Online Measurement of Elemental Yields, Oxygen Transport

Aug 21, 2014 - Mikael Israelsson,* Anton Larsson, and Henrik Thunman. Department of Energy and Environment, Division of Energy Technology, Chalmers ...
0 downloads 0 Views 788KB Size
Article pubs.acs.org/EF

Online Measurement of Elemental Yields, Oxygen Transport, Condensable Compounds, and Heating Values in Gasification Systems Mikael Israelsson,* Anton Larsson, and Henrik Thunman Department of Energy and Environment, Division of Energy Technology, Chalmers University of Technology, SE-412 96 Gothenburg, Sweden ABSTRACT: Biomass gasification produces a wide range of species, from permanent gases to condensable hydrocarbons, with different composition and boiling points. This complicates the mass balance of the system, as multiple techniques are needed to quantify the various components of the produced raw gas. In this study, a high-temperature reactor for thermal conversion of raw gas at 1700 °C was developed to generate a gas stream that consisted primarily of CO, CO2, H2, and H2O. The reactor was experimentally evaluated and subsequently used for measurements of the raw gas from the Chalmers 2-4−MW dual fluidized bed gasifier. The gas stream that exits the reactor is analyzed to obtain the total elemental flows of C, H, O, and N, which facilitate determinations of the fuel conversion and oxygen transport in a dual fluidized bed reactor. The proposed system was operated in parallel with a gas-cleaning system, to determine the yield of condensable species, including tar and GC-undetectable species. A simplified approach is proposed for quantifying the average energy content of the condensable species, thereby allowing the wet raw gas efficiency and lower heating value (LHV) to be calculated.

1. INTRODUCTION Process streams that contain multiple components with different boiling points are inherently difficult to quantify accurately, as multiple measurement techniques must be employed. For example, thermochemical conversion of biomass for biofuel production can be performed using a wide variety of process types that cover a range of operational parameters, i.e., temperature, fuel type, and reactant gas.1 However, the majority of available processes generate streams that comprise multiple components, consisting of species that range from hydrogen and methane to large poly aromatic hydrocarbons (PAHs). The heavier fraction of these components, generally referred to as tar,2 varies in composition and is typically present in concentrations of 1−100 g/Nm3. Given the complexity of the product gas, or raw gas, produced by gasification, it is often described by quantifying the dry cold gas and the tar separately, in addition to measuring the steam content. This allows further determinations of the process efficiency, fuel conversion, tar dew-point, raw gas heating value, and other parameters of interest. The raw gas efficiency and heating value offer valuable information for processes in which the untreated raw gas is combusted. Furthermore, the raw gas efficiency enables determination of the amount of energy that can be retained during catalytic gas cleaning, such as reformation.3 1.1. Quantification of Individual Components of the Raw Gas. The carbon in the gas phase is present either as permanent gas, which is typically monitored using an online instrument such as a micro-GC or NDIR instrument or as condensable species and tar. Tar compounds can be measured using several online methods,4−8 although the most commonly used methods are off-line, such as the tar protocol9 and the SPA method,10−12 both of which allow quantification of individual species. Online methods are typically faster, making them © 2014 American Chemical Society

suitable for monitoring gasification processes. However, off-line methods offer greater accuracy and the ability to determine the tar composition, which facilitate more detailed studies. The use of two different measurement systems for the cold gas and tar, which have different accuracies and time-scales of measurement, makes it difficult to obtain coherent data for transient experiments. Furthermore, all the aforementioned methods have limitations regarding which species they can measure. Thus, certain species, such as GC-undetectable tar13 and semivolatile species,14 may not be adequately quantified. Incomplete quantification of the components of the raw gas will affect the calculated fuel conversion, via the carbon-based mass balance, as well as the determinations of process efficiency and energy balance. The concentration of steam is mainly important due to its effects on the energy balance and heating value of the wet raw gas. Steam is usually measured through condensation due to the difficulties associated with online measurements of raw gas that contain tar and particles. As a result, these measurements are slower than those of the cold gas. Recent work performed by Cherednichenko et al.15 show promising results regarding online steam measurement, although such considerations are not within the scope of the present work. 1.2. Quantification of Total Elemental Flows. To resolve the issues linked to incomplete quantification, Neves et al.16 have proposed a method for the quantification of C, H, O, and N in the raw gas that involves combusting the gas prior to the analysis. This enables determinations of the total elemental flows in the raw gas using comparatively simple equipment, such as a NDIR system or micro-GC. Furthermore, Received: June 27, 2014 Revised: August 20, 2014 Published: August 21, 2014 5892

