COMBUSTION OF SOLID FUEL INA FLUIDIZED BED COMBUSTOR

A Thesis Presented to

The Faculty ofthe

Fritz J. and Dolores H. Russ College ofEngineering and Technology

Ohio University

In Partial Fulfillment

Ofthe Requirement for the Degree

Master ofScience

--,-,*, \ ., by ,m Abu Noman Hossain

June, 1998 i

ACKNOWLEDGEMENTS

I would like to express and extend my heartfelt appreciation and deep gratitude to the many people who have, in one way or other, helped and supported me in my work.

To Dr. David J Bayless, my honorable advisor, whom I respect and admire for the excellent advice and guidance he has given me throughout the entire course ofmy study.

I will always remember him for his advising, academic, and financial support at Ohio

University.

To Dr. H. Pasic, who generously served on my thesis committee. His suggestions and hearty attitude made my work easier and enjoyable.

To Dr. M. Prudich for his keenness and initiative to serve on my thesis committee.

To Len Huffman, for his excellent help and cooperation in the laboratory and mechanical workshop.

To Matt Eckels, Ben Reineck and Kyle Wilson for their collaborative help and work in the design and drafting process.

To Muhammad Turjo and Shatkat Shah for their endless support and continuous cooperation.

To my beloved parents and my family, for their love, support, and encouragement.

To Almighty Allah who has blessed me with the opportunity to come to Ohio University

and has given this chance to work with these people. ii

ABSTRACT

The emissions ofpollutants from power plants have become the subject of increasing public concern. Legislation limiting the amount ofemitted pollutants has made control ofpollutants such SOx, NOx, CO, and particulates a major concern in designing and operating coal fired power plants. Another important environmental issue "addressed here is the recent surge in landfill closure. Environmental concern, cost, and availability ofnew land are major causes for the recent escalation oflandfill costs. An alternative to the waste disposal in landfill is the solid waste incineration. Disposal ofsolid waste through incineration, while offering numerous advantages, produces numerous pollutants.

Operating and initial costs associated with the control ofthese emissions are high. As a result ofboth engineering and regulatory concerns, researchers are looking for

combustion processes with higher efficiency and improved pollutant control. Fluidized

Bed Combustion (FBC), which utilizes the phenomenon offluidization for the purpose of

more efficient combustion, has shown the potential to meet these demands at lower costs.

FBC is studied in this thesis for the possible reduction ofair emissions from coal and

solid waste combustion.

The evaluation ofdesign, development ofa fluidized bed combustor, the analysis

ofthe fluidizing phenomena, and the combustion process for an unspecified fuel are

discussed in this work. In general, the fuel must be fed into a heated bed for combustion.

The entrained particles need to be separated from the exhaust gas and unburned particles

have to be returned to the bed for complete combustion. The whole system must be iii integrated to perform the desired operations. However, the prediction ofthe particle's behavior, the rates ofmass and heat transfer, combustion reactions, and distribution of fuel and sorbent within the bed is difficult to predict accurately. Therefore, the design process requires the modeling ofthe bed in finding the key design parameters. Analytical models are needed to simulate the behavior ofthe bed in the desired operating conditions and be solved by computer codes to find the design parameters. An acrylic prototype model is also needed to verify analytical data. Agreement between experimental data and analytical predictions would indicate that the corresponding data would besuitable for design. Actual construction and equipment specifications necessary to meet constraints, both engineering and economic, are also needed to complete the design description iv

TABLE OF CONTENTS

ACKNOWLEDGEMENTS i

ABSTRACT. ·.. ·.. ·· ii

TABLE OF CONTENTS iv

LIST OF FIGURES vii

LIST OF TABLES ix

NOMENCLATURE x

CHAPTER

1. Introduction 1

1.1 Background 1

1.2 Significance 4

1.3 Combustion Process 6

1.4 Outline ofthe Work 8

2. Fluidization and Components ofthe Fluidized Bed Combustor 10

2.1 Fludization Phenomenon 10

2.2 Circulating Fludized Bed Combustor 12

2.3 Components ofFluidized Bed Combustor 13

2.3.1 Combustor 14

2.3.2 Particulate Collection Device: Cyclones 15

2.3.2.1 Cyclone In.let 17 v

2.3.2.2 Cyclone Body and Cone 17

2.3.2.3 Dust Discharge System (Hopper) 18

2.3.2.4 Cyclone Gas Outlet 19

2.3.3 Particle Fluidizing Device 19

2.3.4 Fuel Feeding Device 20

3. Particle Dynamics 21

3.1 Free Gas Stream Particle Velocity 21

3.2 Minimum Fluidization Velocity 24

3.3 Terminal Settling Velocity 25

4. Numerical Modeling ofthe Particle Behavior in the Bed 27

4.1 Effect ofTerminal Settling Velocity and Minimum FluidizationVelocity 28

4.2 Effect ofGas Velocity 29

4.3 Effect ofGas Density 28

4.4 Effect ofGas Viscosity 29

5. Heat Transfer Modeling 35

5.1 Heat Transfer Process in Fluidized Bed Combuster 35

5.1.1 Convection 36

5.1.2 Conduction 38

5.1.3 Radiation 39

5.2 Thermal Model ofthe Heat Transfer Process 39

5.3 Solution Techn.ique 43 vi

5.4 Result and Discussions 45

6. Experimental Analysis 49

6.1 Validation ofNumerical Model 49

6.2 Experimental Results 50

6.3 Discussion 51

7. Component Design 53

7.1 Combustor 54

7.2 Designing the Cyclone 56

7.3 Fluidizing Device: Blower 59

7.3.1 Bed Pressure Drop 60

7.3.2 Gas Flow Rate 62

7.4 Fuel Feeding Device 62

7.5 Ash Removal System 68

8. Conclusion 69

8.1 Summary- 69

8.2 Conclusions 70

8.3 Recommendations 71 VII

LIST OF FIGURES

Figure 2.1 Schematic Diagram ofBasic Fluidized Bed Combustor System 12

Figure 2.2 Schematic Diagram ofthe Burner Enclosure 15

Figure 2.3 Schematic Diagram ofCyclone 16

Figure 2.4 Schematic Diagram ofthe Air Introduction in the Bed 19

Figure 3.1 Direction ofForces acting on a Particle in the Bed 22

Figure 3.2 Direction ofForces acting on a Free Falling Particle 25

Figure 4.1 Terminal Settling Velocity Behavior for Different Particle 31

Figure 4.2 Particle Velocity Behavior for Different Gas Velocity 32

Figure 4.3 Particle Velocity Behavior for Different Gas Density 33

Figure 4.4 Particle Velocity Behavior for Different Gas Viscosity 34

Figure 5.1 Heat Transfer ofa Single Fuel Particle in the Bed 41

Figure 5.2 Effect ofConduction Surface Area 47

Figure. 5.3 Comparison ofDifferent Modes ofHeat Transfer in the Bed 48

Figure 6.1 Drawing ofthe Acrylic Prototype 50

Figure 7.1 Fluidized Bed Combustor Assembly 53

Figure 7.2 Combustor Detail Drawing 56

Figure 7.3 Standard Cyclone Dimensions 58

Figure 7.4 Detail Drawing ofthe Cyclone 59

Figure 7.5 Bed Pressure Drop along the Flow Path 61 viii

Figure 7.6 Geometrical Dimension ofThread 63

Figure 7.7 Detail Drawing ofScrew Thread 64

Figure 7.8 Detail Drawing ofthe Fuel Hopper 65

Figure 7.9 Detail Drawing ofFuel Feeder Assembly 66

Figure7.10 Detail Drawing ofthe Screw Feeder Assembly 67

Figure7.11 Detail Drawing ofFuel Return Line 68 IX

TABLES

Table 6.1 Comparison ofAnalytical and Experimental Data 51

Table 7.1 Total Pressure Drop in the Fluid Flow Path 60 x

NOMENCLATURE

Ab Cross sectional area ofthe bed Ap Projected area ofthe particle Aev Convective area ofthe particle Aad Conductive surface area ofparticle Arid Radiative surface area ofparticle At Cross sectional area ofthe thread a Length ofthe thread cross sectional area Bi Biot number b Width ofthe thread cross sectional area Cd Drag coefficient Cp Specific heat ofthe sphere CVe Calorific value ofcarbon Dc Cyclone body diameter Dd Diameter ofdust outlet De Diameter ofthe gas exit dm Major diameter ofthe thread dp Diameter ofthe particle Fb Buoyancy force Fd Drag force

Fg Gravitational force H Height ofthe cyclone inlet HCVr Higher calorific value ofthe fuel h Convective heat transfer coefficient hr Radiative heat transfer coefficient K Geometric configuration parameter kr Thermal conductivity ofthe gas kp Thermal conductivity ofthe particle I Distance traveled by thread in one revolution mfr Mass feed rate offuel Lb Length ofthe body Le Length ofcyclone cone me Mass ofthe carbon unburned mr Mass ofthe fuel mp Mass ofthe particle N Rpm ofthe thread Nu Nusselt number p Pitch ofthe thread L1Pb Bed pressure drop Pr Prandtl number xi

Q Fluid flow rate qConv Convective heat transfer rate ofthe bed qCond Conductive heat transfer rate ofthe bed qr Radiative heat transfer rate ofthe bed rp Radius ofthe particle S Length ofthe vortex frnder T Instantaneous particle temperature Tb Bed temperature Ti Initial temperature ofsphere To Ambient temperature Twall Wall temperature t Time Umf Minimum fluidizing velocity V Thread inside volume V p Volume ofthe particle V Volumetric flow rate offuel vg Velocity ofthe gas vp Velocity ofthe particle W Width ofthe cyclone inlet ~U Particle/Fluid relative velocity pg Density ofgas Pp Density ofthe particle Jl Gas dynamic viscosity c Stefan-Boltzman constant Em Apparent value ofemissivity ofsurface 11 co Combustion efficiency CO Angular velocity ofthe screw llc Cyclone collection efficiency 1

CHAPTER 1

Introduction

1.1 Background

An important aspect ofcombustion engineering is the efficient use offuel in the production ofelectrical power. The importance ofefficient fuel use is evident when costs, environmental concerns, and future fuel availability are considered. Fossil fuels account for almost 90% ofthe U.S. energy production, with coal accounting for over 55% ofthe fuel used to produce electrical power [1]. The supply ofcoal in the US is estimated to be

474 billions tons, which is approximately 94% ofthe proven U.S. fossil energy reserves

[1]. However, there are alternatives to coal combustion for the production ofelectricity.

The estimated recoverable domestic natural gas reserve in 1993 was 1,100 trillion cubic feet, which is equivalent to 31 years supply based on the current rate ofconsumption. In recent years natural gas has become the fuel ofchoice for new power plants because ofits lower prices, lower required capital investments and reduced environmental problems associated with combustion compared to coal. Despite its current favor, gas price projection by the Environmental Information Administration (EIA) suggests that advanced coal technology, such as fluidized bed combustor or coal-fueled integrated gasification combine-cycle, will become more economical than gas between the year

2005 and 2010 based on the projected price and the usage ofthe fuel [2]. 2

Oil is no longer considered a viable alternative to coal in producing electric power. The price ofoil is expected to rise, due to the increase in production costs coupled with progressive resource depletion. Ifcrude oil price were to fall to less than $1.40 per

Million Btu, it could replace coal in some existing boilers.

