The motivation of this report is to detail a design to convert thermally activated pyrolytic chars to activated carbon in the most sustainable manner. Activated carbon can have high commercial value if produced to a high grade. The 170 kg/hr feed of char is incrementally fed into fluidized bed reactor heated directly by an encapsulating furnace. An overall carbon conversion of 93% is achieved through manipulation of the random pore model by introducing pore size parameter. The reacting material are designed to enter the reactor 298K and are rapidly heated to the reacting temperature 1232K. A calculated residence time of 38 minutes is found to complete the designed reaction. Through comparing different reaction mechanics and equipment choices the goal of the design is successfully achieved as Activated carbon with a BET surface area greater than 1100m2/g is produced. In addition to safety and process control, annualised costs are of extreme importance and are evaluated thoroughly on Microsoft excel, the conclusions are presented towards the conclusion of this report. This report can be used to understand this gasification reaction in a fluidized bed at a fundamental level and considered when designing a similar system to process pyrolytic char.

Introduction

process overview

Char is a solid material generally produced as a by-product in processes which heat carbonaceous substances to high temperatures. Normally, char is disposed of as waste in industrial operations. However, since the early 1990s environmental sustainability has become a significant concern across all industries. Manufacturers have taken the position to transform waste into useful products, especially those which can yield energy. Activated Carbon is one of these; it has many uses ranging from teeth whitening to filtration. Recently, its price has risen as increasingly more uses are being found for the material in the pharmaceutical industry. As a result, the engineering of activated carbon has received much attention over the last two decades (Tchoffor, Davidsson and Thunman, 2015).

As shown in the process flow diagram in figure 6 Char and oilenter a separator which segregates the solid from the oil with 100% efficiency. The heavy oils are fed to a catalytic reformer to crack the hydrocarbons into more valuable, smaller hydrocarbons. Whereas, char is fed to a heated fluidized bed reactor. The bed is supplied with pre-heated CO₂ (purity 99.9%) a product of syngas combustion which acts as a gasification agent, the high CO₂ purity improves conversion. CO₂reacts with carbon in an endothermic reversible reaction shown in the reaction scheme considered in part 3.0. The activated carbon product is then collected and prepared for sale (Molina-Sabio, González and Rodrı́guez-Reinoso, 2004). There is still considerable uncertainty with regard to profitability of carbon activation processes, therefore a detailed economic evaluation is completed and evaluated in 10.0to establish an overall cost

Reactor conditions determine the conversion. 100% conversion is targeted however; conversion is coupled with residence time. Longer residence times at elevated temperatures causes fragmentation of carbonaceous material. Conversely, time determines the development of pores in the solid. Through comprehensive comparative kinetic study optimal process criterion were found. This led to an achievement 93% carbon conversion. Approaches to scale up these findings are addressed at a later stage of this report.

The extensive literature survey on present research revealed that research is more focused on verification that the process is practicable at laboratory level. Few engineers have attempted to scale-up the process to show economic performance at an industrial level which is put down to lack of understanding of the reaction kinetics and hydronamic fluidized bed behaviour work together to produce a commercial grade activated carbon. However, the Office of coal research and American oil company have showed successful industrial operation (Office of coal Research, 1962).

Design objectives

The principal objective of this report is to outline a detailed design for a fluidized bed reactor which will process pyrolytic char to make the highest amount possible of commercial grade activated carbon. In industry, high product manufacture may not always be ideal because of factors such as; sustainability, safety and cost. Process engineers take these factors into account and seek to optimize operation and select instruments thus an economically worthwhile process can be developed. Through the iterative technique of optimisation and investigation of hydronamic behaviour of the fluidized bed coupled with an enquiry into process control and safety this report will provide a comprehensive demonstration of the use of fluidized beds in industry to make activated carbon.

Design constraints

Temperature constrains the reaction. Temperature drives the reaction, determines conversion and the rate of reaction however, hinders pore formation if heated too high. A maximum temperature of 1473K constrains the reaction as this would begin causing pores to collapse. Further research into the effect of temperature on the reaction kinetics found that at temperatures above 1223K the rate constant changes at a slower rate than is predicted by Arrhenius’s law (Lee and Kim, 1996).

As fluidized bed reactors are often quite long in vertical height therefore the plant height constrains the design. In the case that the vertical height exceeds that of a plant fluidized beds may be placed outside which exposes the reactor to the effects of adverse weather and external damage.

The conversion of the char particle is a constrained by its composition. Although the char is postulated to contains 100% carbon, some of the carbon is present in char in the form of ash. This ash does not react with CO2 it is instead further burnt by the hot temperatures and collected at the solid exit of the reactor.

Choice of equipment

Traditionally, carbonaceous material is thermally activated using fixed bed reactors. Extensive literary research of recent developments in the field of gasification reactors oppose the usage of fixed beds. A lack of understanding of fluidized bed standardised fixed bed employment in industry. This section will provide justification for the design decisions made, comparing circulating and bubbling fluidized bed (BFB) configurations. The circulating fluidized bed (CFB) was considered however, the conclusions demonstrated in figure 1 justified its rejection.

Fluidized beds can perform both activation and gasification of the carbonaceous material. Solid particles are fed into the reactor and the gasification fluid from the bottom. The incoming fluid forces the reactor bed to operate as if it were a fluid. In this case, a uniform particle mixing is achieved, coupled with the expulsion of radial and axial concentration gradients a promotion of higher degrees of gas-solid contact is attained. The increased contact improves the extent of activation achieved and consequently increases the surface area of micro and mesopores of the material accessible through CO₂adsorption. A pitfall of CFB reactors is the formation of concentration gradients along the length of the reactor, which may cause considerable temperature gradients. In industry, gradients catalysts often eliminate these gradients. The proposed catalytic activity is somewhat complex and lacks the research to be replicated for this process.

