<%BANNER%>

Modeling and Simulation of a Fuel Cell Reformer for Control Applications

Permanent Link: http://ufdc.ufl.edu/UFE0021843/00001

Material Information

Title: Modeling and Simulation of a Fuel Cell Reformer for Control Applications
Physical Description: 1 online resource (110 p.)
Language: english
Creator: Nath, Mohua
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: cell, fuel, methanol, modeling, numerical, reformer, simulation
Mechanical and Aerospace Engineering -- Dissertations, Academic -- UF
Genre: Mechanical Engineering thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: The limited success in the hydrogen storage and distribution technology has driven the need for the development of an effective fuel processor. The dynamic performance of a reformer is of critical importance for the successful commercialization of hydrogen as fuel for stationary and transportation applications. The reforming technology is of particular interest to utilities that require a clean and efficient method of generating electricity from fuel cells. As an effort to achieve a better control of the fuel processor parameters, a dynamic model of a generalized reformer is built. The model successfully predicts the transient temperature gradient across a reformer catalyst bed and the reformate exit temperature and is capable of predicting the response of the reformer to disturbances and load fluctuations. The reaction and heat transfer in the catalyst packed bed was analyzed numerically using a generalized physical model. These results provide valuable insight into the transient response of a reformer. The numerical output was compared with analytical results which agreed well with each other. This confirmed the validity of the numerical method. Also as a part of this research, the control logic of a methanol powered fuel cell bus was studied. The overall control scheme shows that the catalyst bed temperature plays an important role determining the fuel flow into the reformer. The successful prediction of reformer parameters can thus be utilized to eventually design a reformer capable of quick starts and faster transient response.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Mohua Nath.
Thesis: Thesis (M.S.)--University of Florida, 2007.
Local: Adviser: Lear, William E.
Local: Co-adviser: Crisalle, Oscar D.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2008-06-30

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021843:00001

Permanent Link: http://ufdc.ufl.edu/UFE0021843/00001

Material Information

Title: Modeling and Simulation of a Fuel Cell Reformer for Control Applications
Physical Description: 1 online resource (110 p.)
Language: english
Creator: Nath, Mohua
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: cell, fuel, methanol, modeling, numerical, reformer, simulation
Mechanical and Aerospace Engineering -- Dissertations, Academic -- UF
Genre: Mechanical Engineering thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: The limited success in the hydrogen storage and distribution technology has driven the need for the development of an effective fuel processor. The dynamic performance of a reformer is of critical importance for the successful commercialization of hydrogen as fuel for stationary and transportation applications. The reforming technology is of particular interest to utilities that require a clean and efficient method of generating electricity from fuel cells. As an effort to achieve a better control of the fuel processor parameters, a dynamic model of a generalized reformer is built. The model successfully predicts the transient temperature gradient across a reformer catalyst bed and the reformate exit temperature and is capable of predicting the response of the reformer to disturbances and load fluctuations. The reaction and heat transfer in the catalyst packed bed was analyzed numerically using a generalized physical model. These results provide valuable insight into the transient response of a reformer. The numerical output was compared with analytical results which agreed well with each other. This confirmed the validity of the numerical method. Also as a part of this research, the control logic of a methanol powered fuel cell bus was studied. The overall control scheme shows that the catalyst bed temperature plays an important role determining the fuel flow into the reformer. The successful prediction of reformer parameters can thus be utilized to eventually design a reformer capable of quick starts and faster transient response.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Mohua Nath.
Thesis: Thesis (M.S.)--University of Florida, 2007.
Local: Adviser: Lear, William E.
Local: Co-adviser: Crisalle, Oscar D.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2008-06-30

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021843:00001


This item has the following downloads:


Full Text

PAGE 1

MODELING AND SIMULATION OF A FUEL CELL REFORMER FOR CONTROL APPLICATIONS By MOHUA NATH A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2007 1

PAGE 2

2007 Mohua Nath 2

PAGE 3

To my parents for their love, sacrifice and steadfast support 3

PAGE 4

ACKNOWLEDGMENTS The completion of this work would not have been possible without the immense contribution of my committee, Dr. William Lear, Dr. Oscar Crisalle and Dr. James Fletcher. They have guided, advised and encouraged me with a lot of patience and supported me in every step during my masters program. I would like to take this opportunity to express my deepest gratitude to them. I would like to thank the most important people in my life, my parents and sister whose faith in me have brought me here. I would also like to mention my friends Alpana Agarwal, Jaya Das, Gaurav Malhotra, Nadeem Islam and Champak Das for being like a family to me and creating a home away from home. A special mention goes to Daniel Betts for sharing his knowledge of fuel cells with me and providing me with valuable suggestions. Last but not the least; I would like to thank Rana Dutta for motivating me to make this last leap possible. You are the light at the end of the tunnel. 4

PAGE 5

TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES...........................................................................................................................7 LIST OF FIGURES.........................................................................................................................8 ABSTRACT.....................................................................................................................................9 CHAPTER 1 INTRODUCTION..................................................................................................................11 1.1 Fuel Cells......................................................................................................................11 1.1.1 Phosphoric Acid Fuel Cells...............................................................................12 1.1.2 Polymer Electrolyte Fuel Cells (PEMFC).........................................................13 1.1.3 Alkaline Fuel Cells (AFC)................................................................................14 1.1.4 Molten Carbonate Fuel Cells (MCFC)..............................................................15 1.1.5 Solid Oxide Fuel Cells (SOFC)........................................................................15 1.2 Hydrocarbons as Indirect Fuel......................................................................................16 1.3 Fuel Reforming.............................................................................................................17 1.3.1 Autothermal Reforming (ATR).........................................................................17 1.3.2 Partial Oxidation Reforming.............................................................................17 1.3.3 Steam Reforming..............................................................................................18 2 BACKGROUND AND LITERATURE REVIEW................................................................19 2.1 Reformer.......................................................................................................................19 2.2 Literature review...........................................................................................................24 2.2.1 Langmuir Hinshelwood Model.........................................................................25 2.2.2 Nakagaki Correlation........................................................................................27 3 MODELING OF A PACKED BED REFORMER................................................................28 3.1 Overview.......................................................................................................................28 3.2 Background of Thermal Model.....................................................................................28 3.2.1 Methanol Steam Fuel Cell Reformer................................................................28 3.2.2 General Description of the Reformer................................................................29 3.2.3 Partial Differential Equation.............................................................................30 3.3 Finite Difference Method..............................................................................................30 3.3.1 Discretization....................................................................................................30 3.3.2 Initial and Boundary Conditions.......................................................................35 3.3.2.1 Dirichlets Boundary Conditions........................................................35 3.3.2.2 Neumanns Boundary Conditions.......................................................35 3.3.2.3 Initial conditions.................................................................................36 5

PAGE 6

3.3.2.4 Treatment of an undetermined boundary condition............................36 3.3.2.5 Special Case of Fictitious Nodes at Neumanns Boundary Conditions...........................................................................................37 3.4 Model Validation..........................................................................................................39 3.5 Non-Dimensional Analysis...........................................................................................42 4 RESULTS AND DISCUSSION.............................................................................................46 5 CONCLUSIONS....................................................................................................................60 APPENDIX A MATLAB CODES.................................................................................................................63 B TEST BED BUS-2 CONTROL LOGIC.................................................................................80 C DESCRIPTION OF CONTROL SCHEME...........................................................................91 LIST OF REFERENCES.............................................................................................................108 BIOGRAPHICAL SKETCH.......................................................................................................110 6

PAGE 7

LIST OF TABLES Table page 4.1 Constants used for finding solution to the reformer model. .............................................46 7

PAGE 8

LIST OF FIGURES Figure page 2.1 Generalized Fuel Cell Schematic.......................................................................................23 3.1 Generalized reformer schematic........................................................................................29 3-2 Discretization along length and radius of the reformer.....................................................32 3-3 Discretization along length and radius of the reformer and along time increments..........32 3.4 Fictitious nodes..................................................................................................................37 4-1 Temperature profile of reformer after 230 mins................................................................47 4-2 Temperature plots at mid-radius of reformer after 230 minutes........................................48 4-3 Temperature profile according to analytical solution........................................................49 4-4 Reformer temperature with zero boundary conditions (Numerical solution)....................50 4-5 Temperature profile of mid-radius according to Analytical Solution................................51 4-6 Transient analytical solution for axial temperature gradient.............................................52 4-7 Transient numerical solution for axial temperature gradient.............................................52 4-8 Analytical heat transfer solution for temperature as a function of axial position and time at radial distance = 0.75R...........................................................................................53 4-9 Numerical solution at different location along x-axis at r = 0.75 m. The temperature profiles are shown at different time instances....................................................................53 4-10 Non dimensional temperature profile using analytical method......................................54 4-11 Comparison of temperature profile of mid-radius according to Numerical Solution with and without heat generation.......................................................................................56 4-12 Temperature profile of reformer after 230 mins (with heat generation)............................57 4-13 Comparison of temperature profile of mid-radius according to numerical solution with and without convection..............................................................................................59 8

PAGE 9

Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Master of Science MODELING AND SIMULATION OF A FUEL CELL REFORMER FOR CONTROL APPLICATIONS By Mohua Nath December 2007 Chair: William Lear Cochair: Oscar Crisalle Major: Mechanical Engineering The limited success in the hydrogen storage and distribution technology has driven the need for the development of an effective fuel processor. The dynamic performance of a reformer is of critical importance for the successful commercialization of hydrogen as fuel for stationary and transportation applications. The reforming technology is of particular interest to utilities that require a clean and efficient method of generating electricity from fuel cells. As an effort to achieve a better control of the fuel processor parameters, a dynamic model of a generalized reformer is built. The model successfully predicts the transient temperature gradient across a reformer catalyst bed and the reformate exit temperature and is capable of predicting the response of the reformer to disturbances and load fluctuations. The reaction and heat transfer in the catalyst packed bed was analyzed numerically using a generalized physical model. These results provide valuable insight into the transient response of a reformer. The numerical output was compared with analytical results which agreed well with each other. This confirmed the validity of the numerical method. Also as a part of this research, the control logic of a methanol powered fuel cell bus was studied. The overall control scheme shows that the catalyst bed temperature plays an important 9

PAGE 10

role determining the fuel flow into the reformer. The successful prediction of reformer parameters can thus be utilized to eventually design a reformer capable of quick starts and faster transient response. 10

PAGE 11

CHAPTER 1 INTRODUCTION The growing world economy and the limited resources of nonrenewable fuels emphasize the need for aggressive development of alternative fuels. Advanced power generation technology utilizing alternative fuels can become a factor in reducing emission of greenhouse gases, improving urban air quality and reducing dependency on foreign oil. Hydrogen can be used in fuel cell which is a promising technology, providing efficient and reliable source of energy for a wide range of applications [1,6]. Although hydrogen is abundantly found in nature, extraction of hydrogen from its compounds remains a challenge before it can be commercially viable as a fuel. Hydrogen can be used for multiple applications, ranging from power generation to transportation applications, or internal combustion engines to fuels cells. Reforming hydrogen from any hydrocarbon carrying fuel such as natural gas, biomass, coal or ammonia provides an attractive solution for hydrogen storage for portable applications [1,2]. The particular case of reforming to produce hydrogen for use in fuel cells for transportation applications is selected here for specification of geometric and thermodynamic parameters to be used in this study. 1.1 Fuel Cells Fuel cells are electrochemical devices that produce direct current electrical energy from chemical energy. Fuel and an oxidant are continually supplied to the fuel cell for the reaction to take place [2]. The main components of a fuel cell are an electrolyte, catalyst and a porous anode and cathode. In the presence of a catalyst, the fuel, particularly hydrogen, splits into a proton and an electron. The electrons flow through an external circuit generating electricity and thereafter combines with protons and oxidants to form by-products at the cathode. Oxygen, acting as an 11

PAGE 12

oxidizing agent in this reaction, combines with the protons and electrons at the cathode to form a molecule of water. The overall reaction is as follows: OHOH22221 1-1 There are many types of fuels cells currently under investigation, including phosphoric acid fuel cells (PAFC), polymer electrolyte fuel cells (PEMFC), alkaline fuel cells (AFC), molten carbonate fuel cells (MCFC) and solid oxide fuel cells (SOFC). The classification of fuel cells is made based on the type of electrolyte. Among these, the PAFCs and PEMFCs hold the most potential for use as an alternative to internal combustion engines for transportation applications [1]. 1.1.1 Phosphoric Acid Fuel Cells Phosphoric acid fuel cells were the first type to be commercially investigated, other than for the U.S. space program. The electrolyte is phosphoric acid, usually contained in a silicon carbide matrix, and the electrodes made of Teflon -bonded platinum or porous polytetradluoroethylene (PTFE) bonded carbon, which is a polymeric binder used to hold the carbon black particles together [1, 7, 5]. In the presence of a catalyst, usually platinum, the positively charged hydrogen ions migrate towards the cathode. Electrons generated at the anode travel through an external circuit towards the cathode, thus creating electric current. The hydrogen ions and electrons combine to form water which is a by-product of this electrochemical process. The phosphoric acid fuel cells operate optimally between 150C to 220C, as at lower temperatures, carbon monoxide poisoning of the anode may occur. PAFC stacks need a heat sink during operation, usually a coolant, either liquid or gas, that passes through channels integral to the membrane electrode assembly [1, 2]. 12

PAGE 13

The chemical reactions are as below: 2Anode: 22HH e 221Catode: 222OHeH O 221Overall: 2 2 H OH OeO The PAFC stack operates at efficiencies between 37% and 40% and since heat is a by-product of the electrochemical process, the overall efficiency of a combined process can reach 80%. PAFCs can tolerate a 0.5% concentration of carbon monoxide and are minimally affected by the presence of carbon dioxide. 1.1.2 Polymer Electrolyte Fuel Cells (PEMFC) Proton Exchange Membrane Fuel Cells are considered most suitable for transportation applications. A PEMFC is based on a solid polymer membrane which may be a thin plastic film of sulphonated fluro-polymers that act as an electrolyte [1]. The two porous electrodes on either side of the membrane are made hydrophobic by coating with a compound like PTFE, thus helping the reactants to diffuse into a platinum layer that acts as a catalyst. The hydrogen ions diffuse through the membrane towards the cathode and the electrons flow through an external circuit thus producing electric current. Oxygen acts as an oxidizing agent and combines with the electrons and hydrogen ions and forms a by-product of water. The chemical reactions are as below: 2Anode: 244HH 22Catode: 442OHeH 222Overall: 22 H OH O PEMFCs have higher volumetric energy density than other types of fuel cells, thus making them compact and suitable for vehicular applications. The optimal operating temperature is around 80C which allows quick start-up and faster transient response. Recent developments 13

