Scholarly article on topic 'Alstom's Regenerative Calcium Cycle - Norcem Derisking Study: Risk Mitigation in the Development of a 2nd Generation CCS Technology.'

Alstom's Regenerative Calcium Cycle - Norcem Derisking Study: Risk Mitigation in the Development of a 2nd Generation CCS Technology. Academic research paper on "Chemical engineering"

CC BY-NC-ND
0
0
Share paper
Academic journal
Energy Procedia
OECD Field of science
Keywords
{"Regenerative Calcium Cycle" / "Calcium Looping" / "Carbonate Looping" / "Pilot Testing" / "Model Development" / "Process Performance Predictions"}

Abstract of research paper on Chemical engineering, author of scientific article — Michael C. Balfe, Ola Augustsson, Raul Tahoces-soto, Liv-Margrethe H. Bjerge

Abstract Alstom is a pioneer and industrial leader in the development of post combustion CCS technologies. Alstom's Regenerative Calcium Cycle (RCC) is a 2nd generation post combustion CCS technology utilizing a calcination/ carbonation loop to capture CO2 from flue gas at high temperatures. The CO2 capture cycle is driven by introducing heat required to regenerate sorbent at 900°C (4.0 GJ/t CO2, chemically bound CO2) the regenerated sorbent is then used for CO2 capture from flue gas where the exothermic heat of the reverse reaction (-4.0 GJ/t CO2 captured) is recovered in a power cycle. Due to the option of using natural sorbent materials (limestone), RCC also provides attractive integration opportunities with industrial processes such as cement production. Recently indirectly fired RCC calcination concepts also open the door for game changing performance gains using natural or even synthetic materials. However, the wide range of performance variability characteristic of natural sorbents contributes to uncertainty in defining process performance. Reliable sorbent deactivation models, which describe both cyclic and chemical deactivation, in combination with mechanistic reactor models which predict gas and solids phase conversions are required to lower the uncertainty associated with new reactor concepts and larger geometries required for commercialization for retrofit power-plant flue gas or for integrated solutions for the cement industry. Parallel to RCC process development, Alstom is currently executing a “Derisking” or risk-mitigation study for Norcem, a Norwegian cement producer, who seeks to take a leading position in the development of technologies for CO2-capture from cement production facilities. Norcem is currently heading up a project on behalf of the European cement industry (partners are Norcem, HeidelbergCement and the European Cement Research Academy) investigating various technological options at the pilot scale. In relation to this project, Alstom's RCC technology was selected with the objective to define process performance and the associated uncertainty. Pilot testing was conducted at the Institute of Combustion and Power Plant Technology at the University of Stuttgart (IFK) planned by Alstom focusing on the validation of process models and less the targeting of pilot plant performance, which is often limited by reactor geometries. This paper discusses the characterization of sorbent performance and the gap between the performance expected from thermo- gravimetric analysis (TGA) and that realized during steady state pilot testing under representative operating conditions. Operational stability, material balance closure and the confirmation of sorbent attrition characteristics from pilot testing are also discussed and steady-state operational data, gained from pilot testing are applied to initialize mechanistic process models which allow prediction of reactor performance for larger more efficient designs.

Academic research paper on topic "Alstom's Regenerative Calcium Cycle - Norcem Derisking Study: Risk Mitigation in the Development of a 2nd Generation CCS Technology."

CS)

CrossMark

Available online at www.sciencedirect.com

ScienceDirect

Energy Procedia 63 (2014) 6440 - 6454

GHGT-12

Alstom's Regenerative Calcium Cycle - Norcem Derisking Study: Risk mitigation in the development of a 2nd generation CCS

technology.

Michael C. BALFEa*, Ola AUGUSTSSONa, Raul TAHOCES-SOTOa Liv-Margrethe H. BJERGEb

aALSTOM Carbon Capture GmbH, Lorenz-Schott-Str.4, Wiesbaden, Germany b NORCEM AS, Setrevegen 2, Brevik, Norway

