Formation and Control of Sulfur Oxides in Sour Gas Oxy-Combustion

Article pubs.acs.org/EF

Formation and Control of Sulfur Oxides in Sour Gas OxyCombustion: Prediction Using a Reactor Network Model Dominik Bongartz, Santosh J. Shanbhogue, and Ahmed F. Ghoniem* Department of Mechanical Engineering, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, Massachusetts 02139, United States ABSTRACT: Sour natural gas currently requires expensive gas cleanup before it can be used in power generation because it contains large amounts of hydrogen sulfide (H2S) and carbon dioxide (CO2) that entail a low heating value and highly corrosive combustion products. A potential alternative is to use the gas directly in a gas turbine process employing oxy-fuel combustion, which could eliminate the need for gas cleanup while also enabling the application of carbon capture and sequestration, possibly combined with enhanced oil recovery (EOR). However, the exact influence of an oxy-fuel environment on the combustion products of sour gas has not been quantified yet. In this work, we used a reactor network model for the combustor and the gas turbine together with our recently assembled and validated detailed chemical reaction mechanism for sour gas combustion to investigate the influence of some basic design parameters on the combustion products of natural gas and sour gas in CO2 or H2O diluted oxy-fuel combustion as well as in conventional air combustion. Our calculations show that oxy-fuel combustion produces up to 2 orders of magnitude less of the highly corrosive product sulfur trioxide (SO3) than air combustion, which clearly demonstrates its potential in handling sulfur containing fuels. Unlike in air combustion, in oxy-fuel combustion, SO3 is mainly formed in the flame zone of the combustor and is then consumed as the combustion products are cooled in the dilution zone of the combustor and the turbine. In oxy-fuel combustion, H2O dilution leads to a higher combustion efficiency than CO2 dilution. However, if the process is to be combined with EOR, CO2 dilution makes it easier to comply with the very low levels of oxygen (O2) required in the EOR stream. Our calculations also show that it might even be beneficial to operate slightly fuel-rich because this simultaneously decreases the O2 and SO3 concentration further. The flame zone temperature in the combustor as well as the residence times in the flame zone and dilution zone have relatively little impact on SO3 formation and can hence be designed to enable good CO burnout. depleted oil field to reduce the viscosity of the oil and increase the amount of oil that can be extracted from it. Using oxy-fuel combustion with EOR for generating electricity from sour gas is expected to be beneficial in several ways: • The problem of a low heating value can be overcome by reducing the amount of diluent that is added to the combustor, thus enhancing flame stability. • Oxy-fuel cycles are likely to operate with little to no excess oxygen.6,9 As will be shown in this study, this can alleviate the corrosion issues caused by the sulfurcontaining combustion products. • Through the use of the CO2 in EOR, an additional revenue stream can be generated to counteract the additional cost and complexity of an oxy-fuel power plant compared to a conventional one.10 However, this also entails special requirements on the combustion products, the most important one being strict limits on the allowable O2 concentrations in the stream to be sequestered.11,12 • For high-pressure gas fields, the direct use of sour gas in a gas turbine process could enable recovering some of the pressure energy of the CO2 and H2S in the gas that

1. INTRODUCTION Sour gas is a form of natural gas which is commonly extracted at many gas fields and constitutes a significant fraction of the world gas reserves.1 Apart from small hydrocarbons (the majority of which is methane (CH4)), it contains significant amounts (typically between 0 and 30% by volume each2) of hydrogen sulfide (H2S) and carbon dioxide (CO2). These contaminants currently have to be removed before the gas can be used for power generation or heating. Carbon dioxide lowers the heating value of the gas, making it less economical to transport and more difficult to burn. Hydrogen sulfide is toxic and in its combustion some sulfur trioxide (SO3) is usually formed (besides the main product sulfur dioxide (SO2)), which causes serious corrosion problems in power plants.3−5 Oxy-fuel combustion (or oxy-combustion) is a promising technology to address the problems associated with the use of sour gas. It has recently received increasing attention as one option for implementing carbon capture and sequestration (CCS).6−8 In oxy-fuel combustion, the fuel is burned in pure oxygen (O2) and some diluent that is added to moderate the temperatures. In the most common cycle designs, this diluent is either CO2, water (H2O), or a mixture of the two. The gas exiting the combustor thus consists mainly of CO2 and H2O.9 After running it through a turbine to generate power, the H2O can easily be removed by condensation. The remaining stream of CO2 can be sequestered, which has been proposed as a means to mitigate climate change, or used in enhanced oil recovery (EOR).8,10 In EOR, the CO2 is injected into a © 2015 American Chemical Society

Received: July 28, 2015 Revised: October 6, 2015 Published: November 6, 2015 7670

DOI: 10.1021/acs.energyfuels.5b01709 Energy Fuels 2015, 29, 7670−7680

Article

Energy & Fuels

Figure 1. Generic gas turbine cycles used for the analysis based on the oxy-fuel combined cycle.9,14 (a) Air combustion. (b) Oxy-fuel combustion with CO2 dilution. (c) Oxy-fuel combustion with H2O dilution.

subsequent cooling through expansion in the gas turbine is analyzed, with special emphasis on SO3. The effect of running the entire process slightly fuel-rich or fuel-lean is also discussed briefly.

otherwise might be lost when removing these components from the gas stream. • Aging oil fields are a common source of sour gas. By using this gas close to the field in an oxy-fuel process, the life of the field can be extended while additionally generating electricity from an otherwise unusable byproduct. However, because sour gas is not commonly used as a fuel, very little has been known about its combustion characteristics and it is still unclear what is the exact effect of an oxy-fuel combustion strategy using either CO2 or H2O dilution on the combustion products of sour gas in a gas turbine, whether the integration of a sour gas oxy-combustion cycle with EOR is feasible with or without expensive gas cleanup processes or how a combustor for handling sour gas should be designed. In our previous work, we have taken a first step toward improving our understanding of sour gas combustion by examining the influence of different H2S contents and different combustion modes (i.e., air combustion vs CO2 or H2O diluted oxy-fuel combustion) on the ignition delay time, laminar burning velocity, and flame structure of sour gas by means of chemical kinetics calculations.13 In this paper, we highlight the influence of the combustion mode and some basic design parameters on the concentrations of CO, O2, and SO3 in a gas turbine power cycle with EOR. First, we describe the development of a reactor network model that is representative for the relevant part of the cycle. Next, the characteristics and design possibilities for the flame zone of the combustor are investigated. Finally, the dilution of the hot gases between the flame zone and the turbine inlet and the