PAGE 14

however have elevated the operating temperature of PEMFCs beyond 150C to reduce the effect of CO poisoning, simplify water and thermal management and to recover high value heat. Other advantages are due to the fact that the electrolyte is solid therefore there is no spillage and corrosion thus contributing to its longer shelf life [5, 7]. One of the major disadvantages of the PEMFC is that the membrane is required to be continually hydrated to operate optimally, thus water management becomes a critical issue. PEMFCs use platinum as a catalyst, which has very low tolerance to CO poisoning. They can operate under a maximum of 10 ppm of CO, thus requiring a clean reformate gas to be used as a fuel. Also, increasing the fuel cell temperature beyond 100C can vaporize the water in the electrolyte which is essential for the conduction of ions, thus requiring tight control of fuel cell temperature and pressure [2, 7]. 1.1.3 Alkaline Fuel Cells (AFC) Alkaline fuel cells use a water-based electrolyte solution of potassium hydroxide (KOH) with a concentration that can vary according to the fuel cell operating temperature, which could be from 65C to 220C [2, 7]. The hydroxyl ion (OH-) acts as the charge carrier as in all other fuel cell types, However, the water formed at the anode travels towards the cathode and regenerates to hydroxyl ions. Therefore, in this type of fuel cell the by-product is only heat. The chemical reaction is shown below: Anode: 2 H2 + 4 OH4 H2O + 4 eCathode: O2 + 2 H2O + 4 e4 OH222Overall: 22 H OH O The main disadvantage of AFCs is that hydrogen and oxygen have to be supplied to the cell with negligible concentrations of CO2, CO or CH4 as it can poison the electrolyte. This requirement makes it difficult to be used for transportation applications. The advantage is that 14

PAGE 15

they operate on comparatively lower temperatures does not require noble metals and have very high efficiency. 1.1.4 Molten Carbonate Fuel Cells (MCFC) These cells use a mixture of molten salt carbonates as an electrolyte which is contained in a porous, inert matrix made of ceramic. The mixture is usually of varied percentages of lithium carbonate and potassium carbonate. These fuel cells normally operate at a temperature of around 650C. The high operating temperature indicates that these fuel cells can operate directly on gaseous hydrocarbon fuels [1, 2]. At the high temperature of 650C, the alkali melt and become conductive to carbonate ions (CO32-) which travel towards the anode. The ions flow from the cathode to the anode where they combine with hydrogen to produce water, carbon dioxide and electrons. Thus the by-products are carbon dioxide, water and heat. The chemical reactions are shown below: Anode: CO32+ H2 H2O + CO2 + 2 eCathode: CO2 + 1/2O2 + 2eCO32 Overall: H2 + 1/2O2 + CO2 H2O + CO2 The main advantage is that at the higher temperature, fuel reforming can take place inside the fuel cell stack itself, thus eliminating the need of an external reformer. The disadvantage of these type of fuel cells is the time required to obtain the high temperature which greatly slows the start-up process and makes the response sluggish [7]. 1.1.5 Solid Oxide Fuel Cells (SOFC) The Solid Oxide Fuel Cell (SOFC) operates under the highest temperature conditions, ranging from 600C to 1000C; thus it can operate with a number of different fuels. The electrolyte is a thin, solid ceramic material which conducts the charge carrying oxygen ions. The 15

PAGE 16

efficiency of the fuel cells are around 60% which is the highest among the fuel cell types. The by-product of steam can be utilized for various purposes. The chemical reaction is as follows: Anode: 2 H2 + O2 2 H2O + 4 eCathode: O2 + 4 e2 O2-Overall: 2 H2 + O2 2 H2O The main application of SOFCs is large-scale industrial systems where the demand is higher power, and long start-up times only minimally affect the performance or system requirements. Similar to MCFCs, the high temperature of the solid oxide fuel cells make them capable of operating on impure fuels and reforming occurs inside the fuel cell itself. SOFCs are being developed more than MCFCs because they have higher efficiency and ability to operate under higher temperatures [2, 7]. 1.2 Hydrocarbons as Indirect Fuel Hydrogen acts as an energy carrier; however it is not a source of energy. Thus there is a need for converting other sources of energy to hydrogen before it can be used directly in a fuel cell. Electrolysis of water is a well known method of hydrogen production, but it cannot be used for commercial production of hydrogen as it would be more economical to use the electricity directly as an energy source than to use it to produce an energy carrier in the form of hydrogen. The other method commonly used to generate hydrogen is by the reforming of hydrocarbons. Hydrocarbons in liquid form are easier to carry on-board and thus can be used for transportation applications. The most commonly-used process to produce hydrogen is the steam reforming reaction which is to react hydrocarbons with water at a high temperature. To improve the hydrogen-production efficiency and remove impurities, a water-gas shift reaction usually takes place after 16

PAGE 17

the steam reforming reaction. In this process, carbon monoxide produced from steam reforming reaction can be utilized to further break down water into hydrogen. 1.3 Fuel Reforming 1.3.1 Autothermal Reforming Among all the methods that produce hydrogen, autothermal reforming reaction is considered to be one of the most effective processes as it allows faster start-up and response time. ATR uses liquid hydrocarbons as fuel that undergoes a reaction with steam or air in a single reactor. While operating in ideal condition, with the optimal amount of air, fuel and steam, the reactions efficiency can reach up to 93.9%. The ATR process is also capable of using hydrocarbons such as gasoline and diesel, which can make it more commercially viable. One of the more recent developments is the possible reduction of operating temperatures from 1,200C to 650-900C by reducing oxygen to carbon ratios. The main advantage of reducing the operating temperature is that it allows for simpler reactor design, lowers cost in terms of material complexity and requires less fuel during startup conditions to pre-heat the reactor [8]. The basic reaction is as follows: 422122CHOHCO 4223CHHOHCO 1.3.2 Partial Oxidation Reforming Partial oxidation reaction is not as frequently used for commercial purposes as the other types. It is an exothermic reaction where a hydrocarbon reacts with controlled amount of oxygen. The advantages of this reaction are that it has a fast response time, high efficiency and does not require a catalyst. In addition, the byproduct of heat can be transferred via a heat exchanger for 17

PAGE 18

other applications. The disadvantages of this process are that it requires a high operating temperature and a high fuel /air ratio for the combustion. The overall reaction is as below: 422122CHOHCO 1.3.3 Steam Reforming The majority of hydrogen produced commercially is from steam reforming reaction. This reaction combines steam with hydrocarbon feedstock in a high temperature and pressure reactor. This is an endothermic reaction, thus requiring an external source of to maintain the temperature in the reactor. Generally, a nickel catalyst bed is used to speed up the reaction and increase the efficiency of the process. The advantages of this reaction are that it can achieve efficiency as high as 85% with heat recovery and can achieve the reaction efficiency of 80% [1]. The main drawbacks of this process are its plant size and slow startup. The overall reaction is as below: 4223CHHOHCO 18

PAGE 19

CHAPTER 2 BACKGROUND AND LITERATURE REVIEW 2.1 Reformer The development of a fuel reformer and its optimization may hasten the advent of widespread fuel cell deployment. The limited success so far, in the establishment of an infrastructure for hydrogen supply, can be overcome by the alternative solution of reforming hydrogen rich fuels on-board. Research on storage methods of hydrogen by physical or chemical means for fuel cell based vehicles has not yet provided a satisfactory practical solution, thus further supporting the need of reforming a hydrocarbon fuel such as methanol. The fuel used for reforming may vary according to the application. Whereas methanol, gasoline or ethanol may be a preferred fuel for transportation applications, natural gas or propane may have advantages for stationary applications. [12]. A reformer system typically consists of a premix tank, preheater, a reactor, a gas pre-treatment unit and a burner. A mixture of liquid hydrocarbon and water is fed from the premix tank to the preheater where it is vaporized and superheated before it is sent to the reactor. The energy for the endothermic reaction is supplied by the catalytic burner which heats up the reactor to a preset temperature. The reformed gas is treated in the gas pre-treatment unit to remove impurities like carbon monoxide and to further adapt it to meet the fuel cell requirements. The particular type of reforming taken as an example in this study is a catalytic steam process. It utilizes catalytic steam reforming to process the fuel mixture of methanol and water into a hydrogen rich gas. The power generation is accomplished by the oxidation of hydrogen in the fuel cell stack. The depleted fuel mixture is combusted in the steam reformer burner before is the products are finally liberated into the atmosphere. 19

PAGE 20

Several engineering issues hinder the commercialization of fuel cell-reformer systems for various applications. One of the most important issues is that the steam reformer dynamic response is considerably slower than that of the fuel cell stack. The response of the reformer is limited by heat transfer rates between the burner gases and the catalyst bed. The response can be defined as a corresponding increase in hydrogen flow rate corresponding to a similar step change in the load at the fuel cell stack. A quick change in heat transfer rates is required to meet the variation of power levels. In addition, the temperature of the catalyst bed should remain constant at design temperature conditions. The increase of catalyst bed temperature from its design conditions can cause the permanent degradation of the catalyst. Thus an optimized reformer design and robust control scheme for the corresponding reformer is to be developed for a quick and efficient dynamic response of the system. The present work is to provide a foundation for such a development. Tighter control of the reformer is required in order for the fuel cell performance to follow the load changing conditions of the vehicle. Control of the reformer has to be achieved by keeping some variables as close to the set point as possible, mostly the catalyst bed temperature, while changing the reformate flow in accordance with the demand at the fuel cell stack which is the load-following generation of electric power. This can be achieved by varying the temperature and flow of exit gases from the burner which is the source of heat for the reformer endothermic reaction. Figure 2.1 shows the general schematic of a fuel cell system with an on-board methanol reformer, where the overall output, in the form of electricity can be stored in batteries or can be used to drive the power-train of a vehicle. 20

PAGE 21

The example system considered here for the sake of this study is a 50kW, 30ft Fuel Cell Power Test Bed Bus, often referred as TBB, at the Fuel Cell Research Laboratory at the University of Florida [9]. A 175 cell PACF stack is used in the TBB. Air at slightly above ambient pressure is supplied to the cathode to provide oxygen. Hydrogen rich gas, i.e. reformate, is generated by an onboard fuel processor which converts methanol and water to H2, CO2 and CO. During normal operating conditions, the fuel cell stack is provided excess reformate to ensure sufficient hydrogen is available to react at the electrodes. Approximately 80% of the hydrogen in the reformate is consumed and the remaining 20% is supplied to the reformer burner providing heat to continue with the endothermic reforming reaction. For a pre-mix of methanol and water as reactants, the endothermic energy requirement is around 131 kJ/mole at 298C. Methanol is initially delivered by a pump to fuel the start-up burner, which in turn provides heat to bring the fuel cell stack and reformer to an initial operating temperature. Once the set-point temperature is reached in the reformer, the premix fuel is pumped into the reformer where the steam reforming reaction begins to take place. In this type of indirect methanol system, the fuel flow into the reformer is varied according to the load demands at the fuel cell stack output. In the first part of the reaction inside the reformer, the hydrocarbon, in this case methanol, undergoes a cracking reaction to decompose into carbon monoxide and hydrogen. This is an endothermic reaction where energy is continuously absorbed from the surroundings. As the reactants that enter the reformer are already pre-heated, there is an initial drop in the reactant temperatures, after which the heat required to maintain the reaction is provided solely by the hot gases burned at the catalytic burner. Carbon monoxide produced by the process of steam reforming may poison the noble metal catalysts in the fuel cell stacks. To reduce the concentration of carbon monoxide in the reformate, a water gas shift reaction is used to reduce 21

PAGE 22

the carbon monoxide concentration and additionally increase the hydrogen yield. The overall reaction is as follows: 232HCOOHCH (Cracking reaction) (2-1) 222HCOOHCO (Shift reaction) (2-2) This process produces hydrogen rich reformate which is sent directly to the anode of the fuel cell to be oxidized. The unused hydrogen-containing gas, called the flue gas, is returned via a feedback loop to a burner that burns the remaining hydrogen, providing the heat required for the endothermic reformer reaction. In case of increasing power demands from the fuel cell stack, the reformer must flow more reactants into the chamber and produce more hydrogen. This response is usually slow as the constraints are the convective heat transfer between the burner gases and reformer walls and conductive heat transfer from the walls to the catalyst inside the reformer [3]. Accompanied by an increase in fuel flow inside the reformer, and a subsequent demand for heat to the reformer walls, the fuel flow inside the burner is also increased. This helps to maintain the temperature inside the reformer, keeping the rate of reaction constant, as the pre-mix fuel is maintained at uniform concentration. Since these processes have a large lag time, the result may be low quality reformate being directed to the fuel cell during transient conditions. On the other hand, during ramp-down conditions of low power demand, the premix fuel flow is reduced to adjust with the low hydrogen requirement at the fuel cell stack. Subsequently, the fuel flow to the burner is also reduced as the hydrogen flue gas returning from the anode is adequate to provide heat to the reactor [9]. Due to the time delay in the reformer due to reaction kinetics, heat diffusion time and convection, the effect of the control action is not measurable for a period of time. Sometimes this 22

PAGE 23

long response time can exceed the quick-changing power demands at the fuel cell stack. This causes the feedback loop to be sluggish where feedback signal is crucial to automatic control of the reformer. The control action is thus inadequate since a change in the fuel flow will deliver the required hydrogen in the stack only after a certain time delay, thus already creating a hydrogen starvation at the stack or depleted hydrogen at the anode flue gas, again causing hydrogen starvation at the burner. Combustion Anode Flue Gas products Anode Reformate Air Air Cathode Reformer Steam Excess burner Reformer Heat exchanger plate Combustion products Vaporizer Heat Exchanger Neat Water methanol methanol Air Start-up burner premix air and water Figure 2.1. Generalized Fuel Cell Schematic Modeling is an essential tool to understand the component level interactions of the reformer with the fuel cell stack and its implications on the overall system performance. A reasonable representation of the transient response will enable future development of design and control architectures for the reformer. The reformer model is to predict the concentration of species in the reformate and the reformer catalyst bed. The hydrogen output with respect to time will determine the transient response and time delay of the reformer. The hydrogen flue gas returning to burner also determines the methanol fuel flow rate to the burner. 23

PAGE 24

To meet the needs of future control-oriented studies, another objective of modeling the reformer is to obtain a set of differential equations which will represent a state space equation. The state space equation will enable integration of the model into a simulation environment in order to numerically predict the behavior of the system under varying operating conditions. 2.2 Literature review According to Helms and Haley, a quick starting and fast transient response are the most important characteristics of a fuel cell power plant for transportation applications. By far the most popular reforming technology for on-board transportation application is catalytic steam reforming [13]. Geyer et al indicated that the methanol steam reforming technology is superior in its steam reforming technology. However, steam reforming transient response is slow in comparison to other components of a fuel cell system, thus limiting the overall effectivity of a plant in terms of dynamic response [17]. Thus it is crucial to increase the dynamic response of a fuel cell reformer. The first step towards improving the design and response characteristics of a reformer is to develop a model that can be utilized to study efficiency and thermodynamics of the system. The model should also be able to be used in conjunction with a controller design that can help achieve faster response to the system dynamics and disturbances. Kumar et al, Vanderborgh et al., and Geyer et al developed steady state models of the steam reformers, which do not predict the steam reformers transient behavior [16, 18]. Ohl et al developed a first principles dynamic model of a steam reformer for use in parametric design studies. The reformer was assumed to be a well mixed tank reactor. A series of rate reactions for each constituent of the reformate and fuel mixture consisted of the state variables of the system. The heat transfer equation consisted of the change in specific enthalpy due to the reaction kinetics within the reformer. These equations were then represented in a state 24

