Output voltage AC ripple waveform   72 6.3.2 Power Efficiencies The power efficiencies are crucial for the overall system design. Compared to the linear voltage regulation, the switching power converter can archive higher level of power conversion efficiency. The total power efficiencies for this synchronous buck converter under different operating points (duty cycle and load) are measured and presented in Figure 6.7. 10 20 30 40 50 60 70 80 90 20 30 40 50 60 70 80 90 100 Duty Cycle (%) E ff ic ie n c y  ( % )   Efficiency (RLoad = 10Ohm) Efficiency (RLoad = 4.7Ohm) Maximum Efficiency =~93% At Duty Cycle 80%  Figure 6.7 Measured power efficiency under different duty cycles and loads Generally the synchronous buck converter shows a reasonable high efficiency up to 93% at 70-80% duty cycle. However, the output diode D1 contributes a large portion of total power loss due to its voltage drop ranging from 0.2-0.43V depending on the output current, as shown in Figure 6.8. The maximum power loss on diode can be up to 800mW with 1.5A output (15W power output).  Therefore the total efficiency can be improved further if the diode can be replaced by active controlled MOSFET (MOSFET-OR controller) in next phase of design.   73 10 20 30 40 50 60 70 80 90 0.2 0.3 0.4 0.5 Duty Cyle (%) D io d e  V o lt a g e  D r o p  ( V )   10 20 30 40 50 60 70 80 90 0 0.2 0.4 0.6 0.8 P o w e r  L o ss  ( W )   Voltage Drop on Diode (RLoad = 10ohm) Voltage Drop on Diode (RLoad = 4.7ohm) Power Loss on Diode (RLoad = 10ohm) Power Loss on Diode (RLoad = 4.7ohm)  Figure 6.8 Voltage drop and power losses on diode 6.3.3 Dynamic Responses A fuel cell current step reference can be programmed in MSP430 to test the dynamic response of FCCL. The PI compensator parameters are set according to the Eq. (5.32). The fuel cell current dynamic responses under step signals are shown in Figure 6.9. It can be observed that the fuel cell current follows the step reference successfully, although the rising time of real fuel cell current (about 10ms) is larger than previous simulation (about 5 ms), which is acceptable due to the difference between ideal models in simulation and practical hardware.   74 -0.25 -0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 0.25 -6 -4 -2 0 2 4 Time (S) V o lt a g e (V )   Measured Fuel Cell Current I FC I FC  Step Reference Signal  Figure 6.9 Measured step response of fuel cell current loop (Kp = 0.2, Ki = 500) 0 5 10 15 20 25 30 35 -0.5 0 0.5 1 1.5 2 2.5 Time (s) C u rr en t (A )   Battery Current FC Current10.8 seconds At t=16.3s the DMFC Current reaches its max. A step load is added at t = 5.5s, BCCL I controller started to accumulate and the DMFC Current output was adjusted by BCCL smoothly to minimize the battery current  Figure 6.10 Step response of BCCL (Ki_BCCL = 0.2) Again the step responses of BCCL are of interest here. According to the analysis in section 5.4.2, a maximum 10 seconds response time will be enough for DMFC System and will lead to I controller parameter _ 0.2i BCCLK  . A fixed battery reference is given   75 and the step response of BCCL is measured (Figure 6.10).  As can be seen that at 5.5sect  a load was added to the DC bus and battery current soared up immediately. Thus a large error signal between measured battery current and its reference will pass to the BCCL and its integral controller started to accumulate, which will increase the fuel cell reference smoothly until it reaches its maximum value (1.2A). After about 10.8 seconds the system reached its steady state. After that battery was still being discharged by load though DMFC was controlled to maximum power output to compensate for the load power. The final measured BCCL step response is very close to its simulation result (Figure 5.15), which proves the validity of system modeling and simulation. 6.3.4 Pulse Load Test Experimental tests using the two different load profiles as shown in Pulse Load Test 1 (Figure 5.17) and Pulse Load Test 2 (Figure 5.19) have been conducted. Figure 6.11 shows the measured current output using the Pulse Test 1. It can be seen that the fuel cell current increases slowly following the average value of battery current and reaches the steady state after about 20 seconds after start-up. During the steady state, the fuel cell supplies the average load current and battery is being discharged during peak load and charged when load current is low. In this test the battery current reference _BAT REFI  is set to zero and CCP mode is not triggered in this test due to the low battery charging current. Experimental results for Pulse Load Test 2 (Figure 6.12) prove the feasibility of CCP mode design. The CCP mode is triggered when load is disconnected and fuel cell current rapidly drops to prevent high inrush charging current. The test result shows that the battery charging current is higher than -600mA (1C). Generally the experimental tests show the feasibility and accuracy of hybrid control design stated in Chapter 4 .  