Electric generation is the single largest use ofcoal in the United States. In 1995, electric utilities consumed 82.26% oftotal coal production with 55.2% ofthe net electric power generation in U.S. [3]. Projections indicate that coal will continue to be a major source offuel for electrical power generation beyond the year 2010. New coal-fired power plants are expected to account for 25% (42 GW) ofall new generation through

2010. Estimates ofcoal's share ofthe total power generation in the year 2010 will be 45­

58% [2]. This additional power generation does not necessarily require the construction ofnew power plants. Repowering or retrofitting can improve the efficiency and increase the generating capacity. Thermal efficiency ofmost coal-fired electrical power generation is only 33%. Thus, 67% ofthe thermal energy ofthe fuel is wasted per every ton offuel burned. Any process that will provide higher utilization ofthe fuel burned in an electric­ power generating station will result in larger fuel savings. This concern initiated the research work for finding fuel-efficient coal combustion technology.

Attempts to bum coal in gas turbine combustors started in the late 1940s, and started again in the 1970s. These attempts were based on the conventional coal combustion technology, such as pulverized coal combustion process. However, these attempts resulted in failure due to the problems associated with erosion and ash deposition on the turbine blade at the high combustion temperatures. The researchers 3 concluded that direct coal firing ofgas turbines could not be seriously considered until there was a radical new development in coal combustion technology.

Other problems associated with current coal combustion technology are air polluting emissions and high initial costs for their emission controls. Control ofSOx,

NOx, C02, particulate, and other air toxins are a determining factor in design and operation ofcoal-based power plants. New state and federal legislation are posing more restriction on power plant emissions. Coal-fired power plants produce 70% ofthe total

SOx in the U.S. and are under mandate to significantly reduce these emissions [4]. Global warming is also a major concern due to the CO2emissions from the coal combustion.

Therefore, utilities are now seeking inexpensive, easy-to-implement emission control techniques.

There are other aspects ofenvironmental concern that can be addressed by a more efficient process. One example would be the growing problem ofsolid waste disposal.

Landfilling is no longer a lucrative solution for waste management. Environmental regulations, increasing costs associated with siting new landfills, and lack ofavailable space continues to increase the costs associated with landfill waste [5]. As a result, communities have sought alternatives for waste dumping. Combustion ofwaste fuels, such as shredded tires, municipal solid waste, and even cow manure, is restrained by environmental and other factors. Waste combustion produces varying concentrations of hazardous and polluting gases which often results in incomplete combustion (due to the poor combustion characteristics oflow-grade fuels) which in turn creates hazardous wastes, such as 4

• Acid gases including SOx, HC1, HF and NOx • Major gaseous products ofincomplete combustion: CO, THC • Heavy metals in various chemical forms, including Pb, Cd, Ti, As, Co, Ni, Se, Re, Sb, Cr, Cu • Polyaromatic Hydrocarbons • Polychlorinated Biphenyls • Polychlorinated Dibenzodioxins and Polychlorinated Dibenzofurans [6].

Moreover, the variable composition and high moisture content ofthe waste stream contribute more combustion difficulties, including reduced flame stability, lower levels of carbon burnout, and reduced combustion enthalpy [6].

Fluidized bed combustion (FBC) holds much promise for both reducing the amount ofmunicipal solid waste and improving emissions from coal combustion. Compared to traditional forms ofcombustion (including pulverized coal combustion for coal and rotary kiln incineration for waste), fluidized bed combustors produce lower levels of unburned carbon in the ash, use no moving parts, reduce formation ofmost airborne pollutants, and attain high combustion efficiency at temperatures below the sintering point ofmost ash. Fluidized bed combustion can also be the most viable means ofwaste management, as it can be used for a variety ofwastes and is cost effective compared to other disposal methods.

1.2 Significance

Because fluidized beds can gain much better combustion through improved mixing, FBC's offer the promise ofhigher combustion efficiencies, lower C02 emissions, and few gaseous pollutants per pound offuel burned. Fluidized bed combustion offers 5 significant advantages in terms ofelectrical power production. The relatively low, uniform combustion temperature (1050 K-1300 K) ofFBC provides a desirable heat source for gas turbines. Lower temperature gas streams make particulate removal less formidable and are better suited to turbine blading. Several systems have successfully used both pressurized and atmospheric fluidized bed combustion with gas turbine electrical energy production [7]. The use ofgas turbines to generate electricity is projected to become more widespread, as thermal efficiencies for new gas turbines systems have been reported to be as high as 55% to 60% [8].

Fluidized bed combustion ofcoal also offers the advantages ofexcellent mixing due to the turbulent swirl. Not only does the mixing promote a higher completion of combustion, it can be used to reduce SOx emission. The high degree ofmixing between the gas and solid phases in the bed promotes a high degree ofsulfur capture [9].

Experiments conducted by American Electric Power on the pressurized fluidized bed unit at Tidd (Brilliant, Ohio) determined 95% sulfur capture was possible for a sorbent to sulfur ratio (Ca:S on a molar basis) of 1.4 to 1.0 [8]. It was found that it not only eliminates the cost ofexpensive wet scrubbers but also reduces operating costs.

Absence ofcoal pulverization equipment results in lower capital and operating costs. Fluidized beds can be designed to incorporate the boiler within the bed. Such a design increases volumetric heat-transfer rates 10 to 15 times and the surface heat transfer rates by 2 to 3 times over a conventional boiler [10]. Due to less S02 in the stack, the gas temperature can be lowered, and this results in increases in overall plant efficiency. 6

Combustion ofwastes in fluidized beds is a relatively new technology, with many ofits scientific methods adapted from the fluidized bed combustion ofcoal. Several studies documented by Saxon and Jotshi have shown that fluidized bed combustion offers several advantages for solid waste incineration [6]. For example, a wide variety oflow­ calorific fuels can be combusted under more stable conditions in fluidized beds. Many of the problems associated with the combustion oflow-grade fuels, such as solid waste, can be addressed by blending with higher-grade fuels [11] or reburning via combustion with natural gas [12]. Fuel blending and sorbent addition to the combustion bed, can influence the characteristics ofthe combustion products and thus, influence the formation of harmful emissions. Further, co-firing coal with natural gas in a drop furnace was shown to increase the transient conversion of802 to S03' thus increasing the reactivity ofthe sulfur gas with the sorbent material in the ash [13].

1.3 Combustion Process

Combustion refers to a rapid chemical reaction between substances with the evolution ofheat and flame. As the temperature ofthe combustion material achieves the ignition point at a specific pressure, combustion begins. This chemical process spreads from the ignition source to adjacent layers. Each burning layer acts as an ignition layer and propagates to the next layer and so on until an equilibrium is attained between the total heat energies ofthe reactants and the total heat energies ofthe products. In addition to the chemical process, mass and energy also transfer in the combustion process. The standard equation ofstoichiometric combustion process with air can be represented by 7

(1.1)

The major components ofcombustion are

Fuel: Normally an energy enriched organic hydrocarbon consisting oflong chain

carbon-hydrogen and carbon-carbon bonds. In the combustion process, these bond

structures are broken up and heat energy is released. Fuels are available in nature

in solid, liquid, or gaseous form.

Oxidant: A substance that aids combustion by breaking the C-C or C-H bonds in

fuel and releasing the chemical energy as heat. The most common example ofan

oxidant is oxygen. Ambient air, containing approximately 21% oxygen (by

volume) is used as the oxidizer.

Products: The reaction products ofthe combustion process are collectively called

combustion products. The combustion products are: (1) oxidized products of

carbon-carbon and carbon-hydrogen bond in the fuels, are namely CO2and H20.

In case ofincomplete combustion, CO and other partially oxidized substances

may be formed; (2) oxidized products ofthe other substances in the fuel, i.e., S02,

S03, P20S, etc. and (3) excess oxygen and nitrogen gas.

Solid fuel combustion proceeds first with devolatilization then remained char is

oxidized. When the raw fuel is heated in an inert or oxidizing environment, the moisture

ofthe fuel will evolve and the particles ofthe fuel will undergo a transformation begins

which may result in the softening ofthe fuel. At further elevated temperatures, gases and

heavy tarry substances are emitted in a process known as pyrolysis. The mass evolved 8 during pyrolysis can vary from a few percent up to 70-80% ofthe total particle weight and may last from a few seconds to several minutes, depending on the composition ofthe fuel. These reactions are considered to be homogenous reactions. The residual particle, known as char, contains hydrogen, carbon, nitrogen, sulfur, and mineral matter.

The residual is then oxidized at higher temperatures with the direct contact of02 , and this reaction is considered to be heterogeneous with gaseous 02 diffusion to and into particles absorbing and reacting on the surface ofthe char. This process is much slower than devolatilization. Other reactants, such as H2, CO2, or steam will also react with the residual but at a much slower speed.

1.4 Outline ofthe Work

This thesis consists ofeight chapters and relevant appendices. Chapter Two presents the fundamentals ofthe fluidization phenomena, operation ofthe fluidized bed combustor, and the basic components ofthe fluidized bed combustor. Chapter Three discusses the particle dynamics ofthe bed. It defines different particle velocities ofthe bed mathematically by analyzing the dynamics ofthe bed. Chapter Four illustrates the numerical models ofthe particle velocity behavior in the bed, which simulates the particle velocity behavior for different boundary conditions and explains these by graphical presentations. In Chapter Five a numerical model ofthe heat transfer process in the bed is explained. The model is solved by finite difference method and the results are discussed by the aid ofgraphing. 9

Chapter Six discusses the experimental analysis, explaining the experimental work performed to validate the numerical models. Chapter Seven describes the design work of the fluidized bed combustor. The design process is analyzed in detail. Chapter Eight concludes with a summary and outlines recommendations for future work. 10

CHAPTER 2

Fluidization and Components ofthe Fluidized Bed Combustor

The fluidized bed combustion process consists ofa various number ofinterrelated operations. In this chapter, the basic principles ofthese interrelated operations are described. Fluidizing operation is introduced first, and then basic fluidized combustion processes and the advantages ofcirculating fluidized bed are evaluated. Finally, a

detailed description ofeach component ofthe combustor is discussed.