dx.doi.org/10.1021/ef501433n | Energy Fuels 2014, 28, 5892−5901

Energy & Fuels

Article

controlled.14 The rotary valve, which introduces fuel to the gasifier, is purged using dried flue gas from the boiler; the amount is determined on the basis of the level of nitrogen in the gas-conditioning system. Bed material is circulated between the boiler and gasifier to allow transport of heat to the gasifier, as well as transport of unconverted char to the boiler where it is combusted. Additionally, metal-containing bed materials oxidize in the boiler and are subsequently reduced in the gasifier, resulting in oxygen addition to the raw gas.21 The hightemperature reactor (HTR) and a gas-conditioning system, for the separation of condensable species, are used to carry out parallel measurements on a slip-stream of raw gas. 2.1. Measurements Using the HTR. The proposed system allows online measurements of elemental yields through comparison of the molar flow rates of the fuel feed and the gas leaving the HTR:

as steam is the only condensable component that remains after combustion, online moisture determination of the gas is possible,17 thereby enabling complete quantification of the hydrogen and oxygen contents of the wet raw gas. The system developed by Neves et al.16 was used successfully to measure the raw gas from the Chalmers 2-4−MW dual fluidized bed (DFB) gasifier. Furthermore, the performed experiments raised the possibility of obtaining even higher levels of accuracy using this type of measurement. The nitrogen present in the air needed for combustion dilutes the species of interest, thereby compromising to some extent the accuracy of the analysis system. Nitrogen dilution can be avoided by using oxygenenriched air for the combustion, although there are serious safety issues concerning the usage of pure oxygen. In the present work, a method that does not require a reactant gas was developed by utilizing thermal cracking at 1700 °C. This induces the decomposition of larger molecules into primarily CO, CO2, H2, and H2O, which are more readily measured than the entire raw gas spectrum. The thermal decomposition of various tar components in argon has been thoroughly investigated,18,19 revealing significant conversion at temperatures in the range of 700°−1000 °C and residence times of 5 s. Similar measurements, in which soot formation was also determined, were performed using steam with shorter residence times at higher temperatures.20 Jess20 achieves complete conversion of naphthalene at 1300°−1400 °C, with maximum yields of other tar components and soot at 1100 and 1250 °C, respectively. At 1400 °C, the amount of soot decreased, but it was still significant. These findings imply that the temperature and residence times needed for satisfactory conversion to light gases are not dictated by the conversion of tar, but rather by the subsequent gasification of soot. Nearcomplete conversion of the soot is crucial, as all the carbon that remains as soot will cause an error in the mass balance, resulting in a seemingly lower yield of carbon. The purpose of the present work was to establish a robust and highly accurate online measurement method for the quantification of C, H, O, and N in gas streams that contain multiple components. The accuracy and robustness of the proposed method was investigated in validation experiments designed to evaluate the levels of conversion and soot formation. The method was validated for large-scale applications, using the Chalmers 2-4−MW DFB gasifier. In addition, a simplified approach for determining the energy content of the raw gas is presented, which allows determinations of the efficiency and heating value of the wet raw gas.

ni̇ ,HTRdrygas ni̇ ,fuel =

(1 ± Etot, i) =

ni̇ ,HTRdrygas(1 ± Emeas, i) ni̇ ,fuel(1 ± Efuel, i)