Good temperature distribution eliminates hot spots and improves heat exchange contributing to the high thermal efficiency displayed by fluidized bed reactors. Although CFB reactors exhibit high heat transfer, it is found to increase with increased fluid superficial velocity, as well as solid circulation rate. Thus, increasing fixed cost is a major drawback of using this technique. Furthermore, CFB are generally much larger than BFB in industry. Findings of a wall – to-suspension heat transfer investigation by figure1suggests, that assuming the same material is used for both reactors, the influence of increased reactor length increases surface area. Subsequently, leading to higher energy requirement.

CFB’s are more expensive to install and maintain than BFB. Assuming of identical conversion in both beds, the relatively small feedstock promotes BFB usage. But, future scale up would advocate usage of a circulating fluidized bed because the low velocities adopted in the bubbling regime will not be able to fluidize large masses efficiently.

Parameter	Bubbling Fluidized Bed (BFB)	Circulating Fluidized Bed (CFB)
Temperature distribution	Uniform	Temperature gradients in direction of solid transport
Heat transfer	High heat transfer	High heat transfer rate, less efficient than BFB
Conversion	High conversion possible	High conversion possible
Fluidization gas velocity	Low	High, leads to erosion
Investment: Installation and maintenance cost	Medium	High

Figure 1: Comparison of bubbling fluidized bed and circulating fluidized bed.

Kinetic study

Comprehensive study on the tendencies of carbon gasification provides a thermochemical explanation through reaction kinetics for the conversion of polymer chars to activated carbon. It is fundamentally apparent that the gasification is reaction controlled and takes place on active sites specifically surfaces situated within micropores.

CO2+Cf↔CO+CO#1a

CO→CO+Cf#1b

CO2+C→2CO#1c

Where C_f, C(O) and CO symbolize free carbon active site, occupied carbon active site and activated carbon respectively. The first step represents the adhesion of an oxygen atom to the carbon active site. The second step shows desorption of activated carbon CO from the carbon matrix. This is essentially an oxygen exchange reaction which is normally modelled using the Langmuir-Hinshelwood equation. However, its findings for equation set 1 are often inaccurate. Thus, much work on the reaction kinetics of char gasification with CO₂ has been carried out using equation 1c, which outlines the overall reaction. Yet, there is still considerable uncertainty on which model best describes the complex chemical and physical behaviour of this reaction. A number of methodologies, developed by different researchers were investigated; Volume Reaction Model (VRM), Modified Volume Reaction model (MRVM) Changing Grain Size Model (CGSM) and the Random Pore Model (RPM) (Bhatia and Perlmutter, 1980). VRM proposes that all pores are of the same size and length and assumes there is uniform gas diffusion which occurs homogeneously instead of heterogeneously. This oversimplification contradicts the heterogeneous reaction mechanism and was not discussed further. The reaction scheme shown above is simplified to:

dXdt=kv1-X#2

MVRM improves upon the VRM by acknowledging the change in rate of reaction as carbonaceous material is converted. The apparent rate constant equation shown below changes as the reaction advances, this provides a more accurate evaluation for the reaction kinetics. Furthermore, incorporation of modified volume parameters α and β provide more structural information about the complex solid structure.

dXdt=kM1-X#3a

km=α1ββ-ln⁡1-Xβ-1β #3b

GCSM is widely used for gas-solid reactions; it is derived from the unreacted core model (UCM) which models the rate of reaction for individual particles, grains. GCSM is based on the presumption that all grains are uniformly sized and spherical in shape. The kinetic model expression shown below assumes the chemical reactions occur on the material surface, it constitutes that that the reactant is converted into another product through gasification hence leaving behind an unconverted core of material. In theory, however, some pores are longer than others hence may extend beyond the centre of a solid (Bhatia and Perlmutter, 1980). Also, a high pore concentration near the core is possible which shows that there are accessible pores towards the centremost point of porous structures. Nevertheless, GCSM present a viable option to model char gasification as it acknowledges the porous nature of char even though it models the pores as spheres.

dXdt=3τ1-X223#4a

τ=rgbvBkCGSMCn#4b

rg=3SBETρt#4c

Out of the models analysed it was found that RPM best represents experimental data. The model shown below, developed by Bhatia et. Al, is especially attractive because it considers structural changes and instantaneous pore overlapping caused during gasification. This can increase and reduce pore area accessible for reaction, simultaneously. The included appraisal of arbitrary pore size distributions not accounted for in the preceding models is fundamental to RPM’s accurate depiction of the gasification of porous materials (Everson, Neomagus and Kaitano, 2011). Using Microsoft Excel a residence time of 38 minutes was calculated to attain a reactant conversion of 93%.

dXdt=kRPM1-Xn1-ψln⁡1-X#5a

kRPM=Aexp-EactRT#5b

X=1-exp⁡2-ψ2ψ#5c

ψ=4πL01-v0S02#5d

Where

ψrepresents a structural parameter which encapsulates the sizing variables shown below. The pore volume distribution

frmay be determined from data collected by experimental trials extended by this model.

v0=∫0∞fr dr#6a

S0=2∫0∞frr dr#6b

L0=1π∫0∞frr2 dr#6c

The pore volume distribution is pivotal when it comes to sizing any porous material. Porosity alone does not say much about the structure of pores within a solid as their size distribution is so large. Several experts have noticed that heat as well as the reaction with CO₂ changes the pore shape. The modified RPM accounts for the evolution of pore volume distribution during gasification, which is a great weakness of the other models previously discussed.

