Energy Fuels 2010, 24, 4693–4702 Published on Web 03/11/2010
: DOI:10.1021/ef901249g
Incorporating Energy Generation into Volatile Organic Compound (VOC) Emission Treatment Using a Solid Oxide Fuel Cell: A Model-Based Approach† Dhananjai Borwankar, Michael Fowler, and William A. Anderson* Department of Chemical Engineering, University of Waterloo, 200 University Avenue West, Waterloo, Ontario N2L 3G1, Canada Received October 30, 2009. Revised Manuscript Received February 19, 2010
Industrial processes that use solvent-based coatings emit volatile organic compounds (VOCs), which when released into the environment are smog precursors. The purpose of this work is to develop a VOC abatement technology that not only destroys such VOCs with high efficiency but also extracts the energy found in these compounds to do useful work elsewhere in the facility. To this end, a model-based design approach using Aspen HYSYS was used to develop and optimize an abatement system that consisted of three separate technologies: an adsorber, a reformer, and a solid oxide fuel cell (SOFC). A model was developed that integrated the technologies, allowing for optimization of the overall operating conditions and performance. The reformer and SOFC portion of the model was validated by a comparison to published literature results for methane as a feedstock. After optimization, the model indicated that this system could achieve 95% VOC removal efficiency, with an electrical efficiency of 49%. When heat integration is considered, the portion of energy recovered increases to 85%. the combustion chamber temperature. For example, a facility that emits 100 tons of VOCs annually at an exhaust rate of 28.3 m3/s (60 000 cfm) will require $658 500 of natural gas annually to operate a regenerative-type thermal oxidizer [calculated using Environmental Protection Agency’s (EPA’s) Air Pollution Cost Control Manual].4 Therefore, it is imperative that new technologies are developed that can reduce the cost burden typically associated with VOC abatement. One way to reduce costs is to add value during the treatment process by either creating a secondary product or commodity for use within the facility. Because VOCs contain a significant amount of energy, a logical first goal would be to generate energy from the process. If this energy could be harnessed, it could provide a substantial benefit to the operations within the plant. In this research, a model-based design is used for the development of hybrid solid oxide fuel cell (SOFC) abatement technology. The basic processes in the hybrid technology include (1) isolation and concentration of VOCs from solventbased coating process emissions using adsorption recovery technologies, (2) reforming of the VOCs to form a mixture consisting primarily of hydrogen and carbon monoxide, and (3) oxidation of hydrogen and carbon monoxide to water and carbon dioxide by air through a SOFC to capture energy for usable work.
Introduction Industrial solvent-based coating and painting processes emit significant amounts of volatile organic compounds (VOCs). When emitted into the environment, these compounds will react with nitrogen oxides in the presence of sunlight to form ground-level ozone, a component of smog. Ground-level ozone can adversely affect climate change, human health, and the growth of plants.1,2 For this reason, governments around the world have implemented legislation to guide industry in responsibly handling these problems. In many cases, this new legislation will decrease previously acceptable threshold values for these compounds, causing facilities that were once compliant to fall out of compliance. For existing facilities with established processes, the quickest solutions are those that treat the emission after it has been released or “end-of-pipe” technologies. The problem with these technologies is that they have high operational costs with low operational flexibility. For facility management, it becomes a choice of spending money without seeing a return on investment. This is because this type of investment, although important, will appear only as rising operational costs. An example of a widely used and accepted technology for VOC abatement is the regenerative thermal oxidizer (RTO). The RTO converts VOCs by raising and maintaining their temperature above their auto-ignition temperature in the presence of sufficient oxygen to complete their conversion to carbon dioxide and water.3 Although effective, operational costs are high because of the natural gas required to maintain
Model Development The software package being used for the model design of the hybrid SOFC abatement system under development in this project is Aspentech HYSYS,5 which is a commercial process
† This paper has been designated for the Bioenergy and Green Engineering special section. *To whom correspondence should be addressed. Telephone: (519) 888-4567, ext. 35011. E-mail:
[email protected]. (1) Farley, J. M. Annu. Rev. Pharmacol. Toxicol. 1992, 32, 67–88. (2) Volatile Organic Compounds in the Atmosphere; Harrison, R. M., Hester, R. E., Eds.; Royal Society of Chemistry: Cambridge, U.K., 1995; Issues in Environmental Science and Technology, pp 1350-7583. (3) Moretti, E. C. Chem. Eng. Prog. 2002, 98, 30–40.