PAGE 25

equation form to describe a dynamic model of a steam reformer. Model results were then coupled with an optimization process to determine the design parameters of a steam reformer [8, 20, 21]. The study however, does not take into consideration the heat transfer between the walls of the reformer and any external heating agent such as burner hot gases or electric heaters. Convective heat transfer between the gases and the reformer wall temperatures are neglected. Ahmed et al developed a performance model of a reformer that predicted the performance and temperature of a reformate gas mixture. A complete conversion of fuel was assumed as there was a lack of chemical kinetics equation for the partial oxidation reaction for which the model was built. The result of the analysis showed that there existed a linear relation between the exit gas temperature and the inlet temperature of the fuel gas mixture. This model was however specific to only partial oxidation reaction and was not a general reformer model that could be applied for other types of fuel reforming [14]. Choi et al showed the results of kinetics of methanol decomposition along with methanol steam reforming and the water gas shift reaction. A non linear least squares optimization method was used to obtain expressions for rate of reactions [15]. Numerical analysis concluded separate rate of reactions for methanol steam reaction, water gas shift reaction and CO selective oxidation. The three reactors were then integrated and modeled in MATLAB. This study helped in observing the behavior of the reactor by changing its volume and temperature. Chois study however did not take into account the heat transfer in the reactor or the dynamic response of the reformer. 2.2.1 Langmuir Hinshelwood Model Ohl et al developed a dynamic model of the methanol reformer using the Langmuir Hinshelwood (LH) reaction rate for the methanol decomposition reaction. The use of this reaction rate is particularly advantageous because of the wide range of pressures that it covers. 25

PAGE 26

As the pressure increases the adsorption by the surface of catalyst bed and products slow down the reaction. This reaction rate takes all this into account. According to B. A Pepply, at constant temperature the rate of reaction increased with operating pressure although the final conversion decreased with pressure at thermodynamic equilibrium [17]. LH reaction kinetics is one of the simplest reaction mechanisms and it describes most catalytic surface reactions. The LH mechanism assumes that all reactants are adsorbed before the actual reaction can take place. Reactions occur between the adsorbed molecules following a fast diffusion process. The adsorbed molecules then are desorbed. The LH mechanism consists of many reaction steps taken together, ranging from 2 and may extend upto 30. Each step is assumed to be an elementary step, meaning that the reaction is supposed to occur exactly as it is written. Although the general success of the LH mechanism has been accepted, inaccuracies can apparently occur due to certain reactions that may be autocatalytic. Since the water gas shift reaction is itself an autocatalytic reaction, the appropriateness of the use of this rate equation for the water gas shift reaction is debatable. In addition to it, there are speculations that the water gas shift reaction may not occur in the presence of methanol on the catalyst bed. Since the LH mechanism assumes that one step is the rate limiting step while the others are at equilibrium, thus it reduces the rate of the overall reaction defined by the rate of the decomposition of methanol. Ohl expressed the methanol decomposition rate as follows: wwmmmmmcpbpbpbkmr1 (2-3) 26

PAGE 27

where r is the reaction rate for the decomposition of methanol, mc is the mass of the catalyst bed, is the effectiveness factor, km is the methanol decomposition rate, bm and bw are the methanol and water adsorption equilibrium constants respectively, pm and pw are the partial pressures of methanol and water, respectively. 2.2.2 Nakagaki Correlation Another way to express the rate of reaction has been developed by Nakagaki et al at Toshiba Power Systems. They carried out tests for evaluating the reaction rate of methanol decomposition with varying mass flow rates. Results showed that the diffusion resistance were more significant for lower mass fluxes thus the reaction rate varied at lower mass fluxes [11]. However it was found to be constant at higher mass fluxes, greater than 0.14kg/sm2 Also a correlation was found between pressure and the reaction rate. It was found that the reaction rate became lower at higher pressures, thus they expressed the rate of reaction as power law of the total pressure. Dependence of reaction rate to temperature was found to adhere to Arrheniuss law for temperatures, T<513 K. However for temperatures, T>513 K, conclusive results were not drawn. Based on these experimental observations, Nakagaki et al., derived the reaction rate of decomposition on Methanol, rm on Zn/Cu catalyst to be: (2-4) NRTEmlmOHCHeTPkr3/0)513/( where, ko = 1.35 X 106 mol / (gcat.s.atm) l = 0.13 E = 1 X 105 J/mol N = 1.3 M = -10 if T > 513 K else = 0 27

PAGE 28

CHAPTER 3 MODELING OF A PACKED BED REFORMER 3.1 Overview This chapter describes the basic principles behind the modeling of physical reformers. The theory governing the equations and the application of initial and boundary conditions are explored. In addition, the text explains the method of finite differences adopted to solve numerically the underlying partial differential equation. An independent analytical solution is presented to solve the heat transfer equation for a special case and thus validate the numerical model. The rate of reaction and the conversion of hydrocarbons to hydrogen depend on the temperature of the catalyst bed of the reformer. Consequently, the catalyst bed temperature and the outlet temperature of the product gases are the most critical process variables affecting the performance of the reformer [4]. The purpose of building the numerical model is to predict the temperature profile and reformate compositions as a function of time and space for a reactor of any cylindrical geometry under different operating conditions. This model should be useful for developing a controller that will manipulate the required process parameters of a methanol fuel-cell reformer or a biomass gasifier reactor to optimize their performance. 3.2 Background of Thermal Model 3.2.1 Methanol Steam Fuel Cell Reformer The conversion of a methanol and steam mixture to hydrogen and other product gases takes place inside the reformer. This reformer is usually a cylindrical packed-bed reactor with Cu/Zn catalyst pellets [10]. The fuel mixture is preheated in a superheater before being injected into the reformer with a feed pump. The heat for the endothermic reaction is supplied through the 28

PAGE 29

walls of the reformer by a catalytic burner. The rate of reaction or decomposition of methanol depends on the temperature of the reformer and the concentration of methanol. The motivation for developing a thermal model for a methanol fuel-cell reformer is to provide a tool for designing a controller which will maintain the reformer at an optimal operational temperature. The thermal model consists of a heat transfer equation with conduction and convection terms. 3.2.2 General Description of the Reformer Due to the similarity between methanol reformers and other hydrocarbon reformers and the common goal of developing a controller, a single model of a general reformer is proposed. A cylindrical packed-bed reformer where the reactions take place on the catalyst pellets is assumed [19]. Figure 3.1 shows the mass and energy flows, where r and x are respectively the radial and longitudinal axis of the reformer. The total length of the reformer in the axial direction is L and that in the radial direction is R. External heating is supplied through the walls of the reformer. The flow of gas mixture takes place in the axial direction and the radial mass flow is ignored. heat r x reactants reformate Figure 3.1. Generalized reformer schematic. 29

PAGE 30

3.2.3 Partial Differential Equation A cylindrical coordinate system is chosen in this problem to convert from three-dimensions to two-dimensions, as the heat conduction and convection is symmetrical about the r axis. The partial differential equation relevant to this model in the cylindrical coordinate system is x TCudVEd x Tk r Tk r T r ktTCggasgeeeee .2222 (3-2) where T = T(x,r,t) and where e is the effective mass density of the control volume, is the effective specific heat of the control volume, is the temperature inside the reformer, is the effective conductivity of the control volume, eC T ek r is the radial axis, is the axial axis, is the rate of heat generated in the reaction, Vis the total volume of the reformer, uis the velocity of gas, x .gEd gas is the density of the gas, and the void factor is the ratio of mass of gas to the total mass of solid inside the control volume[3]. For simplification of the numerical analysis, the convection term and heat generation term x TCudVEdggasg. gasgTuC x is initially omitted. 3.3 Finite Difference Method 3.3.1 Discretization The first order derivatives of a function f(x+h), in terms of the discrete differences can be expressed using Taylor series expansion as 30

PAGE 31

hffxfkk 1')( (3-4a) or, hffxfkk1')( (3-4b) Adding (3-4a) and (3-4b) yields a third possible approximation, namely kk f ffxh '()12 1 (3-4c) Equations (3-4a), (3-4b), and (3-4c) are finite difference approximations using backward, forward and central differentiation, respectively. Using the central difference scheme in the differential equation (3-2), the first order discretization can be written in the form TT (3-5a) Similarly (3-4a) second order discretization is of the form (3-5b) If the radius of the reformer is divided into M equal discrete elements of size r then represents a node on the radial axis, j r .Similarly if the length of the reformer is discretized into N elements of size x,en i represents a node on axial direction, x. Figure 3-2 shows the discretization of a cross-section of the upper half of the reformer. Only the upper half of the cylinder is considered as the heat transfer is symmetrical about the r-axis of the reformer. Figure 3-3 introduces another axis with time,t, as an independent variable. Though while simulating, discretization in time is not carried out, we will further proceed to make a dynamic model of the reformer with the help of SIMULINK which utilizes an explicit method to find the transients of th r rTjiji 1,, 21,,2,22 2TT rTrTjijiji 31

PAGE 32

the temperature profile every instant of time, denoted as thn nt where is a finite time interval. t r Figure 3-2. Discretization along length and radius of the reformer. Figure 3-3. Discretization along length and radius of the reformer and along time increments. x (0,M) i T j+1 x i,T j i-1, j T T i+1,j i,j-1 T r (0,j) (0,0) (i,0) (N,0) n n+1 x r Ti j +1 1 j i T Ti+1 j Ti j Ti j +1 32

PAGE 33

Thus the governing PD(3-2) can be discretized in theand x r E directions into N and M number of nodes and written in the ODE form. For simplicity of numerical analysis, initially the heat generation term has been excluded. 1,1,,1,2 ,ij ,12,,1,2222nnnnnnnnijijijijijijijijeeeeffeffdTTTTTTTTkCkkdtrrrx (3-6) The above general equation form can be written as a set of ODEs by invoking the value and and the resulting set can be re-arranged in terms of a banded pentadiagonal matrix of co-efficient, as follows: T Ni...3,2,1,0 Mj...3,2,1,0 NMrlrtrlrtrlrtdtdT0 NMNMNMTTTTlrtrlrtsrlrtsrlrtsrlrtsrsrlrtsrlrsrldtdTdtdtdT1431761000000 (3-7a) where, = T similarly, lrtsrlrsrlrdTdtdT0 TsrlrdTdt22 T1 TMNNNTNMNNTTTTTTTTTTTTTTT]...........,,,,..,,,,[],.......,,....,,[,2,2,32,22,11,1,31,21,11321 33

PAGE 34

TMNNNTNMNNdtdTdtdTdtdTdtdTdtdTdtdTdtdTdtdTdtdTdtdTdtdTdtdTdtdTdtdT],...,,,..,,,,[],.......,,....,,[,2,2,32,11,1,31,21,11321 The nodes representing i = 0 and j = 0 have not been included in this matrix because i=0 is the boundary region at the entrance of the reformer. Since at this point, the boundary condition is assumed to be constant always, which is equal to the inlet flow gas temperature, it will have a fixed numerical value. Storing N number of numerical values for every computational iteration will take up a lot of computer storage space, thus the nodes are not taken into consideration in this matrix, which was built solely for iteration purpose. All the nodes representing i=0 is represented by one value in the computer program and is used every time the boundary nodes are required to calculate the temperature of its adjacent node. Similarly for j=0 which represents the radius of the cylindrical reformer, it is assumed that there is a no flux condition across the radius of the reformer, these radial nodes will have the same value as its adjacent nodes. In order to save computer space by storing duplicate values, these nodes are not included in the matrix g the iterations. The treatment of these boundary conditions will be dealt with more detail in a later section. Let the N x M pentadiagonal matrix in equation (3-7) be denoted as P. Then, while doin d dtPT (3-8) where, T NMNMTTTT,........,121T This equation can be integrated to yield tiiiid nnn 10 0TPTTT (3-9) 34

PAGE 35

where T0T is a vector of initial conditions TNMNM121Thus a new vector of temperature values as a function of time can be found using various algorithms and iteration methods. SIMULINK has been used to solve the above equations usian explicit Runge-Kutta method. 3.3.2 Initial and Bound ,........,00000T ng ary Conditions lar case of interest, so as to define appropriate problems with unique solutions. The initial condition gives the specific temperature distribution in the system at time zero, and the boundary conditions specify erature or the heat flow at the boundaries of the medium. 3.3.2.1 Dirichlets Boundary Conditions ons are fixed boundary conditions on the reformer wall and the inlet of the reformer, where the temperature is held constant by electric heaters supplying heat into the system through the walls or by the inflow of pre-heated gas and. This is expressed atical = at r = R TTTT Boundary conditions and initial conditions are prescribed for a particu the temp The Dirichlets Boundary Conditi mathemly as follows: Txrt(,,) wallT x L0 and = at x = 0, and3.3.2.2 Neumanns Boundary Conditions. Neumann Boundary Conditions are natural boundary conditions in the outlet where there is no heat flux across the radius because of symmetry and the exit of the reformer is assumed to be insulated so that zero heat flux can be assumed in that plane. This is expressed mathematically as follows: t0 Txrt(,,) inT rR0 t0 35

PAGE 36

rTxrt(,,) = 0, at r = 0, x L0 and t0 Txrt(,,) = 0, at x = L, where L is the length of the reformer, x rR 0 and t0 where and rxTxrtTxrtrx TT ghout the reformer. This is expressed mathematically as follows: 0TxrtxLR where is andary condition (,,)(,,) 3.3.2.3 Initial conditions The reformer is heated to a critical temperature by the start-up burner. The initial condition is assumed to be uniform throu (,,), a t 0, 0initTtr initT constant value. 3.3.2.4 Treatment of an undetermined bou Let us consider the governing equation (3-2) again after neglecting the convection term, namely .22 22 g eeeeffeffCkktrrrxdV (3-10) For the first-or dEkTTTT der term in r the Neumannsun bodary conditions at r = 0 do not apply here because of a term 0/0 that becomes undefined. Applying LHospitals rule yields rrTrrrTr 1 (3-11) where the right hand side can be written as, 22r TTr (3-12) rrrReplacing this term in (3-10) yields, 22222 g eeCk effeffdETTTktrxdV (3-13) 36

PAGE 37

Furthermore, where r=0, the term g dE dVWhile bu is neglected. ilding a numerical simulation, this special case is accounted for separately to obtain the correct temperatures of the nodes at the axis of the reformer. .3.2.5 Special Case of Fictitious Nodes at Neumanns Boundarormer requires temperatures of a fictitious node, Since there is zero heat flux across the exit nodes of the reformer, we make use of this to assign a value to the fictitious nodes. 3y Conditions While implementing central difference scheme, the temperature of a node depends on the temperatures of its adjoining nodes. Thus the nodes at the exit edge of the ref nNjT,1 Figure 3.4. Fictitious nodes The no-flux condition is used to find the value of the fictitious nodes. First the x-direction derivative is written as N jNjT TT x x ,,11 (3-14) which can be solved for to yield, NjT,1 nNjT,1 nNjT ,1 nNjT,1 nNjT ,1 37

PAGE 38

NjNj TTxT x ,,112 (3-15) Now at the zero-flux condition, T x 0 so it follows that 1,1,NjNjTT (3-16) Hence the temperature at fictitious node at ),1(jN can be replaced by the temperature at. A similar relationship can be developed at the other edge having fictitious nodes Finally after implementing all the boundary conditions, the governing pentadiagonal n the form, ),1(jNwhich lie at 0r matrix 3-6 can be re-written i MNMNMTlrdt)1(1)1()1)(100 ldTT2102 rlr NTs1 lrrlr ltldTdt72002 MNNNTTTTTtrlrttrlrtrtstsltltsltssslslslsdtdTdTdtdTdtdTdtdtd4321)1((1600200200200000200200020000 (3-7b) 38