∑j nj̇ ,HTRdrygasYi , j(1 ± εmeas, j) ni̇ ,fuel(1 ± Efuel, i)

(1)

where ṅ is a molar flow [mol/s], ε describes the degree of error of a specific measurement, and E describes the lumped error of a specific process stream or element. The different elements (C, H, O, and N) are represented by i, while j denotes the various gas components, which include CO, CO2, H2, and CH4. Yi,j is the molar content of element i in gas component j [mol/mol]. The measurement error (εmeas,j) is mainly dependent upon the uncertainty concerning the composition of the calibration gases, which is determined to within 1% relative to the given concentration for all the species. Similar to all measurements wherein a measured parameter is related to the fuel feed, uncertainties related to fuel composition can significantly affect the calculated yields of C, O, and H. Consequently, the fuel feed, moisture content, and composition, as well as the composition of the char, need to be determined during the measurements. In the present work, helium was premixed with the steam feed of the gasifier and used as a trace gas to determine the molar flows of the dried gas according to nj̇ ,drygas =

Cj ,drygas C He,drygas

× nHe ̇

(2)

where C is a measured molar concentration [mol/m3]. Implementation of eqs 1 and 2 allows determinations of the carbon-based fuel conversion to raw gas and the char conversion. However, the char conversion calculation requires accurate data concerning the primary char yield and its elemental composition. This necessitates dedicated measurements, as the char yield can vary significantly for different particle sizes and heating rates.22 In the present work, the char yield and composition were obtained from experiments in which fuel from the same source was exposed to operating conditions similar to those in the gasifier, as reported elsewhere.23 The gasifier is fluidized using steam, which also participates in numerous reactions, such as char gasification and the water− gas shift reaction. As a result, unlike the carbon yield, the hydrogen and oxygen yields determined from eq 1 can be significantly higher than 100%. High oxygen yields can also be the result of oxygen addition due to air being fed to the gasifier

2. METHODOLOGY The different measurements and streams associated with the gasifier are shown in Figure 1. The inward flows of fuel, steam, and trace gas (helium) are continuously monitored and

Figure 1. Overview of the flows in the gasifier. HTR, high-temperature reactor. 5893

dx.doi.org/10.1021/ef501433n | Energy Fuels 2014, 28, 5892−5901

Energy & Fuels

Article

or may be due to leaks or oxygen transport via the bed material. Oxygen addition attributable to fed air can be determined by the nitrogen mass balance. However, if a nitrogen measurement system is not readily available or if oxygen is transported by the bed material, as in the case of DFB gasification, the amount of added oxygen can be determined as follows: ⎡O⎤ nȮ −add = ΔH fuel,HTR × ⎢ ⎥ − ΔOfuel,HTR ⎣ H ⎦H O 2 ⎛⎡ O ⎤ ⎡H⎤ ⎡O⎤ ⎞ + ΔCfuel,HTR × ⎜⎜⎢ ⎥ − ⎢ ⎥ × ⎢ ⎥ ⎟⎟ ⎝⎣ C ⎦UC ⎣ C ⎦UC ⎣ H ⎦H2O⎠

Figure 2. Flow of the data that are included in the mass balance. (3)