2.1 Fluidization Phenomenon

Fluidization can be visualized by considering a cylindrical vessel partially filled

with solid material resting on a perforated distribution plate. Gas is forced to pass through

the bottom ofthe distribution plate. As the gas passes upwardly through the bed, particles

offer resistance to fluid flow. As the fluid velocity further increases at a certain point,

particles rearrange themselves to offer less resistance to flow and lift the particles ofthe

bed. The fluidization operation can be considered in several steps. First, at low gas

velocity, gas flows through the interstitial space between the particle [14]. Though the

bed acts as a porous medium, the particles still remain stationary at this gas velocity and

pressure drop rises with the increase offlow rate. With an increase in gas velocity,

particles gain fluid-like properties, and this stage is known as incipient fluidization. Here

pressure drop reaches its maximum value. The corresponding gas velocity is referred to

as the minimum fluidization velocity (umr). Beyond this velocity, the pressure drop across 11 the bed is approximately equal to the weight ofthe bed per unit area. This correlation can be shown mathematically

(2.1)

The increase in pressure drop across the bed depends upon how firmly the bed is packed before any fluid passed through it. At incipient fluidization, extra pressure is required to unlock the particles from their pattern ofpacking. Once they have been released from this pattern, the pressure drop then falls back to a value sufficient to support the weight ofthe particles.

With a further increase in velocity, the bed tends to expand in a stable fashion until the particles are thrown from the bed. The uniform expansion behavior is then lost.

Here instabilities develop and up-moving bubbles appear. The velocity at this point is called minimum bubbling velocity.

A further increase in gas velocity produces a slug in the form oflong, narrow columns that result from bubbles coalescing and the cavities occupying the whole cross- section. At higher velocities, much ofthe material is carried away from the bed. Pressure fluctuation is generated due to bubble bursting, and a spray ofparticles is thrown into the freeboard above the bed. The amount ofparticles that are carried away in the gas stream is related to the particle size distribution and the gas velocity in the freeboard region above the bed. This regime is called fast fluidization or lean-phase fluidization, and pneumatic conveying is attained. Now fuel can be injected into the bed where combustion 12 takes place with the fluidizing media. The uprising velocity ensures rapid and relatively uniform mixing ofthe injected fuel material within the bed.

2.2 Circulating Fluidized Bed Combustor

There are two main types offluidized bed combustion processes. The circulating fluidized bed combustor (CFBC) has the advantage over the bubbling bed combustor due to its high fluidization gas velocity and higher solid fuel handling capacity. It is, therefore, a more efficient and viable technology for the combustion oflow-grade solid fuel. The term 'circulating' means that the solid particles are returned to the fluidized bed by way ofa solid separator. In circulating beds, air velocities are much greater, and the solids are blown out ofthe combustor, separated in a cyclone, and returned to the combustion chamber. In Figure 2.1 the components ofcirculating fluidized bed combustor are shown schematically.

To Cyclone ~-- SeporotOt~

FI ..'I?I Co(V)bustOt~ Hopper~

Figure 2.1: Schematic diagram ofbasic fluidized bed combustor system 13

Typical air velocities in bubbling beds are usually in the range of0.9-3.1 m/s, and for circulating beds air velocities are between 10-30m/s. Typical bed temperatures for both types are in the range of 1000-1200 K. A pilot gas burner is located underneath the combustor unit. As the air passes though the gas flame, it is heated. Sand is heated to the particle ignition temperature by this flowing air. Fuel is then injected into the bed. The large amount ofsand within the furnace is a heat sink that provides significant thermal inertia within the system. This allows the furnace to shut down with minimal heat loss.

Heat is transferred from the sand and other bed material into the fuel. High turbulent mixing ofsand with the fuel helps to obtain a higher rate ofheat transfer between sand and fuel. Continuous bed agitation allows larger particles to break up and to remain in the combustion region ofthe bed until combustion is completed. Because of the high turbulent motion ofthe solid, combustion takes place throughout the combustor, which creates a uniform and low bed temperature profile. Bed temperature is maintained below the melting temperature ofsand to avoid agglomeration ofthe particles. This helps to avoid the ash agglomeration problem (slag) that is generally encountered in other types ofcombustors.

2.3 Components of Fluidized Bed Combustor

From the preceding discussion it can be seen that the most perceived parameter dictating whether a bed is fluidized or packed is the velocity at which the fluid passes upwards through an unrestrained bed ofparticles. Both types ofbeds require a containing vessel with a porous base through which the fluid can be introduced to the bed. This 14 porous base is called a 'distributor'. The containing vessel must be extended to a sufficient height above the free surface ofthe bed to allow space for particles which can be carried upwards from the free surface to disengage from the fluid stream and fall back to the bed. The outgoing exhaust gas contains a significant amount ofunburned particles and ashes and needs to be cleaned. Therefore, it is necessary to add some kind ofgas cleaning device into the system, such as a cyclone separator. Fluidizing the bed also requires a fan or compressor to act as a fluidizing device. Next the important component ofthe system is the fuel feeder. As the bed itselfis at positive pressure, it requires a positive displacement fuel feeder or pressurized injection with variable feed rate options.

In general, a fluidized bed combustor necessarily consists ofthe following components:

• Combustor • Particulate collection device • Fluidizing device • Fuel feeding device

2.3.1 Combustor

The combustor is a cylindrical refractory-lined vertical shell. The material used for the vessel must have high melting point and low thermal conductivity to minimize heat loss.

In general, stainless steel is used with insulating ceramic layer on the outside wall. An additional layer can be added to the inner side ofthe combustor vessel, which in turn acts as an erosion resistance layer.

A pilot burner (see Figure2.2) is placed inside a funnel enclosure at the bottom ofthe combustor. Flowing air is heated up to the ignition temperature ofthe fuel by the pilot 15 gas burner. Heat is then transferred from the primary air to the sand, which acts as the heat sink for the ignition ofthe fuel.

Funnel type enclosure

Funnel enclosure height

Gas flow line

Figure 2.2: Schematic diagram ofthe burner enclosure

2.3.2 Particulate Collection Device: Cyclones

A significant amount ofunburned fuel and ashes exit the combustor with the flue gas. Therefore, it is required to separate the unburned fuel from the exhaust gas and return it to the combustor for complete combustion. Cyclone separators are widely used as a collection device in recirculating fluidized bed systems. Cyclone employs a centrifugal force to separate particles from a gas stream. The device is relatively simple in design and construction and has no moving parts. It is a relatively an inexpensive collector both in construction and operating costs. By using suitable methods and 16 materials, it may be adapted for extreme operating conditions, such as high temperature, high pressure, and corrosive gases.

The basic separation principle ofcyclone is relatively simple. Particles enter the cyclone with the flowing gas. The gas stream is forced to tum, but the larger particles have too much momentum and cannot make the tum (see Figure 2.3). These particles collide with the cyclone wall, slide down the wall, and are collected in a hopper.

Depending on the inlet velocity, particle and gas characteristics, etc., the gas stream actually turns a number oftimes in a helical pattern. Heavier particles cannot make the turns. Thus these particles will break through the streamlines and strike the wall.

=wlt~1 f(iq POt,tlcles

o t t? Cj F'o r tic I E ~. E(I thE t(11 ------

Figure 2.3 Schematic diagram ofcyclone 17

There are four major parts to a cyclone: the inlet, the cyclone body, gas outlet, and cone. Each part affects the overall efficiency ofthe cyclone. Each component is discussed in more detail in the following subsections.

2.3.2.1 Cyclone Inlet

The cyclone inlet accelerates the gas in the cyclone to attain the required tangential velocity. The shape ofthe inlet assists the transformation ofthe incoming gas from a linear flow to a circular vortex pattern. The gas is then accelerated and combined with the already rotating gas in the cyclone. The pressure drop is increased, and more power is required to move the gas through the system. Inlet length and width are also important. As the inlet is made smaller, the inlet velocity increases. This gives higher removal efficiency at the expense ofadded pressure drop. A poorly designed inlet will create turbulence, which will ultimately decelerate the tangential gas velocity and in turn lead to the development ofthe flow eddies at the inlet. So the pressure drop will increase substantially and collection efficiency will decrease.

2.3.2.2 Cyclone Body and Cone

The removal efficiency ofa cyclone for a given particle size depends on cyclone dimensions. The diameter ofthe cyclone has a large effect on the pressure drop for a given volumetric flow rate. The relationship ofbody diameter to collection efficiency is illustrated in the classic cyclone design equation ofLeith and Licht, which is discussed in 18 the cyclone design in Chapter Seven. The overall length determines the number ofturns ofthe vortex; as the number ofturns increase, the removal efficiency increases.

The motion ofa gas in a cyclone is not simple. Two vortices are formed - one descending and the other ascending. The descending vortex often is called the main vortex. For a properly designed cyclone, the vortex will change direction at the bottom of the cone and start ascending. This ascending vortex is smaller in radius with faster tangential velocities than the descending vortex.

The cone primarily serves as a mechanism for removing particulate matter from the walls ofthe cyclone and sending it to the hopper. However, the vortex formed in a

cyclone sometimes deviates from the vertical axis. Because ofthis, it has been found that

bottom ofthe cone should have a diameter ofat least ~ th ofthe cyclone diameter.

Otherwise, the outer vortex may touch the cone wall entraining already captured particles

in the ascending vortex.

2.3.2.3 Dust Discharge System (Hopper)

Ifthe discharge bin is immediately below the cone and nothing is added to the

bottom ofthe cone to arrest the vortex, the vortex will extend into the discharge bin. Dust

can be reentrained from the hopper into the vortex. Ifleaks exist in the bin, dust can be

sucked back up into the cyclone. In general, any kind ofash removal system is attached

to the bottom ofthe cone to avoid particulate reentrainment. 19

2.3.2.4 Cyclone Gas Outlet

Another important consideration for the design ofany cyclone is the exit tube.

The length ofthe tube must extend beyond the inlet so that eddies created in the annulus between the tube and cyclone walls do not mix and particulate can move up and exit the tube.

2.3.3 Particle Fluidizing Device

Blowers are widely used for fluidizing the particles. Air is introduced in the combustor in two positions (see Figure 2.4). Primary air is introduced at the bottom ofthe combustor through a distributor plate. Secondary air enters between the lower reducing zone and the upper oxidizing zone through a set ofnozzles. To ensure good mixing and to prevent wall impingement, nozzles are usually placed in such a way that they are diametrically opposed to each other. Exhaust gas

Combustor Secondary Cyclone air fuel return

Air Blower blow line Pri marv air

Figure 2.4 Schematic diagram ofthe air introduction in the bed 20

2.3.4 Fuel Feeding Device

Type and composition ofthe fuel plays an important role in the design consideration ofthe combustor. Ifthe solid fuel is a heterogeneous mixture, such as municipal waste, and has a relatively low heating value « 8000 Btu/lb.), the shredding, sorting, and drying operation will be more complex. Waste streams containing glass, metal, and plastics make the combustion difficult; fouls heat transfer surfaces, and produces several hazardous gases. Some ofthe combustion process may require incineration with co-fired coal or gas. The products ofcombustion will also be more complex and include various concentrations ofhazardous and polluting gases, such as

SOx, NOx , Pb, Cd, Ni, etc. Therefore, it will require extra effort to clean up these gases from the exhaust. Ifthe fuel is homogenous it will facilitate overall system design, and feeding will be much easier.

Feeding ofthe fuel into the combustor requires efficient design efforts. For

recirculating fluidized bed combustor fuel must be forced into the combustor vessel.