Varying

ψand estimating the effect its effect on conversion yields the graph shown in figure 2. From this we see that conversion increases initially however reaches maximum around X=0.38. This low conversion is offset by introducing empirical parameter x to equation 5a. Its manipulation of gaseous reactant, porosity and shape structure of solid reactant information extends the application range of the RPM. The relationship between X and

ψfor the modified RPM equation is shown in figure 3.

Figure 2: Show the relationship between structural parameter

ψand conversion

The results shown in figure 3 detail the effect which the structural parameter n has on conversion, as this value decreases conversion increases. The dimensions and mechanical properties of pyrolytic char used in this report give a value close to n=0.2, taking a calculated value of =80 a conversion of 0.93% is obtained.

Figure 3: Shows the relationship between structural parameter and conversion.

Using equation 5bfor apparent rate constant, the rate of reaction was evaluated at various temperatures to gain an understanding of the range of reaction rates which can be obtained. The rate of reaction is important because if the reaction is too quick it can cause a large percentage of unconverted char to be produced. Additionally, the small amount of converted char will be of a lower quality. On the other hand, if the reaction is too slow a destructive effect takes place on carbonaceous material with porous structures.

Table 1: Calculated apparent rate constant data for the RPM model.

Temperature (K)	K_RPM
850	1.61×10^-5
950	6.72×10^-5
1050	1.90×10^-4

It can be seen from the data in figure 4 that increasing temperature increases the rate of reaction, resulting in a faster conversion. Naturally, this reduces the residence time which particles have in the fluidized bed. However, to be commercially effective the BET surface area must be above 1000 m²/g this is obtained by heating char particles for a longer period of time to maximize pore development (Mahamud, Menéndez and Sarquís, 2014). 1323 K is not chosen because at this temperature mechanical destruction of the pore structure is witnessed (Matsui, Kunii and Furusawa, 1987). These chosen temperature 1223K correlates favourably with evidence found in literature which selects 1150-1300K as the optimum temperature range for char gasification. Furthermore, it is found that a reaction rate between 0.0003 and 0.0008s^-1 is used frequently at an experimental level which can be maintained assuming that experimental tests are scaled at proportional rate which permits the rate to remain the same (Roberts and Harris, 2006).

Figure 4: The effect of varying temperature on the rate of reaction.

Figure 5 illustrates the time taken to achieve the desired conversion and is hence taken as the reactor residence time. Time taken for particles to fall out of the reactor is nominal and is not considered in the residence time. The relationship shows that for a conversion of 93% the residence time is approximately 38 minutes when the reaction takes place at 1223K.

Figure 5: The relationship between conversion and residence time at 1223K.

Unit description

Process flow diagram

Thegeneral flow of material can be found in the process flow diagram shown below. A simple outline of the engineering process illustrates the initial separation of char from oil using the Oil Water Solid separating system (V101), char is then directly fed to the fluidized bed reactor (V102) where it is heated by the encapsulating furnace system (V103). Minor details and ancillary equipment are not depicted in this flow diagram. They can be found in part 11.0 process control.

Figure 6: Process flow diagram of Char gasification.

Mass balance

Crushed char particles enter the reactor with a calculated mass flowrate of 170 kghr^-1. The reaction which converts char into activated carbon requires 323.2 kghr^-1 of CO₂to achieve the desired conversion. The reaction scheme detailed in equation 1c produces 368.9 kghr^-1of activated carbon. Unconverted char is minimal hence is disposed of, which further negates the necessity for using a circulated fluidized bed to recycle unreacted char for re-gasification. 7.35 kghr^-1of unused CO₂ occurs due to the particle size distribution exhibited by the feed. Some particles have a lower density than others thus instead of adsorbing the gasification fluid, glide off the streamline leaving CO₂to rise to its reactor outlet.

Table 2: Mass balance of fluidized bed unit.

	Mass flowrate In (kg/hr)	Mass flowrate out (kg/hr)	Molar flow (mol/hr)
C	170	11.90	14.17
CO₂	323.21	7.01	7.35
CO	0	368.90	13.18

Energy balance

The reactant streams enter the reactor system at different temperatures. Char enters the reactor at room temperature from the separating system; it is then heated to 1232K and fluidized for 38 minutes. Evaluation of reaction kinetics found 950K as the ideal reacting temperature (T₂). The reactor is designed to operate isothermally to maintain a consistent conversion. Again, using Excel, the adiabatic inlet temperature for the CO₂stream was calculated as 623 K, this is achieved through pre- heating the stream. However, in optimisation, as mentioned in part 7.0 it was found that it was more cost effective to abstain from pre-heating thus inlet temperature was chosen to be 298K.

Table 3: Stream enthalpy energy balance of fluidized bed unit

	Inlet T (K)	Outlet T (K)	C_p kJ/(kgK)	Stream Enthalpy (kW)
C	298	1232	1.84	81.15
CO₂	298	1232	1.22	102.30
CO	0	1232	0.32	40.40

Individual stream enthalpies were found through determination of specific heat capacity values; the specific heat capacity of char and CO₂ were found using the heat capacity equations shown in equation set 6. The wide variety of sources from which char is derived complicates confirmation of a universal specific heat capacity. Hence its C_p value is taken from a process which produced pyrolytic chars as a by-product of polymer pyrolysis.

∆H=CpT2-T1#7a

CpCO2=24.997+55.187T-33.691T2#7b

CpC=-0.218+3.807×10-3T-1.758×10-6T2#7c

Q=ṁCpT2-T1#7d

Rate Based Equipment sizing

Fluidized bed reactor

In order to assess the reactor size, it is essential to determine the range of velocities which permit fluidization. As shown below, the ergun equation effectively evaluates the drag exerted on particles by a gasification fluid. Bed length (L_mf) nor

∆p are yet to be determined thus pressure drop per unit length is calculated to allow for later determination.