ni̇ ,CS = ni̇ ,HTR (1 ± εi ,HTR,meas) + ni̇ ,HTR,H2O

Here, Δifuel,HTR denotes the differences in the molar flow rates of carbon, oxygen, and hydrogen, respectively, as determined by the levels in the fuel feed and the dried gas exiting the HTR. The bracketed terms in eq 3 denote the molar ratios of the unconverted part of the fuel (UC) and the oxygen to hydrogen ratio of water (1/2). The first two terms in eq 3, which relate to hydrogen and oxygen, simply describe the differences in oxygen and hydrogen levels between the fuel and the HTR gas. The third term, concerning carbon, describes the effects of the oxygen and hydrogen within the char that is exiting the gasifier. If the difference in hydrogen levels between the fuel and the HTR gas is twice the difference in the oxygen levels, the increase in oxygen level can be explained by the water−gas shift reaction, and therefore, oxygen addition is not occurring. Conversely, if the difference in hydrogen levels between the fuel and the HTR gas is less than twice that of the difference in the oxygen levels, oxygen addition is taking place. The importance of quantifying the oxygen balance for DFB gasifiers has been emphasized previously.21 It should be noted that eq 3 is accurate even at very high levels of oxygen addition, such as those used in CLC combustion, 24 provided that the composition of the unconverted fuel is known. The present work focuses on validating the proposed measurement method, using DFB gasification, and the composition of the unconverted fuel is equivalent to that of the previously mentioned char. A positive aspect of char is that the hydrogen to oxygen ratio is around 2 for a wide range of temperatures.25 Consequently, the expression in parentheses in the third term of eq 3 often has a low value, which means that char conversion has little effect on the perceived oxygen addition. Nevertheless, if a process exhibits incomplete devolatilization or uses other types of fuels, information regarding the unconverted fuel composition is required. 2.2. Combined Measurements. Operating the proposed system in parallel with a gas-conditioning system permits the acquisition of useful additional information. When the two systems are synchronized, the measurements can be compared to yield indirect measurements of the amount and average composition of condensable species, which consist of all raw gas species that are not found in the cold gas. In combination with known process parameters, such as the fuel and steam feeds, the two systems can be operated (as shown in Figure 2) to monitor the C, H, O, and N molar balances in the gasifier. Comparison of the data from the gas-conditioning system, G, and the data from the HTR can be done on two levels: with and without SPA analysis of the tar. If the tar measurement is omitted, the comparison is fast and yields information on the amount and average C, O, and H composition of all the condensable species (CS), which are not measured by the gas conditioning system, as follows:

− ni̇ ,G(1 ± εi ,G,meas) − ni̇ ,G,H2O

(4)

In this equation, H2O represents the condensed steam after the HTR and gas-conditioning system. Errors in the measured amounts of condensate after both systems will affect the determined oxygen and hydrogen content of the condensable species. The errors related to the characterization of the condensable species are also dependent upon the gas measurement. However, as two separate gas measurements are used, i.e., one for the gas-conditioning system and one for the HTR gas, the impact of the analysis error depends on the differences between the two systems in the measured concentrations of a specified component. For instance, if there are low levels of tar and other decomposable components in the raw gas, the difference in the volumetric helium concentration between the two systems will be small. Similar values for the measured helium concentrations will entail almost identical systematic errors of analysis, provided that the two systems were calibrated using the same gas. Consequently, the resulting total error for helium will be small. Conversely, for large differences in the concentrations of helium, the resulting error can increase, as the systematic error of the two measurements may differ. In the present work, the accuracy of the measurement of the condensable species is estimated by varying the concentrations of all species randomly, assuming normally distributed probabilities, based on the given accuracies of the calibration gases. This does not include the previously mentioned dampening effects on the overall error of similar concentrations. Nevertheless, it does yield information concerning the detection limit of the current system. The main purpose of the proposed system is to quantify product streams that are relevant for the overall mass and heat balances. Therefore, while a low detection limit is desirable, other methods will be more suitable for the quantification and identification of low levels of condensable species.9,10 If the tar is measured, using the SPA method, an additional level of comparison is possible. However, the rate of this comparison is low and is dependent upon the performed tar analysis. Subtracting the level of measured tar from the level of condensable species enables determination of a group of species that was identified by Larsson et al.14 As they are considered to be semivolatile, these species are not measured using the SPA method and are too heavy to be measured by the gas-conditioning system. Presumably, this group also contains a fraction of species, ranging from benzene to xylene, which is not completely quantified using the 500 mg of aminopropylbonded silica adsorbent in the SPA columns.11,26 Regardless of 5894

dx.doi.org/10.1021/ef501433n | Energy Fuels 2014, 28, 5892−5901

Energy & Fuels

Article

Table 1. Calculated CHmin and Oxygen-Based LHV Values for a Variety of Species Present during Pyrolysis and Gasification CHmin