Gravitational forces are usually insufficient to complete this process due to positive

pressure in the vessel. A positive displacement screw feeder is a good selection for the

fuel feeding, as it is consistent with the distribution ofmaterial. Hard piping and positive

pressure in the fuel container are required to prevent feedback ofthe fuel particles from

the bed. Therefore, in the opening ofthe bed there must be some valve arrangement,

which will ensure positive flow from the feeder to the bed. 21

CHAPTER 3

Particle Dynamics

Particle properties and shapes play an important role in the fluidization process.

Different particle configurations and densities result from different pressure drops in the bed; thus the bed experiences different velocities ofthe gas and particles. Therefore, particle properties and shapes should be seriously considered while modeling the particle fluidizing process. The important characteristics ofthe particles include the particle's size and size distribution, shape, density, viscosity, reactivity, toxicity, the particle drag force, and the terminal settling velocity.

3.1 Free Gas Stream Particle Velocity

Particle velocity in the bed depends upon the different parameters that incorporate the force balance equation ofthe particle flowing through the bed. For calculating the particle velocity in free gas stream, an equation ofmotion was developed for a particle

flowing through a rapidly accelerating fluid stream in the combustor. By applying,

Newton's second law ofthe motion on the forces acting on the particle and assuming

gravitational, buoyancy and drag are only forces acting on the particle, then

dv m .:». =t+"F (3.1) p dt L...J I

(3.2) 22

Gas velocity Particle velocity

Figure 3.1 Direction offorces acting on a particle in the bed

Now particle drag force acts on the particle due to the particles' motion relative to the fluid surrounding it. The drag force exerted by stagnant gas on a moving, isolated particle is given by [15]

(3.3)

Buoyancy force acts opposite to drag force; i.e., along the fluid motion, and can be expressed as[15] (3.4)

Gravitational force is expressed as [15]

(3.5)

So Equation 3.1 can be written as 23

(3.6)

Considering the following assumptions

• All the particles are spherical • Flow is steady • Velocity ofthe particle is constant

Equation 3.6 can be simplified as

(3.7)

and the mass ofthe particle ofa sphere can be written as

(3.8)

Replacing the value ofmass into the equations

(3.9)

(3.10)

(3.11)

Cd is an experimentally determined drag coefficient, and it is a function ofthe particle's

Reynolds number. Reynolds number is expressed as

(3.12) 24

Many correlation have been proposed to approximate the Cd-Re relation - one typical case is [15]

24 C -­ (Re < 0.1) (3.13) d - Re

24 (3 9 2 ) c, = Re 1+ 16 Re+ 160 Re *log(2Re) (0.1< Re < 3) (3.14)

(3 < Re < 500) (3.15)

C, = 0.44 (Re> 500) (3.16)

Plugging these values in Equation 3.7 yields

Rearranging the above equation and dividing by 1"( d;pgg , both sides yields 6

4(p - Pg)gd p p (3.18) 3CdPg

3.2 Minimum Fluidization Velocity

Minimum Fluidization Velocity (MFV) is the velocity at which fluidization starts.

At this stage, particle velocity becomes zero. MFV determines the lower limit ofthe

operating gas velocity for any fixed particle. Thus, it sets the minimum flow rate ofthe

blower for a given pressure drops ofthe bed. Setting vp = 0 to Equation (3.18) we have 25

4(pp - Pg)gdp vg = (3.19) 3CdPp

3.3 Terminal Settling Velocity

Terminal settling velocity (TSV) is the velocity reached by a free-falling particle in a stagnant medium. The gas velocity, which is higher than the TSV, will carry out the particle from the bed.

Particle velocity Gas velocity

a Figure 3.2: Direction of forces acting on a free falling particle

For the derivation ofthe equation for terminal settling velocity, the forces acting on a single free-falling particle was balanced accordingly:

dv m .:». =i+~F (3.20) p dt L.J I

(3.21) 26

(3.22)

dv p For steady-state conditions, -;jf=O.

(3.23)

Plugging the value ofFs, Fg, Fd and simplifying yields

4(p - Pg)gd p p (3.24) 3CdPg

For TSV vg=0 . So Equations (3.23) yields as

4(pp - Pg)gdp vp = (3.25) 3CdPg

The value ofparticle density, particle diameter, gas density, gas viscosity, and gas temperature must be known in order to calculate terminal settling and minimum fluidization velocity. These velocities are different for different boundary conditions.

Therefore, to find the velocities for different boundary conditions, it was required to simulate each set ofboundary conditions in the numerical model. 27

CHAPTER 4

Numerical Modeling ofthe Particle Behavior in the Bed

To optimize the design process and to predict design parameters for a broad range ofoperating conditions, it is necessary mathematically, to model particle behavior in the combustor. Particle velocity in the bed depends upon many parameters, such as particle diameter, gas velocity, particle density, gas density and viscosity, and gas temperature.

To simulate particle velocity behavior in the bed, numerical models were developed by balancing the forces acting on a particle. These models were developed based on numerous assumptions including ignoring the pressure drop and the frictional losses.

Computer programs were developed in C for solving the differential equations ofthe models.

These programs are included in the appendices. Particle velocity behavior was

analyzed as a function ofthe following parameters

• Particle diameter • Superficial gas velocity • Gas viscosity • Particle density • Gas density

For each case, terminal settling velocity and minimum fluidizing velocity were

calculated using the model. Experimental analysis was done for comparison with the

predicted data. 28

4.1 Effect ofTerminal Settling Velocity and Minimum Fluidization Velocity

Assuming particle density, gas density and viscosity, and gas temperature are constant, terminal settling velocities for different particle diameters were calculated by the model. The behavior ofthe terminal settling velocity is shown graphically in Figure

4.1. In general, the larger the particle diameter, the higher the terminal settling velocity.

Minimum fluidization velocity is the velocity at which fluidization starts. So it limits the lower limit ofthe gas velocity. By comparing Equation 3.19 and 3.25 it can be said that absolute values ofterminal settling velocity and minimum fluidizing velocity are the same. Therefore both velocity profile are identical.

4.2 Effect of Gas Velocity

In Figure 4.2, assuming particle density, gas density and viscosity, and temperature

are constant, particle velocity behavior for gas velocities of 18 mls and 32 mls are

represented. In the model gas velocity was varied between 12m1s to 42m1s, as in this

range all testing particles become fluidized in our experiment. The results indicate that a

higher gas velocity leads to a higher particle velocity for the same diameter particle.

From Figure 4.2 it also can be shown that there are sharp transitions in the velocity

profile. This are due to the transition ofCd in the Re-C, correlation for the corresponding

particle diameters. 29

4.3 Effect ofGas Density

In Figure 4.3 assuming particle density, gas velocity, viscosity, and temperature

are constant, the effects ofgas density are shown in a graph ofparticle velocity versus particle diameter. Particle velocities are calculated by the model for different diameters

by considering gas densities as 1.142 and 0.353 kg/nr', A higher gas density leads to a

higher Reynolds number and results in a higher drag coefficient Cd.

Therefore, the particle velocity increases with an increase in gas density. From the

Figure 4.3, it also can be shown that for both gas densities there are sharp transitions in

the velocity profile. This is the same fact as it was discussed earlier i.e, due the transition

in the Re-C, correlation. Gas density is dependent on gas temperature. The higher the

temperature, the lower the gas density. Therefore, at higher temperature particle

velocities are lower for a constant gas velocity. However, the lower the gas density, the

higher the gas velocity, and a higher gas velocity leads to a higher particle velocity.

4.4 Effect ofGas Viscosity

In Figure 4.4, by considering particle density, gas density and temperature are

constant, particle velocities are shown for a range ofdp at two different viscosities,

4.65xl0-5 and 1.8x 10-5 N/m-s. Particle drag force is one ofthe determining factors in

calculating particle velocity. It is a function ofthe drag coefficient, Cd, which is in turn, is

a function ofReynolds number. Again, Reynolds number is a function ofgas viscosity.

Hence, a higher gas viscosity leads to a lower Reynolds number. From the empirical

correlation it can be shown that this results in a lower drag coefficient. Therefore, at a 30 higher gas viscosity, particles experience a higher particle velocity for the same gas velocity (see Figure 4.4). As the combustion begins, the temperature ofthe bed rises sharply, which results in an increase in gas viscosity. Therefore, at an elevated gas temperature, the particle velocity increases from the effect ofgas viscosity. 70 T'------

60

~ fI.} 50 ! ~ u 40 e ~ ~30 U ~ ~= 20 - Terminal settling velocity(m/s) 10

o ~ ' 0.0000 0.0030 0.0060 0.0090 0.0120 0.0150 0.0180 Particle Diameter(m)

Figure 4.1 Terminal Settling Velocity Behavior for Different Particles w ..... 35 -r-i ------

30 -+- Gas velocity at 18m/s ~ 25 ] Gas velocity at 32m/s '-'" --.- C .-u 20 o ~ > ~~ 15 u ~ ... ~= 10

5

o , --..p , o 0.002 0.004 0.006 0.008 0.01 0.012 0.014 Particle Diameter(m)

Figure 4.2 Particle Velocity Behavior for Different Gas Velocity w tv 35

30

~ 25 ~ a'-" - Gas density(1.142 kg.m/S) Eu 20 .....o ~ --..- Gas density(O.353 kg/m''S) > ~ u 15 ! ~= 10

5

o -l------o 0.0021 0.0042 0.0063 0.0084 0.0105 0.0126 0.0147 0.0168 Particle Diameter(m)

Figure 4.3 Particle Velocoity Behavior for Different Gas Density VJ VJ 35 ~,------

30

,.-...25 rIJ - Gas viscosity (4.68e-5 N/m-s) 5 ...... --- Gas viscosity (1.8e-5 N/m-s) £u 20 ....e >~ ....~ 15 .,...u 1: ~= 10

5

o , , J , o 0.0005 0.001 0.0015 0.002 0.0025 0.003 0.0035 0.004 0.0045 0.005 Particle Daimeter(m)

Figure 4.4 Particle Velocity Behavior for Different Gas Viscosity VJ ~ 35

CHAPTERS

Heat Transfer Modeling

In this chapter, an analytical expression for the transient heat-transfer ofa particle in the bed is derived, which is used to determine particle temperature as a function of time. The model is then solved by a Finite Difference Analysis (FDA) method. To show the effect ofthe conduction and radiation mode in the local heat-transfer process explicitly, the problem is solved in three cases: one without the conduction mode, one without the radiation and one with or without all three modes (conduction, convection and radiation). The solutions ofthese analytical models are discussed with the aid ofa graphical presentation.

5.1 Heat Transfer Process in Fluidized Bed Combustor

Understanding the heat-transfer mechanism in a fluidized bed combustor is important due to its influence on the required bed surface area, bed height, and fuel feed rate. The prediction ofthe particle temperatures is needed to calculate the burning rate of the fuel and the production rate ofpollutant gases, such as NOx, CO and SOx. Even though considerable analytical and experimental research have been conducted, few of them represent the local heat-transfer ofa single solid particle accounting for effect of direct environment [16].