2. MODELING In this section, a description of the generic gas turbine power cycles considered herein is given, followed by an explanation of the relevant performance measures for evaluating the different options. Finally, the reactor network model and the reaction mechanism used in this study are described. 2.1. Power Cycle and Boundary Conditions. For the analyses, three versions of a generic gas turbine power cycle similar to the oxy-fuel combined cycle analyzed by Kvamsdal et al.9 and Sanz et al.14 were considered (see Figure 1). Air or a diluent (either CO2 or H2O) are compressed to the combustor pressure of pcomb = 40 atm. In the combustor, sour gas (and in the oxy-fuel cases also O2) is added and burned to raise temperature to the turbine inlet temperature of Tti = 1500 K. The gas is then expanded in a gas turbine down to pout = 1 atm and Tout = 750 K and further cooled in a heat recovery steam generator (HRSG) which is powering a steam turbine. In the oxy-fuel cases, the flue gas then enters a condenser where all the H2O in the gas is condensed out, leaving a stream of CO2 with other noncondensable species that is to be used in EOR. Part of either the H2O or the CO2 is recycled to the combustor. 2.2. Performance Measures. Three primary performance measures can be identified regarding the combustion products of sour gas for a gas turbine based oxy-fuel power cycle with EOR. 7671


Article

Energy & Fuels

gas turbine, we set up the reactor network model shown in Figure 2.

First, both CO and O2 at the combustor outlet should be low for good combustion efficiency. Excessive CO signifies a waste of fuel and excess O2 a waste of energy for air separation. We expect CO to be about four times more important for the combustion efficiency than O2 (on a molar basis): The industry standard cryogenic air separation processes consume on the order of 0.2 kWh/kg O2, mostly in the form of compression work.15 Assuming a thermal efficiency for the generation of this work of 33%, this corresponds to an energy penalty of about 70 kJ/mol O2 in the exhaust gas, which is a quarter of the heating value of CO of roughly 280 kJ/mol. Second, the concentrations of CO and O2 in the CO2 stream to be used in EOR are also restricted. The concentrations in the EOR stream can be approximated with concentrations in the dry products because H 2 O is condensed out before sequestration.6,7 They differ considerably from those in the exhaust gas, in particular in the case of H2O diluted oxy-fuel combustion with the associated high water content in the gas stream. The CO mole fraction in the EOR stream is restricted because it increases the so-called Minimum Miscibility Pressure (MMP) required for injecting the products into the oil well and because of safety concerns for pipeline transportation of the products. Oxygen also increases the MMP, but more importantly can cause local overheating and raise the viscosity in the oil well due to oxidation as well as enhance bacterial growth.12 For our analysis, we refer to the requirements specified for the Weyburn EOR project,11,12 which imposes a limit on the CO and O2 concentrations of 1000 and 50 ppm, respectively. However, it should be noted that the limits on these species vary between different EOR projects and there is no uniform industry standard yet. Third, the concentration of SO3 must be kept as low as possible in the entire cycle because of its role in low temperature hot corrosion and cold end corrosion. The former commonly occurs in gas turbines at temperatures of 925−1075 K5 and hence depends on the SO3 mole fraction from the combustor outlet to the turbine outlet. The latter occurs during further cooling of the gases as virtually all SO3 present in the mixture reacts with H2O to form sulfuric acid (H2SO4), which starts to condense and cause damage to the equipment around T ≲ 500 K.4,16,17 The sulfuric acid dew point increases with the concentrations of both SO3 and H2O.18−20 The lower the sulfuric acid dew point, the further the gas can be cooled to extract more energy before the acid starts to condense. Because in an oxy-fuel cycle the working fluid has to condense at some point, this determines from which point in the cycle expensive acid resistant materials have to be used or where a flue gas desulfurization unit should be installed. These considerations have important implications for both the efficiency and the economics of the cycle.21,22 Nitrogen oxides (NOx) are not considered in this study because they are not particularly problematic in oxy-fuel combustion of gaseous fuels thanks to the very low concentrations of nitrogen in the oxidizer (at most a few percent due to imperfect air separation).23−25 2.3. Reactor Network Model. The parts of the cycle that are strongly influenced by chemistry are the combustor and possibly the turbine. Because the kinetics of both CO oxidation and SO3 formation essentially freeze at temperatures below T = 1000 K,26,27 no further reactions are expected to occur in the HRSG and the condenser. To model the chemistry in the relevant part of the cycle, that is the combustor and through the

Figure 2. Reactor network model used in this study. For all reactors in the combustor the energy equation is solved without heat losses. For the PFR representing the cooling in the turbine, a linear temperature and pressure profile is prescribed.

Fuel, oxidizer, and diluent (in the case of oxy-combustion) enter the combustor at a temperature of Tin = 670 K and a pressure of pcomb = 40 atm. For simplicity, sour gas was assumed to be a mixture of CH4 and H2S only. The H2S mole fraction in the fuel was varied between 0% and 30% to cover the range of common sour gas compositions. For the oxy-fuel cases, the oxygen purity was taken to be 100%. As commonly encountered in gas turbines, the combustor was assumed to consist of two distinct zones.28,29 In the flame zone (or primary zone), fuel is burned at a temperature chosen to guarantee stable combustion and quick fuel burnout. In air combustion, this is achieved by adjusting the equivalence ratio (typically close to stoichiometric in this zone), while in oxy-fuel combustion, the equivalence ratio is fixed at Φ = 1 (unless otherwise noted) and the flame temperature is controlled by adjusting the amount of diluent (CO2 or H2O) that is mixed with the O2 entering the primary zone. In the dilution zone, the remaining diluent (in the case of oxy-combustion) or air is added to lower the gas temperature to the turbine inlet temperature of Tti = 1500 K. The flame zone is represented by a single well-stirred reactor (WSR) to model the recirculation zone with intense mixing that is commonly found in this zone. Heat losses to the environment or to the dilution streams are not considered because they are usually small.30 The approach of using a single WSR for the flame zone followed by one or more plug flow reactors (PFR) is based on the work by Bragg31 and Beér and Lee32 and is widely used for modeling combustion in different systems. Many variations and refinements of this model have been proposed. One of the most common is the introduction of a reactor accounting for macroscopic flow recycling through the inner or outer recirculation zones found in many swirlstabilized combustors.33 The introduction of such a reactor has advantages for predicting blowout phenomena,34−37 and NOx formation in pulverized coal combustion.38−40 However, simple models without a recycling reactor have been shown to give good results for predicting both CO and NOx emissions from the combustion of gaseous fuels in gas turbine combustors.41−47 Because the focus of the present study is on giving general trends for CO, O2, and SO3 concentrations and we are not aiming at predicting NOx formation or blowout accurately, the choice of this simplistic model is justified. The dilution zone of the combustor is modeled by three PFRs of equal residence time in series. The additional diluent or air is added to these three reactors in equal parts. The 7672