Abstract

Alstom is a pioneer and industrial leader in the development of post combustion CCS technologies. Alstom's Regenerative Calcium Cycle (RCC) is a 2nd generation post combustion CCS technology utilizing a calcination/ carbonation loop to capture CO2 from flue gas at high temperatures. The CO2 capture cycle is driven by introducing heat required to regenerate sorbent at 900 °C (4.0 GJ/t CO2, chemically bound CO2) the regenerated sorbent is then used for CO2 capture from flue gas where the exothermic heat of the reverse reaction (-4.0 GJ/t CO2 captured) is recovered in a power cycle. Due to the option of using natural sorbent materials (limestone), RCC also provides attractive integration opportunities with industrial processes such as cement production. Recently indirectly fired RCC calcination concepts also open the door for game changing performance gains using natural or even synthetic materials. However, the wide range of performance variability characteristic of natural sorbents contributes to uncertainty in defining process performance. Reliable sorbent deactivation models, which describe both cyclic and chemical deactivation, in combination with mechanistic reactor models which predict gas and solids phase conversions are required to lower the uncertainty associated with new reactor concepts and larger geometries required for commercialization for retrofit power-plant flue gas or for integrated solutions for the cement industry.

Parallel to RCC process development, Alstom is currently executing a "Derisking" or risk-mitigation study for Norcem, a Norwegian cement producer, who seeks to take a leading position in the development of technologies for CO2-capture from cement production facilities. Norcem is currently heading up a project on behalf of the European cement industry (partners are

* Corresponding author. Tel.: +49 6134 712 319; fax: +49 6134 712 590. E-mail address: michael.balfe@power.alstom.com

1876-6102 © 2014 Alstom Technology Limited. Published by Elsevier Limited. This is an open access article under the CC BY-NC-ND license

(http://creativecommons.Org/licenses/by-nc-nd/3.0/).

Peer-review under responsibility of the Organizing Committee of GHGT-12

doi: 10.1016/j.egypro.2014.11.679

Norcem, HeidelbergCement and the European Cement Research Academy) investigating various technological options at the pilot scale. In relation to this project, Alstom's RCC technology was selected with the objective to define process performance and the associated uncertainty. Pilot testing was conducted at the Institute of Combustion and Power Plant Technology at the University of Stuttgart (IFK) planned by Alstom focusing on the validation of process models and less the targeting of pilot plant performance, which is often limited by reactor geometries.

This paper discusses the characterization of sorbent performance and the gap between the performance expected from thermo-gravimetric analysis (TGA) and that realized during steady state pilot testing under representative operating conditions. Operational stability, material balance closure and the confirmation of sorbent attrition characteristics from pilot testing are also discussed and steady-state operational data, gained from pilot testing are applied to initialize mechanistic process models which allow prediction of reactor performance for larger more efficient designs.

© 2014 Alstom Technology Limited. Published by Elsevier Limited. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).

Peer-review under responsibility of the Organizing Committee of GHGT-12

Keywords: Regenerative Calcium Cycle; Calcium Looping; Carbonate Looping; Pilot Testing; Model Development; Process Performance Predictions

1. Introduction

Alstom's Regenerative Calcium Cycle (RCC) is a 2nd generation post combustion CCS technology utilizing a calcination/ carbonation loop to capture CO2 from flue gas at high temperatures. The CO2 capture cycle is driven by introducing heat required to regenerate sorbent at 900 °C (4.0 GJ/t CO2, chemically bound CO2) the regenerated sorbent is then used for CO2 capture from flue gas where the exothermic heat of the reverse reaction (-4.0 GJ/t CO2 captured) is recovered in a power cycle. Utilizing natural materials such as limestone for make-up sorbent, the RCC also provides attractive integration opportunities for industrial processes such as cement production. While larger scale testing of 2nd generation RCC technology remains a focus of the development effort, indirectly fired calcination concepts provide a new game changing perspective for both power generation and industrial integration. By improving performance predictions of the 2nd generation, Alstom is well positioned to fast track development of future indirect calcination schemes at low risk.

Norcem, a Norwegian cement producer, seeks to take a leading position in the development of technologies for CO2-capture from cement production and is heading up a project on behalf of the European cement industry (partners are Norcem, HeidelbergCement and the European Cement Research Academy). Norcem has received funding from Gassnova through the CLIMIT program to build a small scale CO2 capture test center at the cement plant in Brevik, Norway. The Norcem test facility will be used to study and compare different post-combustion CO2 capture technologies and determine how suitable they are for implementation in modern cement kiln systems. The current project phase is scheduled from May 2013 to March 2017 and Alstom's Regenerative Calcium Cycle (RCC) has been selected among other technologies for investigation. Alstom is currently executing a "Derisking" or risk-mitigation study to utilize state of the art sorbent characterization techniques combined with pilot testing and process modeling to increase the level of precision associated with process calculations and performance predictions. One aspect of the Derisking Study is the execution of pilot testing at the Institute of Combustion and Power Plant Technology at the University of Stuttgart (IFK), focusing on the validation of process models.