The heat-transfer in a fluidized bed is an extremely complex phenomenon due to the wide variety ofconditions and regimes. In a circulating fluidized combustor, heat- 36 transfer involves interaction ofconduction in and between particles and convection in moving fluid, and radiation. The relative significance ofthese modes ofthe heat-transfer process depend on the combination offluid temperature, flow pattern ofthe fluid and solids, surface area ofthe particles, wall surface area ofthe combustor and concentration, and size distribution ofparticles suspended in the gas. The spatial and temporal variations ofthe thermal conditions at the solid-fluid interface are not always known, as they are a function oftime ofthe thermal boundary conditions ofthe adjacent solid surfaces. To fmd the transient temperature distribution and to measure the significance ofall modes of the heat transfer process, it was necessary to analyze the energy equations at every instance oftime. This model approximates particle behavior as perfect spheres. As the heat-transfer process is very complex in form and is very difficult to simulate, the model was developed with some approximate boundary conditions. The model was simplified so that the contributions ofthe different variables in the heat-transfer process could be represented in a more straightforward manner.

5.1.1 Convection

Convection is the mode ofheat-transfer, which takes place in a moving fluid due to the combination ofconduction and energy transfer ofthe flowing fluid [17]. The fluid velocity must be known for calculating the convective heat-transfer coefficient. Fluid boundary layer has an important contribution to the convective heat-transfer process due to the viscous property ofthe fluid. The relative thickness ofthe thermal boundary layer 37 and the velocity boundary layer depend on the magnitude ofthe Prandtl number ofthe liquid. The Prandtl number represents the relative importance ofmomentum and energy transport by diffusion process. Prandtl number can be expressed as [17] ep Pr=-P- (5.1) kf

Because ofthe low thermal conductivity ofthe gas, heat energy diffusion is much greater than the momentum diffusion. On the other hand, the high gas velocity region near the surface ofthe particle, in which most ofthe velocity varies, is considered to be quite thin. Therefore, this simplification ofignoring the boundary layer influence in the heat-transfer process is quite reasonable.

Deriving an equation for the convective heat-transfer process involves the principles ofheat conduction, fluid dynamics, and boundary layer theory. But these complexities may be lumped together by the introduction ofNewton's law ofcooling

[17]

(5.2)

The term h, which is known as convective heat-transfer coefficient or film coefficient, is a complex function ofthe fluid properties and geometry ofthe solid and the hydrodynamics ofthe fluid motion. Many researchers had developed many correlation for calculating the value ofh; none are regarded as universal due to the complexities of the bed heat-transfer process. The correlation developed by Ranz for spherical surface was chosen to apply the typical particle surface. According to Ranz, the heat-transfer 38

coefficient (h) at the surface ofthe sphere ofdiameter (dp ) passing at velocity (vp ) through a gas follows[18]

( 5.3)

hd Nu = _P- = 0.03 Re~3 Rep <100 (5.4) kf

The value ofthe film coefficient (h) can be found from the above equations.

5.1.2 Conduction

The conductive heat-transfer mechanism in the fluidized bed combustor is complicated by the bed's complex boundary and initial conditions and by the varying nature ofthe conduction path. These boundary and initial conditions are described later in this chapter. Conductive heat-transfer process in the bed is described by a multi- directional heat diffusion equation with appropriate boundary and initial conditions.

However, conduction was considered unidirectional only by assuming that heat is transferred in the radial direction only in and between particles.

Therefore, heat flux qcond within the particle is given by the well-known Fourier equation[17]

(5.5)

For a small distance the above equation can be simplified as

(5.6) 39

5.1.3 Radiation

The fundamental law ofheat transfer process is the fact that temperature ofa body, hotter than the surrounding, tends to decrease with time. Physical medium is not always necessary for the transports ofthe thermal energy. Heat may also be transferred by the another mechanism without any aid ofmedium, and it is called radiation. When radiant energy ofa hotter body strikes a cooler body, energy is absorbed in the form of thermal internal energy. The radiant heat transfer from a surface, qr, is expressed as [17]

(5.7)

Radiative heat-transfer coefficient can be expressed as [17]

(5.8)

The above relations were used to estimate the radiative heat transfer from the wall to the particles. Radiation mechanism was treated as independent as it was assumed that hot gas surrounding the particles is radiatively nonabsorbing and nonemitting,

5.2 Thermal Model of the Heat Transfer Process

The development ofa heat transfer model for fluidized bed combustor is complex in form and very difficult to include all the affecting components in the model. It is quite impossible to simulate the heat transfer process ofthe bed without extensive experimental work. Instead this model was developed to examine the transient temperature distribution ofthe bed before the particles are totally burnt out and to see the influence ofeach heat 40 transfer mode in the bed heat transfer process. The model was simplified by assuming the following conditions

• Fluid flow is constant • Bed temperature is uniform over the combustor • Physical properties offuel and fluid are constant • Fuel particles are surrounded either by the sand or by the other fuel particles • Solid feed ratio is constant • The bed is well insulated so that there is no heat-transfer from or to the bed • All particles have the same mean diameter (dp), and heat-transfer obeys the correlation developed by Ranz • Fluid is nonemitting and nonabsorbing • Convective and radiative area are same • Temperature ofthe wall and bed are same

It is perhaps simplest to consider that a fuel particle is being brought into the bed, residing there, and then being replaced by another fresh particle. The combustion chamber is maintained at a fixed temperature level (650-900 °C). The fuel particle was initially at a uniform room temperature, T, and suddenly exposed to the bed temperature,

Tb. Ifthe particle enters the bed at t=O, the temperature ofthe solid particle will increase for t>O until it reaches the bed temperature, Tb. Heat is transferred by convection by the flowing hot gas at the particle surface-gas interface, and by radiation between wall and particles. From the Figure 5.1 it can be shown that after entering into the bed the particle itselfis surrounded by the hot particles (mainly sand and existing fuel particles) and heat must also be transferred by conduction from the surrounding hot particles to the new fuel particle. Therefore, in the heat balance equation necessarily it was required to add the conduction term. 41

Hot gas

Hot gas

Surrounding sand particle Particle

Figure 5.1 Heat Transfer of a Single Fuel Particle in the Bed

The meaning ofthe Biot number less than unity is the assumption that the temperature ofthe particle is spatially uniform at any instant during the transient heat- transfer process. This implies the existence ofinfmite thermal conductivity and this type ofassumption is quite impossible. However, this condition can be closely approximated ifthe conduction resistance ofthe particle is small compared to the film resistance. In neglecting the conduction resistance, it can no longer be considered the heat-transfer process within the framework ofFourier's heat equation.

To determine the relative effect ofconduction, the Biot number was calculated by considering the following operating conditions:

dp = 0.008 meter kp= 0.116 w/m-k Cp = 1.14 kJ/kg_OC vg = 30 m/s Pp = 2600 kg/rrr' Pg = 0.353 kg/rrr'

J.1g = 4.68xlO -5 kglm-s To=300K r, = 1000 K

Now Biot number is expressed as [17] 42

hr Bi=-P ». (5.9)

By applying the above mentioned value in the Equation (5.9) yields Bi = 1.05

It was also found that for all limiting cases Bi> 1 which showed that temperature gradients exist within the particle, i.e., dT *o. Therefore, the use ofthe lumped dr capacitance method for solving the heat-transfer process in the bed will not be appropriate. Therefore, to determine the time-dependence ofthe temperature within a solid spherical fuel particle, the energy balance between the particle and its surrounding was developed as

(5.10)

(5.11)

In order to show the effect ofthe conduction term in the local heat-transfer process; the heat-transfer model was solved by three different cases:

1. Heat-transfer process without radiation term:

(5.12) 43

2. Heat-transfer process without conduction term:

4 dT = hA - T) + a&4 CT: - T ) cvC1;, rad all (5.13) dt mpCp

3. Heat-transfer process with conduction, convection and radiation modes:

(5.14)

In the above equation, Acd is surface area ofconduction, determined by assuming that particle is surrounded by the sand. Such condition is impossible and transient (non- constant) in a real case. Therefore, to see the effect offractional area to the total area, this area was also varied in the heat-transfer model.

To check the significance ofradiative mode in the heat transfer process, temperature was also calculated without including the radiative term in the energy balance equation.

5.3 Solution Technique

To obtain the solutions ofthe above transient ordinary differential equations,

finite difference approximation (FDA) method was applied. A question may arrive why

FDA was used instead ofdirect integration method. FDA was used as it was already

found that dT "* 0 and Acv in the Equation C5.11) is not constant. Therefore, it is not dr

possible to further integrate the above equation. 44

The main objective ofFDA method is to transform a differential equation into an algebra problem by discretizing the continuous physical domain. The time derivative in the above first order ordinary differential equation has the following form:

Y'=: =!Ct,y) , YCto)=Yo (5.15)

The solution ofthe above derivative is the function y (t), which must satisfy the initial condition at t = to, y (to) = yo. Here t has an unspecified fmal value.

Several numerical methods had been developed for solving the ODE. The fourth- order Runge-Kutta method was selected to solve the models. Because it is a more accurate multi-step method, it should obtain higher-order accuracy by introducing

intermediate points between n and n+ 1. According to this method

(5.16)

A computer code was developed in C to solve the above process, which is attached in

Appendix B. 45

5.4 Results and Discussion

The results ofthe models depict the heat-transfer process and illustrate the

contribution ofthe different parameters in the heat-transfer process. Considering the

following boundary conditions, these numerical models were solved as follows

0 T, = 21 C (room temperature) vg = 30 m/s Tb= 540 0 C fl = 4.68xlO-s kg ms w Cp = 1.14 kJ/kg-K kf = 0.06752 - mk =2600 kg w Pp 3 kp = 0.116 - m mk _ kg Ps - 0.353 -3 TWall = 1000 K m

Em == 0.85

In Figure 5.2, the effect ofconduction surface area in the heat-transfer process is

shown in the plot oftemperature versus time. As the particle was considered perfectly

spherical, the validation ofthis assumption is also questionable. To check the validation

ofthis assumption in Figure 5.2, two temperature profiles ofthe bed were plotted. One by

considering the total surface area ofthe sphere as conduction area, the other by

considering the halfofthe total surface area as conduction area. From Figure 5.2, it can

be said that the effect ofnon-contact surface area in the heat transfer process is

negligible. Since both ofthe temperature profile (when the contact area is halfoftotal

surface area and when the contact area is the full oftotal surface area) attained the bed

temperature in a short time (within 0.1second). Although within this heating period, there 46 are large differences, they exist for a short period oftime compared to the residence time in the combustion unit.

In Figure 5.3, the effect ofeach mode ofheat transfer is shown by considering heat transfer process with all the modes, without conduction mode and without radiation mode. Hence a graph was plotted, temperature versus time. Conduction and radiation initially playa significant role in the heat transfer process. It requires more time to attain any temperature ifonly convection is considered as the mode ofheat transfer. From

Figure 5.3, it can be shown that at 0.10 second bed attain 1000 °C for convective, conductive and radiative heat transfer, 830°C for convective and conductive heat transfer, and 620 °C for convective and radiative transfer. However, it was also found that the effect ofthe conduction and radiation is less when particle temperatures reached

bed temperature (1000 °C - 1100 °C).