PAGE 39

In the above equation, the fictitious node N+1 and its multiples, which represent the exit end of the reformer has been included in the pentadiagonal matrix. Its value is calculated exactly as the value of N-1 node and its multiples. The nodes along the radius of the reformer form the undetermined boundary conditions. Here, at the value of j=0, the equation (3-13) is implemented. Thus all along the radius, the node temperatures have different co-efficients than that of the rest of the discretized reformer. As shown above, in the initial first N nodes of the reformer this boundary condition is imposed. This equation represents the governing matrix that was used to numerically model the generalized refiormer. 3.4 Model Validation To validate the accuracy of the numerical results, a comparison is made by solving the partial differential equation (3-2) using an analytical method. The physical model is simplified by neglecting the convection term to yield, 222211TTTrrrxt T (3-17) T= initTT and where initT is the initial temperature. where, rR0, x L0and t0, Equat ion (3-17) is subject to the boundary conditions, 0T, at 0rRxL (3-18a) 0T at 0, 0 x rR (3-18b) 0, at ,0T x Lr R x The problem (3-17) can be solved analytically by applying boundary conditions (3-18a) to (3-18c) to yield TxrtCRrx(,,)()( (3-18c) xe())22011 (3-19) mptmpmpmp 39

PAGE 40

where the value of constant ishe principle of orthogonality. Thus mpC determined by using t multiplying both sides by drrrRmR)(00 and dxxxLp0),( yields FdrdxxXrrRNNCpm RL1 rxpmmp)()()()(000 (3-20) Substituting (3-20) into (3-17) yields mp t RLmpeTxrtRrXxrFRrXxdrdx()(,,)()()()()220 (3-21) Now, it is necessary to find the eigenfunctions mpmpmprxNN()()01100 )(0rRm the norm)(mN and the e igenvalu em From Table 3-1, Case 3 in reference [ondition in question 10], it follows that for the boundary c )(0rRm )(0rJm = (3-22a) and )(2)(1202RJRNmm (3-22b) where m are the positive roots of 0)(0 RJm and J is the Bessel function of order 0. To find the eigenfunctions (xXp 0 ) the norm ) (pN and the eigenvalue p we refer to Table 2-2, Case 8 in [10] to find that ()sinpPXxx (3-23a) and LNp2)(1 (3-23b) 40

PAGE 41

where p are the positive roots of cos(Lp ) = 0. Substituting (3-22a) (3-23b) into (3-21) yields mptRLmpmpmpmprxeTxrtJrsinxrFJrsinxdrdNN x()()()()()()()22001100 (3-24) Separating the integrals produces the equation (,,) mptRLmpmpmpmprxeTxrtFJrsinxrJrdrsinxdxNN()(,,)()()()()()()22001100 (3-25) Integrating analytically the rightmost integral, and using the expressions (3-22a), (3-22b), yields (3-23a) and (3-23b) mptRmpmmpm pr e1TxrtFJrsinxrJrdrLRJR4 ()'(,,)()()()()220022110 (3-26) 0 mptRmpmmpmpre1TxrtFJrsinxrJrdrLRJR ()'(,,)()()()()22002211004 (3-27) where from [1], )(1)(10rrJdrrrJmm (3-28) Substituting (3-28) into (3-27), the generalized final analytical solution reduces to, mptmpm mpmmpLRJR()2111 4FeJrsinxJRTxrt()()()()(,,)2201 (3-29) 41

PAGE 42

3.5 Non-Dimensional Analysis A non-dimensional an alysis is presented here to determine the heat transfer process in the reformer. The formulation of the problem in terms of non-dimensional terms helps in studying the physics of the process and should provide a useful tool for data analysis and comparison. Consider the general partial differential equation 222211TTTtrrrx T In order to non-d imensionalize along radius and axial direction, let Rrr/ (3-30a) L x-30b) and x (3 0TTwall 0TTT (3-30c) Subst ituting these new variables into the governing equation yields t TTTxxxTxTTrrrTRrTTrTTTTwallwallwall wall 02 (3-31) or )()()())((00002 t TT x T L r T R r r T R 22222 022111 (3-32) aking 2 L tt T equation (3-32) can be expressed as t TTT xrrRLrTRL 222221 (3-33) 2 42

PAGE 43

Using the method of separation of variables )(),(),,(txrtxrT (3-34) Equation (3-32) can be represented as 22222)()(111tdtdtxrrzrz (3-35) where, .Then, taking the separated equations 2 )/(RLz 0)()(2ttdtd (3-36a) and 02222 x r r z r z (3-36b) xXr Again invoking separation of variables ),(Rxr )()( quation (3-35) becomes e 01 12zRdz 2222xdXdXrddRrrdR (3-37a) R and X Separating in 2221 x dX(3-37b) nd Xd a 222221rvdrdRrzrdRdzR (3-37c) ads to two equations that can be solved individually to yie leld 0222X x d Xd (,):sincosXxxandx (3-38a) 43

PAGE 44

0020202R r ddR r z r dRdz (3-38b) 0)()(2ttdtd (3-38c) Note: as equation has no dependence, it follows that, v=0 Thus the solutions of (3-38,a-c) taken as, XrRepmt ),(x ),,(,02 e expressed as XrRCtxrT(),(),,(011 can b tpex2), (3-39) mmpmp Thus the final solution to equation (3-32) is rRrxXrRNNFetxrTmrxpmmppmt ,(),(),()()(),,(101000112Now to find the eigenfunctions from reference [1], rdxdxXp ),() (3-40) )()(00rJrRmm (3-41a) )( 21 )(JRNm 202Rm (3-41b) here ,m are the positive roots of 0),(0RJm w )(sin)(xxXpp (3-42a) LNp2)(1 (3-42b) where, p are the positive roots of 0)(cos Lp wh ere R and L are assumed to be 1. Substituting equations (3-41a) to (3-42b) in (3-40), 44

PAGE 45

rdxdxrJrxrJRJLRFetxr pmrxpmmpmt ),(sin),(),(sin),()(4),,(101000112022ion in non-dimensional form, T (3-43) Solving the remaining integrals gives the final solut ,()(4),,(011212mmppmmtJJFetxrT )(),(sin)1mpJxr (3-44) 45

PAGE 46

CHAPTER 4 RESULTS AND DISCUSSION This chapter summarizes the computational results obtained by solving the physical models of the reformer numerically and partially validating the results with analytical solution. The steady state solution of the numerical model equation is solved using Matlab. The transient solution is obtained by using SIMULINK, which uses the explicit method to find solution in the next time instant. A major purpose of this study was to be able to develop a model that would correctly predict the reformer behavior in terms of temperature at any particular location. This temperature would then act as an input to a controller that would determine the flow of premix fuel into the system. The validity of the numerical model is established as will be shown in the subsequent sections. For the sake of the study the value of the reformer constants were taken same as the values of a methanol reformer located at UC-Davis. The following table shows the values that were estimated to match the UC-Davis reformer parameters. Table 4.1 Constants used for finding solution to the reformer model. (Source: Table 4.1, Daniel Betts, 2005) Name Value Used How it was determined Catalyst Bed Heat Capacity, Ce 900 J/kg-K Inferred Catalyst Bed Thermal Conductivity, ke 5.00 W/m-K Estimated from Data Catalyst Bed Density, e 1983 kg/m3 Measured The catalyst heat capacity was assumed to be equal to an average of the heat capacities of copper and zinc. The void factor used was 50% The catalyst density was measured via a water displacement method for a single pellet 46

PAGE 47

Figure 4-1 shows the numerical solution of (3-7) to find the temperature profile for a simple case when the reformer boundaries are kept at constant temperature. The upper boundary or the cylindrical wall of the reformer is kept at 580 K whereas the left side or the face of the reformer is kept at 520 K. Length of reformer [m] 5 10 15 20 25 5 1 0 1 5 2 0 2 5 530 535 540 545 550 555 560 565 570 Radius of reformer [m] Figure 4-1 Temperature profile of reformer after 230 mins 0.5 Radius of reformer, r [m] 0 Length of reformer, x [m] 0.5 The temperature at different location inside the reformer is shown with color variation. The initial temperature of the reformer was 520 K. As the time increases the temperature inside the reformer gradually increases as can be seen from the plot. The profile shown above is only for one half cross section of the reformer. Now if this transient formulation is run for longer period of time, it gives a steady state solution. Figure 4-2 shows such time histories for different location along the center-line of the reformer. Figure 4-2 A shows the temperature plot with respect to time at the inlet of the reformer. It shows constant temperature at the beginning and then it rises and attains a steady state temperature of 567 K whereas for figure 4-2 B and C the temperature rises from the beginning 47

PAGE 48

(A) 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2x 104 523 523.5 524 524.5 525 525.5 526 526.5 Time [s]Temperature [K] (B) 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2x 104 520 525 530 535 540 545 550 555 560 Time [s]Temperature [K] (C) 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2x 104 520 525 530 535 540 545 550 555 560 Time [s]Temperature [K] Figure 4-2 Temperature plots at mid-radius of reformer after 230 minutes at (A) Inlet (B) Center (C) Exit. 48

PAGE 49

Plotting equation (4-3) in Matlab, we get a temperature profile at r = 0.75, and at different location of distance (x) axially along the reformer. Figure 4-3 Temperature profile according to analytical solution Tem p erature T [ K ] Time, t [s] Since for the analytical solution we used the Dirichlets boundary conditions as 0K, a numerical simulation with similar boundary conditions was carried out by shifting the wall and inlet gas temperature conditions to 0. The initial reformer temperature is kept at 520 K. As seen from the plot, as the axial length increases the temperature approaches towards the steady state of 0 K at much slower rate. Figure 4-4 shows the temperature distribution plotted after 230 seconds with zero boundary conditions. Figure 4-5 shows the temperature profile along the center line of the reformer for the analytical solution. The temperature profile for the inlet shows constant temperature at the beginning before gradually falling and reaching a steady state temperature of 0 K. 49

PAGE 50

R] eformer length [m Reformer Radius [m] 5 10 15 20 25 5 10 15 20 25 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Radius of reformer, r [m] 0.5 Figure 4-4 Reformer temperature with zero boundary conditions (Numerical solution) 0 Length of reformer, x [m] 0.5 50

PAGE 51

A) 0 5000 10000 15000 0 100 200 300 400 500 600 Time [s]Temperature [K] (B) 0 5000 10000 15000 0 100 200 300 400 500 600 Time [s]Temperature [K] (C) 0 5000 10000 15000 0 100 200 300 400 500 600 Time [s]Temperature [K] Figure 4-5 Temperature profile of mid-radius according to Analytical Solution at (A) inlet (B) center (C) exit Figure 4-6 shows the plot of temperature at r = 0.75 and different locations along the axial center of the reformer with time. This plot represents the analytical solution obtained directly from equation 3-21 whereas figure 3-11 shows the same temperature plot at similar location along the axial direction for finite difference solution. The plots match closely with each other when plotted together. Here they are plotted differently for brevity. 51

PAGE 52

Figure 4-6 Transient analytical solution for axial temperature gradient 0 2 4 6 8 10 12 14x 104 0 100 200 300 400 500 600 t [s]T [k] x=0.1 x=0.25 x=0.50 x=0.75 x=0.90 x increasing Figure 4-7 Transient numerical solution for axial temperature gradient. Figure 4-8 shows the temperature plot at different axial location along the reformer when solved analytically. As the axial distance increases, the temperature flattens out. 0 2 4 6 8 10 12 600 14 x 10 4 0 100 200 300 400 500 x = 0.1 x = 0.25 x = 0.5 x = 0.75 Temperature, T [K] Time, t [s] 52

PAGE 53

Figure 4-8 Analytical heat transfer solution for temperature as a function of axial position and time at radial distance = 0.75R Figure 4-9 Numerical solution at different location along x-axis at r = 0.75 m. The temperature profiles are shown at different time instances Temperature, T [K] Length of reformer, x [m] Temperature, T [K] Length of reformer, x [m] 53

PAGE 54

Figure 4-10 shows the results of the non-dimensional analysis using the analytical method. The x-axis represents the time 2 L tt the y-axis represents the scaled reformer temperature. Figure 4-10 Non dimensional temperature profile using analytical method Temperature, T [K] Length of reformer, x [m] So far the results were obtained without taking into consideration the heat generation term and convection term. However, it is important to include the heat of reaction in the current analysis as the reforming reaction is endothermic and apart from the heat transfer between the fluid and reformer wall, energy is consumed to maintain the reaction. The rate at which methanol is consumed is calculated using the Nakagaki rate of reaction as is discussed in detail in Chapter 2. The following results show the comparison between the reformer bed temperatures by addition of the heat generation term. .gEd 54

PAGE 55

A) (B) Figure 4-11 Comparison of temperature profile of mid-radius according to Numerical Solution with and without heat generation at (A) exit (B) center (C) inlet 55

PAGE 56

(C) Figure 4-11. Continued Figure 4-12 shows the temperature profile of the reformer with the heat generation term. The upper boundary or the cylindrical wall of the reformer is kept at 580 K whereas the left side or the face of the reformer is kept at 520 K. As shown in Figure 4-11 and Figure 4-12, it can be seen that due to the endothermic reaction the reformer reaches a steady state at a lower temperature. Thus it can be concluded that the reformer exhibits temperature fluctuations with the change in methanol flow rate. Since the rate of reaction is directly proportional to the number of moles of methanol consumed, any increase of flow of pre-mix fuel will cause a substantial drop in reformer catalyst bed temperature. Thus there is an immediate requirement to ramp up the heat supply into the reformer in terms of burner hot gases. This condition will be especially noticeable in a steep increase in the load demand at the fuel cell stack. 56

PAGE 57

Figure 4-12 Temperature profile of reformer after 230 mins (with heat generation) Radius of reformer, r [m] 0.5 0 Length of reformer, x [m] 0.5 Betts indicated that the convection term was small compared to the conduction in the heat transfer equation. The authors conclusion was backed by obtaining experimental data from UC-Davis reformer which showed that the convection was below 0.4% of the value of the conduction term. These results are expected as the axial temperature gradient is much lesser than the radial temperature gradient and most of the flow occurs in the axial direction [4]. The current analysis includes the convection term and as can be seen from Figure 4-12, the comparison shows negligible contribution to temperature gradient by the convective heat transfer. 57

PAGE 58

A) (B) Figure 4-13 Comparison of temperature profile of mid-radius according to numerical solution with and without convection (A) exit (B) center (C) inlet 58

PAGE 59

(C) Figure 4-13. Continued 59

PAGE 60

CHAPTER 5 CONCLUSIONS The methodology and techniques used for dynamic modeling of a reformer are presented. The model is further simulated numerically to predict the transient and spatial temperature distribution of the reformer and the composition of the reformate. Steady state steam reformer models have been developed in the past, but most of them are applicable to particular reformer geometry and parameters. The reformer model developed in this research can be used to study the transient behavior of a reformer of any geometry and size. The purpose of this research was to develop a model that could be used in conjunction with a control model to ultimately design a faster responding fuel processing system. The most desired characteristic of a reformer is a quick transient response to load fluctuations. It should be able to ramp up or down the reformate flow according to load demands while maintaining an economic use of primary fuel. The numerical model built in this study helps in understanding the reformer parameters that play an important role in deciding the reaction rate and subsequently the species concentration of reformate. This model in conjunction with a controller model should help build a controller design that will lead to a quick and efficient dynamic response of a reformer. A sensitivity of the various design parameters and their reasonable representation obtained on the basis of the simulation will help in selecting the desired level of control capabilities of various controllers and in evaluating alternative designs of the reformer. From this study, the following conclusions may be drawn: a. At typical boundary conditions and initial conditions of a steam reformer, the model predicts that steady state is reached after approximately 65 minutes. b. Possible location of sensors preferably at the mid-radius of the reformer. At boundaries, temperatures dont predict accurate reactions rate fluctuations. Aim 60