r 2010 American Chemical Society
(4) Vatavuk, W. M.; Klotz, W. L.; Stallings, R. L.; van der Vaart, D. R.; Spivey, J. J. Handbook of Control Technologies for Hazardous Air Pollutants; Office of Research and Development, United States Environmental Protection Agency (U.S. EPA): Washington, D.C., 2002; pp 195-388. (5) Ng, C.; Smith, P.; Teh, A.; Brenner, S.; Gierer, C.; Strashok, C.; Jamil, A.; Nguyen, N.; Chau, A.; Sachedina, M.; Hugo, L.; Lowe, C.; Hanson, K. HYSYS 3.1 User Guide; Hyprotech: Calgary, Alberta, Canada, 2002; p 458.
4693
pubs.acs.org/EF
Energy Fuels 2010, 24, 4693–4702
: DOI:10.1021/ef901249g
Borwankar et al.
Table 1. Breakdown of Concentrated VOCs substance
annual mass (kg)
xylene MEK toluene nBA total
20000 30260 35094 14463 100000
molecular weight (g/mol)
annual moles (kg mol)
current flow (kg mol h-1)
adjusted flow for continuous operation (kg mol h-1)
106.18 72.12 92.15 116.18
188 419 380 126
0.0495 0.109 0.099 0.032 0.290
0.0232 0.0518 0.0471 0.0155 0.137
Figure 1. Main flowsheet for the hybrid SOFC abatement system model.
Figure 2. Adsorption subsystem flowsheet.
annually. The VOCs are assumed to be comprised of 15% xylene, 35% methyl ethyl ketone (MEK), 40% toluene, and 10% n-butyl acetate (nBA) (Table 1). The air emissions temperature was assumed to be 298 K, with 50% relative humidity. The hybrid SOFC model consists of one main model (shown in Figure 1) separated into four flowsheets: (1) adsorber subsystem (Figure 2), (2) heat-exchanger subsystem (not shown), (3) reformer subsystem (Figure 3), and (4) SOFC stack subsystem (Figure 4). To examine the overall model, follow the “VOC contaminated gas” stream from its entrance into Figure 1 (bottom left corner) until its final conversion to CO2 and H2O as the “system exhaust”. Before discussing the adsorption subsystem, it is important to understand that modeling an adsorption system in HYSYS presents interesting complexities. To begin, there are interphase transport issues, uncertainty in equilibrium effects, and uncertainty in the effect that each component may play upon one
simulator that contains a rigorous thermodynamic and physical property database. It also provides comprehensive built-in process models that address a wide range of steady-state and dynamic operations. The method developed here is to fully use the functions within HYSYS to create all system components. There is no use of userdefined code, and the system will run using only the HYSYS models. The advantage of this methodology is a reduction in computation times, ease of use, and the ability to perform rapid parametric, thermodynamic, and statistical analysis on the overall system. Overall, the result of this work will be a very simple model that can provide a holistic analysis of VOC abatement by a SOFC hybrid abatement system. The model and analysis were developed for a base case representing typical emissions from a moderately sized automotive parts painting facility. The total air flow rate is 28.3 m3/s for 240 days/year, at 16 h/day, with 100 tons of VOC emitted 4694
Energy Fuels 2010, 24, 4693–4702
: DOI:10.1021/ef901249g
Borwankar et al.