Article

Energy & Fuels

strategy for oxy-fuel combustion yet,48 this complete range of flame temperatures was considered. 3.1.1. Equilibrium Trends. For evaluating the emissions of the species of interest (CO, O2, and SO3), we first calculate the concentrations at thermodynamic equilibrium. This corresponds to the case where the combustor is designed so that the residence time in the flame zone is sufficient for CO burnout to go to completion. Although it is unlikely that this will be achieved in an actual combustor (except maybe at very high flame temperatures), this analysis does provide some insight into the underlying trends. At chemical equilibrium, CO2 dilution leads to CO concentrations that are 1 order of magnitude higher than for H2O dilution and O2 concentrations that are about twice as high (see Figure 3). The concentrations of both species

amount of diluent added in the different cases, that is, air combustion and oxy-combustion using CO2 or H2O as a diluent, is such that the combustor exit temperature is the same (noting that the specific heats of air, CO2, and H2O are different). The added streams account for both dedicated dilution jets and entrainment of cooling air/diluent from the combustor walls. The importance of modeling the dilution zone for predicting CO concentrations from realistic gas turbine combustors is well established.30,35,41,48 Three PFRs with addition of diluent at the beginning of each reactor (which is assumed to mix instantly with the products of the preceding reactor) have been found to be an effective way of modeling this zone.49 After the combustor, the flue gas is quenched rapidly during the adiabatic expansion in the gas turbine. This is modeled by a PFR in which temperature and pressure are reduced linearly and simultaneously with time down to 750 K and 1 atm. The cooling rate is taken to be 100000 K/s, which was estimated assuming a typical turbine length on the order of one meter and a constant axial velocity corresponding to a Mach number of 0.3 at the outlet. All calculations were conducted using CHEMKIN-PRO.50 The reaction mechanism employed in this study is described in detail in our previous work.51 It contains a total of 157 chemical species and 1011 reactions. On the basis of the availability of experimental data, it has been validated (among others) for experiments on CO concentrations in oxy-fuel combustion of CH4 and in the presence of sulfur species and formation of SO3.51

3. RESULTS AND DISCUSSION From a design perspective, the parameters that can be varied are the fraction x of the diluent or combustion air that is added to the flame zone to control the flame temperature (cf. Figure 2), the residence time of the flame zone WSR, and the residence time in the dilution zone. The total amount of diluent added or the overall equivalence ratio (for the case of air combustion) is constrained by the requirement to meet a fixed turbine inlet temperature. In the following, we first discuss the influence of the flame zone temperature and residence time on the combustion products for different combustion modes. Next, the product concentrations in the dilution zone and the turbine are investigated, also discussing the concentrations of CO and O2, formation of SO3, and the influence of varying the equivalence ratio in the case of oxy-combustion. The ranges of residence times and flame temperatures considered are based on typical values for gas turbine combustors.29,52,53 3.1. Flame Zone Design. As mentioned previously, the purpose of the flame/primary zone is to stabilize combustion and provide the necessary conditions to allow for fast burnout of the fuel.28 If the concentrations of the undesired species CO and O2 in this zone get too high, poor mixing could result in high emissions because of insufficient burnout in the dilution zone. There are two design parameters for achieving these goals, namely the flame temperature and the residence time of the WSR. Increasing the flame temperature by adding less diluent or air to the flame zone leads to a higher burning rate and hence better combustion stabilization.54 Typical flame temperatures in gas turbines range from 1700 K for lean premixed systems to 2200 K in diffusion burners.29 Because there is no established

Figure 3. Equilibrium mole fractions of CO and O2 in the flame zone are larger for CO2 dilution than for H2O dilution and increases with the flame temperature. The equilibrium mole fraction of SO3 is much larger in air combustion than in oxy-fuel combustion using either diluent. Equilibrium at constant pressure and enthalpy for Tin = 670 K and p = 40 atm.

increase with temperature because of the dissociation of CO2. Fuel-lean air combustion is shown for comparison and leads to much lower CO and much higher O2 concentrations due to the excess O2 available at the low equivalence ratios required for achieving these temperatures. In oxy-combustion, the flame temperature is controlled by the diluent concentration instead of the equivalence ratio, so that no excess O2 is required. For SO3, the H2S content in the fuel has the strongest influence on the results (see Figure 3). For H2O and CO2 diluted oxy-combustion, the SO3 mole fraction rises slightly with increasing flame temperature, while for fuel-lean air combustion it is more than an order of magnitude higher and decreases with increasing flame temperature. The behavior of the SO3 concentration at equilibrium is caused by the combined effect of temperature and O2 concentration, with 7673