PAGE 61

is to have lesser number of sensors and less data analysis for the control. Preferably have three sensors at x=0.1L, 0.5L and 0.9L and r=0.75R. Controller input to be average of the three sensors. c. The temperature near the inlet reaches steady state faster than near the exit. However, the steady state temperature reached near the inlet is lower (closer to that of reformate gas temperature boundary condition) than that near the exit. Reactions occurring at the exit are lower as most of the reforming has already taken place near the inlet and heat required for the endothermic reaction is lower at the exit. Thus for higher efficiency and for optimal utilization of premix fuel, it is preferred that the aspect ratio of the reformer closer to 1. d. The effect of pre-heating the fuel and reformer From the results, it can be concluded that for obtaining faster start-up times, it is advisable to preheat both the premix fuel and reformer catalyst bed. e. The convective heat transfer is negligible compared to conductive heat transfer and can be eliminated in further related studies. f. Non-linear behavior associated with packed bed steam reformers may be noticed during ramp up conditions of the fuel cell stack load conditions due to the endothermic reaction. Since the reformer heat transfer rates are considerably slower than the rest of the system, an immediate increase in premix fuel rate may not produce desired amount of hydrogen in the reformate gas due to a drop in the reformer bed temperature. To avoid this condition of hydrogen starving at the fuel cell stack, designing a feed forward controller is suggested, which will increase the fuel flow into the burner as soon as ramp up condition occurs at the fuel cell stack. Controlling the burner fuel flow with the reformer catalyst bed temperature as the input to the controller will result in sluggish response and incomplete conversion of premix fuel. Further work 1. Input disturbance into the system by making the reformer wall temperature as a Neumann B.C instead of Dirichlets B.C. 2. Gather substantial experimental data to completely verify the numerical model built here. 3. Design a model-based controller The input variables to the controller can be the reformer catalyst bed temperature, reformate gas temperature and reformate 61

PAGE 62

composition. The manipulated control variables can be the methanol premix flow and the air flow into the burner. 62

PAGE 63

APPENDIX A MATLAB CODES Reformer Input File % reformer_inputfile.m % Mohua Nath 04/23/2006 % Revision 2 % This program specifies all the reformer parameters % This program is called by the m-file temp_cal1 %========================================================== % Catalyst bed Gas mixture effective parameters %========================================================== ro_eff = 1983 ; % effective mass density of catalyst gas mixture in [kg/m3] C_eff = 900 ; % effective specific heat of catalyst gas mixture in [J/kgK] K_eff = 5.00 ; % effective conductivity of catalyst gas mixture [W/mK] %========================================================== % Reformer Dimensions (Cylindrical) %========================================================== Radius = 0.5 ; % Radius of reformer [m] Length = 0.5 ; % Length of reformer [m] %========================================================== %Discretization Parameters %========================================================== M = 30 ; N = 30 ; n = (N-2)*(M-2) ; del_r = Radius / N ; % Size of finite element along Radius del_x = Length / M ; % Size of finite element along Length %========================================================== % Co-efficients of PDE and ODE %========================================================== alpha = K_eff / ( ro_eff C_eff ) ; % co-efficient to have Tdot on LHS for i = 1 : M-2 p(i) = alpha / ( del_r i* del_r ) ; % Co-efficient of del r term of ODE end q = alpha / ( del_r del_r ) ; % Co-efficients of del r2 term of ODE 63

PAGE 64

r = alpha / ( del_x del_x ) ; % Co-efficients of del x2 term of ODE %=========================================================== % Nakagaki reaction rate constants %=========================================================== k0=1.35e6 ; % reaction rate constant [mol/(g_cat.s.atm)] P=1 ; l=0.13 ; E=1e5 ; % [J/mol] nn=1.3 ; R=8.3143 ; % Universal gas constant %=========================================================== % Ohl's reaction rate constants %=========================================================== pm=10 ; % partial pressure of methanol pw=20 ; % partial pressure of water cm=30 ; % partial pressure of methanol %mc=40 ; % partial pressure of methanol %=========================================================== % Methanol properties %=========================================================== Er=201e3 ; % heat of reaction per mole of methanol consumed %=========================================================== for i = 1 : M-2 dia1(i) = -(p(i) +2*q+2*r) ; dia2(i) = (p(i) +q) ; end dia11 = -2*(del_r*p(1)+q)-2*r ; dia22 = del_r*p(1)+q ; %=========================================================% 64

PAGE 65

Ordinary Differential Equation (Including Heat Generation Term) % matrix_reformer3.m % Mohua Nath 04/23/2006 % Revision 2 % This function executes the S-function that solves the N set of ODE's % Invoked by the SIMULINK program matrix_sim_reformer2.sim %=========================================================== % : t= time ; x= vector of stateS ; u= vector of I/Ps ,... % flag= from simulink : sys = to simulink, x0 = I.Cs %=========================================================== function[sys, x0] = matrix_reformer3(t,x,u,flag) %=========================================================== % Call the inputfile for parameters %=========================================================== reformer_inputfile2 ; %=========================================================== % Flag 0 : Sends sizes and initial state vector to Simulink %=========================================================== if flag == 0 size_states = n ; % continuous time states size_disc_stats = 0 ; % number of discrete time states size_outputs = n ; % number of outputs to view size_inputs = 0 ; % number of inputs size_disc_roots = 0 ; % always 0 for now size_feedthrough = 0 ; % algebraic feed required? (no) sys = [size_states, size_disc_stats, size_outputs, size_inputs, size_disc_roots, size_feedthrough] ; % column vector of sizes from... % m file to simulink at flag=0 x0 = [523*ones(n,1)] ; % column vector of initial conditions %=========================================================== % Flag 1 : Brings T (x = state vector) from Simulink and Sends Tdot.. %(sys = state derivative) to Simulink %=========================================================== 65

PAGE 66

elseif abs(flag) == 1 %_________________________________________________________________ % Call the reaction_rate2 file to calculate rate of reaction %-------------------------------------------------------------------------------------------------[r1] = reaction_rate2(x,0); % send 'T' for Nakagaki rate, and n %_________________________________________________________________ % Calculate the heat of reaction at every node %-------------------------------------------------------------------------------------------------n =(N-2)*(M-2) ; E(n) = 0 ; % initialise E reaction at all nodes as 0 for j = 1:n mc = 2 pi r1(j) del_r del_x ; %mass of control volume E(j) = mc*Er*r1(j)/(ro_eff*C_eff) ; % energy consumed at all node end %_________________________________________________________________ % Imposing of B.C's at reformer wall = 400 C = 673 K %------------------------------------------------------------------------------------------------wall = 573 ; % create a vector of wall boundary nodes %________________________________________________________________ % Imposing of B.C's at reformer input --> 100 C = 373 K %-----------------------------------------------------------------------------------------------inlet1 = 523 ; % create a vector of inlet boundary nodes %________________________________________________________________ % Impose the Nuemann condition at the exit of the reformer %------------------------------------------------------------------------------------------------st = 1 ; for i = N-1 : (N-2)*2 mid_rad(st) = x(i) ; st = st + 1 ; end j = 0 ; for i = 1 : M-2 cc = (N-2)*i 1 ; 66

PAGE 67

j = j+1 ; exit1(j) = x(cc) ; end %_________________________________________________________________ % calculate the Tdot...i.e temperature gradient at every node %------------------------------------------------------------------------------------------------Tdot(1) = dia11*x(1) + r*x(2) + dia22*x(N-1) + dia22*mid_rad(1) + ... r*inlet1 E(1) ; % Tdot for 1st row %-----------------------------------------------------------------------------------------------for j = 2:N-2 Tdot(j) = r*x(j-1) + dia11*x(j) + r*x(j+1) + dia22*x(N-2+j) + dia22*mid_rad(j) E(j) ; %Tdot for 2nd four rows if j == N-2 Tdot(j) = r*x(j-1) + dia11*x(j) + dia22*x(N-2+j) + dia22*mid_rad(j) + ... r*exit1(1) E(j) ; else end end %----------------------------------------------------------------------------------------------d = 1 ; for j = (N-1):(N-2)*(M-3) Tdot(j) = (q*x(j-N+2) + r*x(j-1) + dia1(d)*x(j) + r*x(j+1))+... dia2(d)*x(j+N-2) E(j) ; % Tdot for middle rows for f = 1 : M-3 if j == ((N-1)*f)-(f-1) d = d+1 ; Tdot(j) = q*x(j-N+2) + dia1(d)*x(j) + r*x(j+1) + ... dia2(d)*x(j+N-2)+r*inlet1 E(j) ; elseif j == (f+1)*(N-2) Tdot(j) = q*x(j-N+2) + r*x(j-1) + dia1(d)*x(j) + dia2(d)*x(j+N-2) + ... r*exit1(f+1) E(j) ; end end end %------------------------------------------------------------------------------------------el=60; for j = (N-2)*(M-3)+1:(N-2)*(M-2) 67

PAGE 68

if j == (N-1)*d d = d + 1 ; else end if j < (N-2)*(M-2) Tdot(j) = q*x(j-N+2) + r*x(j-1) + dia1(d)*x(j) + r*x(j+1) + ... q*wall + p(d)*wall E(j) ; % Tdot for last 5 rows else end if j == (N-2)*(M-3)+1 Tdot(j)= q*x(j-N+2) + dia1(d)*x(j) + r*x(j+1) + q*wall+p(d)*wall+r*inlet1 E(j) ; elseif j == (N-2)*(M-2) Tdot(j) = q*x(j-N+2) + r*x(j-1) + dia1(d)*x(j) + q*wall+p(d)*wall + ... r*exit1(M-2) E(j) ; end end sys = Tdot(1:n)' ; %========================================================== % Flag 3 : Choose the outputs from the state vector 'T' (or x) %========================================================== elseif flag == 3 for j = 1:n y(j) = x(j) ; % We choose the entire T at all nodes end sys = y' ; % send column vector of outputs %========================================================== % Any other Flag : undefined....so return an empty vector %========================================================== else % All other flags are irrelevant to the problem sys = [] ; end %%======================================================%% 68

PAGE 69

Driver File % driver_reformer2.m % Mohua Nath 04/23/2006 % Revision 4 % This function executes the Simulink function... % 'matrix_sim reformer2' and calls the plot function %====================================================== reformer_inputfile2 % call inputfile to initialize parameters %====================================================== para = p ; par = [q,r,del_r,N,M,n,dia1,dia2,dia11,dia22, Er,ro_eff,C_eff] ; % pass parameters to matrix_reformer3 %====================================================== sim('matrix_sim_reformer') %plots_reformer_temp2(y) ; %====================================================== Banded Matrix Creation File % temp_cal2.m % Mohua Nath 04/23/2006 % Revision 2 % This function creates a banded matrix co-efficient of the discretized ODE %=================================================================== % : N = no. of nodes in x, M = no. of nodes in r : C = Banded matrix %=================================================================== function[c] = temp_cal2(N,M) n = (N-2) (M-2) ; % gives the length of the T vector %=================================================================== % Call the inputfile for parameters %=================================================================== reformer_inputfile2 % call reformer_inputfile for parameters %=================================================================== % Build first row of c matrix %=================================================================== 69

PAGE 70

s = N-1 ; c = [] ; % initiate the matrix as null matrix first_row = zeros(1,n) ; % populate the 1st row with zeros first_row(1) = -(p+2*q+2*r) ; first_row(2) = r ; first_row(s) = p+q ; c = [c;first_row] ; % update c matrix 1 row %=================================================================== % Build 2nd four row of c matrix %=================================================================== u = N-2 ; for j = 2:u %populate the 2nd row to(N-2)th row second_row = zeros(1,n) ; % populate the rows with zeros second_row(j-1) = r ; second_row(j) = -(p+2*q+2*r) ; second_row(j+1) = r ; second_row(j+u) = p+q ; c = [c;second_row] ; % update c matrix 2-5 rows end %=================================================================== % Build middle rows of c matrix %=================================================================== s = N-2 ; u = (N-2)*(M-3) ; v = N-1 ; d = 2-N ; for j = v:u %populate the middle rows mid_row = zeros(1,n) ; %populate the middle rows with zeros mid_row(j+d) = q ; mid_row(j-1) = r ; mid_row(j) = -(p+2*q+2*r) ; mid_row(j+1) = r ; mid_row(j+s) = p+q ; temp_row = mid_row ; c = [c;temp_row] ; % update c matrix 6 -> n-6 rows end %=================================================================== % Build 2nd last 5 rows of c matrix %=================================================================== 70

PAGE 71

for j = ((N-2)*(M-3))+1:((N-2)*(M-2))-1 %populate the 2nd last rows last_ros = zeros(1,n) ; %populate the 2nd last rows with zeros last_ros(j-N+2)= q ; last_ros(j-1) = r ; last_ros(j) = -(p+2*q+2*r) ; last_ros(j+1) = r ; temp_row = last_ros ; c = [c;temp_row] ; % update c matrix 2nd last rows end %=================================================================== % Build the last row of c matrix %=================================================================== last_ro = zeros(1,n) ; % populate the last row with zeros last_ro(((N-2)*(M-2))-N+2) = q ; last_ro(((N-2)*(M-2))-1) = r ; last_ro((N-2)*(M-2)) = -(p+2*q+2*r) ; temp_row = last_ro ; c = [c;temp_row] ; % update c matrix last row %%===============================================================%% Calculation Of Rate Of Reaction % reaction_rate2.m % Mohua Nath 04/23/2006 % Revision 2 % This program calculates the rate of reaction according... % to Nakagaki or Ohl's model % This function is called by matrix_reformer2.m %============================================================== % : brings in T vector, flag chosen as Nakagaki, n... % : Rate of reaction %=============================================================== function [r1] = reaction_rate2(x,flag) %=============================================================== % Call the inputfile for parameters %=============================================================== reformer_inputfile2; 71

PAGE 72

%=============================================================== % FLAG 0: Calculate rate of reaction according to Nakagaki %=============================================================== n = length(x) ; r(n) = 0 ; %initialize rates at 'n' nodes to be zero if flag == 0 % Nakagaki's model chosen for j = 1:n if x(j) > 513 %specify limit of temp for 'm' m = -10 ; else m = 0 ; end r1(j) = k0*(P^l)*((x(j)/513)^m)*(exp(-E/R*x(j)))*(cm^nn) ; % calculate rate at each node end %=============================================================== % FLAG 1: Calculate rate of reaction accoding to Ohl (we dont use... % this for time being) %=============================================================== elseif flag == 1 % Ohl's model chosen for j = 1:n bm(j) = 0.154256e-6*(exp(32342.474 /(R*x(j)))) ; %calculate adsorption rate of methanol as fctn of T bw(j) = 0.623242e-8*(exp(34883.8183 / (R*x(j)))) ; %calculate adsorption rate of water as fctn of T nu = 0.5 ; X(j) = -102,679.506/(R*x(j)) ; r1(j) = (mc*nu*km*bm(j)*pm)... /(1+bm(j)*pm*bw(j)*pw) ; % calculate rate at each node end 72