begins, the reaction rates of individual hydrocarbons can vary substantially from compound to compound, and even though most higher hydrocarbons react faster than methane (and at lower temperatures), they are susceptible to non-catalytic thermal cracking.8 At the high temperatures associated with fuel cells, thermal reactions begin competing with catalytic reactions, causing the formation of olefins (precursors to coke formation) from these higher hydrocarbons.8 Coke deposits will inactivate the catalyst, causing irreversible harm and deactivating the anode electrode.9 In general, larger hydrocarbon molecules (more carbons) result in slower reaction rates, increasing the risk of thermal cracking.8 For this reason, the model was developed to include an external reforming process. This protects the fuel cell from coke deposition and also provides for a more complete conversion process of the VOCs to fuels suitable for use in the fuel cell. The external reforming process being modeled in the reformer subsystem is autothermal reforming. The process involves two separate reactions: partial oxidation (POX) and SR occurring at different locations in the reactor. POX occurs at the front end of the reactor and involves the reaction of the fuel hydrocarbon with a sub-stoichiometric ratio of oxygen to form a reformate mixture of carbon monoxide and hydrogen.10 In principle, the low ratio of oxygen ensures that the stoichiometry will not allow for full combustion to occur, consuming all of the oxygen but not all of the fuel. One of the main attractions of this type of system is that the process is highly exothermic, operating at approximately 800-1000 °C;10 thus, no heat input is required to sustain the reaction.11 The general reaction is presented below as eq 2.
Figure 3. Reformer subsystem flowsheet.
another during multi-component adsorption. These are only a few of the issues, without even considering the dynamic aspects of the process, which is inherently a batch operation. For this reason, much of the background rigor was simplified and calculated outside of HYSYS using standard adsorption system design methods,11 with the results of these analyses being substituted into simple HYSYS system models made to represent the overall steady-state adsorption-desorption system, as discussed above. In the adsorption subsystem flowsheet (Figure 2), the HYSYS component splitter represents the steady-state adsorption cycle and is used to separate excess air from the VOC contaminated gas. The concentrated VOC stream is now labeled “VOC vapour” and is sent to a mixer model, which combines steam with the concentrated VOC stream to simulate the anticipated desorption stream. A HYSYS condenser model is then used to remove the excess water from the desorption stream. This minimizes the flow rate of feed to the reformer subsystem but, more importantly, increases the VOC concentration to approximately 40% of the remaining gaseous feed. The overall result is a concentrated and humidified stream of VOCs. In this particular case, the painting operation cycle is 16 h/day, whereas the SOFC should be operated continuously (24 h/day). This will prevent having to thermally cycle the SOFC, which could cause damage to many of the components, such as the ceramic electrolyte or interconnect materials. For this reason, a HYSYS Tee model is used to simulate the removal of some VOCs for storage (and use during shutdown periods), while the rest of the desorption flow is sent to the reformer subsystem. The nature of the storage system was not examined in this work; however, a simple example would be another adsorption that could be desorbed at a later time or a pressurized tank. Up to this point, the VOCs have been humidified and concentrated (labeled as HVOC1 in Figure 1). The next step is to convert them into a usable fuel source for the SOFC. In principle, an SOFC can use any combustible fuel, because the high-temperature operation supports internal reforming.6 Internal reforming involves the conversion of the source fuel to a reformate mixture that is compatible with the internal electrochemical reactions of the fuel cell.7 For the SOFC, this mixture will consist of hydrogen and carbon monoxide. The internal fuel reforming step uses the process of steam reforming (SR) to convert hydrocarbons into the reformate mixture. The overall stoichiometry of the reaction is displayed in eq 1. y Δ Cx Hy Oz þ ðx -zÞH2 O f xCO þ x -z þ H2 ð1Þ 2
1 y Cx Hy Oz þ ðx -zÞO2 f xCO þ H2 2 2
ð2Þ