The fuel cell current was actively controlled exactly as predicted in the system modeling and simulation (Figure 5.18Figure 5.20) and battery charging current is limited successfully.   76 0 5 10 15 20 25 30 35 40 45 50 -1 0 1 2 3 C u rr en t (A )   I FC I BAT 35 36 37 38 39 40 41 -0.5 0 0.5 1 1.5 Time (s) C u rr en t (A ) Zoom In  Figure 6.11 Experimental results for Pulse Load Test 1 0 1 2 3 4 5 6 7 8 -1 -0.5 0 0.5 1 1.5 Time (s) C u rr en t (A )   I FC I BAT Max. Charge Current: 1C (-600mA)The CCP mode is triggered when I BATCH >600mA In CCP the FC current dropsrapidly to protect the battery from high charging current  Figure 6.12 Experimental results for Pulse Load Test 2 (battery charging current limitation)   77 More tests with different types of load profiles have been extensively tested on this hybrid system to ensure that the hybrid controller could tolerate complicated load characteristics and provide accurate hybrid control and protection for both DMFC and battery. In the Figure 6.13 three stages of constant current loads are used to demonstrate the hybrid behavior between battery and DMFC. The DMFC system starts up at time (a) (ignore the cold start procedure here and initially the battery charging is enabled), battery voltage is less than 8.4V (not fully charged) thus the constant current charging mode is selected. The fuel cell output is adjusted to output a small amount of current for the battery constant charging. However, CC charging is interrupted by a 2A constant current load at time (b). But the control loop BCCL tries to maintain the same charging current for the battery so the fuel cell current increases until it reaches the maximum value. During the constant 2A load period (c), the fuel cell is trying to support its best to supply the load but it fails since the load current is higher than _FC MAXI . The remaining power is supplied by the battery. In the period (d) when the load jumps to 4A, the fuel cell current remain stable and this time battery provides all the load changes. During the last period (e), the load current reduces to 1A but the fuel cell current is not reduced because the battery charging current hasn’t reached the maximum limit. At the moment when the constant current load is switched, the load current is zero for a short time due to the existing problem of programmable electronic load system. The hybrid system entered into CCP mode during that time since the charging current is too large during the disturbance and some spikes can be observed. The detailed operation status of DMFC and battery is listed in table below: Table 6.2 DMFC and battery operation status under different loads Load  DMFC Output Battery Status No load (a) - (b) Small. Supply for battery charging Charging Medium 2A Load (c) Maximum 1.2A Discharging Peak 4A Load (d) Maximum 1.2A Discharging Low 1A Load (e) Maximum 1.2A Charging   78 50 100 150 200 250 300 350 400 -1 0 1 2 3 C u rr et  ( A )   0 50 100 150 200 250 300 350 400 6 8 10 12 14 16 Time (s) V o lt a g e (V )   v FC v BAT I FC I BAT I LOAD = 4AI LOAD =2A (c) (d) (a) (b) (e) I LOAD = 1A CCP during load swithcing  Figure 6.13 Constant current load test (CC 2A@1min, 4A@1min, 1A@1min) Another complicated and random load test has been conducted as well. The random load is implemented by a resistor bank which is switched on and off with different duty cycle and frequency for the simulation of complicated load profile. The results (Figure 6.14) show a good hybrid power management between DMFC and battery. The fuel cell output current (blue lines in upper figure) always supplies the average load current neglecting short and pulsating load disturbance while the battery provides all the load transients.   