Article

Energy & Fuels

corresponding to the lower carbon content in the fuel and the lower amount of O2 needed for oxidation. Carbon dioxide dilution leads to CO concentrations that are an order of magnitude higher than in the case of H2O dilution and O2 concentrations that are about twice as high (see Figures 4b and 5b). Fuel-lean air combustion, which is shown for comparison, leads to much lower CO and much higher O2 concentrations for a given flame temperature due to the availability of excess O2. The higher CO concentrations for the case of CO2 dilution have been reported in previous studies on oxy-fuel combustion of CH4.23,24,52,56−58 They can be explained by increased dissociation of CO2 and slower kinetics of CO burnout. The SO3 mole fraction in the flame zone WSR corresponds to the equilibrium trends for oxy-combustion above Tflame = 2000 K (see Figure 6a). At lower temperature, increasing

lower temperature and higher O2 favoring SO3 formation thermodynamically.53,55 3.1.2. Flame Zone Residence Time. Because both CO oxidation and SO3 formation are mostly limited by chemical kinetics and equilibrium is unlikely to be achieved in the flame zone, the effect of varying the residence time in the flame zone WSR was investigated. The residence time in the WSR (τres,WSR) represents the size of the recirculation zone in a typical swirl stabilized combustor.33 If it is too small, blowout may occur because the reactants do not have enough time to get mixed with hot products and ignite. At the same time, the residence time should not be too large in order to allow for sufficient residence time in the rest of the combustor, namely the dilution zone (and to limit NOx formation in the case of air combustion). Higher residence times lead to lower CO and O 2 concentrations in the WSR because there is more time for CO burnout (see Figures 4a and 5a). For a fixed residence time,

Figure 6. SO3 concentration in the flame zone as a function of flame temperature: (a) CO2 diluted oxy-combustion at different residence times and H2S contents in the fuel. (b) Different combustion modes at τres,WSR = 20 ms and 30% H2S in the fuel.

Figure 4. CO concentration in the flame zone as a function of flame temperature: (a) CO2 diluted oxy-combustion at different residence times and H2S contents in the fuel. (b) Different combustion modes at τres,WSR = 20 ms and 0% H2S in the fuel.

residence time leads to either more or less SO3, depending on the conditions. The reason for this complex behavior is discussed in Section 3.2.2. Carbon dioxide dilution leads to higher SO3 concentrations, corresponding to the higher O2 content. Fuel-lean air combustion leads to SO3 concentrations that can be more than an order of magnitude higher and decrease with increasing flame temperature. 3.2. Dilution Zone Design. To investigate the influence of the cooling rate in the dilution zone, which is needed to bring the products’ temperature down to values acceptable by the gas turbine, the residence time in the flame zone WSR was fixed at τres,WSR = 20 ms for all following calculations. This value was chosen similar to refs 23,56 to balance the requirements of low CO and O2 and blowout safety on the one hand and allowing enough time for the dilution zone on the other hand. The flame temperature in the WSR was found to have only relatively little effect on the species concentrations after dilution, so only the results for Tflame = 1900 K are shown. 3.2.1. Dilution Cooling Rate. The cooling rate in the dilution zone is varied by changing the residence time (τres,dil), which is equally distributed between the three PFRs (cf. Figure 2). The cooling rate in the turbine is not expected to be a design variable and remains fixed at 100000 K/s. Increasing the residence time (i.e., slower cooling) generally leads to lower CO and O2 concentrations because more time is available for CO oxidation (see Figure 7). Because of the faster kinetics, H2O dilution compares even more favorably to CO2

Figure 5. O2 concentration in the flame zone as a function of flame temperature: (a) CO2 diluted oxy-combustion at different residence times and H2S contents in the fuel. (b) Different combustion modes at τres,WSR = 20 ms and 0% H2S in the fuel.

there is always an intermediate temperature (around Tflame = 1900 K) for which the CO and O2 concentrations are minimal. At high temperatures, the concentrations increase because of the increasing equilibrium values, while at low temperatures they increase because the kinetics get slower and do not allow sufficient burnout (products quenching). Increasing H2S content in the fuel leads to lower CO and O2 concentrations, 7674


Article

Energy & Fuels

Figure 7. CO and O2 mole fractions at the combustor outlet both decrease with increasing residence time in the dilution zone (i.e., slower cooling). Water dilution leads to lower O2 and much lower CO mole fractions than CO2 dilution.

Figure 9. SO3 concentration at the combustor outlet and after the turbine are up to 2 orders of magnitude larger for air combustion than for oxy-combustion. It increases slightly with the dilution zone residence time (i.e., slower cooling).

dilution than it does for the flame zone WSR, indicating better combustion efficiency (see Figure 7). Increasing H2S content in the fuel again leads to slightly lower CO and O2. However, as mentioned in Section 2.2, CO and O2 are not only measures for combustion efficiency, but their concentrations are also restricted for EOR. In the CO2 stream for EOR (i.e., in the dry products), the comparison between H2O and CO2 dilution gives a different result than for combustion efficiency. For H2O dilution, about 90% of the exhaust gas is steam, so that the condensation process leads to a strong enrichment of CO and O2 in the gas phase. Therefore, in the EOR stream, CO concentrations are only two to four times lower for H2O dilution than for CO2 dilution, while O2 concentrations are higher (see Figure 8). Comparison with

SO3 mole fraction increases slightly with increasing residence time in the dilution zone. While the SO3 concentration at the combustor outlet is important for hot corrosion in the gas turbine, sulfuric acid corrosion is determined by the SO3 and H2O concentrations after the turbine. For oxy-fuel combustion, the SO 3 concentration decreases during the gas expansion in the turbine (see Figure 9). For air combustion, the concentration increases during the expansion, which is similar to what is commonly encountered in power plants and oil boilers, where most of the SO3 is formed during flue gas cooling.4,55 In fact, by comparing the SO3 profiles along the gas flow path for air combustion and oxy-fuel combustion, one can observe that the behavior differs widely between the two combustion modes. In air combustion (see Figure 10), about 1.5% of sulfur

Figure 8. Concentrations in the CO2 stream for use in EOR correspond to the concentrations in the dry products. Therefore, CO2 diluted oxy-combustion leads to less O2 in this stream than H2O dilution. Values shown are for pure CH4. The horizontal dashed lines represents the limits specified for the Weyburn EOR11,12

Figure 10. For air combustion, the SO3 mole fraction increases from the flame zone to the turbine outlet. The calculations shown are for 30% H2S in the fuel, TFlame = 1900 K and a dilution zone residence time of 40 ms.