In the second half of the reactor, SR takes place. SR externally is different than SR internally (in the fuel cell), because externally, the reactor temperatures can be tightly controlled. This means lower temperatures can be used (approximately 250-300 °C), eliminating the risk of carbon formation.8 The lower temperatures also permit the use of Group VIII metals as the reformer catalyst, normally nickel.8 The overall process becomes autothermal because the exothermic POX reaction fuels the endothermic SR reaction. In the model, the HVOC1 stream is sent to the reformer subsystem, which is modeled as two separate reactors: one Gibbs reactor for the POX reaction and one Gibbs reactor for the SR reaction. This allowed greater flexibility during optimization and also allowed us to control the temperature of each reaction separately. In HYSYS, the Gibbs reactors calculate energy and material balances by minimizing the Gibbs energy function, where it is assumed the reactions will go to equilibrium. More specifically, in this model, air exiting the heat-exchange subsystem and humidified VOC (HVOC1) coming from the adsorber subsystem (Figure 2) are mixed together in a HYSYS mixer operation and sent to the POX reactor. The outlet of the POX reactor is combined with excess steam to form the inlet reaction mixture for the SR reactor, where the SR reaction and the water-gas shift reaction occur. The presence of excess steam drives the water-gas shift reaction in the forward direction, thus creating more hydrogen from the reaction mixture. To ensure the endothermic SR reactions are sustained only by the energy generated from the POX reactions, the HYSYS set function is used to link them. In the previous step, the humidified VOC stream was transformed into a mixture of hydrogen, carbon monoxide, nitrogen,
SR of higher hydrocarbons occurs in two steps: first, there is an irreversible adsorption onto the catalyst, and second, there is a subsequent cleavage of C-C bonds one by one, until finally only simple single-carbon compounds are remaining.8 Once cleavage
(9) Song, C. Catal. Today 2002, 77, 17–49. (10) Ahmed, S.; Krumpelt, M. Int. J. Hydrogen Energy 2001, 26, 291– 301. (11) Ming, Q. M.; Healey, T.; Allen, L.; Irving, P. Catal. Today 2002, 77, 51–64.
(6) McIntosh, S.; Gorte, R. J. Chem. Rev. 2004, 104, 4845–4866. (7) Ormerod, R. M. Chem. Soc. Rev. 2003, 32, 17–28. (8) Joensen, F.; Rostrup-Nielsen, J. R. J. Power Sources 2002, 105, 195–201.
4695
Energy Fuels 2010, 24, 4693–4702
: DOI:10.1021/ef901249g
Borwankar et al.
Figure 4. SOFC subsystem flowsheet.
basis of the assumption that 20.5% of the air used is composed of oxygen. The air flow to the SOFC is preheated by exchanging heat with the cathode exhaust stream. Preheated, the air stream is sent through a component splitter, which is used to simulate the movement of the oxygen anion across the electrolyte. Once through the component splitter, the air stream is separated into two streams: one consisting only of oxygen (called cathode O2 in) and another consisting of nitrogen, unused oxygen, and argon (excess air). The cathode O2 in stream is then sent to the Gibbs reactor representing the anode. The excess air stream, which has been heated through a HYSYS heater model, simulates unused air moving through the SOFC, absorbing heat from the fuel cell and exiting at the cell temperature. Once through the heater, the excess air stream is termed the cathode exhaust stream. On the anode side of the SOFC, the reformed fuel proceeds to a HYSYS Tee model, which is used to set the amount of fuel used in the reaction. The used fuel portion is sent to the Gibbs reactor labeled as the anode. Here, the conversion of the reformed fuel to the exhaust components (primarily carbon dioxide and water) is simulated. Meanwhile, the unused portion of the fuel is sent to a HYSYS heater model to simulate its heat absorption as it moves through the SOFC. Finally, the unused fuel (unused fuel 1) is mixed with the anode exhaust to become the true anode exhaust (labeled as anode exhaust 1), which must now be split via a HYSYS Tee model to allow for the recycling of a percentage of the unused fuel. The anode exhaust is then sent to the heat-exchange subsystem to preheat the air and water streams being used in the reformer subsystem. The fuel to be recycled is sent through a HYSYS recycle function, which installs a theoretical block in the process stream5 to assist with model convergence calculations. The feed into the block is termed the calculated recycle stream, and the product is the assumed recycle stream. The following steps take place during the convergence process:5 (1) HYSYS uses the conditions of the assumed stream and solves the flowsheet up to the calculated stream. (2) HYSYS then compares the values of the calculated stream to those in the assumed stream. (3) On the basis of the difference between the values, HYSYS modifies the values in the calculated stream and passes the modified values to the assumed stream. (4) The calculation process repeats until the values in the calculated stream match those in the assumed stream within specified tolerances.