PAGE 73

%=============================================================== % ANY OTHER FLAG : Null %=============================================================== else error('wrong flag'); end %%==============================================================%% Input File By Including Convection Co-Efficients % reformer_inputfile.m % Mohua Nath 04/23/2006 % Revision 2 % This program specieies all the reformer parameters including convection terms % This program is called by the m-file temp_cal1 %================================================================ % Catalyst bed Gas mixture effective parameters %================================================================ ro_eff = 1983 ; % effective mass density of catalyst gas... %mixture in [kg/m3] C_eff = 900 ; % effective specific heat of catalyst gas... %mixture in [J/kgK] K_eff = 5.00 ; % effective conductivity of catalyst... %gas mixture [W/mK] C_g = 1911.6; ro_g = 0.8640; vel = 0.01 ; %[m/s] %================================================================ % Reformer Dimensions (Cylindrical) %================================================================ Radius = 0.5 ; % Radius of reformer [m] Length = 0.5 ; % Length of reformer [m] %================================================================ %Discretisation Parameters %================================================================ 73

PAGE 74

M = 30 ; N = 30 ; n = (N-2)*(M-2) ; del_r = Radius / N ; % Size of finite element along Radius del_x = Length / M ; % Size of finite element along Length %================================================================ % Co-efficients of PDE and ODE %================================================================ alpha = K_eff / ( ro_eff C_eff ) ; % co-efficient to have Tdot on LHS for i = 1 : M-1 p(i) = alpha / ( del_r i* del_r ); % Co-efficient of del r term of ODE end q = alpha / ( del_r del_r ) ; % Co-efficients of del r2 term of ODE r = alpha / ( del_x del_x ) ; % Co-efficients of del x2 term of ODE s = (vel*0.5*ro_g*C_g)/ (ro_eff*C_eff); %================================================================ % Nakagaki reaction rate constants %================================================================ k0=1.35e6 ; % reaction rate constant [mol/(g_cat.s.atm)] P=15 ; l=0.13 ; E=1e5 ; % [J/mol] nn=1.3 ; R=8.3143 ; % Universal gas constant %================================================================ % Ohl's reaction rate constants %================================================================ pm=10 ; % partial pressure of methanol pw=20 ; % partial pressure of methanol cm=30 ; % partial pressure of methanol %mc=40 ; % partial pressure of methanol %================================================================ % Methanol properties %================================================================ 74

PAGE 75

Er=201e3 ; % heat of reaction per mole of methanol consumed %================================================================ for i = 1 : M-2 dia1(i) = -(p(i+1) +2*q+2*r+s) ; dia2(i) = (p(i+1) +q) ; end dia11 = -2*del_r*p(1) (2*q)-(2*r) s ; dia22 = del_r*p(1)+q ; rs = r+s ; %%===============================================================%% Differential Equation Calculation Including Convection Term % matrix_reformer2.m % Mohua Nath 04/23/2006 % Revision 2 % This function executes the S-function that solves the N set of ODE's % Invoked by the SIMULINK program matrix_sim_reformer2.sim %=================================================================== % : t= time ; x= vector of stateS ; u= vector of I/Ps ,... % flag= from simulink : sys = to simulink, x0 = I.Cs %=================================================================== function[sys, x0] = matrix_reformer5(t,x,u,flag) %=================================================================== % Call the inputfile for parameters %=================================================================== reformer_inputfile5 ; %=================================================================== % Flag 0 : Sends sizes and initial state vector to Simulink %=================================================================== if flag == 0 size_states = n ; % continuous time states size_disc_stats = 0 ; % number of discrete time states size_outputs = n ; % number of outputs to view size_inputs = 0 ; % number of inputs size_disc_roots = 0 ; % always 0 for now 75

PAGE 76

size_feedthrough = 0 ; % algebraic feed required? (no) sys = [size_states, size_disc_stats, size_outputs, size_inputs, size_disc_roots, size_feedthrough] ; % column vector of sizes from... % m file to simulink at flag=0 x0 = [523*ones(n,1)] ; % column vector of initial conditions %=================================================================== % Flag 1 : Brings T (x = state vector) from Simulink and Sends Tdot.. %(sys = state derivative) to Simulink %=================================================================== elseif abs(flag) == 1 %____________________________________________________________________ % Call the reaction_rate2 file to calculate rate of reaction %-------------------------------------------------------------------[r1] = reaction_rate2(x,0); % send 'T' for Nakagaki rate, and n %____________________________________________________________________ % Calculate the heat of reaction at every node %-------------------------------------------------------------------n =(N-2)*(M-2) ; E(n) = 0 ; % initialise E reaction at all nodes as 0 for j = 1:n mc = 2 pi r1(j) del_r del_x ; %mass of control volume E(j) = mc*Er*r1(j)/(ro_eff*C_eff) ; % energy consumed at all node end %____________________________________________________________________ % Imposing of B.C's at reformer wall = 400 C = 673 K %-------------------------------------------------------------------wall = 560 ; % create a vector of wall boundary nodes 76

PAGE 77

%____________________________________________________________________ % Imposing of B.C's at reformer input --> 100 C = 373 K %-----------------------------------------------------------------------------------------------------inlet1 = 520 ; % create a vector of inlet boundary nodes %____________________________________________________________________ % Impose the Nuemann condition at the exit of the reformer %-----------------------------------------------------------------------------------------------------st = 1 ; for i = N-1 : (N-2)*2 mid_rad(st) = x(i) ; st = st + 1 ; end j = 0 ; for i = 1 : M-2 cc = (N-2)*i 1 ; j = j+1 ; exit1(j) = x(cc) ; end %_____________________________________________________________________ % calculate the Tdot...i.e temperature gradient at every node %------------------------------------------------------------------------------------------------------Tdot(1) = dia11*x(1) + rs*x(2) + dia22*x(N-1) + dia22*mid_rad(1) + r*inlet1 E(1) ; % Tdot for 1st row for j = 2:N-2 Tdot(j) = r*x(j-1) + dia11*x(j) + rs*x(j+1) + dia22*x(N-2+j) + dia22*mid_rad(j) E(j) ; %Tdot for 2nd four rows if j == N-2 Tdot(j) = r*x(j-1)+ dia11*x(j) + dia22*x(N-2+j) + dia22*mid_rad(j) + rs*exit1(1) E(j) ; else end end d = 1 ; for j = (N-1):(N-2)*(M-3) 77

PAGE 78

Tdot(j) = (q*x(j-N+2) + r*x(j-1) + dia1(d)*x(j) + rs*x(j+1))+ dia2(d)*x(j+N-2) E(j) ; % Tdot for middle rows for f = 1 : M-3 if j == ((N-1)*f)-(f-1) d = d+1 ; Tdot(j) = q*x(j-N+2) + dia1(d)*x(j) + rs*x(j+1) + dia2(d)*x(j+N-2)+r*inlet1 E(j) ; elseif j == (f+1)*(N-2) Tdot(j) = q*x(j-N+2) + r*x(j-1) + dia1(d)*x(j) + dia2(d)*x(j+N-2) + ... rs*exit1(f+1) E(j) ; end end end el=60; for j = (N-2)*(M-3)+1:(N-2)*(M-2) if j == (N-1)*d d = d + 1 ; else end if j < (N-2)*(M-2) Tdot(j) = q*x(j-N+2) + % Tdot for last 5 rows else end if j == (N-2)*(M-3)+1 Tdot(j)= q*x(j-N+2) + dia1(d)*x(j) + r*x(j+1) + ... q*wall+p(d)*wall+r*inlet1 E(j) ; elseif j == (N-2)*(M-2) Tdot(j) = q*x(j-N+2) + r*x(j-1) + dia1(d)*x(j) + q*wall+p(d)*wall + ... r*exit1(M-2) E(j) ; end end sys = Tdot(1:n)' ; 78

PAGE 79

%=================================================================== % Flag 3 : Choose the outputs from the state vector 'T' (or x) %=================================================================== elseif flag == 3 for j = 1:n y(j) = x(j) ; % We choose the entire T at all nodes end sys = y' ; % send column vector of outputs %=================================================================== % Any other Flag : undefined....so return an empty vector %=================================================================== else % All other flags are irrelevant to the problem sys = [] ; end %%===============================================================%% 79

PAGE 80

APPENDIX B TEST BED BUS-2 CONTROL LOGIC S10.6 TURN OFF P-02NEAT CH3OH TO RF Bar S 7.1 Set Parameters for P-02 Calculate p for P-01,Premix Pump Set SV of P-01=p Set Parameters of P-01 Start time t-70For CHOPPER = 0.15 T-700.15 over? Turn CHOPPER on Set Timer t-71 to 0.1s T-710.1sup? S7.2 H-01 Constant Loop heater on H-01Started? Start Timer t-72 =20 sec T-72 20s up? A X N NN A S 7.3 H-02 heater on H-02Start? Start Timer t-73 =15 sec T-73 15s up? X S 7.4 H-03Start? Start Timer t-74 =15 sec T-74 15s up? H-03 heater on S 7.5 H-04Start? Start Timer t-75 =60 sec H-04 heater on X T-75 60s up? Set P-01 premix P10 parameters S 7.6 TCA 601Base Temp 1oC NN X NYYYNNY S 7.7 S10.9 NYY 80

PAGE 81

81 START Set Premix Flow Turn OFF Neat Methanol to burner (p-02) Turn on CHOPPER Turn on H-01-04 heaters in coolant loop one by one Set Premix Flow to RF Check Burner BUTemp >1 oC STOPNY

PAGE 82

SOV-930 CLOSE, NO CO2PURGING IN RF S 6.4 SOV-210 OPEN, LET AIR TO CATHODE CHECK FC VOLTAGE > 150 V? CALCULATE B-01SV PARAMETER TURN ON BLOWER TO CATHODE B-01 SET BLOWER B-01PARAMETERS T-64 40 SEC UP? S 6.5 START TIMER T-65 TO 28 SECNYYN S 6.41 ES-1 FC VOLTAGE >150 V? YN S 6.6 S10.6 T-6528 SEC OVER? NY A SOV-920 CLOSE, NO CO2PURGING OF FC S 6.1 START TIMER t-60 TO 1 SEC T-601 SEC OVER? NY S 6.2 SOV-160 OPEN, OPEN FLUE GAS LINE SOV-150 OPEN, OPEN REFORMATE LINE START A TIMER t-61 TO 1 SEC T-611 SEC OVER? NY S 6.3 SOV-170 CLOSE, CLOSE ANODE VENT FROM FC SOV-180 CLOSE, CLOSE BYPASS VALVE TO FC START TIMER t-63 to 2 sec t-64 to 40 sec T-632 SEC OVER? NY A S10.5 82

PAGE 83

START WAIT 1 SEC STOP CO2TO FC OPEN FLUE GAS LINE OPEN REFORMATE LINE WAIT 1 SEC IF FC VOLTAGE > 150 V NY CLOSE FC VENT DONT BYPASS FLUE TO FC WAIT 2 SEC NO CO2TO RF BU SUPPLY AIR TO CATHODE CALCULATE BLOWER FOR AIR TO CATHODE PARA SET PARAMETERS WAIT 1 SEC A A NO CO2TO RF BU ES-1 WAIT 1 SEC S10.6 IF FC VOLTAGE > 150 V ES-1 83

PAGE 84

S10.5 Trigger S13.1 Turn off Igniter -1 Turn off Igniter -2 Turn off Neat CH3OH(P-01) Burner 1 Set Parameter of P-02 Turn off Neat CH3OH Burner 2 Set Parameter of P-04 Close Neat CH3OH Valve S120 Close Premix Value S130 Set Blower (RF) Parameter -02 Set CH3OH heater H-05 Parameters Turn off H-05 Set P-03 Neat CH3OH Parameters Turn off P-03 Set start up blower -3 Paramters Set P-01 premix pump parameters Turn P-01 OFF No Premix. .Blower -01 parameter set ??? To cathode .Blower -01 Turn OffNo Air to Cathode Trigger S13.2 Y S13.3 INCREASE COOLANT PUMP P-05 SETTINGS ??? COOLANT LOOP TEMP > 60 C Start Timer t-130 to 5 min IS 5 min of T-130 over? Trigger S13.5 Close SOV 170,180, FC VENT, FC BY PASS FC TEMP <120 CTCA -700 TCA -101RF TEMP< 170C S13.6 PROCESS Set B-03 BLOWER Startup Parameters Turn off B-03 Turn off P-05 S13.7 ?????? Blower B-02 Parameters B-02 Off CHECK S13.6 Wait for Some time? Y CHECK S13.7 S10.1 S13.8 Trigger Set P-05 coolant pump control X X 84

PAGE 85

If Coolant Temp > 60 C NY Turn off Igniter 1 & 2 Stop Neat CH3OH to Burner 1 & 2 Stop Premix to Reformer Increase Blower to Burners Parameters Stop Air to FC Cathode S Increase Coolant Pump Settings Start Timer Wait for 5 Mins Close FC Vent, By Pass Valve to FC Is temp of FC < 120 C? Turn off Startup Blower Coolant Pump Is temp of FC < 120 C? Turn off RF Blower Y YNNY NORMAL SHUTDOWN S Wait Till FC Temp< 120 C Wait Till RFTemp<170C Wait for Some Time S10.1 NNNYYY X X 85

PAGE 86

S10.9 S8.1 Set Timer T-80 to 15 Sec SCS COMMAND DOWN Is 15 Sec of T-80 up? S8.2 SCS Command up? SCS COMMAND STAY S 8.1 STAY CONTROL STEP DOWN CONTROL S 8.3STEP UP CONTROL S 8.5 H.01 OFF? H.01 OFF SET T.81 = 5 SEC T.81 5 SEC UP? S 8.6 H-02 OFF? H-02 OFF SET T-82 =5SEC T-82 -> 5 SEC UP? STEP UP CONTROL S 8.6 H-03OFF? B A A H.03 OFF Set T-3 = 5 sec S 8.8 T-3 5 sec up? B T-3 5 sec up? H-04 Off Set T-84 = 5 sec T-84 5 sec up? S 8.9 Set up Control Parameter SET FC power >14 KW Set T-85 = 24 sec <= FC power? T-85 24 sec up? C Chopper Error FC Current <0A T-703 sec up? T-70=3 sec C S 8.31CHOPPER S7.2 NY B YNNYNYYNYNYNYNNY YYN YY NNNN YYNNNNYY 86

PAGE 87

FC POWER < 11 Kw? STEP DOWN CONTROL S 8.11 H-04 TURN ON H-04 ON? SET T-84 = 5 SEC T-845 sec? S 8.12 H-03 ON? H-03 turn ON SET T-85 = 5 SEC S 8.13 H-02 ON? T-855 sec? H-02 turn ON A A NY NYNN YN A T-82 5 sec up? STEP DOWN CONTROL PARAMETERS IS H-02 ON? S 8.14 H-01TURN ON C Set T-82 = 5 Sec A1 YNYNYY STAYCONTROL FC CURRENT 30A S 4.5 STANDBY REQ ON? S 8.17 | STOP S 10.10 Flg S 8.17 Set T-76 =3 Sec T-76 5 sec up? C8 =3? C8 UP S 7.2 NY STOP isSCS command? YNNNY Z ES -1.2 NYY 87