Reynolds number also significantly influences the heat transfer process. Nusselt

number is correlated with the Reynolds number, and convective heat transfer coefficient

is expressed in term ofNusselt number. Therefore gas and particle velocities, particle

density, particle diameter and gas viscosity affects the heat transfer process. 1200

1000

~ 800 U '-'"~ ...... E= 600 -+- Temperatute at full surface area(C) ~ c.

a~ --.- Temperature at half surface ~ 400 area(C)

200

o ~- o 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 Time (Sec)

Figure 5.2 Effect of the Conduction Surface Area ~ --..J 1200 ------

1000

800 ,-..... U '-"" ..~ ~ = 600 ..~ Q.. e~ ~ 400 --e--- Heat transfer by cond and cony ---- Heat transfer by cony and rad mode --+- Heat transfer by cond,conv and rad 200

o --+-I------r------.--- o 0.05 0.1 0.15 0.2 0.25 0.3 0.35 Time (sec)

Figure 5.3 Comparism of Different Modes of Heat Transfer in the Bed ~ 00 49

CHAPTER 6

Experimental Analysis

Several experiments were performed to check the validity ofthe numerical models using an acrylic prototype that was previously made as a mechanical engineering senior design project. Minimum fluidizing velocities for different particle configurations were measured in these experiments. These data were correlated with the numerical data.

6.1 Validation of Numerical Model

This experiment was performed for the evaluation ofthe numerical model. Five spherical balls ofdifferent diameters were used for this experiment. Volumes and weights ofthe balls were measured first. Then densities ofthese balls were calculated by dividing the weights ofthe balls by the volumes ofthe balls. Each ofthe balls was fluidized separately in the acrylic prototype (see Figure 6.1). A velocity transducer was placed at the bottom ofthe prototype and the corresponding minimum fluidizing velocity was measured by this velocity transducer. Experimental boundary conditions were then applied to the numerical model, and minimum fluidizing velocity was calculated.

Experimental and numerical velocities are represented in the table in the following section. 50

Velocity Me s u I~ e r'je II t F',~ cl!Je

Figure 6.1 Drawing ofthe Acrylic Prototype

6.2 Experimental Results

The experimental results are shown in the Table 6.1. It was found that the ball having a diameter of0.0043 meter started fluidizing at the gas velocity of25 mls and calculated minimum fluidizing velocity for this ball was found as 16 meter/second. The diameter ofthe next ball was found as 0.0063 meter and experimental MFV and calculated MFV were measured as 24 mls and 15 mls respectively. Similarly, for the ball diameter of0.00631, 0.0098 and 0.10161 meter experimental and calculated, MFV's were found as 50 mis, 43 mis, 47 mls and 33 mis, 30 mis, 35 mls respectively. From

Table 6.1, it can be shown that balls having larger diameter have larger MFV. 51

Table 6.1 Comparison ofAnalytical and Experimental Data

Particle Diameter Particle Density Experimental MFV Calculated MFV

(m) (kg/rrr') (mls) (mls)

0.0043 1002 25 16

0.0063 1306 24 15

0.00631 2535 50 33

0.0098 7714 43 30

0.0161 8141 47 35

6.3 Discussions

Table 6.1 illustrates the effect ofthe particle density in minimum fluidization velocity in real condition. The data for this table are taken from the above-mentioned experiment. From data ofthe Table 6.1 it can be found that all the experimental values are less than the calculated values. This is due to the roll effect ofthe particle while the bed is being fluidized. A wire mesh was used as the distributor plate at the bottom ofthe prototype bed. But when air is introduced to the bed this wire mesh lost it's flatness property and all the particles rolled to the edge ofthe wall ofthe prototype where the velocity was not the same as at the bottom center ofbed. Furthermore, the values ofthe laboratory data, such as room temperature, gas viscosity, gas density, etc., which were 52 inserted for solution ofthe equations were insufficiently accurate due to the instrumental precision. Therefore, experimental velocities are lower than the calculated velocities.

From the table above it can be deduced that for a given particle diameter, the higher the particle density and the greater the required minimum fluidization velocity. 53

CHAPTER 7

Component Design

As discussed in Chapter 2, a FBC necessarily consists offour major components with necessary accessories. They are combustor vessel, fluidization device, particulate collection device, and fuel feeding device. A FBC was designed and constructed and a detail drawing ofthe fluidized bed combustor system with the necessary piping and connectors are shown in Figure 7.1. Design considerations ofeach component are discussed in detail with necessary figures and equations throughout this chapter. CAD figures ofeach ofthe components which were used for the fabrication process are also shown with the technical specifications.

Cyclone

COMbustor

Fuel Hopper Screw Feeder~~~1 d Cyclone Return

Pilot Burner

Figure 7.1 Fluidized Bed Combustor Assembly 54

7.1 Combustor

The combustor is the main component ofthe fluidized bed combustion system.

The combustor was designed to be tall, shell-type component with an inner erosion resistant layer that can facilitate less friction. On the other hand, the combustor shell and the friction layer must be able to withstand high temperature and should not lose a lot of thermal energy. Therefore, for designing the combustor, the following parameters were considered:

• Height, diameter, and thickness ofwalls • Material for walls and lining • Placement ofthe burner • Position ofprimary air inlet and flue gas outlet

The height ofthe combustor was determined 2 meters to facilitate one second residence time for the particle while assuming particle velocity as 2 mls. The material should be easily machined because several ports will be added for airflow, thermocouples, and pressure gauges. Due to the ofexcellent properties ofhigh wear resistance and easily machinable qualities, schedule 40 stainless steel was selected for the outer layer. Alumina (AI20 3) is a ceramic, hard shell that has an excellent property to withstand high temperature, a thermal expansion coefficient and thermal conductivity.

Therefore, it was selected as an erosion-resistant inner layer. Three bars of304 stainless steel were welded to the bottom ofthe combustion cylinder to support the stainless steel wire mesh. A pilot burner was placed under the support plate. The function ofthis burner is to heat up the moving air, which will transfer the heat to moving air. A cone enclosure

(see Figure 2.2} was designed for protecting the flame from the high gas velocity. 55

Primary air was designed to be introduced at the bottom ofthe combustor through an L shaped elbow, which expands from a four-inch pipe to a six-inch combustor. Flue gas outlet was designed at the top ofthe combustor through a six-inch to four-inch L shaped reducer, which connects to the cyclone inlet (see Figure 7.2).

Due to budget constraints, it was decided to make the combustor in such a way that it would minimize unnecessary fabrication and would be easy to assemble. A custom made part is expensive to fabricate and non-replaceable. On the other hand, commercial parts are less expensive, easily assemble and replaceable, and eliminate unnecessary fabrication. Thus, it was decided to use commercial parts as much as possible. A cyclone inlet port is four inches in diameter. Blower outlet pipe is also four inches in diameter.

Therefore, two six inch-four inch reducers were required to connect the combustor with cyclone and blower. Two commercials six-inch to four-inch reducers were used respectively for the gas inlet and exhaust gas outlet. However, a six inch inner diameter commercial shell was used for the combustor with required fabrication for the fuel inlet port and cyclone return line port. A small burner, which is usually used as a home water- heater pilot burner, was also welded inside the combustor shell and a wire mesh (able to

screen particle size less than 100Jlm) was also welded inside the combustor just above the

reducers. The designed combustor is made ofthe following component (see Figure 7.2)

[19]:

• 6" ill and 6' long straight pipe with slip on flanges • 6"- 4" reducers with welded neck flanges • 4" 90° elbows with welded neck flanges • 2" threaded saddles welded to walls • 100Jlm wire mesh. 56

«» w~ H

r--r--~ wire tv1esh

~,,- BI owef'~lirle

Figure 7.2: Combustor Detail Drawing

7.2 Designing the Cyclone

Design ofthe cyclone for fluidized bed combustor requires the establishment ofa specified collection efficiency ofthe cyclone. It can be stated here that there is no way of predicting from available equations the mass ofunburned particles, which will be returned to the combustor under specified conditions with sufficient accuracy. Some semi-empirical design equations have been applied successfully to cyclone design. These equations are generally derived by considering the particle trajectories under more realistic assumptions concerning the flow in the cyclone. The design theory developed by

Leith and Licht has proven useful in practical cyclone design. According to their theory, cyclone efficiency can be expressed as [20] 57

(7.1)

p M = 2[_(~_Q_fJp_(_n+_1_)))2 318p (7.2) Dc

1 p=-- (7.3) (n+ 1)

n =1-(1- 0.67d 0.14)( 1;, ) (7.4) p 283

It was focused to attain 90% cyclone efficiency for collecting particle sizes of diameter twenty-five micrometers or higher. To fmd the cyclone body diameter, Equation

7.1 was solved by considering the following boundary conditions

•T == 850 °C • llc(dp) == 0.90 •Q == 0.146 m3/second • J.l == 4.68 xl0-5 kg/m-s • K== 402.9

By considering the above mentioned boundary conditions, Equation 7.1 was

solved and cyclone body diameter was calculated as 6". Classical cyclone design

equations ofShepherd and Lapple were used for determining other geometric

specifications [20]. These equations are as follows:

H=!D (7.5) 2 c

S=!D (7.6) 2 c W=!D (7.7) 2 c 58

1 De = 2 Dc (7.8)

L, = 2D e (7.9) 1 D =-D (7.10) d 8 e

Cyclone standard dimensions are shown in Figure 7.3 graphically.

Figure 7.3 Standard Cyclone Dimensions

A detailed drawing ofthe cyclone is shown in Figure 7.4. The cyclone consists ofthe following components [18]:

• 12" ill schedule 40-304 SS commercial pipe with one slip-on flange and a 12" to 4" reducer fabricated to one end w/welded neck flange. • Tangential inlet was made from 4" ill schedule 40-304 SS commercial pipe. One end had a slip-on flange and the other was welded to the cyclone body. • 7" ill schedule 40-304 SS commercial pipe was welded to a bolted flange exiting the cyclone. Slip-on flanges the exiting end. 59

• 7"- 4" schedule 40-304 SS commercial reducer w/welded neck flanges on both ends. • 4" ill schedule 40-304 90° -elbow w/welded neck flanges on both ends.

Exho.ust

InLet. Body

( Return line to COMbustor

Figure 7.4 Detail Drawing ofthe Cyclone

7.3 Fluidizing Device: Blower

Depending upon the properties and type ofthe particles, gas velocity and pipe size, different regimes offlow can develop in the bed. Hence, the flow rate ofthe gas predominantly determines the flow patterns and the combustion reactions. Therefore, variables, such as bed pressure drop, particle geometric configurations, and gas flow rate ofthe equipment, needs to be carefully analyzed. The selection ofthe fluidization device; i.e., blower size depends on the gas flow and pressure drop ofthe system. 60

7.3.1 Bed Pressure Drop

When gas is introduced at the bottom ofthe fixed bed, gas flows through the particles interstitial space. The bed pressure drops increase linearly until the gas velocity exceeds the minimum fluidizing velocity. Bed pressure drop was calculated by the following equation:

(7.11)

There are also other drops associated with the gas flow in the pipe, elbows, and other

pipe connectors. They were calculated by applying Bernoulli's equations. All the

calculated pressure drops are shown in Table 7.1.