steam, and minor quantities of CO2. This converted stream (now called the reformer out in Figure 1) is sent to the SOFC subsystem for final conversion and energy extraction. The basic operation of the SOFC is outlined in the following steps: (1) At the cathode, oxygen from air is transformed into oxide ions by receiving electrons from the external load. (2) The oxide ion travels from the cathode, through the ceramic electrolyte, and to the reaction site of the anode. (3) In the anode, the oxide ion combines with the reformed fuel (hydrogen and carbon monoxide) to form water and carbon dioxide, releasing electrons to the external circuit and venting gaseous water and carbon dioxide. (4) The electrons travel through the external circuit, through the load, and back to the cathode to begin the process again with fresh oxidant and fresh fuel. The electrochemical reactions taking place are outlined in eqs 3-7. Anode Reactions H2 þ O- f H2 O þ 2eð3Þ CO þ O2- f CO2 þ 2e-
ð4Þ
Cathode Reactions O2 þ 4e- f 2O-
ð5Þ
Overall SOFC Reactions 1 H2 þ O2 f H2 O 2
ð6Þ
1 ð7Þ CO þ O2 f CO2 2 Modeling of a SOFC in HYSYS is complicated because there are no unit operations that have been built to represent the electrochemical reactions occurring in the fuel cell. As a result, the reactions taking place above are simulated using a combination of HYSYS unit operations. In this model, the reformer outlet stream (consisting primarily of nitrogen, hydrogen, carbon monoxide, and steam) was sent to the anode side of the SOFC. This mixture is sent to a component splitter that is used to set the fuel use factor. The fuel being used is sent to a HYSYS Gibbs reactor with the appropriate stoichiometric ratio of oxygen from the incoming air stream. The appropriate molar flow of air required is calculated via the spreadsheet function in HYSYS. This is performed by first setting a desired air use factor and using it to determine the required oxygen content to ensure complete conversion. The air flow is calculated on the 4696
Energy Fuels 2010, 24, 4693–4702
: DOI:10.1021/ef901249g
Borwankar et al.
Fuel Cell Performance Calculations The main purpose of this model is to determine the amount of power that can be produced from a fuel cell using VOCladen process fumes as its fuel source. The main relationship between the reactions taking place in the fuel cell and the power developed by the fuel cell is the Nerst equation. RT ln Q ð8Þ nF The Nerst equation represents the reversible potential of an ideal fuel cell. In it, E0 represents the potential at the standard temperature and pressure, R is the universal gas constant, T is the cell temperature, and Q is the reaction quotient. The reaction quotient consists of the ratio of products over reactants at the outlet of the fuel cell. It can be represented in several ways, the first of which is to convert the reactants and products into a function of total pressure using mole fractions. In the equation below, the reaction quotient Q is the term in the square brackets. 2 3 ΔE ¼ ΔE 0 þ
E ¼ E0 þ
RT 6 XH2 OAnO 7 7 ln6 1 15 nF 4 XH2 AnO XO2 CaO PT 2 2
Figure 5. Representation of the effect of pressure and temperature on the ideal cell voltage for the oxidation of carbon monoxide, hydrogen, and methane.