the limits specified for the Weyburn EOR11,12 (dashed lines) shows that CO concentrations are within the allowable range for both diluents (for residence times in the dilution zone of more than ≈15 ms), while O2 concentrations exceed the allowable values in either case. The concentration of SO3 at the combustor outlet is up to 2 orders of magnitude higher for air combustion than for oxy-fuel combustion (see Figure 9). Carbon dioxide dilution produces about 50% more SO3 than H2O dilution, corresponding to the higher O2 concentrations which favor SO3 formation.27 The

is converted to SO3 in the primary zone (the rest being almost exclusively SO2). In the dilution zone, more SO3 is being formed as more air is added because the temperature drops and more O2 is available, leading up to a final mole fraction of 800 ppm, which corresponds to 4.3% of the total sulfur. This qualitative behavior as well as the magnitude of the percentages of sulfur present as SO3 agree well with previous experimental and modeling studies of SO3 formation in conventional gas turbines and aircraft engines.3,53,59 For CO2 diluted oxycombustion, the behavior is more complex and SO3 is either 7675


Article

Energy & Fuels

If there is ample excess O2 (1% O2, which corresponds to an equivalence ratio of Φ = 0.5, where Φ is calculated based on CO oxidation), at chemical equilibrium (i.e., infinitely slow cooling) all SO2 gets converted to SO3 as the temperature is decreased because its formation is thermodynamically favorable (see Figure 12). In this case, higher cooling rates lead to less

produced or consumed in the different reactors representing the dilution zone and the turbine (see Figure 11). Together

Figure 11. For oxy-combustion with CO2 dilution, the SO3 mole fraction decreases from the flame zone to the turbine outlet. The calculations shown are for 30% H2S in the fuel, TFlame = 1900 K and a dilution zone residence time of 40 ms. Figure 12. In the presence of excess O2, the SO3 concentration increases with decreasing temperature. Higher cooling rates lead to quenching of SO3 formation. Initial conditions: equilibrium of 1% SO2, 1% CO, 1% H2O, and 1% O2 in Ar at T = 1700 K and p = 1 atm.

with further dilution of the mixture, this leads to a significant reduction of the SO3 mole fraction from that produced in the flame zone (even here the values are already much lower (0.1% of sulfur is present as SO3) than for those found in the case of air combustion) to the end of the turbine (here, only 0.05% of sulfur is present as SO3). Note that the jumps in Figures 10 and 11 are because of the changes in the local concentrations due to the mixing with extra air/diluent as the flow passes from one reactor to the next. This comparison clearly shows the advantage of handling sulfur containing gaseous fuels in an oxy-fuel process rather than through air combustion. This is opposite to what is generally reported for oxy-fuel combustion of coal, where SO3 concentrations are expected to be higher than in air combustion.55,60,61 The main difference is that coal power plants cannot operate as close to stoichiometric conditions as gas fired plants because of the heterogeneous reactions that generally require a few percent of excess O2.6 Moreover, Fleig et al.55 attribute most of the increase in SO3 for oxy-combustion as compared to air combustion to the recirculation of flue gas containing SO2. In our calculation, we did not consider SO2 in the diluent, which corresponds to the case where a flue gas desulfurization unit is installed in the cycle. This has been found to be the most economical way of handling the sulfur combustion products in a gas turbine cycle.21,22 Preliminary calculations show that considering SO2 in the diluent stream (i.e., without flue gas desulfurization, corresponding to the “acid resistance” case in ref 22) does increase SO3 by a factor of 5− 10. Even in this case it is still much lower than in air combustion. 3.2.2. Formation of Sulfur Trioxide During Cooling. The significant differences between air combustion and oxycombustion with respect to SO3 formation can be explained by the fact that SO3 formation depends strongly on both temperature and excess O2 concentration.27,55 To demonstrate the thermodynamic and chemical effects that determine the observed behavior, we constructed a simplified model for a homogeneous mixture of 1% SO2, 1% CO, 1% H2O, and varying amounts of O2 in an argon (Ar) bath. The mixture is first brought to chemical equilibrium at T = 1700 K and is then cooled down to T = 300 K at different cooling rates.

SO3 because its formation gets quenched as the temperature drops. Even with a cooling rate of 1 K/s, the SO3 concentration drops by nearly 2 orders of magnitude from the equilibrium value. Further increase in the cooling rate leads to lower SO3 concentration, but the rate of reduction in its production diminishes quickly. However, regardless of the cooling rate, SO3 is only formed and never consumed during the cooling process. This continuous production of SO3 in the presence of excess O2 is similar to the case of fuel-lean air-combustion (cf. Figure 10). Under stoichiometric conditions with respect to CO oxidation (0.5% O2), the situation gets more complex because SO3 formation has to compete for O2 with the oxidation of CO. At equilibrium (i.e., infinitely slow cooling), virtually all CO gets oxidized as the temperature is decreased, consuming all the O2 (see Figure 13). Accordingly, SO3 (which is already low at T = 1700 K) gets consumed as the temperature is reduced (Figure 14). This shows that thermodynamically, CO oxidation is

Figure 13. At stoichiometric conditions, higher cooling rates lead to quenching of CO oxidation. Initial conditions: equilibrium of 1% SO2, 1% CO, 1% H2O, and 0.5% O2 in Ar at T = 1700 K and p = 1 atm. 7676


Article

Energy & Fuels

is based on an estimate.62 Under these conditions, the S2 involved in this reaction is mostly being produced by the reaction S2O + SO = SO2 + S2, which also uses an estimated rate. This very high reactivity of S2 and S2O at low temperature is unexpected, and a sound theoretical or experimental investigation of the rates of the aforementioned reactions would thus be desirable. In summary, at stoichiometric conditions (relative to CO), SO3 is either produced or consumed during cooling, depending on the cooling rate. This corresponds to the case of CO2 diluted oxy-combustion (cf. Figure 11). Under the present conditions, faster cooling leads to more SO3, not less (but the concentrations remain very low, on the order of 1 ppm). This shows the strong influence of both the equivalence ratio and the time−temperature history for the formation of SO3 and demonstrates how the presence of a reducing agent such as CO can suppress the formation rate of SO3. 3.3. Influence of the Equivalence Ratio. As shown in the previous section, it may not be possible to reach the low O2 concentrations required for EOR with either diluent while burning under stoichiometric conditions. However, because CO concentrations are well below the allowable limits, it seems promising to operate the cycle slightly rich instead of stoichiometric to achieve lower O2 concentrations. To assess whether this approach is feasible, similar calculations to the ones presented in Section 3.2.1 for CO2 and H2O dilution with 30% H2S in the fuel were conducted for different equivalence ratios close to stoichiometry. Both CO and O2 mole fractions are highly sensitive to the equivalence ratio, especially for H2O dilution (see Figure 16).

favored over SO3 formation. In fact, CO oxidation occurs as SO3 is being reduced.