Table 7.1 Total Pressure Drops in the Fluid Flow Path

Flow Velocity Static Pressure Drop Rate (cfm) (fpm) (Bed) (pipe) (Elbow) (Cyclone) (Total) 0 0 10.5 0 0 0 10.5 100 510 1 0.004 0.054 0.104 1.162 200 1020 1 0.019 0.18 0.346 1.545 300 1531 1 0.042 0.288 0.553 1.883 400 2041 1 0.069 0.576 1.106 2.751 500 2551 1 0.096 0.918 1.763 3.777 600 3061 1 0.138 1.44 2.765 5.343 700 3571 1 0.174 2.16 4.417 7.751 800 4082 1 0.21 2.7 5.184 9.094

The table was plotted by considering following assumptions:

• The following fluid is air and at temperature of 550 °c. • System Design pressure should be 12" capability. • Frictional pressure drops in the 4 " section pipe is too low to be considered. 61

A fan performance curve was plotted by considering the pressure drops, which is shown in Figure 7.5. It can be shown from Figure 7.5 that bed experiences highest- pressure drop when fluidization starts. As the velocity further increases, particles rearrange themselves to offer less resistance and, hence, bed pressure decreases.

However, as the flow increases, static pressure drops in the elbows, pipe, and in the other connectors increase gradually. It was seen from Figure 7.5 that at gas velocity 30 meter/second (desired operating gas velocity) total pressure drop is 0.24 meters. While selecting the commercial blower, this fan curve was tested with the manufacturer's table to ensure that the blower will not be either underrated or overrated.

0.3 'k -l ...-~ ~ 0.25 e '-"" c.. 0.2 0 ~ 0.15 --Total Pressure -~ Drop(meter) :I fIl 0.1 fIl -~ ~ 0.05 'C- ~ = 0 0 5 10 15 20 25 30 35 40 Gas Velocity (meter/second»

Figure 7.5 Bed Pressure Drop along the Flow Path 62

7.3.2 Gas Flow Rate

The gas flow rate should be such that solid particles do not form a fixed bed on the bottom ofthe combustor. Fluidization starts when the gas velocity exceeds the minimum fluidization velocity. At the most elementary level ofconsideration, the air velocity must exceed the minimum fluidizing velocity considerably in order to achieve a high firing rate, yet it must be less than that which gives excessive elutriation ofunburned

fuel particles with its attendant loss ofcombustion efficiency.

For the desired operating conditions, a velocity range of IOmls-40mls is practical

and economical. The frictional losses increase approximately as the square ofthe flow

velocity. It is, therefore, desirable to maintain the velocity as low as possible.

7.4 Fuel Feeding Device

In any process, the taking ofa feedstock at atmospheric pressure and passing it

into a pressurized reactor imposes complexity and loss ofefficiency. The main objective

in the design ofthe feed system was to ensure positive fuel feed flow and to use a simple

existing feed system. Therefore, developers search for a proven, already existing fuel­

feed system, which can be modified to meet the required feed rate. There was not much

option available for this small combustor system. Considering the required offuel feed

rate and budget constraints, a specially designed screw feeder was designed. A screw

feeder is a dependable, proven controllable low-rate fuel feeder and is a good option for

this purpose, because it offers one directional fuel feed with variable feed rate. As the bed

is under positive pressure and required fuel feed rate is very low 50 gm/m; a screw feeder 63 is suitable for this purpose. However, commercially available screw feeders were cost prohibitive and did not meet required feed rates. So it was decided to design and fabricate the feeder at Ohio University facilities. In order to reduce cost and ease manufacturing, the design process was simplified as much as possible.

It was decided that the average feed rate ofthe fuel would be 50 gm/min. The following calculations were made to fmd the pitch (P) and major diameter ofthe custom thread. Thread cross sectional area is trapezoidal which is shown in the Figure 7.6 and is expressed as

At = (a+b)h (7.12) 2

b ! /

Figure 7.6 Geometrical Dimension ofThread

It was assumed a= p 2

b = P 4 By plugging these values Equation 7.12 yields as

(7.13)

Major diameter was considered as

dm = O.886p (7.14)

Now volume ofthe thread for one revolution is expressed as 64

V=At I (7.15)

Angular velocity ofthe thread is expressed as

0) = 1tdmN (7.16)

Following assumptions were considered for this calculation

N= 10 rpm mfr = vPp = 50 gm/minute Pr = 2600 kg/rrr'

Now distance traveled by one revolution is calculated by

(7.17)

To find for the assumed fuel feed rate the following equation was solved

(7.18)

Pitch was found as p = 1.2 mm and major diameter ofthe thread was calculated as dm =

O.86603p = 1.045 mm. The detail drawing ofthe screw thread is shown in Figure 7.7.

I i

I I ~- SCt~~ew hr~eod L- I --[.,------~_r------<~~ -. - --+ ;o~~ - --- - $3--

i -~ceni:er line Por bearing ! I

Figure 7.7 Detail Drawing ofthe Screw Thread 65

The combustor pressure is slightly higher than the atmospheric pressure.

Therefore, it was required to keep the fuel hopper pressure slightly greater than the combustor pressure in order to maintain a positive displacement fuel feed rate. Again, commercial piping was used as the components ofthe hopper (see Figure 7.8).

HC)~Jpet'~'

T CJ C C) r~Tl kJ l.·j 5 t 0 t'~

I ! i i IN

Figure 7.8 Detail Drawing ofFuel Hopper

The components are as follows (see Figure 7.9 for details) [18]:

• 10" ill commercial carbon steel pipe w/slip on flange on one end and a 10"­ 4" reducer fabricated on the other end • Welded to the reducer is a 4" ill carbon steel commercial pipe w/slip on flange bolted to a cap • 2" threaded saddles welded to 4" pipe • 10" flange cap bolted to top 66

~-+----,.~Fuel Hopper

----COMbustor

Screw feeder

Fuel feedllne

Figure 7.9 Detail Drawing ofFuel Feeder Assembly

The screw feeder assembly can be joined to the hopper through the 2" saddles.

The feed screw carries the fuel in the space between its threads and the enclosing casing.

A custom made thread was developed for the required feed rate. This screw thread has two threads per inch, which enable the screw to carry more volume offuel per pitch. The thread size was calculated in relation to the design feed rate.

An AC motor drives the feed screw. The required fuel feed rate requires a low rpm (8-10) motor. Fuel is carried by the screw to the end ofthe screw casing and reaches the inlet port ofthe combustor. 67

The combustor is at positive pressure, and this may counteract the effects of gravity transport. Therefore, the fuel line needs to be pressurized along with the hopper.

A blow line was installed at the fuel inlet port to help force the fuel into the combustor.

Since the fuel hopper and fuel line are pressurized, there must be a sealing device on the shaft ofthe screw. A gland seal has been chosen for use with a rotating shaft and has the advantage ofbeing easy to fabricate. The gland seal has two components: a sealing component and a mating component. Within the sealing component, there is a close reduction in diameter from one size to another. The mating component is a pipe with an outer diameter a little smaller than the inner diameter ofthe pipe on the sealing component. The sealing material, such as felt, wraps the shaft around. The two parts are drawn together, mating components inside ofsealing component using a nut and bolt.

The screw shaft is supported by two bearings at opposing ends ofthe hopper. The bearings are supported by a frame with common plane alignment, as shown in the

Figure 7.10.

~~_---; ~-- Fuel Hopper ->

<;C t~ e w fee d e t~ -

L.....-----Glcuid Seal

Figure 7.10 Detail Drawing ofthe Screw Feeder Assembly 68

7.5 Ash Removal System

The cyclone separates particles to be returned to the combustor. These particles include sand, fuel, and ash. Any fuel particle with a diameter higher than 100Jlm contains a significant amount offuel and is required to recirculate for complete combustion.

Therefore, unburned fuel particles with a diameter greater than l00J.1m and sand need to be separated from the ash and should be returned to the combustor. To remove the ash from the fuel and sand, an ash removal system was placed at the bottom ofthe dust hopper ofthe cyclone (see Figure11). This system consists oftwo-inch stainless piping with elbows and tees. Particles less than 100Jlm diameter are then removed from this point, as the end ofthe section is open to the atmosphere at a lower pressure. Airlines are tapped at various points in the system to blow the particles to the combustor. A separate

airline is aimed at the wire to prevent build-up, which would block the gap ofthe air

mesh causing a flow restriction.

Re-tunllne FrOM Cyclone

,r----- COMbus-tor~-+--t--_Screen

'-+++------Ash F tow ---Unburned Fuel Flow

Figure 7.11 Detail Drawing ofFuel Return Line 69

CHAPTER 8

Conclusion

8.1 Summary

This thesis evaluated various aspects ofthe fluidized bed combustion process.

Although fluidized bed combustor systems offers many advantages, they are not a panacea, and each application must be carefully considered. Fluidized bed combustion process is likely to be one ofthe best combustion processes for low-grade and sulphurous fuel. It has some major advantages over the conventional combustion process, such as being less selective in fuel quality, higher heat transfer coefficient, and easier pollutant emission control.

As with most engineering equipment, it is not realistic to attempt to design a fluidized bed combustor entirely from theoretical concepts. Throughout its history, fluidized bed technology has developed largely from experimentation. The modeling of the fluidized process has to reflect this, and because ofthe complexity ofthe real process, caution must be exercised in applying it. Therefore, a good design is the product of fundamental understanding, acquisition ofdata, and many practical experiments with the equipment.

A review ofthe design procedures has been made throughout the thesis including numerical modeling to experimental verification. Rough assessments ofsome parts ofthe design work, such as diameter and length ofthe combustor, bed pressure drop, frictional loss, etc., were also made. On the other hand, equal importance was directed to the ancillary equipment, such as feeder, cyclone, blower, valves, control system, and safety 70 and environmental protection features for proper functioning ofthese components. All these considerations were encompassed in the design process so that the system was capable ofmeeting the desired criteria.

8.2 Conclusions

A fluidized bed combustion system can satisfy many differing requirements simultaneously, ranging from the burning oflow-grade fuels to low emission of pollutants. It is hoped that this thesis has given, in an elementary way, some indication of the potentiality offluidized bed combustion process and its development process.

Numerical models ofthe bed also describe the importance ofdifferent parameters in designing and operating the combustor system.

Particle fluidization depends on the gas velocity and gas velocity in side the combustor vessel depends on the gas flow rate and bed pressure drop ofthe system. From the particle dynamics models, it can be said that particle velocity depends on the particle and gas properties, particle geometric configurations and gas velocity. It also can be concluded from the calculation that bed pressure drop depends on the gas velocity which is higher at the beginning (for static bed) and decreases linearly with the increases ofgas velocity. Therefore, particle and gas properties necessarily affected the design process.