quantified and applied to a plot of voltage versus current density within the cell is called the polarization curve. For a fuel cell running on pure hydrogen and oxygen, at 25 °C, and operating at atmospheric pressure, the ideal fuel cell voltage will be -1.23 V.12 When no current is running through the cell, this ideal potential difference is realized [in Figure 5, this is called the theoretical electromotive force (EMF) or ideal cell voltage]. Once current begins to flow through the cell, the actual work being extracted deviates from the ideal case, as depicted by the curve in Figure 5. This curve represents losses from three main sources, each of which exerts pronounced effects during different regions of the curve. At low current densities, the reaction is slow, reflecting sluggish electrode kinetics.12 At this point, the reaction is trying to overcome the activation energy of the reaction, which results in the pronounced drop in voltage, as outlined above. Stated another way, the activation potential is the extra potential required to overcome the kinetic barriers of the reaction. It can be quantified using the Tafel equation and is called activation polarization (eq 15). RT i ð15Þ ln ηAct ¼ RnF i0
ð9Þ
In eq 9, XH2OAnO is the mole fraction of water at the anode outlet, XH2AnO is the mole fraction of hydrogen at the anode outlet, XO2CaO is the mole fraction of oxygen at the cathode outlet, and PT is the total pressure of the cell. Equation 9 can be further manipulated to obtain an expression that will yield a value for the ideal potential as a function of fuel use. Equation 10 defines fuel use (UfH2), and eqs 11 and 12 define the respective mole fractions in terms of fuel use. Equation 14 combines eqs 8-13 to obtain the overall expression of the ideal potential of a fuel cell running on hydrogen and air. XH2 AnO XH2 AnI
ð10Þ
XH2 AnO ¼ 1 -UfH2
ð11Þ
XH2 OAnO ¼ UfH2
ð12Þ
UfH2 ¼ 1 -
λ -UfH2 λ -UfH2 0:21 2 1=2 3 λ -UfH2 6 UfH2 7 7 RT 6 0:21 7 ¼ E0 þ ln6 6 7 1=2 nF 4ð1 -UfH2 Þ½ðλ -UfH2 ÞPT 5 XO2 CaO ¼
Eideal
ð13Þ
In eq 15, ηAct is the drop in voltage because of activation polarization, R is the electron-transfer coefficient, i is the current density, and i0 is the exchange current density. Higher exchange current densities reflect reactions with fast kinetics, meaning that a smaller activation polarization needs to be used to overcome the kinetic barriers. Conversely, small exchange current densities result in larger activation polarization values. On the extreme right of the curve, the reaction rate is very high and reactants are being used much quicker than can be supplied to the reaction sites.12 Essentially mass transfer of reactants is the controlling step, resulting in a severe drop in cell potential.12 The mathematical expression quantifying the contribution to voltage losses from this phenomenon is called concentration polarization and presented in eq 16. RT i ln 1 ð16Þ ηConc ¼ nF iL
ð14Þ
In eqs 13 and 14, λ refers to the ratio of excess air (not oxygen) being supplied to the cell. Equation 14 is an important one because it relates temperature, pressure, and fuel use together in one simplified relationship, which will yield the reversible potential of the fuel cell. Although very useful, eq 14 is limited because it does not account for voltage loss as a result of mass transfer, kinetic effects, or resistances within cellular components. The contribution to voltage loss from these three terms when (12) Winkler, W.; Nehter, P. Thermodynamics of fuel cells. In Modeling Solid Oxide Fuel Cells, Methods, Procedures and Techniques; Bove, R., Ubertini, S., Eds.; Springer Science þ Business Media: New York, 2008; Vol. 1, pp 13-50.
In eq 16, ηConc refers to the polarization effect on voltage as a result of mass-transfer issues and iL refers to the limiting current density, which is a measure of the maximum rate at 4697
Energy Fuels 2010, 24, 4693–4702
: DOI:10.1021/ef901249g
Borwankar et al.