Figure 14. At stoichiometric conditions, SO3 is either formed or consumed during cooling, depending on the cooling rate. Initial conditions: equilibrium of 1% SO2, 1% CO, 1% H2O, and 0.5% O2 in Ar at T = 1700 K and p = 1 atm.

On the other hand, the kinetics of SO3 formation are faster than those of CO oxidation under the present conditions and over the entire temperature range considered, meaning that it takes longer for CO burnout to go to completion than for SO3 formation (see Figure 15). The combination of these two effects leads to the complex dependence of SO3 formation on the cooling rate seen in Figure 14.

Figure 15. The chemical time scale for SO3 formation at stoichiometric conditions is significantly smaller than that of CO formation. We define it as the time to reach 90% of equilibrium SO3 or 110% of equilibrium CO, respectively. Initial conditions: 0.5% O2, 1% H2O, and either 1% CO or 1% SO2 in Ar.

Figure 16. CO and O2 mole fractions in the dry products for CO2 and H2O diluted oxy-combustion with 30% H2S at TFlame = 1900 K and a dilution zone residence time of 40 ms. At an equivalence ratio around Φ = 1.0025, both the CO and O2 concentrations are below the limits for the Weyburn EOR (horizontal dashed lines).

As the cooling rate is raised, CO oxidation is increasingly quenched as the temperature drops and more and more O2 is left (see Figure 13). Therefore, at higher cooling rate, more SO3 is formed initially because the temperature drops while the O2 concentration is still relatively high (CO oxidation has not started yet). As O2 starts to drop due to CO oxidation (which is slower than SO2 oxidation to SO3), SO3 is consumed again. Once CO oxidation stops at around T = 1200 K, SO3 rises again because the O2 concentration remains constant and the temperature falls. Around T = 1000 K, SO3 formation freezes as well. The reduction of SO3 at lower temperatures observed for cooling rates of 10000−100000 K/s is uncertain because it is caused by the reaction SO3 + S2 = S2O + SO2, the rate of which

Operating slightly fuel-rich can lead to significantly lower O2 concentrations but at the expense of higher CO and thus a lower combustion efficiency. In fact, there is a very narrow window of the equivalence ratio around Φ = 1.0025 within which the Weyburn EOR requirements for both CO and O2 concentrations are satisfied. However, in order to meet this window, the equivalence ratio would have to be controlled within a very narrow range and the system would thus be highly vulnerable to the influence of poor mixing. Operating fuel-rich also has the added benefit of further reducing SO3 formation because of the lower O2 concentrations53,55 (see Figure 17). Around Φ = 1.0025, the SO3 7677


Article

Energy & Fuels

lower than for fuel-lean air combustion. Unlike for air combustion, the concentration of SO3 decreases in the dilution zone of the combustor and in the turbine as the oxidation of CO consumes more O2. The exact behavior depends strongly on the equivalence ratio and the time− temperature history. 2. Carbon dioxide and H2O dilution are likely to be comparable in terms of cold-end corrosion. While the SO3 concentrations are up to 50% higher for CO2 dilution, the higher H2O content of the gas when using H2O dilution will increase the sulfuric acid dew point as well. However, it should be noted that the existing correlations for predicting the sulfuric acid dew point may not be valid for the conditions encountered in oxyfuel combustion. 3. The combustion efficiency, measured by the concentrations of CO and O2 in the exhaust gas, is higher for H2O dilution than for CO2 dilution. On the other hand, CO2 dilution seems more promising for keeping CO and O2 mole fractions in the EOR stream within the allowable limits. This is mainly due to the large change in concentrations in the gas stream for H2O dilution as water is condensed before sequestration. 4. For either diluent, it may not be possible to reach the very low levels of O2 required in the EOR stream when operating the cycle at stoichiometric conditions. However, because CO limits are usually less strict, it may be beneficial to operate the combustor slightly fuelrich. This also significantly decreases SO3 formation, thus further alleviating the associated corrosion issues. 5. The temperature in the flame zone of the combustor has only little impact on SO3 formation and can hence be chosen to enable good CO and O2 burnout. In the dilution zone of the combustor, slower cooling leads to better CO and O2 burnout but also increases SO3 formation somewhat under most conditions. It should be emphasized that this study aims only at identifying key trends and trade-offs of the system. To establish the best possible design, future studies should address specific thermodynamic cycles, also considering the fate of impurities as they pass other components and are potentially recycled to the combustor. Moreover, more elaborate CFD models of the combustor are needed to determine the impact of the mixing patterns used in the different sections on the formation and consumption of the different gaseous species in the products stream.

Figure 17. SO3 mole fraction at the combustor and turbine outlets can be reduced significantly by operating slightly fuel-rich. At stoichiometric and fuel-rich operation, SO3 decreases during cooling, while in fuel-lean operation it increases. CO2 diluted oxy-combustion with 30% H2S at TFlame = 1900 K and a dilution zone residence time of 40 ms.

concentration is reduced by a factor of 3−6 compared to the stoichiometric case. This calculation shows again the importance of the equivalence ratio in determining the distribution of SO3 formation: at stoichiometric and fuel-rich conditions, the SO3 mole fraction decreases from the flame zone to the combustor outlet and further to the turbine outlet. The strong reduction in SO3 at the turbine outlet above Φ ≈ 1.01 is again caused by the reaction SO3 + S2 = S2O + SO2 and might thus be questionable (cf. dotted line in Figure 17 and the discussion in Section 3.2.2). At fuel-lean conditions, more SO3 is formed in the dilution zone of the combustor, just like in air combustion. If the equivalence ratio gets too high, formation of solid sulfur may occur.63−65 However, in the range considered here, the concentrations of the disulfur molecule (S2), which is considered the first precursor of solid sulfur formation, remains well below 1 ppm. In principle, the strong influence of the equivalence ratio on SO3, in particular close to stoichiometric conditions, is wellknown and has been demonstrated both numerically and experimentally.3,27,53,55 However, in the classical air combustion strategies for gas turbines, it is not possible to take full advantage of this effect because large amounts of excess air are generally required for limiting the turbine inlet temperature to levels acceptable to the materials used. In oxy-fuel combustion, the fact that temperature control is achieved through addition of a diluent and is thus decoupled from the equivalence ratio enables operation of the entire combustor close to or even slightly below stoichiometry. As demonstrated by these calculations, this offers the potential for significant reductions in SO3 formation.