∆pLmf=gρf-ρε1-ε#8a

∆pLmf=150μvmfDp21-εmf2εmf3+1.75ρfvmf2Dp1-εmfεmf3#8b

The minimum fluidization velocity expresses the velocity required for fluidization to begin. It is found by combining equation 8a and 8b to obtain the quadratic equation below where

Re= Dpρfvmfμ. Through mathematical manipulation V_mf is then found to be 0.0043 ms^-1 which gives a Reynolds number value of 0.401 which supports the use of equation (4) as it is applicable only to Re<20 systems.

1.75εmf3 Dpρfvsμ2+1501-εmfDp3 Dpρfvmfμ=Dp3ρfρp-ρfgμ2 #9a

vmf=Dp2ρp-ρfg1650μ#9b

Determination of the maximum fluidization velocity which is also known as settling velocity in other literature, V_set, is necessary to prevent particle elutriation. The low particle density derived from small size and porous structure of char means that conditions must be imposed to prevent char particles being carried out with the gasification agent. This occurs when the gas superficial velocity is equal to that of settling particles. Using stokes law the maximum fluidizing velocity is calculated to be 0.40 ms^-1. This is consistent with literature which details that

vsetvmf≈90thus, the calculated fluidisation range is appropriate.

Vset=Dp2ρp-ρfg18μ#10

The superficial velocity must be sufficient to homogeneously fluidize the reactor bed. Since pressure drop is strongly correlated to fluidization velocity a value was chosen to ensure that a sensible pressure drop for the fluidized bed was calculated. A superficial velocity of 0.3 ms^–¹ was chosen which produced a pressure drop value of 6.71 KPa for an estimated 3 m bed length, as the bed experiences a 10% expansion during fluidization the bed length was taken to be 3.3m. The freeboard length above the bed is 2.7m and was found using a well know correlation from Sinnott and Towler this gives a total reactor length of 6m. Using a further correlation from literature, the optimal length-to-diameter ratio was found to be 8 which gave a diameter of 0.75m therefore, the total reactor volume was 2.65 m³.

Centrifugal Fan

It was decided that a centrifugal fan would be used to propel gasification fluid into the bed. With the selected fluidizing velocity of 0.3ms^-1 and assumption of adiabatic operation the power required was found to be0.001kW, when operating with backward curved blades which increased efficiency to 83% from standard blades which achieved 80% efficiency. The fan outlet diameter, 0.197m, was calculated from the fluid mass flowrate. Furthermore, when calculated, the temperature increase of the fluid due to friction was found to be insignificant enough to have an apparent effect on the process so it was ignored. The equation used to calculate power consumption is as follows.

P=ΔPqη#11

Furnace

Normally, heat exchangers are used to provide heat to fluidized beds with steam as the heating fluid. 1232K is too high to use steam, superheated steam was proposed but the requirement of special to withstand significant heat and pressure stress in the heat exchanger was deemed too expensive. Instead, a furnace was used to heat the fluidized bed. The reactor bed will be placed inside of the furnace and heated externally, which has been shown to work on a small scale. This is because fuel combustion will be release H₂, CO, CH₄, N₂and S which will all react with char to make the reaction extremely complex. To operate the bed for 8000 hours (efficiency 85%) the power requirement was found to be 2680 kW. The steel walls were designed to withstand high temperature damage yet still transfer heat. The wall thickness is 50.2mm and thermal conductivity is 30 Wm^-2K^-1. Using Microsoft excel, an area of 5.50 m²was found to suffice the energy demand.

Q=hs A∆TLM#12

Cyclone separator

The cyclone separator was designed to remove fines from the outgoing flue gas to prevent abrasion of the equipment internals. Cyclones must have substantially large inlets to allow axial rotation which will generate a centrifugal separating force. The particle distribution was 40 – 500

µm hence qualifying as Geldhart group B particles. Using the equation shown below the lower boundary of particle diameter was used to size the cyclone. The small particle size meant a higher centrifugal velocity was used which was found to be 1.02 ms^-1. Only one internal cyclone was required which operated with 96% efficiency.

vradial= ρp-ρairRw2d218μ#13a

vradial=Qin2πRL#13b

Discussion of Assumptions

Bed height is expected to experience a 10% increase from minimum fluidization to the point where superficial velocity controls operation. At full operation superficial velocity 0.3 m/s, the void fraction 0.43 is calculated for the expanded bed and assumed to remain constant despite the increase in bed height. As diameter stays constant and height increases volume must increase therefore, for the same number of particles the void fraction must increase but it is assumed that this stays constant. Most of the literature analysed makes the same assumption with limited explanation. It is thought the assumption is made for simplification purposes.

The reaction causes char particle density to increase because of the adsorbed oxygen particles but, for simplicity the change in char density is postulated to be insignificant therefore, a model to account for the variation of density need not be included in this report.

All carbonaceous material experiences carbon burn-off which is not mentioned in this report because it is assumed that carbon burn-off occurred during the plastic waste pyrolysis preceding the gasification reaction. Carbon burns begins to burn off at around 573K, given the ultra-high furnace temperature used to pyrolyze plastic. The char entering the fluidized bed is taken to be inconsistent of particles waiting to burn off.

The mass balance compositions were made on the idealistic assumption that char contains no volatiles, sulfur, chlorine or plastic waste material. This assumption was made to simplify reaction mechanism because there would be several reactions going on at the same time which would be difficult to model.