LHV [kJ/mol O2]

included species

Milne pyrolysis excluding SPA SPA tar

0−2.0 (0.79)

418.7−460.3 (436.9)

ethene, acetaldehyde, acetic acid, acetone, acrylic acid, furan, 2-butanone, cyclopentanone, furfural, furfuryl alcohol

0.5−1.25 (0.78)

414.5−431.9 (421.9)

miscellaneous

0.5−3.0 (2.07)

408.1−438.4 (417.0)

benzene, phenol, toluene, o-cresol, styrene, benzofuran, m/p-xylene, indene, naphthalene, 1naphthol, 2-methylnaphthalene, dibenzofuran, biphenyl, acenaphthylene, acenaphthene, xanthene, fluorene, phenanthrene, anthracene, fluoranthene, pyrene, triphenylene, coronene ethane, ethanol, propane, propene, propanol, butane, butadiene, butanol, 1,4-butanediol, diacetyl, pentane, 1-pentene, pentanol, n-hexane, cyclohexane, 1-hexanol, heptane, octane, nonane, decane

or 2.8%, assuming equal amounts of all the species. The accuracy of this approach for determining a heating value is debatable, although it offers a fairly narrow range within which the correct value can be expected. As the condensable species most probably comprise a mixture of compounds, large deviations from the determined mean value are unlikely. Furthermore, the energy contained in the condensable species accounts for roughly 10% of the energy in the fuel.14 Thus, errors as large as 10% in the oxygen-based LHV will only induce an error of the order of ≤1% in the overall energy balance. The calculated average energy content of the condensable species can be combined with that of the dried cold gas to determine the raw gas efficiency according to

the nature of this group, its quantification allows a deeper understanding of the gas phase chemistry in gasification. 2.3. Average Composition of Condensable Species. Both methods for CS determination, i.e., with or without SPA analysis, require fast and accurate measurements of the steam in the raw gas, to determine with accuracy the oxygen and hydrogen fractions of the indirectly measured components. If the steam is not measured, the condensate terms in eq 4 can be omitted. As a result, the average oxygen and hydrogen content of the organic components will include an unknown fraction of H2O: CiHjOk = CiHj − 2xOk − x + x H 2O

(5)

The lowest possible ratio of hydrogen to carbon of the condensable species can be determined by setting x = k, thereby removing all the oxygen as water. The minimum ratio of hydrogen to carbon, hereinafter referred to as CHmin, contains information on the average size of the unknown components, as well as their average heating value: CH min =

ηrawgas = ∑ nj̇ ,G × LHVj ,G + (nC,CS + (n H ̇ ̇ min,CS /4)) × LHVO2,CS ṁ fuel × LHVfuel (8)

j−2×k i

Similarly, the theoretical raw gas efficiency can be determined by calculating the energy in the converted fraction of the fuel, using eq 9. The resulting efficiency describes the maximum amount of energy in the raw gas that can be recovered from the energy in the fuel.

(6)

A CHmin value in the range of 2−4 implies that the condensable mixture consists mainly of alkanes, while CHmin values in the range of 0.5−1.0 suggest the presence of aromatic species. However, species with high O/C ratios, such as acrylic acid (C3H4O2) and furan (C4H4O), have CHmin values of 0 and 0.5, respectively. Thus, low CHmin values may result from large PAHs, small oxygenated species, or both. 2.4. Average Energy Content of Condensable Species. The lower heating values (LHV), on a mass basis, vary significantly for different hydrocarbon species, making it difficult to estimate the energy content of the condensable species. As an alternative, the amount of released energy per reacted O2 molecule needed for combustion [kJ/mol O2] can be determined for compound A (CiHjOk) according to LHVO2,A =

ηrawgas,theor =

ṁ fuel × LHVfuel − ṁchar × LHVchar ṁ fuel × LHVfuel

(9)