Table 2. Comparison between the Staite et al. Power-Generating SOFC Model versus the SOFC Hybrid Model (Retrofitted To Run on Natural Gas)
reformer heat duty (kW) fuel volumetric flow rate (m3/s) gross DC power output (kW) recirculation fuel (%) exhaust-fuel heat-exchanger duty (kW) exhaust-air heat-exchanger duty (kW) stack radiation heat available to air (kW) recovery water heat exchange available for co-generation fuel use (%) air use (%) reformate composition after pre-reformer (%)
SOFC hybrid model at 67% recycled fuel
Staite et al.14
system model parameter
SOFC hybrid model at 40% recycled fuel
3.9 3.12 10-4 (18.7 slpm) 5.95 67 0.18 6.6 2.26 5.3
N/A 3.12 10-4 (18.7 slpm) 6.30 67 0.19 6.27 3.24 N/A
N/A 3.12 10-4 (18.7 slpm) 5.91 40 0.19 5.88 2.81 N/A
83 40 CO, 23.6; H2, 39.1; steam, 27; CO2, 10.3
80 40 CO, 5.13; H2, 15.1; steam, 34.0; CO2, 12.4; N2, 33.1
80 40 CO, 7.5; H2, 26.9; steam, 22.1; CO2, 10.0; N2, 33.1
which a reactant can be supplied to an electrode.13 The term ln(1 - i/iL) represents the ratio of the reactant in the surface of the electrode versus the reactant in the bulk solution, which permits the determination of how mass transfer will affect fuel cell efficiency (voltage) as the reactant is depleted around the electrode surface. In the intermediate regions, the drop in potential difference is related to the resistance of the various materials within the cell (electrodes, electrolytes, and interconnects).12,13 The loss is linear because resistance of these materials is constant at constant temperatures. This is called ohmic polarization and is mathematically expressed as an Ohm’s law relationship in eq 17 below. ð17Þ ηOhm ¼ iR
methanol. A summary of the results that they had achieved from their model (for natural gas only) is compared to the values obtained from the SOFC hybrid abatement model (altered to run on natural gas) in Table 2. The comparison indicates that the SOFC hybrid model provides results similar to the published model. The main difference between the two models lies in the reformate composition, reformer duty values, and to a lesser degree, percentage of fuel recycled. The first two factors can be attributed to differences in the reforming system used in Staite et al.’s model and the SOFC hybrid model. The difference in the percentage of fuel recycled would be dependent upon the physical characteristics of the fuel cell. At first glance, it was difficult to ascertain what type of reformer was used in Staite et al.’s model, because no steam or oxygen stream was sent into the reformer. In fact, only two streams were sent to the reformer subsystem, which were the fresh fuel and the recycled fuel streams. The recycled fuel came from the anode side of the SOFC, meaning it would be composed of unused fuel, carbon dioxide, and steam. With the inputs composed of only these materials, the reformer subsystem could only be acting as a steam reformer. In the case of the SOFC hybrid model, the reformer was developed to resemble an autothermal reformer, in which two separate reactions are working simultaneously in two separate reactors. The POX reaction requires oxygen, and in the model, oxygen is supplied as air. This is significant because it means that a substantial portion of the reformate would not only consist of reformed fuel but also nitrogen. Furthermore, the autothermal reformer developed in the SOFC model was designed so that the exothermic reactions taking place in the POX reactor could generate the heat required by the endothermic reactions of the steam reformer. This means that, in an idealized case, the net heat duty would be zero and, therefore, be substantially different from the 3.9 kW identified for the steam reformer designed in Staite et al.’s model. Accurate calculation of gross direct-current (DC) power requires physical cell parameters, such as active cell area. Without the active cell area, current density could not be calculated and, therefore, the voltage drop as a result of polarization effects would not be available. Thus, increasing the percentage of fuel recycled decreased the molar composition of hydrogen and carbon monoxide in the reformate while increasing the total molar flow of reformate. The overall effect was a slight increase in the equivalent molar flow of hydrogen to the fuel cell. An increase in molar flow of fuel will increase the current produced. This alone will increase power. However, increasing
In eq 17, ηOhm refers to the voltage lost as a result of resistance within the fuel cell components and R refers to the sum of the resistances of the various materials in the cell. Overall, these effects can be summed and subtracted from the ideal voltage to determine the actual voltage. Eactual ¼ Eideal -ðηAct þ ηConc þ ηOhm Þ
ð18Þ
Model Validation No system was found in the literature that used VOCs as a fuel source. Therefore, to validate the performance of the SOFC hybrid model, it was modified to use compounds similar to those systems found in the literature. The most common fuels used in systems published in the literature were methane or natural gas. The published model used for a comparison was “feasibility analysis of methanol fuelled SOFC systems for remote distributed power applications”.14 The purpose of that study was to determine through both experiments and models the feasibility of converting a 5 kW natural gas SOFC power generator system to run on methanol. For the purposes of the model validation, only the modeling portion of this publication will be examined. The model developed by Staite et al. was completed in Aspentech HYSYS and made to run on natural gas and (13) Chen, E. Thermodynamics and electrochemical kinetics. In Fuel Cell Technology Handbook; Hoogers, G., Ed.; CRC Press: Boca Raton, FL, 2003. (14) Staite, M.; Marcazzan, P.; Ghosh, D.; Stannard, J.; Chong-Ping, C. Feasibility analysis of methanol fuelled SOFC systems for remote distributed power applications. In Solid Oxide Fuel Cells (IX); Singhal, S. C., Mizusaki J., Eds.; The Electrochemical Society, Inc.: Pennington, NJ, 2005-2007; Vol. 1, pp 216-228.