This paper discusses the characterization of sorbent performance and the gap between the performance expected from thermo-gravimetric analysis (TGA) and that realized during steady state pilot testing under representative operating conditions. Operational stability, material balance closure and the confirmation of sorbent attrition characteristics from pilot testing are also discussed and steady-state operational data, gained from pilot testing are applied to initialize mechanistic process models which allow prediction of reactor performance for larger more efficient designs.

Nomenclature

X single particle sorbent activity (from TGA), [mole species/ mole Ca]

XAVE particle population sorbent activity (extracted from pilot plant), [mole species/ mole Ca]

Xr residual activity (fitting parameter), [mole species/ mole Ca]

k deactivation constant, (fitting parameter), [-]

N number of calcination/ carbonation cycles, [-]

XN particle sorbent activity after N cycles, [mole species/ mole Ca]

FO sorbent make-up rate, [mole CaO/h]

Fr sorbent circulation rate, [mole CaO/h]

,N fraction of sorbent having made N cycles, [-]

U Superficial velocity, [m/s]

Umf Superficial velocity at minimum fluidization, [m/s]

do Initial particle size, [m]

t Time, [s]

dt Particle size at time t, [m]

Ka Attrition constant, [s/m2]

D50 Diameter associated with 50% of the particle mass, [^m]

2. Analysis of Sorbent Materials

The carbonation rate of calcined limestone is characterized by a rapid kinetically controlled initial reaction stage followed by a much slower diffusion controlled second stage, due to the formation of a CaCO3 product layer [1]. Successive sorbent calcination and carbonation cycles cause sintering, and a loss of surface area [2], Eq. (1) has been shown to fit the deactivation of natural sorbents under many conditions for up to 500 cycles [3].

- + X,

1 - X,,

- + kN

Fitting parameters, k, and, Xr, can be determined by TGA, but the thermal program and gas composition environment can have a significant influence on the values of these parameters [4,5]. While TGA conditions are selected to be representative of process operating conditions, limitations in machine programming and capability (thermal ramping rates, reaction times, control of feed gas composition, heat flux, measurement resolution and sample sizes) lead to approximations of the true reaction environment and associated inaccuracies in parameter estimation. Following this line of thought drastically different measurement techniques might be expected to magnify differences in the estimated parameters. Recent investigations for various sorbents comparing TGA and a fluidized bed device [6] indicate such differences. Since test conditions between methods also vary, the authors conclude that the type of differential reactor has a lesser impact on deactivation than the achieved reaction environment.

2.1. TGA apparatus and programming

A Perkin Elmer Pyris 1 thermo-gravimetric analyzer (manuf. in 2007) with an external high temperature furnace (having a maximum heating rate of 50 °C/minute, a peak temperature of 1400 °C and continuous operation up to 1200 °C) was used for this work. The apparatus is equipped with a gas system that allows programming the reactive gas by switching of up to four different sources. Typical reactive gases include O2, SO2 mixtures, CO mixtures, or HCl mixtures which are metered at typically between 10-50 cmVminute. The inert gas flow to purge the balance is

typically between 10-50 cmVminute and can be selected as N2, Air, or CO2 mixtures. Sample sizes are typically selected to be between 1 and 50 milligrams, with sensitivity of 0.1/1.0 micrograms (low/high range).

The thermal program applied consisted of a 3 minute carbonation step at 650 °C with 15 vol. % CO2 followed by a calcination step above 900 °C for 3.5 min in nitrogen. Although calcination in nitrogen occurs at lower temperatures the sorbent was heated to above 900 °C in each cycle in an effort to capture a representative sintering effect. While a shorter calcination time was desired, since according to [7] 3.5 minutes is notably longer than necessary for calcination in pure CO2, additional time was required to allow the balance to settle out above 900°C leading to a minimum time span of 3.5 min.

2.2. Sorbent composition

Of four limestone sources investigated (three locally accessible to Brevik and one from Germany) two were selected for testing in the IFK pilot plant. Rheinkalk, in Wülfrath, Germany supplied a Baseline sorbent which was suggested to be globally readily accessible, and Franzefoss in Verdal, Norway supplied a locally accessible material which was shipped to Rheinkalk in Germany for further processing to achieve a suitable size distribution. A representative portion of each material was pulverized in a laboratory ball mill, to further homogenize the samples, and analyzed according to the applicable ASTM standard methods. The chemical analysis of the "as-received" materials is summarized in Table 1.