The fluidized bed is heated unorthodoxically from the side of the reactor instead of from the bottom which is seen normally in industry. Only the bed is required to be heated hence the set of burners can be visually imagined to be similar to a cooling jacket however instead of a cooling medium pressurised fuel is burnt in a number of burners and used to heat the vessel walls horizontally. The pressure allows the fuel to heat the vessel horizontally. Radial and axial conduction and conversion heat the reactor bed.

The heat produced by the exothermic nature of this reaction is stipulated to act as a source of internal heat and accounts for the heat lost though the walls of the freeboard section of the vessel. It is assumed that char particles are evenly distributed in the bed hence heat is released homogeneously throughout the bed at the same rate however this would cause hot spots at the reactor walls and a radial temperature gradient. However, a further assumption that the reactor bed is very well-mixed minimizes the temperature distributions. In addition, temperature increase decreases the density of the gasification fluid which will lead to a lowering in the superficial velocity which reduces the quality of fluidization. But, it is assumed that the CO₂is incompressible hence its density remains constant.

Due to lack of specific factor and costing data for fluidized bed reactors, the vessel was costed using a data for a vertical pressure vessel.

Char particles are assumed to be spherical, although in reality can range in shapes and sizes.

The 7% of unconverted char is assumed to be ash and is sieved out of the activated carbon collector.

Optimisation

Particle diameter

Particle diameter was found to have a profound effect on the process economics. The particle diameter size is shown to predict an increase in residence time, it is clear from figure 7 that especially for larger particle sizes the time required to heat the particles up to 1232 K is longer. In addition, to the cost assumed by increased heating time it is also acknowledged that an increase in individual particle size increases their volume therefore also increases the bed volume. This causes further costs and optimisation discussion as either the reactor length or diameter must then increase to suffice for the increase in volume.

Figure 7: The relationship between reactor residence time and particle diameter at 950 K.

Gasification fluid adiabatic inlet temperature

The CO₂ reactor adiabatic inlet temperature was considered during optimisation. The duty of this feed was measured at various feed inlet temperatures. The results presented in figure 8 highlight that the feed duty decreases as inlet temperature increases because less furnace heat is needed to be transferred to the pipe which carries CO₂. This heat is instead to be supplied by a pre-heater but due to the short time needed to heat the gas in the furnace to 1232K a cost evaluation of the proposed heat exchanger was made to discover whether installation of this heat source will be worthwhile compared to heating the stream to reacting temperature upon entry to the reactor.

Figure 8: The relationship between CO2 heat exchanger inlet temperature, heating duty and cost.

The optimisation results showed that it would be more power effective to pre-heat the stream but on taking into the cost of installation and future maintenance costs the savings were not large enough to justify the construction of an additional heat exchanger system. Furthermore, the plant is expected to produce Activated carbon for the next 20 years however, heat exchangers have an average life expectance of 15 to 20 years so there is a possibility it may need changing which will increase costs. Using equipment cost correlations mentioned in part 10.0 the heat exchanger cost was calculated to be £ 6223 which when added to operating cost produced a minimum combined cost of £106,00 which was far too high to be considered further.

Sensitivity study

Bubble size

Accurate predication of bubble size in a gasification reaction is essential to the hydronamic understanding and of the bed. This calculation is important because it determines whether the flow in the vessel will cause slugging which occurs when the bubble diameter is greater than that of the equipment. A study by Cai et al (1994), found that increasing temperature inside of the reactor causes an increase in bubble size. Bubble size can be found through the equation shown below which was used to predicted bubble size in this study. Its findings are illustrated in figure 9.

dB=(U-Umf12h+h034)g14#14

Figure 9 : The effect of bed temperature on bubble size.

The findings of this study clearly show a positive correlation between the two variables. From the results it is seen that a 1000K increase in temperature causes bubble diameter to increase by 0.0019m which concluded to be an insufficient amount of variation to require further analysis or prevention.

Fluidization velocity was also found to affect bubble formation. Keeping the minimum fluidization velocity constant it shown by figure 10 that increasing the velocity of gasification fluid causes an increase in the particle diameter. It is possible there may be fluctuations in the velocity of fluid up to 10% hence it was important to determine whether the system would be capable of operating in case of such an event. The reactor diameter is 0.5m however was constrained to 0.45m for these calculations in order reduce the possibility of slugging to minimum. Bubble mechanics the system maintains sensitivity to superficial velocity but it is clear that no further precautions need to be taken the bubble diameter remains substantially below the boundary which may begin to cause slugged flow.

Figure 10: the relationship between fluid velocity and bubble size.

Effect of Temperature and Pressure

Increasing pressure has two effects; increasing the density of gasification gas which increases the power required to drive the fan as it is dependent on the mass flow rate of the constituent fluid. Furthermore, increases the convective effect which the gas has on particles. Additionally, increasing the heat transfer coefficient as the increase of pressure increases the suppression of bubble formation. Consequently, improving fluidization at the interaction between heat source and bed, heat transfer mechanisms; conduction and convection extend this effect across the bed.

The increase of temperature counteracts the effects of pressure increase. The gas experiences a decrease in density which is strongly coupled with convective heat transfer which also decreases. Hence both parameters must be controlled in a simultaneous manner. Further discussed in 10.0.

Superficial velocity

The system is very sensitive to changes in gasification flowrate this is thought to be because of the small particles used. As the range between fluidization velocities is small, changes considered insignificant by other processes will have a large effect on the pressure drop in a fluidized bed. The char particles to be gasified are measured in microns, although the particles are rather dense not much vertical force is required to displace the effect of weight and gravity acting on an average sized char particle (150 micron diameter) which could cause larger particles to enter the cyclone where they may be trapped because the cyclone is designed to remove the smallest of particles which range between 300-500 microns. In the graph shown below we see how an 0.05 m/s increase from 0.6 m/s to 0.65 m/s can cause an increase 4500 kPa rise in pressure drop. Unwarranted changes in pressure drop may necessitate re-evaluation of the bed dimensions. To prevent this event from occurring exceptional effort is made in the process control of this fluidized bed to ensure that the superficial velocity remains at 0.3 m/s with very little deviation a maximum of

∓0.05% is recommended.