Because the size ofthe blower depends on the required gas flow rate and bed pressure drop. 71

From the experimental model, it was found that all the experimental data were higher than the calculated data. This was due to the inaccuracy ofthe prototype. The wire mesh which was used as distributor plate, lost its' flatness when gas velocity was increased. Consequently, all the balls rolled back to the edge ofthe bed where gas velocity was lower than the center ofthe bed and it was not possible to measure the velocity at the side ofthe prototype shell. Therefore, this deviation was found from the experimental results.

It can be concluded that several mechanism contribute to heat transfer in fluidized beds and that estimates ofheat transfer coefficient depend upon empirical correlation.

Care must be taken to ensure that such correlation is not applied outside their range of validity or conditions under which they were obtained. It also can be said that conduction plays a significant role for a while in the heat transfer process, and after that, heat is mainly transferred by convection.

8.3 Recommendations

This section presents recommendations for future experimental work using this combustor system. This combustor system can be used as an experimental setup for the study ofthe fluidized bed combustionprocess, The following experimental studies can be performed by using this system: 72

To establish high combustion efficiency irrespective offuel types the factors affecting combustion efficiency can be examined by this laboratory scale fluidized bed. The factors affecting combustion efficiency, mainly, are air-fuel ratio, moisture content in the fuel, ash in the flue gases, incomplete combustion ofcombustible gases, and carbon. For the evaluation ofcarbon burn-up, combustion efficiency (l1co ) for combustion ofsolid fuel can be defined as mAHCV )-mc(Cr:,) f (8.1) n: = mAHCV ) f

The mass ofcarbon unburned (111c) must be determined by collecting the mass ofsolids carried over while a mass offuel mr is burnt, analyzing these solids for carbon content and, hence, determining the mass ofcarbon unburned for a given mass offuel.

The devolatilization and combustion characteristics ofhigh sulfur Ohio coals and solid waste could be studied by this pilot model. Several optical ports should be incorporated in the bed so that temperature can be measured by a technique, such as optical pyrometry. Sampling probes also need to be integrated with the system to extract solid and gaseous samples for later analysis

This bed could be used to test the pollutant formation during the combustion of various fuel blends. Small amounts ofmore desirable fuels, such as natural gas or high- volatile, high rank coal, can be blended with less-utilized fuels, such as low-rank coals, biomes, municipal solid waste, and refuse-derived fuels. Data gained from these experiments should help clarify the formation ofpollutant gases. BIBLIOGRAPHY

1. Coal Production 1992, Energy Information Administration(EIA), US Department of

Energy,DOEIEIA-0118,1992,pp. 10-15

2. Coal Energy for the Future, National Research Council, National Academy

Press,1995, pp. 288-297

3. US Statistics Year Book 1996, US Statistical Society, 1997, pp. 34-35

4. "Clean Coal Technology Demonstration Program Update 1992", U.S. Department

of Energy Report DOEIFE-0284, 1993, Microfiches E 1.8:0284, Fiche 1-2

5. Franklin, W., and Franklin, M.; "The Rise ofRecycling as a Solid Waste

Management Option", MSW Management, Mar/Apr 1995, pp. 16-23

6. Saxena, S.C., and Jotshi, C.K.; "Fluidized-Bed Incineration ofWaste Materials",

Progress in Energy Combustion and Science, (20), 1994, pp. 281-324

7. Marrocco, M., and Bauer, D.A.; "American Electric Power Pressurized Fluidized

Bed Combined Cycle Technology Status", DOE Third Annual Clean Coal

Technology Conference, Chicago, IL, Sept. 1994, pp. 7-31

8. Valenti, M.; "Breaking the Thermal Efficiency Barrier", Mechanical Engineering,

(117)7, 1995,pp. 86-89

9. Bayless, D J; Project Proposal for Fluidized Bed Combustion Project to Ohio

University., 1997, pp. 10-23

10. EI-Wakil, M.M.; Power Plant Technology, McGraw-Hill, Inc. 1984, pp. 142

11. Frazzitta, S., and Annamalai, K.; "Performance ofa Burner with Coal:Manure

Blends", Proceedings ofthe Joint Meeting ofthe Central and Western Sections of

the Combustion Institute, San Antonio, TX, April 1995, pp. 614-619 12. Bayless, D.J., Schroeder, A.R., Johnson, D.C., Peters, J.E., Krier, H., and Buckius,

R.O.; "Effects ofNatural Gas Cotiring on Ignition ofCoal and Coke Particles",

Combustion Science and Technology, (98), 1994, pp. 185-196

13. Bayless, D.J., Schroeder, A.R., Olsen, M.G., Johnson, D.C., Peters, J.E.,

Krier,H.,.and Buckius, R.O.; "The Effects ofNatural Gas Cotiring on Sulfur.

Retention in Ash", Combustion and Flame, 1996, pp. 123

14. Cuenca, M Alvarez, and Anthony, E.J.; Pressurized Fluidized Bed Combustion,

McGraw-Hill, 1996, pp. 48-56

15. Bayless, D.I; "Experimental Studies ofSingle Particle Coal Combustion, Igition,

Sulfur Retention, and the Contributions ofMorphology", Department ofMechanical

Engineering, University ofIllinois at Urbana-Champaign,1995, pp. 41

16. Howard, J.R.; and Hilge, Adam; Fluidized Bed Technology Principles and

Application, Mc Milan Publishing, NY, 1989, pp. 82

17. Holman, Alan J.; Conductive and Radiative Heat Transfer., Macmillan Publishing

Company, pp. 5,13

18. Kunni, Diezo; levenspiel,and Octave; Fluidization Engineering, Addison-Wesley

Publishing Company, 1994, pp. 156

19. Wilson, K.; Eckels, M. and Reineck, B.; Fluidized Bed Combustion Process,

Senior design project, Department ofMechanical Engineering, 1997 pp. 12-18

20. Nevers, De; Air Pollution Control Engineering, McGraw-Hill Publishers, 1995,

pp.213 APPENDIX A

/* This program calculates particle velocity by solving the particle dynamics equation of the bed.*/

# include # include # include # define g 9.81 # defme rowp 1000 /* Particle Density */ # define mewg 4.68e-6 /* Gas Viscosity • / # define rowg 0.353 /* Gas Density */ # define pi 3.142 maine) { FILE *thefile; float cd; /* Drag Coefficient */ float vg; /* Gas Velocity ale / float vpl; /* Particle Velocity */ float vp=O; 1* Initial Guess ofParticle Velocity */ float rep; /* Reynolds Number *1 float dp=.0006; 1* Diameter ofthe Particle *1 float temp; 1* Temperature ofthe Gas *1 int trial,i = 0; thefile = fopen("mfvl ","w"); vpl=.16;vp=2;trial=1;

printf("Enter the value ofvg : \nn); scanf("%f",&vg); for (dp = O.OOOI;dp <= O.02;dp+=0.0001) 1* Particle Diameter Range *1

{

trial = 0;

do { rep=(fabs(vp-vg)*rowg*dp)/mewg;

if((rep>O.1 )&&(rep<2»

cd=(24/rep)*(1+«3/16)*rep+(9/16)*pow(rep,2)*log(2*rep»»

else if«rep>2) && (rep<500»

cd=(24/rep)*(1+0.15*pow(rep,O.687»;

else if(rep>500)

00=0.44;

else

cd=24/rep;

vp1=vg-sqrt(«rowp-rowg)* g*4*Dp)/(3*cd*rowg»;

temp = vp;

vp=vpl;

tria1=trial+1;

[whileufabsrvpl-tempp-u.Ool ) && (trial

return (0);

} APPENDIXB

I· This program calculates temperature ofthe particle by solving the transient heat transfer process ofthe bed. •I

# include # include # include # define g 9.81 # define rowp 1000 1* Particle Density *1 # define mewg 4.68e-6 /* Gas Viscosity *1 # define rowg 0.353 1* Gas Density *1 # define pi 3.142 # defme kf.067 /* Thermal Conductivity of Gas */ # defme kp 1.83 /* Thermal Conductivity of Particle lie I # define Cp 1.14 /* Specific Heat ofGas *1 # define Tb 540 /* Bed Temperature *1 # define Ti 24 1* Initial Temperature of Particle *1 # define vg 30 /* Gas Velocity */ # defme t 10 float repCal(float); float dp,rep; maine) { FILE *thefile; int m,j,delt,count = O,z=O; float temp T, a, mp, Pr, Nu, delr, h, T[IO], deITI[4]={O}; thefile = fopen ("tlO","w"); fprintf (thefile," t T\n");

for (dp = O.Ol;dp <= O.2;dp+=O.Ol)

{ fprintf (thefile,"dp=%f\n",dp);

rep=repCal(dp); 1* Reynolds Number *1

Pr=(Cp*mewg)/kf; /* Prandtl Number *1

a=pi*(dp*dp); mp=(pi)*(pow(dp,3»*ro\\tp/6.0; /* Mass ofthe Particle */

if(rep>100) /* Nusselt Number */

Nu = 2+0.6*«pow(rep,0.5»*(pow(pr,.3334»);

else

Nu = 0.03*pow(rep,1.3);

h=(Nu*dp)/kp; 1* Convective Coefficient *1

de1r== dp/2; delt = 1; count = 0; T[O] = 24; while(count <= 24) /* Temperature Calculation by Using Runge-Kutta Methods */ { delTl[0] = delt * «a*h*(Tb - T[z]»/(mp*Cp»; deITl[l] = delt * «a*h*(Tb - (T[z] + O.S*deITl[O]»)/(mp*Cp»;

delTl [2] = delt * «a*h*(Tb - (T[z] + O.5*deITl [l]»)/(mp*Cp»;

delTl [3] = delt * «a*h*(Tb - (T[z] + O.S*deITl[2]»)/(mp*Cp»;

T[z] = T[z] + (deITl[O] + 2*deITl[1] + 2*deITl[2] +

delTl [3])/6.0;

count++;

fprintf (thefile,tt%d %f\n",count,T[z]);

}

}

fclose(thefile);

return(O);

} float repCal(float Dp ) 1* Function for Calculating Reynolds number *1

{ float cd; 1* Drag Coefficient *1 float vp 1, vp=O, temp, k; int trial; vp 1=.16;vp=2;trial=1; 1* Intialization *I trial = 0; do { k=(fabs(vp-vg)*rowg*Dp)/mewg; 1* Reynolds Number Calculation */ if«k>O.l)&&(k<2» 1* Drag Coefficient Calculation */ cd=(24/k)*(1+«3/16)*k+(9/16)*pow(k,2)*log(2*k»);

else if«k>2) && (k<500»

cd=(24/k)*(1+O.15*pow(k,0.687»;

else if(k>500)

cd=O.44;

else cd=24/k;

vpl=vg-sqrt«(rowp-rowg)*g*4*Dp)/(3*cd*rowg»;

temp = vp;

vp=vpl;

tria1=trial+1;

}while«fabs(vpl-temp»O.OOl) && (trial<1000»;

return k;

}