79 0 10 20 30 40 50 60 70 80 90 100 110 5 6 7 8 9 10 Battery (DC Bus) Voltage Time (s) B a tt er y  V o lt a g e (V ) 0 10 20 30 40 50 60 70 80 90 100 110 -1 0 1 2 FC/Battery Current C u rr en t (A ) Fuel Cell Output Current Battery Current  Figure 6.14 DMFC/battery output under random load test 6.3.5 No Load Test (CCP) A No Load Test that demonstrates the effect of CCP is shown in Figure 6.15. When the system starts up, the integral controller in BCCL is accumulated by using _i slowK  and fuel cell current follows the load very slow. However, once the system detects a sudden load disconnection or decrease (at about 20.5 seconds) the battery will be charged by a very large amount of current. If the battery charging current exceeds 1C, the CCP will be enabled and the integral controller in BCCL will be calculated by _i fastK . Thus the fuel cell current can be reduced to protect the battery and keep the battery charging current at 1C within several milliseconds.   80 8 10 12 14 16 18 20 22 24 -1 -0.5 0 0.5 1 1.5 Time (s) C u rr en t (A )   i FC i BAT Max. 600mA Charging System Started up System Reached Steady-state Load disconnected  CCP Mode  Figure 6.15 No load test (CCP Mode) 6.3.6 Battery CC-CV Charging 0 500 1000 1500 2000 2500 3000 3500 4000 -0.2 0 0.2 0.4 0.6 0.8 Time (s) 6 6.5 7 7.5 8 8.5 Battery Charging Current (-600mA) Battery Voltage Constant Current Charging (1C) V max =8.4V Constant Voltage (8.4V) Current (A) Voltage (V)  Figure 6.16 Experimental test results of constant current and constant voltage charging scheme The Figure 6.16 shows measured test results of CC-CV battery charging control algorithm, as illustrated in Figure 4.2. The battery was discharged to empty (6V) at first   81 and then charged using constant current charging (600mA, 1C) and lasts for about 50min until the battery pack voltage reaches 8.4V (4.2V per cell). In the Constant Voltage Charging Mode, the battery charging current was reduced to ensure that the battery voltage never exceeds maximum upper limit. The battery charging will be disabled once if the charging current that keeps the battery voltage at 8.4V reduces to preset charging stop limit (0.1C, 60mA). However, the load disturbance may affect the battery charging due to the voltage drop on internal resistance. Thus according to the battery charging control algorithm (Figure 4.2), once the battery is marked as fully charged, the battery charging will restart only if its voltage drops below battery charging start threshold _ 8BC STARTV V  so that frequent charging of battery can be avoided. To verify this, the battery is fully charged and then a small pulse load is used to simulate a load disturbance. The results (Figure 6.17) shows that despite of several times of voltage drops caused by load, the battery charging status is still full and DMFC never turns on, which avoided the frequent switching on of DMFC for battery charging. 0 0.5 1 1.5 2 2.5 3 3.5 4 -0.5 0 0.5 1 C u rr en t (A ) 0 0.5 1 1.5 2 2.5 3 3.5 4 8 8.1 8.2 8.3 8.4 V o lt a g e (V ) DMFC is always off Voltage drop caused by load  Figure 6.17  Battery charging control algorithm can avoid false triggering of battery charging caused by load disturbance   82 6.3.7 Hybrid System Performance As can be seen in Table 6.3, the overall DMFC hybrid system performance in terms of power and energy densities has been improved dramatically by incorporating with a lithium polymer battery. Assuming that the BOP scheme as shown in Figure 2.9 is selected for the practical design, the weight of DMFC system can be estimated based on the weight of stack plus 500cc 100% methanol fuel tank (the weight of pumps and small mixing chamber is ignored). Thus the total weight of this hybrid system is about 1.15kg (1.12kg for DMFC system and 30g for 2-cell 600mAh lithium polymer battery). The energy density of hybrid system is 372.6/kg, more than two times higher than pure lithium polymer battery system (154Wh/kg). The power output capacity of hybrid system has also been improved by battery. The power density of hybrid system is about 40W/kg, which is 4 times higher than pure DMFC System. Although it is still much lower than Li-Po battery power system, it provides a simple and immediate solution for power output improvement for DMFC system. Moreover the hybrid system power density can be further improved by using a larger battery pack or reducing the weight of stack and BOP. Table 6.3 Energy and power density comparison of DFMC, Li-Po battery and hybrid system   Total Weight (kg) Rated Power (W) Power Density (W/kg) Energy Density (Wh/kg) Pure DMFC System (Stack + Fuel Tank) 1.12 12 10.7 378.5 Pure Li-Po Battery 0.0306 29.2 967.3 154 DMFC/Li-Po Hybrid 1.15 43.8 38.1 372.6   83 7  Conclusions and Future Work 7.1 Conclusions The DMFC system has high energy density and can be recharged within seconds, which has become the potential