The difference between the above efficiencies can be considered as the enthalpy change within the gasifier due to various reactions, including heat from the bed material that is chemically bound within the raw gas. The combined energy contents of the dried cold gas and the condensable species can be used together with the total flow of raw gas to determine the LHV [in MJ/Nm3] of the wet raw gas according to

Δf H g0,A − i × Δf H g0,CO2 − (j /2) × Δf H g0,H2O

LHVrawgas =

i + (j /4) − (k /2) (7)

∑ nj̇ ,G × LHVj ,G + (nC,CS + (nH ̇ ̇ min,CS/4)) × LHVO2,CS ̇ Vrawgas

Implementation of the oxygen-based LHV makes it possible to determine the energy content of the condensable species using only the amount of carbon and the CHmin value. Furthermore, it is irrelevant if the “true” component A contains oxygen, i.e., x ≠ k in eq 5, as this will not affect the amount of oxygen required for combustion. The calculated heating values of compounds derived from pyrolysis27 and gasification, as well as those of various alkanes, alkenes, and alcohols, are presented together with their calculated CHmin values in Table 1. Overall, the mean value of the oxygen-based LHV for all three groups is 422.9 kJ/mol O2, with a standard deviation of 11.7 kJ/molO2,

(10)

where V̇ is the total volumetric flow of the raw gas [in Nm /s], consisting of the cold gas flow, as measured by the gasconditioning system, the steam flow, determined using the mass balance, and the flow of condensable species, which are assumed to be free of oxygen. Furthermore, it is assumed that the average tar molecule contains six carbon atoms. However, this assumption is of little relevance, as the contribution to the total flow is small. 3

5895

dx.doi.org/10.1021/ef501433n | Energy Fuels 2014, 28, 5892−5901

Energy & Fuels

Article

to separate the combustion of remaining gas and soot attached to the pipe walls from the soot captured by the filter. Due to the fast combustion of the soot, the produced CO and CO2 were measured using a NDIR instrument (Rosemount MLT) to gather data once per second. Once a stable, atmospheric CO2 background was obtained, the filter temperature was increased to >500 °C, to allow combustion. As a result, the quantified levels of soot should be considered indicative of the total soot yield. Soot formation was determined at temperatures of 1500, 1600, and 1700 °C. Furthermore, the measurement at 1700 °C was used to determine the stability, accuracy, and reproducibility of the reactor system. 3.3. Applied Measurements. The gasifier measurements were performed in the Chalmers 2-4−MWth DFB gasifier, which is fluidized using steam with a known amount of helium, usually at 20−50 NL/ min, to allow the quantification of gaseous species.14 The helium was added to the steam prior to it entering the gasifier to facilitate an even distribution throughout the gasifier. The resulting volumetric fraction of helium in the raw gas was around 0.5−1.0%. The gasifier was operated with wood pellets at 1−2 kPa subatmospheric pressure and 820 °C with fuel and steam feeds of 295 and 130 kg/h, respectively. The mass fraction of moisture in the wood pellets was 8.2%, and the contents of C, H, and O were 50%, 6%, and 43%, respectively, on dry mass basis. During the period of measurement, the gasifier was operated using bauxite at low levels of bed material regeneration, in order to determine aging effects in a separate study. As a result, the system exhibited significant char conversion and oxygen transport between the boiler and gasifier. Furthermore, due to the aging of the bed material, it was possible to perform measurements during certain periods of transient behavior of the fuel conversion. This enabled an appraisal of the system’s ability to monitor small changes in real time. The instrumental setup is depicted in the left panel of Figure 4. Raw gas (1) was continuously sampled through a heated ceramic filter (2),

The LHV of the raw gas, determined using eq 10, requires input data (Figure 2). However, if a gas-conditioning system is not available, the data obtained using only the HTR can be used to determine equivalent LHVs. As an example, the methane-equivalent raw gas LHV is determined by rearranging the gas composition that exits the HTR into CH4, H2O, and CO2. Although the choice of equivalent species is dependent upon the process type, the equivalent LHV nevertheless presents a simple means for process monitoring and control.