4698
Energy Fuels 2010, 24, 4693–4702
: DOI:10.1021/ef901249g
Borwankar et al.
Table 3. Adsorption System Stream Characteristic Summary gas composition (mole %) stream ID process emission VOC vapor aira (N/A) HVOC HVOC1 HVOC2 wastewater HVOC stored HVOC3 cathode exhaust water exhaust steam
temperature (K)
pressure (bar)
molar flow (mol/h)
O2
N2
toluene -5
MEK
xylene -5
-5
nBA -6
298 385.5
1.013 1.013
4224 0.2874
0.209 0
0.781 0
2.3 10 0.342
2.6 10 0.376
1.2 10 0.169
8 10 0.113
378.7 359.0 359.0 359.0 359.0 359.0 1071 288 364.7 385
0.993 0.983 0.983 0.983 0.983 0.983 0.980 1.013 0.970 1.003
1.787 1.787 0.7109 1.0761 0.3739 0.3370 3.219 1.500 3.219 1.500
0 0 0 0 0 0 0.102 0 0.102 0
0 0 0 0 0 0 0.892 0 0.892 0
0.055 0.055 0.134 0.003 0.134 0.134 0 0 0 0
0.061 0.060 0.015 0.001 0.015 0.015 0 0 0 0
0.027 0.027 0.64 0.003 0.64 0.64 0 0 0 0
0.018 0.018 0.043 0.001 0.043 0.043 0 0 0 0
H2O 0 0 0.839 0.839 0.608 0.992 0.608 0.608 0 1 0 1
a This stream would simply be air at atmospheric temperature and pressure. The system does not precisely represent the actual adsorption-desorption process, just the outcome.
current also increases current density, which decreases voltage. Without the knowledge of the active cell area, current density is unavailable, and therefore, to compare the two models appropriately, two conditions were considered: (1) compare the models if both are run at the same recycle rate and (2) compare the models if both are run such that reformate compositions are similar. Therefore, although the current density was unknown, a comparable performance was achieved by the SOFC hybrid model to Staite et al.’s model. Differences between the models can be attributed to the difference in reformer designs as well as unknown SOFC active cell area. For the purposes of this SOFC model, a current density of 250 mA/cm2 was assumed on the basis of literature reports.18
Table 4. Heating System Duties unit operation
heat duty (kW)
cathode HX condenser HX
19.82 -13.12
There are two ways in which this can be performed. The first method is to split the VOC emission stream after the adsorption-desorption cycle, such that a portion is being sent to a storage tank for off-production use, while the rest is being treated. The second method is to size the system for the 417 kg of VOCs/16 h and then run natural gas through the system during shutdown periods. In this work, only the first method was investigated. Optimized Operation of the Adsorber. The adsorption system was designed such that excess air from the process stream would be eliminated, leaving only concentrated VOCs, which are then desorbed using steam as a purge gas. However, after desorption (using a typical value of 2 kg of steam/kg of VOC desorbed), the concentration of VOCs was still low because of the diluting effect of the steam. To further concentrate the system, a condenser unit was placed after the desorbing bed to eliminate a substantial amount of moisture from the emission stream. It was predicted that the condensed water would contain