alternative to the widely used lithium battery for consumer and industrial application. However, one big drawback of current DMFC technology is its low power output/density and slow start-up response. The ideal application for DMFC system is the environment where the load is low and constant. Due to the pulsating peak load profile for most of current battery-based electronic products, an extra battery is needed to provide cold start and boost up the system performance by supplying peak load power. This thesis presented a hybrid control design and prototype implementation of microcontroller-based DMFC/battery hybrid energy management system. First of all, different DMFC/Battery hybrid system configurations were reviewed and compared. According to the comparison in Table 3.2, it was concluded that among several possible options, the hybrid configurations with only fuel cell active output control with battery attached to the load with battery protection circuit (Figure 3.14) is the optimum selection for DMFC/Battery hybrid system. Then based on the simulation and experimental results and analysis, It can be proved that by combining a small lithium polymer battery with DMFC system for BOP start-up and high peak transient load compensation under proper active load sharing hybrid control, the overall hybrid system shows a significant improvement in terms of power density and dynamic response, as shown in Table 6.3. 7.2 Contributions The main contributions of this research project are listed as below:   84  A novel Battery-Current-Based Hybrid Control (BCBHC) is proposed to provide the optimum load sharing control between DMFC and battery power source. The DMFC will follow the load power while battery provides transient high peak load.  Battery protection is implemented by monitoring the battery current and active the fuel cell output control. DMFC overcurrent protection is inherently implemented in the hybrid system controller and DMFC operating voltage is simply limited to the battery voltage since the diode will automatically disconnect DMFC from the load if the voltage of the DMFC is lower than that of the battery.  A customized CC-CV battery charging algorithm is implemented in the hybrid EMS based on BCBHC. Under no load or constant load condition the lithium battery can be charged properly by DMFC to ensure that battery will not be undercharged next time.  Good compatibility to old battery-powered system. From the load side, the characteristics of this hybrid system are almost as same as previous pure battery power supply since the battery is directly attached to the DC bus without any power conditioning. Thus this DMFC hybrid system can be applied to any previous battery-powered system without caring about the characteristics of DMFC. 7.3 Future Work Several areas for future work were identified throughout this thesis, as listed below: 7.3.1 Overall System Efficiency and Performance Improvement The system efficiency and performance can be further improved by reducing the component size and optimizing DMFC BOP Control. The power loss of BOP can be further reduced by utilizing new architecture and dynamic control under different levels of load profile. The influence of cathode air and anode fuel flow rate on the DMFC performance is still not completely clear for this DMFC system. More tests have to be done to find out the optimum cathode air flow rate and minimum methanol solution flow for DMFC system under different load levels. As a matter of fact, the main purpose of high methanol fuel flow rate even for low power DMFC is to flush the CO2 generated   85 inside the anode flow channel since the actual methanol consumption rate for DMFC is much lower than current flow rate. Thus the size and power of circulating pump can be further decreased if the CO2 Removal can be improved by new stack design. 7.3.2 Better Temperature Control of DMFC Stack by Self-heating As one can see that for the prototype development the heating problem is not considered. According to the conclusion stated in section 2.2.2, the stack performance is very sensitive to the temperature changes. Thus a fast heating up and constant temperature distribution is desired for constant and stable performance. A heater may be used to warm up the stack for cold start of DMFC but its high power loss will make the final system inefficient. Another way to warm up the stack is the self-heating of stack. The self- heating is mainly caused by methanol crossover so the stack temperature will go up faster under high methanol concentration, which is good for the system start-up. On the other hand the high methanol crossover degrades the DMFC performance because of large portion of methanol crossover. Therefore an optimum operating methanol concentration can be determined by a series of tests. Under this concentration the DMFC system will be self-heated fast while maintaining a good power output. 