3. EXPERIMENTAL SECTION 3.1. Reactor Design. The high-temperature reactor (Figure 3) consists of a ceramic reactor and oven inside a gastight steel casing.

Figure 3. Schematic of the high-temperature reactor (left), with an enlarged image of the connections (right). Gas, at a temperature of 350 °C, is introduced to the top of the reactor via a stainless steel adaptor (1). The adaptor is connected to the reactor by a flange (2) using graphite packing to avoid leaks. The other end of the adaptor is connected to an 8 mm alumina (Al2O3) tube (3) using a stainless steel fitting with graphite packing, to create a leakagefree joint without breaking the alumina tube. The lower part of the reactor contains a larger 35 mm alumina tube with a closed bottom (4), surrounded by four heating elements (5) (Kanthal Super 1800). The top part of this tube is connected to the reactor ceiling using a pack box (6) with graphite packing. The outer shell of the reactor is composed of stainless steel and is designed to be gastight at operational pressures (80−101 kPa). In the event of excessive internal gas exchange between the large alumina tube and oven, the gas in the oven can be continuously evacuated to ensure minimal back-mixing into the reactor. During operation, gas is drawn through the adaptor and is heated during its transport to the bottom section via the narrow alumina tube. The narrow tube ensures minimal residence times at temperatures that promote high soot yields but that are too low to support soot gasification. The gas then enters the larger alumina tube and is slowly transported upward through the high-temperature section of the reactor. The gas exits the reactor via an outlet (7) that is positioned 10 cm below the inlet adaptor to avoid excessive convective heating of the upper graphite packing. 3.2. Validation Experiments. Synthetic gas measurements were performed to determine the overall degree of conversion and soot formation in the reactor. Synthetic gas of a known composition was supplied from a gas bottle and mixed with steam to 50 vol % before entering the reactor. After the reactor, soot particles were collected in an uncoated diesel particulate filter (DPF), which was maintained at 150 °C during operation. The particle-free gas was then cooled to condense the remaining steam before it was analyzed in a micro-GC (Varian CP4900), described elsewhere,14 that was capable of analyzing all the species in the supplied gas. The collected soot was quantified at the end of each measurement by introducing a known flow of air into the system while maintaining the reactor temperature. The oxidation of the system was performed at an initial filter temperature of 150 °C,

Figure 4. Schematics of the HTR system (left) and the gas-cleaning system (right). The different components are 1: raw gas; 2: ceramic filter; 3: HTR; 4: SPA sampling point; 5: Peltier cooler; 6: filter; 7: gas pump; 8: flow meter; 9: micro-GC; 10: cooler.

which was maintained at 350 °C and used to remove particles from the gas before it entered the HTR (3). Samples for the SPA analysis were collected directly at the outlet of the HTR (4), as described previously,12 to determine the degree of reformation of the SPAdetectable tar fraction. The gas flow was cooled, and steam was condensed in a Peltier cooler (5), after which the aerosols were separated using a filter (6). The dry gas passed through a pump (7) and a flow meter (8) before reaching the micro-GC (9), which was calibrated 1 day prior to the measurement. The start-up procedure for the measurements consisted of initiating a temperature ramp a few hours before operation, to allow the alumina tubes to heat up slowly and thereby avoid cracking as a result of thermal expansion. When the operational temperature was reached, the reactor was purged with nitrogen before the introduction of a flow of raw gas. Parallel measurements were performed on the untreated raw gas, using the gas-cleaning system (right panel in Figure 4) to determine the dry raw gas composition, concentration of steam, and amount of tar. SPA samples (4) were collected directly after the ceramic filter (2), after which the gas was quenched with isopropanol in two coolers (10), to condense the tar and steam. The gas was further cooled in a Peltier cooler (5), after which it was passed through a wool filter (6), 5896