PAGE 88

Is SCS command UP START WAIT 15 SEC Turn off H-01 to 04 after 5 sec interval Wait till FC o/p power 11 kW Wait 24 Sec If FC Current0A S 7.2Chopper Error Is SCS command DOWN IS FC power < 11 KW? Turn on H-01 to 04 after 5 secons Is SCS CommdSTAY Is FC Current 30A Check if StdbyOP commdfrom SCS ES 1.2 Is any heater already on? Step Down Control parameter Wait 3 Sec IF counter = 3 S 7.2. STOP NNNNYYYY NNYYNYNY 88

PAGE 89

S1.10 PROCESS T-23 5 sec up? S2.1 PROCESS RF Set B-02 PID Settings X Start at Timer(t-23) to 5 sec Set B-02 PID Again S 2.2 Process IG -01 IgnitorON Start Timer T-26 to 10 sec T-2610 sec up? S 2.3 Start Timer T-21 to 30 sec Turn P-02 ON,Neat Methanol to Burner Set P-02 Parameters T-25 2 sec up? Start T-25 to 2 Sec A 89

PAGE 90

T-21 ,30 Sec up? Set Blower to RF Burner, B-02 param. A S 2.4 Burner Temp TCA 601 200C Start a Timer T-22 to 3 sec T-22,3 Sec up? Burner Temp TA 601 200C S 2.5 Turn off Ingitorin RF Burner S 2.6 RF Burner TCA 101 > 200C S10.4? S 2.15 RF B Turn off Ignitorin RF Burner S 2.20 COUNTER C2 ON Turn off MethanolIn F02 RF 8V Set P-01 parameters Start timer t-24 = 30 sec S 2.21 Set RF burner param C2 UP? S 2.40 ES-1 X S 2.7 Turn off Neat CH3OH to burner (p-02) S10.4? X Set P-02 Parameters Set Blower to RF BUB-02 Parameters S 2.8 RF temp TCA 101 225C B N NYNYYYNNN NYY 90

PAGE 91

APPENDIX C DESCRIPTION OF CONTROL SCHEME Main Routine Introduction The main routine, called the S10 routine, summarizes the entire process of operating the bus. This routine calls for and activates other subroutines that carry out their individual sub processes. The main functions in this routine are start up of the bus, check for emergency shut down conditions, chopper control and normal operation and shut down. This process starts with initiating RS-232 serial port communication between FCIC (Fuel Cell Internal Controller) and SCS (Subsystem Control System). After ensuring that all pumps and solenoids are in their initial normal position, It turns on the APU (Auxiliary Power Unit) and calls a sub routine to start the Fuel cell and Reformer. It waits for 24 minutes to turn on the reformer and bring up the temperature of catalyst bed to its minimum operating temperature. Supply of air to cathode is done and load is increased through chopper control. Throughout the process, in every scan, check is made for shut down request or other abnormal condition, in which case the process is interrupted by calling a sub-process that walks the system through a shut-down. Pseudo code description A short description follows of the main process flow in the form of a sequential pseudo code: MAIN PROCESS (S 10) Initialize RS232 Communication Turn on SOV PUMP o Returns a failure flag if CO2 purge is ON Stop CO2 purge on failure Check if FC ON is requested Turn ON APU (Auxiliary Power Unit) Start Reformer (S1) o Reformer Blower B-02 turn ON o Igniter IG-01 turn ON 91

PAGE 92

o SOV for Neat methanol SOV-120 turn ON o Vent for FC SOV-170 turn ON o By-Pass Valve SOV-170 turn ON o Pump from Neat methanol, P-02 turn ON o When temperature of burner is adequate, igniter IG-01 turn OFF Start Fuel Cell (S 3) o Start up Blower B-03 turn ON o Igniter IG-02 turn ON o Adjust coolant flow through P-05 pump o Start neat methanol flow by P-04 turn ON o Put B-03 on PID control o When temperature is adequate, Igniter IG-02 turn OFF o When temperature of coolant is steady, neat methanol pump P-03 turn OFF o If temperature of coolant falls restart procedure S 3 Reformer gas cooler, E-03 turn ON Heat exchanger for neat methanol to reformer burner, H-05 turn ON Try to reach reformer operating temperature for 24 minutes o On failure ES-01 sub routine (S 12) is called for emergency shutdown Check if reformer temperature is steady. o On failure restart S 1, S 3 Air to Cathode Blower B-01 turn ON Coolant loop heaters H-01 to H-04 turn ON Chopper turn ON Wait till reformer temperature reaches above 200C Check if FC temperature> 150 C o On failure adjust chopper, turn on H-01 to H-04 o Else Reformer and Fuel Cell turn OFF and restart process S 1 and S 3 at step 2 On shutdown request start S12 emergency shut-down If standby is requested, wait 10 minutes and restart S 1 and S 3 at step 2 Discussion of Main Process The process begins when the power ON button is pressed by the operator. This action first initiates the RS232 serial port communications between FCIC which is a MICREX 140th PLC and SCS which is a Windows based Lab View Controller and DAS. The solenoid valves and pumps are checked. If they are not normally ON or Normally OFF condition then check is made whether CO2 purge is being made on the system and corrected if required. A sub process is then initiated whose main function is to bring the temperature of the reformer to its operating temperature. The Blower to reformer burner, igniter, and solenoid valve from Neat Methanol Tank, Vent for Fuel Cell are turned on. Also the by-pass valve which 92

PAGE 93

returns the tail reformate gas back to the Fuel Cell instead of the burner is turned on. It also starts neat Methanol flow into the burner and turns off the igniter if burner temperature is adequate. At the same time, another sub process is started that starts the Fuel Cell and brings it up to its operating temperature. Its main function is to turn the Blower of start up burner and Igniter on. And turn off the igniter and stop fuel flow when adequate temperature is reached. Again at the same time as above two processes, cooler is turned on to cool the reformate going to the Fuel Cell. Along with this heat exchangers are turned on in the neat methanol line in order to vaporize the methanol before it reaches the burner. The process allows 24 minutes for the reformer temperature to reach operating condition of which if exceeded leads towards emergency shutdown. Once the temperature is steady in reformer, air is supplied to cathode. The chopper is adjusted to increase the load. If the temperature reaches 150 C, the system is shutdown and restarted. An operator instigated shut down request leads the system to a normal shutdown process. S1 Sub-Routine Introduction The S1 sub-routine, describes the process of turning on the reformer. Its man function is to warm-up the reformer catalyst bed upto its minimum operating temperature. This entire process takes 24 minutes. This process is instigated by the main process S10. It turns on the blower and igniter and the valves for fuel flow. It adjusts the air flow through PID control so that the temperature of the reformer is brought upto 180C and maintained at the same temperature throughout the entire operation of the Bus. 93

PAGE 94

Pseudo Code Description A short description follows of the main process flow in the form of a sequential pseudo code: Reformer startup Process (S 1) Blower to reformer Burner, BU01 turn ON Adjust PID settings on Blower, B02 Igniter IG-01 turn ON Neat Methanol valve SOV-120 turn ON FC Anode vent solenoid valve SOV-170 turn ON By pass valve SOV180 turn ON Neat Methanol pump P-02 turn ON Adjust PID settings on Blower, B-02 Check burner, BU-01 temperature above 200? o If no, wait for 30 sec then instigate ES-1, S 12 process after IG01, o P-02 turn OFF and adjustment of PID settings of B-02 o If yes, next step Igniter IG-01 turn OFF Wait till bottom of reformer temperature >180 C Neat methanol (liquid)flow to burner by Pump P-02 turn OFF Adjust PID of Blower to maintain RF temperature below 225 C Discussion of Process S1 When instigated by main process S10, sub process S1 starts. It first turns on the blower of the reformer burner by energizing an auxiliary relay coil in the FCIC. The PID settings of the blower are adjusted to give 40% output. After a wait of 30 seconds the blower PID settings are again adjusted, this time reduced by 16.7%. The igniter then is turned on which starts the flame in the burner in order to raise the temperature inside the super heater which carries the fuel-water premix. After a wait of 10 seconds, solenoid valve from the neat methanol tank is opened. The FC anode vent is opened and also the FC by pass valve. Initially the residual reformate coming out of the FC called the tail gas is returned to the Fuel Cell in order to raise the temperature of the fuel cell to its operating temperature. After a wait of 1 sec, the pump that supplies neat methanol to the reformer burner is turned on. Settings for neat methanol pump are adjusted to 50% output. After an interval of 2 sec the 94

PAGE 95

Blower PID settings are again changed. At this point if the burner temperature is above 200C then the ignition is assumed successful. If the flame temperature is steady for 3 seconds after this, the igniter is turned off. System waits till reformer temperature reaches 180 C after which pump of liquid methanol is turned off to the burner and now the burner continues combustion with flue gas from FC. Settings are adjusted for the pump of liquid neat methanol to 0 and reformer blower PID settings are re-adjusted for hydrogen fuel. The temperature of the reformer is maintained at 225C by adjusting the blower settings. Completion of this step instigates sub-process S2. In case of burner not being able to build flame for 30 seconds, the igniter, neat methanol is turned off. Liquid neat methanol pump is adjusted to 0 and blower PID settings are increased to 70%. This step next instigates S12 or emergency shutdown sub process. S2 Sub-Routine Introduction The S2 subroutine, describes the completion of S1 process of turning on the reformer. Its main function is to increase the temperature of the reformer catalyst bed by adjusting the blower parameters. Increase of the reformer temperature to its maximum operating temperature optimizes the fuel combustion efficiency to produce more hydrogen. This process is instigated by the sub process S1. It turns on the blower and the igniter and the valves for fuel flow. It adjusts the airflow through PID control so that the temperature of the reformer is brought upto 225C and maintained at the same temperature throughout the entire operation of the Bus. Pseudo Code Description A short description follows of the main process flow in the form of sequential pseudo code: Reformer temperature increase sub process (S2) 95

PAGE 96

Initiated by the end of process S1 Blower B02 PID setting adjusted Igniter IG01 turn ON Check if burner BU01 temperature is >200 C o If no, then instigate ES-1, S12 o If yes, next step Igniter IG-01 turn OFF Wait till bottom of reformer temperature >240 C Neat methanol (liquid) flow to burner by pump P-02 turn OFF Adjust PID of the Blower to maintain RF temperature below 24O C Discussion of Process S 1 When instigated by main process S1, sub process S2 starts. It first turns on the blower of the reformer burner by energizing an auxiliary relay coil in the FCIC. The PID settings of the blower are adjusted to 16.7% output. The igniter then is turned on which starts the flame in the burner in order to raise the temperature inside the super heater which carries the fuel-water premix. After a wait of 10 seconds, solenoid valve from the neat methanol tank is opened. The settings of this pump are adjusted to 50%. The blower settings are changed is 45%. If the burner temperature does not rise at this point after 30 seconds the system is lead to emergency shutdown. If the burner temperature rises and is steady for 3 seconds then the neat methanol flow is turned off and the system waits till reformer temperature reaches 225 C. S3 Sub-Routine Introduction The S1 sub-routine, describes the process of running on the start-up burner BU-02. Its main function is to start the flame in the start-up burner. The purpose of the start up burner is to heat the fuel cell stack upto its minimum operating temperature of 150C. The coolant loop which takes away heat from the Fuel Cell during normal operating condition, at this point acts as a heating loop as the same coolant liquids adds heat to the cathode in order to raise its temperature. 96

PAGE 97

This heating loop also derives heat by 4 electric heaters H01 to H04 which are adjusted according to the load at the output of the fuel cell. This process is instigated by the main process S10. It turns on the blower and igniter and the valves for the fuel flow. It adjusts the air flow through PID control and maintains the temperature of the loop below 165C, this process ends when FC has reached a temperature of 150C. Pseudo code description A short description of the main process flow in the form of a sequential pseudo code: Start up Burner on Process (S3) Start-up Blower, BU-02 turn ON Check if coolant temperature>=60C? o If <=60C decrease coolant flow o If yes, increase coolant flow Wait 30 seconds Adjust Blower B-03 Igniter IG-02 turn ON Wait 5 secs Start neat methanol flow to start up burner by P-04 turn ON Adjust blower B-03 PID settings Check if temperature of start-up heat exchanger is between 30C-400C? o If >400C, wait 15 secs, If still above 400C instigate ES-1, S12 and restart o If yes, next step Igniter IG-02 turn OFF Check coolant temperature and adjust flow of coolant If coolant temperature> 165C, then P-02 turn OFF Adjust blower B-03 till coolant<135C Wait till FC reaches 150C Wait 15 secs and restart procedure if temperature is unsteady Discussion of Sub process S3 When instigated by main process S10, sub process S3 STARTS. It first turns on the start-up blower of the heat exchanger. It then checks for the temperature of the coolant loop. If the temperature is above 60C the coolant flow is decreased and if it is below 60C, the coolant flow is increased by adjusting the pump P-05. 97

PAGE 98

After a wait of 30 seconds, the Blower B-03 parameters are adjusted and Igniter is turned on. After a wait of 5 seconds, neat methanol tank solenoid valve, SOV-120is turned on. The blower PID settings are again adjusted to maintain the temperature of the heat exchanger between 300C to 400C. Upon attaining this temperature the igniter is turned off and coolant temperature is checked and flow adjusted accordingly. System waits till coolant temperature reaches 165C. When this temperature is reached the neat methanol flow to start up the burner is closed. After a wait of 30 secs, the blower parameters are adjusted till FC reaches 150C. If the heat exchanger takes more that 15seconds to reach temperature between 300C and 400C since the neat methanol is supplied, then the system undergoes an emergency shutdown and after a wait of 15 secs, the process restarts. S4 Sub Routine Introduction The S10.3 flag in the main process S10 instigates S4 sub routine. This describes the process of hot-standby required during start-up. Its main function is to make the system wait for 30 minutes and maintain the temperature of fuel cell upto 150C. The S10.3 flag is raised when a hot standby is requested in the main process. Pseudo code description A short description follows of the main process flow in the form os a sequential pseudocode. START-UP BURNER ON PROCESS (S3) Start-up timer to 30 mins Check if hot standby command is requested If yes, wait 30 ins to restart procedure For 30 ins, heaters H01 to H04 turn ON Maintain FC temperature to 150C 98

PAGE 99

Discussion of Sub Process S4 When instigated by main process S10, sub process S4 starts. It first turns on a timer to 30 seconds. It then checks the FC temperature to be 150C. If not, it turns on heaters in the coolant loop to increase or decrease the temperature. After a wait of 10 mins, it checks for hot standby off request where-in the system continues normal operation, else it returns to normal operation after a wait of 3 seconds. S6 Sub routine Introduction The S1 sub routine, describes the process of turning on operation of system from start-up to normal mode. Its main function is to stop the purging of CO2 and stop the return of gas to the fuel cell and maintain a steady voltage output at the fuel cell load. Pseudocode description A short description follows of the main process flow in the form of a sequential pseudocode: START-UP BURNER ON PROCESS (S6) Purging of CO2 to the FC turn OFF Flue gas line to reformer burner turn ON Reformate line turn ON Wait 1 sec FC vent OFF By-pass valves that returns tail gas to FC turn OFF CO2 purging to reformer and burner turn OFF Air supply to cathode turn ON Calculate blower, B01 parameters for air requirement to cathode Check if FC vtg >150V? o If no, wait 40 seconds, if still no, then lead to ES-1, S12 o If yes, next step Check if FC vtg>150V? o If no, then lead to ES-1, S12 o If yes, next step o Instigate hot-standby 99