7.3.3 Searching for New Hybrid System Scheme Fuel Cell Stack Battery B Li-ion Charging IC DC/DC Conv. Battery A iFC Load iBAT Hybrid Controller Bat. Discharge Selection Bat. Charge Selection Bat. charge control FC output Control  Figure 7.1 Dual battery hybrid system for better battery charging control   86 The battery life and hybrid system performance can be further improved if some new system schemes are used. For example as shown in Figure 7.1, a dual battery hybrid system can be used to guarantee that all the time only one battery is being discharged thus the battery charging on the other battery will not be interrupted. The hybrid controller can monitor the battery condition and switch the other battery to the load when the capacity of battery being used is too low. 7.3.4 Hybrid Optimization for Long Term Lifetime Improvement  The current research and development of hybrid control haven’t considered the long term lifetime issues of DMFC and battery. Thus some optimizations and more advanced control methods considering the BOP control, DMFC and battery lifetime need to be studied to further improve stability of DMFC power system.    87 References [1] J. D. Larminie, Andrew, "Fuel Cell Systems Explained (2nd Edition)," John Wiley & Sons. [2] B.-D. Lee, D.-H. Jung, and Y.-H. Ko, "Analysis of DMFC/battery hybrid power system for portable applications," Journal of Power Sources, vol. 131, pp. 207-212, 2004. [3] J. Han and E.-S. Park, "Direct methanol fuel-cell combined with a small back-up battery," Journal of Power Sources, vol. 112, pp. 477-483, 2002. [4] K. Lee, N. Chang, J. Zhuo, C. Chakrabarti, S. Kadri, and S. Vrudhula, "A fuel-cell- battery hybrid for portable embedded systems," ACM Trans. Des. Autom. Electron. Syst., vol. 13, pp. 1-34, 2008. [5] P. Thounthong, S. Rael, B. Davat, and I. Sadli, "A Control Strategy of Fuel Cell/Battery Hybrid Power Source for Electric Vehicle Applications," in Power Electronics Specialists Conference, 2006. PESC '06. 37th IEEE, 2006, pp. 1-7 %@ 0275-9306. [6] J. Zhenhua and R. A. Dougal, "A Compact Digitally Controlled Fuel Cell/Battery Hybrid Power Source," Industrial Electronics, IEEE Transactions on, vol. 53, pp. 1094-1104, 2006. [7] J. Zhenhua and R. A. Dougal, "A hybrid fuel cell power supply with rapid dynamic response and high peak-power capacity," in Applied Power Electronics Conference and Exposition, 2006. APEC '06. Twenty-First Annual IEEE, 2006, p. 6 pp. [8] J. Zhenhua, R. A. Dougal, and R. Leonard, "A novel digital power controller for fuel cell/ battery hybrid power sources," in Applied Power Electronics Conference and Exposition, 2005. APEC 2005. Twentieth Annual IEEE. vol. 1, 2005, pp. 467-473 Vol. 1. [9] J. Zhenhua, G. Lijun, and R. A. Dougal, "Adaptive Control Strategy for Active Power Sharing in Hybrid Fuel Cell/Battery Power Sources," Energy conversion, ieee transactions on, vol. 22, pp. 507-515, 2007.   88 [10] G. Aydinli, N. S. Sisworahardjo, and M. S. Alam, "Reliability and Sensitivity Analysis of Low Power Portable Direct Methanol Fuel Cell," in EUROCON, 2007. The International Conference on "Computer as a Tool", 2007, pp. 1457-1462. [11] S.-C. Yao, X. Tang, C.-C. Hsieh, Y. Alyousef, M. Vladimer, G. K. Fedder, and C. H. Amon, "Micro-electro-mechanical systems (MEMS)-based micro-scale direct methanol fuel cell development," Energy, vol. 31, pp. 636-649, 2006. [12] "Battery Charger Termination Issues with System Load Applied Across Battery While Charging," TI Application Notes. [13] "Datasheet: MAX4173 Low-Cost, SOT23, Voltage-Output, High-Side Current-Sense Amplifier," MAXIM, 2004. [14] R. W. Erickson, Fundamentals of Power Electronics, Second ed.: Springer Science+Business Media, LLC, 2001. [15] "Datasheet: INA170 High-Side, Bidirectional CURRENT SHUNT MONITOR," Texas Instruments. [16] "Datasheet: IRF7329 HEXFET(R) Power MOSFET," International Rectifier.    89 Appendices Appendix A. Schematics of Prototype Hybrid EMS Board    90 Appendix B. PCB Layout (Revision 1.1 02/17/2009) B.1 Top Layer and Silkscreen  B.2 Bottom Layer 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0067644/manifest

Comment

Related Items