■

AUTHOR INFORMATION

Corresponding Author

*Phone: +1-617-253-2295. Fax: +1-617-253-5981 E-mail: [email protected].

4. CONCLUSION A reactor network model has been applied to study CO2 or H2O diluted oxy-fuel combustion of sour gas in a gas turbine power cycle with carbon capture for EOR in order to identify promising combustor design strategies. From the results, we can draw the following conclusions: 1. In terms of hot corrosion and cold-end corrosion, oxyfuel combustion is much more promising than air combustion for handling gas with large H2S content. The absence of excess O2 in oxy-fuel combustion leads to SO3 concentrations that are 1−2 orders of magnitude

Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The financial support by Siemens Energy, Inc. is gratefully acknowledged. D.B. thanks the German National Academic Foundation and the German−American Fulbright Commission for predoctoral grants. S.J.S. and A.F.G. where also supported by King Abdullah University of Science and Technology (KAUST) under the award no. KUS-110-010-01. 7678


Article

Energy & Fuels

■

(35) Hammond, D. C., Jr.; Mellor, A. M. Combust. Sci. Technol. 1970, 2, 67−80. (36) Swithenbank, J.; Poll, I.; Vincent, M. W.; Wright, D. D. Symp. Combust., [Proc.] 1973, 14, 627−638. (37) Sturgess, G. Combustor LBO Design Model Evaluation; United Technologies Corp. Pratt & Whitney, Government Engines & Space Propulsion, West Palm Beach, FL, 1995. (38) Heap, M. P.; Lowes, T. M.; Walmsley, R. Symp. Combust., [Proc.] 1973, 14 (1), 883−895. (39) Van der Lans, R. P.; Glarborg, P.; Dam-Johansen, K.; Larsen, P. S. Chem. Eng. Sci. 1997, 52 (16), 2743−2756. (40) Pedersen, L.; Glarborg, P.; Dam-Johansen, K.; Hepburn, P.; Hesselmann, G. Combust. Sci. Technol. 1998, 132, 251−314. (41) Morr, A. R.; Heywood, J. B.; Fitch, A. H. Combust. Sci. Technol. 1975, 11 (3−4), 97−109. (42) Michaud, M. G.; Westmoreland, P. R.; Feitelberg, A. S. Symp. Combust., [Proc.] 1992, 24, 879−887. (43) Feitelberg, A. S.; Lacey, M. A. J. Eng. Gas Turbines Power 1998, 120, 502−508. (44) Straub, D. L.; Casleton, K. H.; Lewis, R. E.; Sidwell, T. G.; Maloney, D. J.; Richards, G. A. J. Eng. Gas Turbines Power 2005, 127, 36−41. (45) Gokulakrishnan, P.; Fuller, C. C.; Joklik, R. G.; Klassen, M. S. Chemical Kinetic Modeling of Ignition and Emissions from Natural Gas and LNG Fueled Gas Turbines. Proceedings of the ASME Turbo Expo 2012, Copenhagen, Denmark, 2012. (46) Göke, S.; Paschereit, C. O. Influence of Steam Dilution on NOx Formation in Premixed Natural Gas and Hydrogen Flames. 50th AIAA Aerospace Sciences Meeting including the New Horizons Forum and Aeospace Exposition, Nashville, TN, 2012; pp AIAA-2012−1272; 10.2514/6.2012-1272. (47) Göke, S.; Schimek, S.; Terhaar, S.; Reichel, T.; Göckeler, K.; Krüger, O.; Fleck, J.; Griebel, P.; Paschereit, C. O. Influence of Pressure and Steam Dilution on NOx and CO Emissions in a Premixed Natural Gas Flame. Proceedings of ASME Turbo Expo 2013: Turbine Technical Conference and Exposition, San Antonio, TX, June 3, 2013; GT2013-94782, 10.1115/GT2013-94782. (48) Lieuwen, T. C.; Yang, V., Eds. Gas Turbine Emissions; Cambridge University Press: Cambridge, UK, 2013. (49) Allaire, D. L.; Waitz, I. A.; Willcox, K. E. A comparison of two methods for predicting emissions from aircraft gas turbine combustors. Proceedings of the ASME Turbo Expo 2007: Power for Land, Sea and Air, Montreal, Canada, 2007; GT2007-28346, 10.1115/GT200728346 (50) Kee, R. J.; Rupley, F. M.; Miller, J. A.; et al., CHEMKIN-PRO 15112. Reaction Design: San Diego, CA, 2011. (51) Bongartz, D.; Ghoniem, A. F. Combust. Flame 2015, 162, 544− 553. (52) Amato, A.; Hudak, B.; D’Souza, P.; D’Carlo, P.; Noble, D.; Scarborough, D.; Seitzman, J.; Lieuwen, T. Proc. Combust. Inst. 2011, 33, 3399−3405. (53) Hunter, S. C. J. Eng. Power 1982, 104, 44−50. (54) Turns, S. R. An Introduction to Combustion, 2nd ed.; McGrawHill Higher Education: Singapore, 2000. (55) Fleig, D.; Normann, F.; Andersson, K.; Johnsson, F.; Leckner, B. Energy Procedia 2009, 1, 383−390. (56) Richards, G. A.; Caselton, K. H.; Lewis, R. E.; Rogers, W. A.; Woike, M. R.; Willis, B. P. Advanced Steam Generators; National Energy Technology Lab: Pittsburgh, PA; NASA Glenn Research Center (US): Morgantown, WV, 2001. (57) Glarborg, P.; Bentzen, L. L. B. Energy Fuels 2008, 22, 291−296. (58) Chen, L.; Ghoniem, A. F. Combust. Sci. Technol. 2014, 186, 829−848. (59) Harris, B. W. J. Eng. Gas Turbines Power 1990, 112, 585−589. (60) Fleig, D.; Andersson, K.; Johnsson, F. Ind. Eng. Chem. Res. 2012, 51, 9483−9491. (61) Stanger, R.; Wall, T. Prog. Energy Combust. Sci. 2011, 37, 69−88. (62) Zhou, C.; Sendt, K.; Haynes, B. S. Proc. Combust. Inst. 2013, 34, 625−632.