PAGE 100

Discussion of Sub Process S6 When instigated by main process S10.5, that requests hot standby, sub process S5 starts. It stops the CO2 purging of the Fuel Cell Stack. CO2 is purged during conditions when bus is not in operation. Pressurizing the catalyst bed and FC by CO2 ensures that no oxidation of the above two takes place. However before the normal operation of the bus, the CO2 purging has to be stopped. Next step is to open the flue gas line so that the un-used hydrogen at the FC can be used at the reformer burner as fuel. This not only improves the efficiency of the system but also saves a lot of fuel in form of neat methanol. The reformate line is opened as the product gases from premix combustion containing hydrogen is supplied to the fuel cell stack. After a wait of 1 second, the FC vent is closed and also the by pass valves which allowed the tail gas to be returned to the fuel cell stack. CO2 purging is stopped into the reformer and burner and air is supplied to the cathode. Air requirement into the cathode is calculated as the parameter settings of the blower and these settings changed too supply correct amount of air. If at this point FC voltage is not steady at 150V for 40 seconds, then emergency shutdown occurs, else hot standby is again instigated. 100

PAGE 101

S7 Sub-Routine Introduction The S7 sub-routine, describes the process of adjusting premix fuel flow according to the change of load at the fuel cell output. This process is instigated at the end of subprocess S6. Its main function is to set premix flow, turn on chopper adjust heaters in coolant loop and maintain burner temperature. Pseudo code description A short description of the main process flow in the form of sequential pseudocode follows: ADJUST PREMIX FUEL FLOW PROCESS (S7) Neat Methanol to burner B-02 by P-02 turn OFF Set premix flow Adjust pump P 01 according to setting parameters calculated Chopper turn ON Heaters H 01 to H 04 turn ON one by one as load gets changed Set premix flow according to load Wait till burner temperature has reached minimum operating temperature End Process by Instigating S 10.9 Discussion of Sub Process S7 When instigated by sub process S6, this sub process starts. First it turns off the neat methanol supply to the reformer burner as at this point hydrogen in the flue gas is enough to maintain the flame. The premix flow rate is calculated and parameters of the pump for premix flow into reformer is changed. The Chopper is turned on that adjusts the load at the output and gradually increases output from 0 to 100%. The heaters in coolant loop are turned on to maintain the FC temperature to 150oC. 101

PAGE 102

As the load varies the premix flow is adjusted to meet the hydrogen requirement at the FC. After ensuring that the reformer burner has maintained its steady temperature the process then instigates the next sub process S8. S8 Sub-Routine Introduction The S8 sub-routine, describes the process of chopper control at the fuel cell output load. Its main function is to take necessary action according to chopper command from the SCS and to adjust the heaters in the coolant. As the chopper increases the load at the output FC draws more current and as a result the temperature goes up. However when the load is suddenly dropped the heat generated is not enough to maintain the temperature at the Fuel Cell stack, due to this reason we need to adjust the heaters to maintain a steady temperature at the FC. Pseudo code description A short description of the main process flow in the form of sequential pseudo code follows: CHOPPER CONTROL PROCESS (S 8) Check if SCS command is UP/DOWN/STAY If UP: o Heaters H 01 to H 04 turn OFF one after another in 5 seconds interval each. o If any heater is on then adjust step up control parameters o Wait till FC O/P power 14 kW o If FC does not have control output show Chopper Error If DOWN: o Check if FC power < 11kW o If yes H 01 to H 04 turn ON one after another in 5 seconds interval between each and restart process o If no, check if any heater is already ON, if ON step-down control parameters are adjusted, else heaters H 01 to H 04 are turned ON and restarted. If STAY: o Check if FC current less than 30 A? o If yes wait 3 seconds and if still less than 30 A then lead to ES-1 102

PAGE 103

o If above 30A, check for standby o If yes STOP and instigate S10.10 and if no, then restart Discussion of Sub Process S8 When instigated by sub process S7, this sub process starts. First it checks for the command from the SCS whether to step up or down the load or to stay at same level. The chopper control at the SCS communicates with the FCIC with RS 232 communications. If the chopper has commanded the load to be stepped up then the heaters in the cooling loop are turned off as there is adequate heat in the Fuel Cell. The system waits till the FC power output reaches 14 kW. If the FC does not show any current output at this point then show copper error. If the chopper on the other hand has commanded a decrease in load then the heaters are turned on one by one to maintain the temperature at 150oC and the process is restarted. If the chopper commands a stay, the FC output current is checked if it is steady at 30A. If it is below 30A even after a wait of 3 seconds then the system goes through emergency shut down. If the FC output current is steady at 30A then check is made if standby is requested. If not then the process restarts after instigating S10.10. S9 Sub-Routine Introduction The S9 sub routine, describes the process of maintaining FC temperature with chopper control scheme. Its main function is to continue with S8 process and turn on/ off the heaters to maintain the FC temperature at 150oC. Pseudo code description A short description of the main process flow in the form of sequential pseudocode follows: CHOPPER CONTROL PROCESS (S 9) Set start-up blower B 03 parameters. 103

PAGE 104

Check if FC temperature > 140oC o If no, wait for 90 seconds, If still no then lead to ES 1 S12 o If yes, wait for 40 seconds and go ahead Set start-up blower B-03 parameters Check SCS UP/DOWN/STAY command If UP: o H 01 to H 04 turn OFF one after another in 5 seconds interval between each o Set step-up control parameters If DOWN o H 01 to H 04 turn ON one after another in 5 seconds interval between each o Set step-down control parameters If STAY: o Restart process after checking FC temp > 140oC Discussion of Sub process S9 When instigated by sub process S8, this sub process starts. First it calculates and adjusts the start-up blower parameters in order to maintain the temperature of the coolant loop. Then it checks if FC temperature is at its operating temperature or not. If not then it leads to emergency shut down. If it has maintained operating temperature then it waits for 40 seconds and goes to next step of checking SCS chopper command, If the SCS chopper command is UP, then it turns off the heaters in the coolant loop as there is adequate heat in the system. It then changes the step-up control parameters in the SCS. If the SCS command is DOWN, then it turns on the heaters to increase the temperature of the FC to 150oC and changes the step down parameters in the SCS. If the SCS command is STAY, then the process is restarted after checking that the FC temperature has reached 140oC. S11 Sub-Routine Introduction The S11 sub-routine, describes the process of normal system level shutdown procedure. Its main function is to turn off the blowers, fuel and turn on the vents and keep the blower on. The 104

PAGE 105

normal shut down procedure is instigated by the operator when he wants to shut down the bus from its running condition. Pseudo code description A short description of the main process flow in the form of sequential pseudo code follows: NORMAL SHUT DOWN PROCESS (S 11) Ignitors IG 01 and IG 02 are turn OFF Neat Methanol flow by P 02 and P 04 are turn OFF Premix flow by P 01 is turn OFF Increase the settings of blowers B 02 and B 03 Air to cathode blower B-01 is turn OFF If coolant loop temperature > 60oC increase the settings of coolant flow pump, P 05 Wait for 5 Seconds FC Vent SOV 170 is turn OFF Wait till FC temp is < 120oC and Reformer temp < 170oC Wait for 3 seconds Instigate S10.1 main process Discussion of Sub Process S11 When instigated by any abnormal condition detected during the main process S10, this sub process starts. The first step is to turn off all fuel supply to the burners. The premix fuel is stopped into the reformer. Air Flow into the cathode is stopped. If the coolant temperature is above 60oC then the coolant pump flow is increased to reduce the temperature to the FC stack. After the system is made to wait for 5 minutes, the FC vents are closed and the bypass valves to the FC are closed. If the temperature of the FC isles than 120oC then the start up blower is turned off and coolant flow is also turned off. If the temperature of the reformer is below 170oC, then the reformer blower is turned off. 105

PAGE 106

When the temperature of the reformer and FC is below 170oC and 120oC the shutdown procedure is considered complete and process is restarted at S10. Then it check if FC temperature is at its operating temperature or not. If not then it leads to emergency shut down. If it has maintained operating temperature then it waits for 40 seconds and goes to next step of checking SCS chopper command. If the SCS chopper command is UP, then it turns off the heaters in the coolant loop as there is adequate heat in the system. It then changes the step-up control parameters in the SCS. If the SCS command is DOWN, then it turns on the heaters to increase the temperature of the FC to 150oC and changes the step down parameters in the SCS. If the SCS command is STAY, then the process is restarted after checking that the FC temperature has reached 140oC. S12 Sub-Routine The S12 Sub-routine describes the process of system level emergency shut down procedure. Its main function is to purge the system with CO2, turn off the blowers, fuel and turn on the vents and keep the blower on. The emergency shutdown procedure is instigated by the main process on detection of any abnormal condition. Pseudo code description A short description of the main process flow in the form of sequential pseudo code follows: NORMAL SHUT DOWN PROCESS (S11) Ignitors IG 01 and IG 02 are turned OFF Neat Methanol flow by P 02 and P 04 are turned OFF Heaters H -01 to H 04 turn OFF Premix flow by P 01 is turn OFF Increase the settings of blowers B 02 and B 03 Neat methanol heater H 05 turn OFF 106

PAGE 107

Reformate line and flue gas line turn OFF Air to cathode blower B-01 is turn OFF If coolant loop temperature >60oC increase the settings of coolant flow pump, P-05 Wair for 5 seconds. FC Vent SOV 170 is turn OFF By pass valve SOV 160 is turn OFF Purge system with CO2 in Reformer, FC and burners. If the temperature of FC < 120oC, start up blower B 03 and coolant flow pump P-05 is turn OFF If Reformer temp > 170oC reformer blowe B-02 is turn OFF Wait till FC temp > 120oC and Reformer temp < 170oC Wait for 3 seconds Instigate S10.1 main process Discussion of Sub Process S11 When instigated by any abnormal condition detected during the main process S10, this sub process starts. The first step is to purge the system with CO2. It then turns off all fuel supply to the burners. The premix fuel is stopped into the reformer. Air flow into the cathode is stopped. If coolant temperature is above 60oC the coolant pump flow is increased to reduce the temperature of the FC stack, After system is made to wait for 5 minutes, the FC vents are closed and the by pass valves to the FC are closed. If the temperature of the FC is less than 120oC then the start up blower is turned off and coolant flow is also turned off. If the temperature of the reformer is below 170oC, then the reformer blower is turned off. When the reformer and FC temperature is below 170oC and 120oC then the shut down procedure is considered complete and process restarted at S10. 107

PAGE 108

LIST OF REFERENCES 1. EG & G Services Parson, Inc., Fuel Cell Handbook, 5th edition, U.S. Department of Energy, National Energy Technology Laboratory, Morgantown, 2000. 2. J. Larminie, A. Dicks, Fuel Cell Systems Explained, John Wiley & Sons, 2003. 3. D. Betts., Transient performance of steam reformers in the context of automotive fuel cell system integration, Ph. D Thesis, University of Florida, Gainesville, 2005. 4. A. G. Stefanopoulou, J. T. Pukrushpan and H. Peng, Control-oriented modeling and analysis for automotive fuel cell systems, Journal of Dynamic Systems, Measurement, and Control, 126 (2004). 5. L. Blomen, M. Murgerwa., Fuel Cell Systems, Plenum Press, NY, 1993. 6. K. Kreuer, W. Vielstich, A. Lamm and H. Gasteiger, Handbook of Fuel CellsFundamentals, Technology Applications, Vol 3, John Wiley & Sons, Chichester, 2003. 7. S. Rohit., Water Balance Considerations in Modeling of PEM Fuel Cell Systems, Masters Thesis, University of Florida, Gainesville, 2005. 8. G. Ohl, S. Jeffrey, S.Gene, A Dynamic Model For The Design Of Methanol To Hydrogen Steam Reformers For Transportation Applications, Department of Mechanical Engineering and Applied Mechanics, Ann Arbor, 2004. 9. B. Daniel., S. Timothy, E. Paul, and R. Vernon., Review of the University of Florida Fuel Cell Bus Research, Demonstration and Education Program, University of Florida, Gainesville, 2000. 10. O. Necati., Heat Conduction, Wiley, NY, 1980. 11. T. Nakagaki, T. Ogawa, K. Murata, Y. Nakata, Development of methanol steam reformer for chemical recuperation, Journal of Engineering for Gas Turbines and Power, 2001. 12. S. Ahmed, M. Krumpelt, Hydrogen from hydrocarbon fuels for Fuel Cells, International Journal of Hydrogen Energy, 2001. 13. H. Helms, and P. Haley, Development of a PEM Fuel Cell System for Vehicular Application, Society of Automotive Engineers, 1992. 14. S. Ahmed, R. Kumar, and Krumpelt, Development of a Catalytic Partial Oxidation Reformer for Methanol Used in Fuel Cell Propulsion Systems, San Diego, 1994, 15. Y. Choi and H. Stenger, Kinetics, simulation and optimization of methanol steam reformer for fuel cell applications, Journal of Power Sources, Bethlehem, 2004. 108

PAGE 109

16. H. Geyer, R. Ahluwalia, M. Krumpelt, R. Kumar, Transportation Polymer Electrolyte Fuel Cells Systems for Different On-Board Fuels, San Diego,1994 17. J. Amphlett, R. Mann, R. Peppley, B. Stokes, Methanol Reformers for Fuel Cell Powered Vehicles: Some design considerations, Fuel Cell seminar, Phoenix, 1991. 18. N. E. Vanderborgh, R. D. McFarland, J. R. and Huff, Advanced System Analysis for Indirect Methanol Fuel Cell Power Plants for Transportation Applications, Phoenix, 1990. 19. E. Santacesaria, S. Carra, Kinetics of Catalytic Steam Reforming of Methanol in a CSTR Reactor, Applied Catlisis, 1983. 20. S. Chan and H. Wang, Thermodynamic and Kinetic Modeling of an Autothermal Methanol Reformer, Journal of Power Sources, 2004. 21. G. Ohl, J. Stein, and G. Smith, Fundamental Factors in the Design of a Fast Responding Methanol to Hydrogen Steam Reformer for Transportation Applications, Journal of Energy Resources Technology, 1996. 109

PAGE 110

BIOGRAPHICAL SKETCH Mohua Nath was born in Lumding, India, in 1979. Mohua completed her Bachelors Degree from Cummins College of Engineering, Pune, India, 2002, in Instrumentation and Control Engineering. After that she worked in Pune Instrumentation, a private limited company, where she obtained hands-on experience in instrumentaion, hydraulic and pneumatic systems. Mohua then started working towards her masters degree at the University of Florida from the Fall of 2004 under the guidance of Dr. W. Lear, Dr. O. Crisalle and Dr. J. Fletcher. She started working full time in Siemens Power Generation, Orlando from January 2007. Upon obtaining her M.S degree, Mohua plans to continue to contribute to the field of energy generation. 110