Figure 11: The variance of pressure drop with superficial velocity.

Mechanical design

Reactor Thickness

Stress-strain behaviour of metals varies depending on the temperature to which it is heated. Essentially, as suggested by the ASM metals handbook, for continuous operation Grade 310 Stainless steel was used to form the vessel shell because it boasts strong mechanical performance at elevated temperatures. The structural properties such as expansion coefficient, young’s modulus and tensile stress of Grade 310 stainless steel are suitable at temperatures reached by this process. Due to the cost of high-performance material thin walls are desirable but the vessel must be able to operate under the stresses explored in part 9.2 (Chasse and Singh, 2013).

Grade 310 was designed especially for high temperature operation, it also possesses impeccable resistance to scaling and corrosion due the 24% chromium content used in the sample chosen for the vessel wall. The material is subjected to a ductility restoration process after each 1000 hours of service above 923K using an annealing technique. At 1223K the metal has a thermal conductivity of 18.7 W/mK (AZoM.com, 2018). To withstand the loads imposed on the reactor vessel. A thickness of 44mm was found to suffice the heat transfer through the vessel walls.

Stresses

There are four principal stresses which arising from the loads burdened by the reacting vessel. Stress calculations form an important part of mechanical design as small inaccuracies can lead to major faults in process engineering. Therefore, the equipment is to be manufactured with a ±0.05% allowance (Towler and Sinnott, 2013).

Longitudinal stress and circumferential stress

Longitudinal and circumferential stress are the axial and radial stresses experienced by cylindrical reactors, their values are 8.4 and 4.2 Nm-2 respectively. Although nominal, the calculation of longitudinal stress caused a slight increase in shell thickness as mentioned in 10.1. The stresses can be determined using the equations below.

σh=PDi2t#15a

σL=PDi4t#15b

Direct stress

The stress due to the weight of the vessel and its contents is known as direct stress. The value 568kNm-2 is calculated assuming the reactor is operational at optimal condition inhabited by the reacting material.

σw=WπDi+tt#16

Bending stress

The predominant factor in bending stresses for tall vessels is wind loads, it is assumed to be the only variant of bending stress on the bed reactor as the wind load acting on the cyclone was found to be negligible. Bending stress is a good indicator of how much support a vessel will need, which is discussed further in this section. The value was calculated to be 684 kNm-2, using the equation below.

σb=∓MIvDi2+t#17a

Iv=π64D04-Di4#17b

Support sizing

Vessel support must provide adequate strength and ability to support maximum system load. A simple saddle design was chosen to support the dead weight imposed by the reactor and its constituents. The reactor dead weight was calculated assuming full operation and inclusion of the cyclone dead weight, it was found to be 2.402 kN whereas the dead weight stress was 0.615 nmm-2. The vessel diameter correlated to an interpolated maximum weight of 32 kN. This produced a saddle with diameter of 0.45. Thermal expansion due to high temperature operation is possible which is accounted for in the support sizing data found (Towler and Sinnott, 2013).

Distributor

The gas distributor is designed to maximize gas-solid contact, the chosen design is shown in figure 12 and made from nickel alloy 900 for its high-temperature resistance. Erosion is not a problem; local erosion of distributor parts is eliminated by reducing the jet length. Pressure drop plays a key role in successful design of gas distributors. It is well known that the pressure drop must be high enough to fluidize the bed so that all areas of the bed are fluidized uniformly. The gas distributor pressure drop is calculated to be 73 kPA using the equation below.

∆Pd=∆Pbvmfvset1b11-εmf#18

Related image

Figure 12: An illustration of the fluid bed distributor.

Economics

Equipment cost

Each component of the design system was evaluated to give an indication of the capital cost associated with the process. Total capital cost was estimated by using hand correlations to effectively value equipment depending on size and equipment type. The relationship between size and cost is given using correlation 19a which provided a guide for pricing the reactor (Grade 310 Stainless steel) and furnace (Carbon Steel). The plant life is set to 20 years and interest rate 0.5% is used to estimate the annualised cost of capital (Towler and Sinnott, 2013).

Ce=a+bSy#19a

CF=F∑CE#19b

ACC=CF×i1+1y1+iy-1#19c

CEPCI2018=CEPCI2006×I2018I2006#19d

Table 3: cost of equipment.

Cost	Total cost	Lang factor	TISBL	ACC £/y
cyclone	311	2.5	777.5	49.87439259
furnace	11254	2	22508	1443.823574
vessel	12412	4	49648	3184.776647
centrifugal fan	842	4	3368	216.0475296
			76301.5	4894.522143

Operating cost

Operating costs can bring success or failure to an engineering project. They depend mainly on utility costs which vary depending on the type of energy source and its current market value, geographical location also affects utility costs. The price of gas which is heated in the furnace is 0.025 £/kWh. There are only two processes which necessitate utilities; the burning of fuel to heat the reactor and energy required to power the centrifugal fan. The centrifugal fan is much smaller than the furnace and requires a relatively small amount of power therefore, uses electrical energy which is priced at 0.12 £/kWh.

Table 4: Operating costs.

kW	kWh	Price per kwh	Annual operating price	Pounds
2680	21440000	0.025	536000	5360
2	16000	0.12	1920	19.2
				537920

Cost of reactants

Char and CO2 are supplied to the process at no added cost, both materials are co-products of the preceding pyrolysis reaction where CO2 is a by-product of syngas combustion and is separate from water in a fractional distillation column.