REFERENCES

(1) Burgers, W. F. J.; Northrop, P. S.; Kheshgi, H. S.; Valencia, J. A. Energy Procedia 2011, 4, 2178−2184. (2) Hammer, G.; Lübcke, T.; Kettner, R.; Pillarella, M. R.; Recknagel, H.; Commichau, A.; Neumann, H.-J.; Paczynska-Lahme, B. Ullmann’s Encyclopedia of Industrial Chemistry; Wiley-VCH: Weinheim, Germany, 2012; Vol. 23; Chapter Natural Gas, pp 739−792. (3) Barrett, R. E.; Hummell, J. D.; Reid, W. T. J. Eng. Power 1966, 88, 165−172. (4) Blythe, G.; Dombrowski, K. SO3 Mitigation Guide Update; Report 1004168; EPRI, Palo Alto, CA, 2004. (5) Eliaz, N.; Shemesh, G.; Latanision, R. M. Eng. Failure Anal. 2002, 9, 31−43. (6) Toftegaard, M. B.; Brix, J.; Jensen, P. A.; Glarborg, P.; Jensen, A. D. Prog. Energy Combust. Sci. 2010, 36, 581−625. (7) Chen, L.; Yong, S. Z.; Ghoniem, A. F. Prog. Energy Combust. Sci. 2012, 38, 156−214. (8) Rubin, E.; Mantripragada, H.; Marks, A.; Versteeg, P.; Kitchin, J. Prog. Energy Combust. Sci. 2012, 38, 630−671. (9) Kvamsdal, H. M.; Jordal, K.; Bolland, O. Energy 2007, 32, 10−24. (10) Davison, J. Energy 2007, 32, 1163−1176. (11) De Visser, E.; Hendriks, C.; Barrio, M.; Mølnvik, M. J.; de Koeijer, G.; Liljemark, S.; Le Gallo, Y. Int. J. Greenhouse Gas Control 2008, 2, 478−484. (12) Pipitone, G.; Bolland, O. Int. J. Greenhouse Gas Control 2009, 3, 528−534. (13) Bongartz, D.; Ghoniem, A. F. Combust. Flame 2015, 162, 2749− 2757. (14) Sanz, W.; Jericha, H.; Bauer, B.; Göttlich, E. J. Eng. Gas Turbines Power 2008, 130, 031702. (15) Pfaff, I.; Kather, A. Energy Procedia 2009, 1, 495−502. (16) Srivastava, R.; Miller, C.; Erickson, C.; Jambhekar, R. J. Air Waste Manage. Assoc. 2004, 54, 750−562. (17) Schweitzer, P. A. Fundamentals of Metallic Corrosion: Atmospheric and Media Corrosion of Metals; CRC Press, Boca Raton, FL, 2006. (18) Verhoff, F.; Banchero, J. Chem. Eng. Prog. 1974, 70 (8), 71−72. (19) Okkes, A. Hydrocarbon Process. 1987, 66 (7), 53−55. (20) ZareNezhad, B. Oil Gas J. 2009, 107, 35. (21) Chakroun, N. W.; Ghoniem, A. F. Int. J. Greenhouse Gas Control 2015, 36, 1−12. (22) Chakroun, N. W.; Ghoniem, A. F. Int. J. Greenhouse Gas Control 2015, 41, 163−173. (23) Lewis, R. E.; Caselton, K. H.; Richards, G. A.; Straub, D. L.; Rogers, W. A. A Comparison of Turbine Combustion Concepts Suitable for Integration with CO2 Sequestration. 2001 Joint International Combustion Symposium, Kauai, HI. 2001. (24) Richards, G. A.; Casleton, K. H.; Chorpening, B. T. Proc. Inst. Mech. Eng., Part A 2005, 219, 121−126. (25) Williams, T. C.; Shaddix, C. R.; Schefer, R. W. Combust. Sci. Technol. 2007, 180, 64−88. (26) Abián, M.; Giménez-López, J.; Bilbao, R.; Alzueta, M. U. Proc. Combust. Inst. 2011, 33, 317−323. (27) Fleig, D.; Alzueta, M. U.; Normann, F.; Abián, M.; Andersson, K.; Johnsson, F. Combust. Flame 2013, 160 (6), 1142−1151. (28) Lefebvre, A. H. Gas Turbine Combustion, 2nd ed.; Taylor & Francis: Philadelphia, PA, 1999. (29) Boyce, M. Gas Turbine Engineering Handbook, 4th ed.; Butterworth-Heinemann: Waltham, MA, 2012. (30) Hammond, D. C., Jr.; Mellor, A. M. Combust. Sci. Technol. 1973, 6, 279−286. (31) Bragg, S. Application of Reaction Rate Theory to Combustion Chamber Analysis; Aeronautical Research Council Pub. ARC 16170; Ministry of Defense: London, England, 1953. (32) Beér, J. M.; Lee, K. B. Symp. Combust., [Proc.] 1965, 10, 1187− 1202. (33) Beér, J. M.; Chigier, N. A. Combustion Aerodynamics; Robert E. Krieger Publishing Company: Malabar, FL, 1983. (34) Hottel, H. C.; Williams, G. C.; Bonnell, A. H. Combust. Flame 1958, 2, 13−34. 7679


Article

Energy & Fuels (63) Nehb, W.; Vydra, K. Ullmann’s Encyclopedia of Industrial Chemistry; Wiley-VCH: Weinheim, Germany, 2006; Vol. 35; Chapter Sulfur, pp 5−71. (64) Selim, H.; Al Shoaibi, A.; Gupta, A. Appl. Energy 2011, 88, 2593−2600. (65) Vervisch, L.; Labegorre, B.; Réveillon, J. Fuel 2004, 83, 605− 614.

7680


Formation and Control of Sulfur Oxides in Sour Gas Oxy-Combustion

Recommend Documents