dx.doi.org/10.1021/ef501433n | Energy Fuels 2014, 28, 5892−5901

Energy & Fuels

Article

to separate the aerosols, a pump (7), and a flow meter (8), before being analyzed in the micro-GC and NDIR instrument (9). Errors can be expected during the measurements from air leakages into hot zones, the escape of soot from the HTR or gasifier, and incorrect synchronization of the equipment. Leakages of air prior to the HTR may be interpreted as oxygen addition (Figure 2), as the leaked air is combusted. The equipment was continuously tested for leaks. However, as heated connections can occasionally fail, resulting in leaks, the He/N2 ratios of the cold gas and the HTR gas were monitored for deviations. When detected, the amount of leaked air could be determined and compensated for, by comparing the nitrogen flows of the two systems. Soot formation within the gasifier could not be determined, and this results in a seemingly lower fuel conversion. The escape of soot from the HTR affects the determined fuel conversion, as well as the yield of condensable compounds. Therefore, complete soot conversion needs to be guaranteed for reliable measurements. Incorrect synchronization of the measurements can further complicate the analyses of transient measurements. This is rarely a problem, as the initiation of helium provides the difference in response time for the two systems.

are summarized in Table 3. The results were calculated according to eqs 1−3 for data obtained during 1 h of operation Table 3. Yields, Variance, and Estimated Errors for the Measurements and Analyses of the C, O, and H Levels and Oxygen Addition during Synthetic Gas HTR Experiments Conducted at 1700 °C

Table 2. Compositions of the Inlet and Outlet Streams during Synthetic Gas HTR Experiments Conducted at 1700 °C

a

feed gas [vol %]

exiting gas [vol %]

H2 CO CO2 CH4 C2H2 C2H4 C3H8 He N2

25.2 39.52 8.94 11.9 0.496 4.99 0.994 4.97 2.99

54.31 29.13 11.45 0.05 N.D.a 0.03 N.D. 2.73 2.30

yield [%]

sample %RSD

Etot [%]

Eanalysis [%]

C O H O addition

99.83 165.3 156.5 102.8

0.33 0.63 0.55 0.51

0.79 1.34 1.25 0.72

0.53 0.86 0.87 0.53

at 1700 °C. Table 3 also includes the estimated errors of the analysis, Eanalysis = f(εmeas), and the total errors of the measurement, Etot = f(εmeas, εfuel). Both the error for the carbon yield (0.17%) and the variations of all the measurements are lower than the corresponding errors for the analysis. This indicates that the accuracy of the carbon measurement is limited by the analysis equipment. The error for oxygen addition (2.8%) is larger than the total error for its measurement, indicating a significant level of added oxygen in the HTR. The oxygen addition could be caused by inward air leakages, the presence of pockets of air inside the reactor system, or outward leakage of hydrogen. To determine the actual cause, accurate measurements of the steam before and after the reactor are needed, which would allow individual quantification of the levels of hydrogen and oxygen. However, the current accuracy of the oxygen addition is limited to that of the reactor system. The soot yields, expressed as fractions of the supplied carbon [mass %], from the validation experiments are presented in Figure 5. The amount of soot collected after the HTR is shown

4. RESULTS AND DISCUSSION The experimental results, presented below, are divided into those obtained: (1) from validation experiments, to evaluate the method, and (2) from applying the method to the Chalmers DFB gasifier. 4.1. Validation Experiments. Synthetic gas measurements were performed on a known gas mixture, to determine the degree of conversion, as well as the levels of accuracy and soot formation as a function of temperature. The average values obtained from the synthetic gas measurements performed at 1700 °C are presented in Table 2, together with the composition of the synthetic gas mixture. A

species

element

N.D., not detected.

sufficient conversion of large species was obtained, with only low levels of CH4 and C2H4 exiting the reactor system. Furthermore, the total volume of dry gas was increased by a factor of 1.82, as determined by the concentration of helium, due to the cracking of larger molecules and the water−gas shift reaction. The concentration of nitrogen is not affected to the same extent due to an exchange of gas between the measured gas and the volumes that were purged with nitrogen prior to the measurement. This exchange of gas is small at