Process control

Temperature control

The source of heat originates from the furnace. The temperature produced in th furnace must be controlled to ensure that enough heat is transferred through the vessel walls. Conversely, control is also important for preventing temperatures from increasing to a point which may begin to cause damage to the metal shell or increase the rate of reaction to an extent which will irreversbly affect the properties of char particles.

The furnace is equipped with a gas furnace thermocouple which is connected to the valve regulating the flow of fuel in to the furnace. The valve reduces gas flow when the temperature rises to temperatures approaching the upper constraint placed on the controller. In the case of system malfunction the thermocouple is designed to shutdown the furnace to prevent a fire from occuring.

Pressure control

Pressure in this system is indirectly controlled by the thermocouple and gasification fluid inlet valve. The variation of temperature causes gas to expand which increases pressure within the vessel whereas increasing the flow rate of entering CO₂ can cause pressure within the vessel to increase as there is more material in the reactor bed at a given time.

The vessel is to be fitted with a pressure relief valve in the case of elevated pressures also a pressure sensor which is to be inter-looped with flow and temperature controllers. The pressure sensor will send the necessary instructions to the flow and temperature controllers.

Flow control

The flow rate of CO₂is controlled by a sensor-valve response system which detects a variation in flow then acts immediately to alter the flow rate as necessary.

Level control

A level indicator is situated at 0.3m above the idle reactor bed to account for the proposed 10% expansion. 3.3m is the maximum bed length for optimum fluidization. If exceeded the level indicator (depending on the cause) feeds this information directly to the valve which regulates CO₂flow rate in order to reduce it or the thermocouple which reduces the fuel entering the furnace. Increase in level can cause larger particles to enter the cyclone which will damage the internal material as it is made to separate very small particles from the flue gas.

Design safety considerations

Start up and shutdown

It is important that the major safety issues in a chemical engineering process are addressed. High temperatures and wear and tear being the main issues in this process. Automatic shutdown has been discussed in the previous section, 11.0. However, for safe operation during start up the following procedure should be followed:

Reactor is purged with nitrogen to remove all impurities and inert the system.
The furnace is activated and used to pre-heat the bed to 1223K
The reactor if filled up to 3m which forms the bed
The CO₂valve is opened to permit fluidization to begin
The system is monitored from the minimum fluidization point until superficial velocity has been reached

A simple shutdown procedure must also be implemented to assure there are no outstanding risks upon ceasing the process. The gasification fluid flowrate is gradually reduced until it reaches 0m/s, when this is reached the bed reaches a standstill. The furnace is then turned off flue gas drained from the system.

Safety and maintenance

Proposed safe manning substantiates that all tasks must be operated by operators who have passed the required safety examinations. Unexperienced members of staff must not be allowed to carry out maintenance unsupervised, supervision by experienced professionals with a proven record of accomplishment should operate all procedures. Recommended staffing arrangements for small-medium sized plants suggest that too few staff are present to control and maintain process safety. Eatec Uk Ltd recommends that a minimum of two qualified operator are present at all times (Cooper and Dolbey Jones, 2002). Furthermore, plants should seek to hire safety staff who are experts on given processes and vessels because this reduces complications or lack of knowledge when unexpected events arise.

All chemical engineering processes require regular maintenance to ensure firstly that the system is operating as it should be for the given point in its lifespan but also as a preventative mechanism to reduce potential downtime due to breakage. Although the metal used has a 15-year lifespan it is necessary to ensure that the reactor shell is not experiencing more wear and tear than it should be. The NDT structural integrity test is maintenance method to be used, it is a comprehensive structural examination which examines the adequacy of a material (Andrzej and Marta, 2014). Auxiliary equipment must also be thoroughly tested every 6-18 months.

Figure 12: Piping and Instrumentation Diagram

Design specifications

Figure 13 shows a schematic diagram representing the physical set up of the fluidized bed with its two-supporting equipment: furnace and cyclone. The auxiliary diagram shows how the heating element provides heat through the walls of the reactor vessel. A metal grate is used to distribute heat equally across the 3.3m height of the bed whilst under operation. The char inlet is shown on by the inverted triangle on the left-hand side of the reactor whereas the inlet for the said gasification fluid is depicted by the inlet tap at the bottom of the reactor. However, it is must be mentioned that this diagram is not to scale and should be used for illustrative purposes only.

Figure 13: Fluidized bed Reactor set up.

	SPECIFICATION SHEET: Fluidized Bed Reactor
Cedrick Ofori 18/03/2018
PROCESS DATA

Solid particle			Char
Particle shape			Spherical
Particle diameter		(micron)	150
Particle density		kg/m³	1490
Bed length		m	3
Bed diameter		m	0.5
Bed volume		m³	0.0
Bed outer surface area		m²	5.2
REACTOR INLET
Fluid flow direction			Up
Adiabatic Temperature		^oC	25
Pressure		atm	1
Solid flow		kg/h	170
Fluid flow		kg/h	323.1
Fluid dynamic viscosity		Ns/m²	0.0000459
REACTOR OUTLET
Adiabatic Temperature		^oC	950
Pressure		atm	1
Solid flow		kg/h	368.8
Fluid flow		kg/h	11.2
Fluid viscosity		Ns/m²	0.0000421
CONSTRUCTION & MATERIALS
Shell material			Grade 310 Stainless steel
Shell inner diameter		m	0.5
Shell outer diameter		m	0.44
Shell length		m	7.42
SADDLE
Saddle material			Grade 310 Stainless steel
Saddle leg thickness		m	0.12
Distributor material		m	Grade 310 Stainless steel
Distributor diameter		m	0.14

Figure 14: Fluidized Bed Reactor Specification table.

Conclusion and Recommendations

In this paper a technique detailing the production of commercial grade activated carbon from pyrolytic chars has been explored. The investigation has concluded that a bubbling fluidised bed reactor heated by a furnace can be used to complete this process with 93% conversion. The results from this report should be used as a basis for understanding the fundamentals of this process and it is suggested that more research should be done to understand the mechanisms involved in the hydronamic performance in the fluidized bed system. Furthermore, heating of the fluidized bed must be improved, perhaps an internal heating system which is able to transfer heat to the reactor bed at high temperatures would be more ideal than the furnace system designed in this report.

Firstly, more laboratory experimentation must be performed to confirm char behaves as depicted by the random pore model which is used to calculate the reaction kinetics. The study which suggested the random pore model for porous carbonaceous material was developed in 1987. In the thirty years since its invention it has been adapted multiple times by new researchers which argues that there is a good probability that there are better models which explore more factors affecting char during its gasification thus, will give a more correct reaction kinetics. More effort must be made to discover the most ideal model or through experiments find a model which models chars reactivity well.

Symbol	Definition	Symbol	Definition
K_x	Rate constant	∆p	Pressure drop
k_PRM	Random Pore Model rate constant	L	Length
α	Modified volume parameters	D	Diameter
β	Modified volume parameters	g	Gravity
X	Conversion	ε	Void space
t	time	v	Velocity
SBET	BET surface area	μ	Viscosity
r_g	Granule radius	P	Power
b	Pore volume parameter	η	Efficiency
τ	Dimensionless time	q	Volumetric flow
vB	Volume of pores	Q	Duty
n	Empirical parameter	h	Heat transfer coefficient
G	Concentration of reactant solid	S	Thickness
ρ	Density	A	Area
ψ	Structural parameter	∆TLM	Log mean temperature difference
Eact	Activation temperature	R	External radius
R	Ideal gas constant	U	Fluidization velocity
T	Temperature	W	Dead weight
A	Pre exponential constant	π	PI
L₀	Pore length	I	Moment of inertia
S₀	Pore surface area	M	Mass
V₀	Average pore volume	F	Installation factor
K	Kelvin	C_F	Total fixed capital cost
C_p	Specific heat capacity	C_e	Total delivered cost
m	Mass flow rate	ACC	Annualised capital cost
		y	Plant life

Figure 15: Nomenclature.

mf	Minimum fluidization
p	Particle
f	Fluid
set	Settling velocity
0	bottom

Figure 16: Subscript Nomenclature.

References

Andrzej, M. and Marta, M. (2014). Modern NDT Systems for Structural Integrity Examination of Concrete Bridge Structures. Procedia Engineering, 91, pp.418-423.
AZoM.com. (2018). Stainless Steel – Grade 310 (UNS S31000). [online] Available at: https://www.azom.com/article.aspx?ArticleID=966 [Accessed 22 Mar. 2018].
Bhatia, S. and Perlmutter, D. (1980). A random pore model for fluid-solid reactions: I. Isothermal, kinetic control. AIChE Journal, 26(3), pp.379-386.
Booksite.elsevier.com. (2018). Elsevier: Towler, Sinnott: Chemical Engineering Design: Principles, Practice and Economics of Plant and Process Design, 2nd Edition Welcome. [online] Available at: http://booksite.elsevier.com/9780080966595/ [Accessed 22 Mar. 2018].
Chasse, K. and Singh, P. (2013). Corrosion Study of Super Ferritic Stainless Steel UNS S44660 (26Cr-3Ni-3Mo) and Several Other Stainless Steel Grades (UNS S31603, S32101, and S32205) in Caustic Solution Containing Sodium Sulfide. Metallurgical and Materials Transactions A, 44(11), pp.5039-5053.
Cooper, W. and Dolbey Jones, D. (2002). Electrical safety engineering. Oxford: Newnes.
Everson, R., Neomagus, H. and Kaitano, R. (2011). The random pore model with intraparticle diffusion for the description of combustion of char particles derived from mineral- and inertinite rich coal. Fuel, 90(7), pp.2347-2352.
Lee, J. and Kim, S. (1996). Gasification kinetics of waste tire-char with CO2 in a thermobalance reactor. Fuel and Energy Abstracts, 37(3), p.202.
Matsui, I., Kunii, D. and Furusawa, T. (1987). Study of char gasification by carbon dioxide. 1. Kinetic study by thermogravimetric analysis. Industrial & Engineering Chemistry Research, 26(1), pp.91-95.
Molina-Sabio, M., González, J. and Rodrı́guez-Reinoso, F. (2004). Adsorption of NH3 and H2S on activated carbon and activated carbon–sepiolite pellets. Carbon, 42(2), pp.448-450.
Office of coal Research, U. (1962). Research and Development Report – Office of Coal Research.
Tchoffor, P., Davidsson, K. and Thunman, H. (2015). Production of Activated Carbon within the Dual Fluidized Bed Gasification Process. Industrial & Engineering Chemistry Research, 54(15), pp.3761-3766.
Towler, G. and Sinnott, R. (2013). Chemical engineering design. Boston, MA: Butterworth-Heinemann.
Warnecke, R. (2000). Gasification of biomass: comparison of fixed bed and fluidized bed gasifier. Biomass and Bioenergy, 18(6), pp.489-497.
Mahamud, M., Menéndez, J. and Sarquís, P. (2014). CO2 activation of chars: Effect of burn-off on texture and fractal properties. Fuel Processing Technology, 119, pp.41-51.
Roberts, D. and Harris, D. (2006). A Kinetic Analysis of Coal Char Gasification Reactions at High Pressures. Energy & Fuels, 20(6), pp.2314-2320.