Table 1. Virgin sorbent composition.

Wet Basis Baseline sorbent Verdal Sorbent

wt % Ca as CaCO3 97.10 97.05

wt % Mg as MgCO3 0.87 0.80

wt % Inerts (difference) 1.96 2.06

wt % Total 100 100

2.3. Characterization of sorbent activity

A tendency of natural sorbents to reactivate under conditions of diffusion controlled carbonation [4,5] complicates determination of the active fraction because measured values require correction. To avoid this correction we chose to adjust the testing program to by shortening the carbonation times to avoid an extended diffusion controlled regime. Since modification of the TGA program does not allow visual localization of the active fraction by extending lines representing the kinetic and the diffusion controlled rates, a percentage of the maximum rate was adopted as transition criteria, allowing reproducible specification without the need to enter the diffusion controlled regime.

Fig. 1. Transition from fast to slow reaction regime.

In this work the transition point criteria was defined to be where the rate drops to 10% of the maximum recorded rate for the carbonation interval. Fig. 1 indicates calculated transition points for X45, the 45th carbonation cycle, applying 10%, 20% and 30% maximum rate criteria to a TGA run on a fine fraction (150 ^m - 250 ^m) of the Baseline sorbent. From Fig. 1 it is apparent that the definition of the transition point has a large influence on the calculated active fraction.

2.4. TGA sorbent deactivation

Fitting parameters from Eq. (1) were estimated applying the 10% maximum rate criteria to TGA runs on fine fractions of sorbent (150 ^m - 250 ^m). The parameter values k = 0.45 and Xr = 0.05, were found representative of both Baseline and Verdal sorbents. TGA deactivation runs are shown in Fig. 2 along with predictions of Eq. (1). Fig. 2a and 2b illustrate the deactivation behavior with and without sulfur for TGA tests on Baseline and Verdal sorbent. Sulfur was tested with 300 mg/Nm5 SO2 included in the flue gas mixture (15 vol.% CO2 and 6 vol. % O2, balance N2) to account for sulfur entering either in the coal and flue gas during each cycle.

Irregularities shown in the measured XN for CO2 are a result of an apparent random error in synthetic flue gas switching. In some cases the transition from the carbonation to the calcination step was not accompanied by a switch from synthetic flue gas to nitrogen, increasing CaCO3 conversions resulting from a prolonged exposure of sorbent to flue gas during the temperature ramp. It is interesting to note the marginal increase in active fraction during the gas switching failures, caused by the diffusion controlled carbonation which decays completely in just two subsequent cycles and does not appear to significantly increase residual activity, Xr.

Cycle number

Cycle number

Fig. 2. (a) Baseline sorbent CO2 deactivation; (b) Verdal sorbent CO2 deactivation.

In Fig. 2 the negative effect of SO2 adsorption on CO2 activity is clearly shown for both Baseline and Verdal sorbents. Measurements conducted with HCl and SO2 in synthetic flue gas showed a similar deactivation effect for both components. Fig. 3 indicates CaO utilization more clearly for Baseline sorbent; indicating accumulated sulfur X SO3 and the resulting X CO2. Although the sulfation pattern varies widely between different sorbent materials, sulfur adsorption always leads to a decrease in CO2 activity and the deactivation is always disproportionately high for low cycle numbers and disproportionately lower for larger cycle numbers. Disproportionately high meaning the molar SO2 activity, X SO2, causes an even larger decrease in the molarX CO2 activity. CaO utilization predictions of an in-house model are also shown in Fig. 3. The Alstom model was found to match well the behavior of both Baseline and Verdal sorbents and was also successfully used to describe sulfation of Imeco sorbent, for up to ~0.6 mol.% sulfation per cycle as studied in [8]. The Alstom utilization model allows reliable predictions of sorbent CO2 activity as a function of sulfur loading and cycle number. The model is implemented into an AspenPlus process simulation tool which is used to evaluate performance and integration efficiency trade-offs.

Cycle number

Fig. 3. Baseline CaO Utilization model.

3. Pilot Testing

Both Baseline and Verdal sorbent materials were tested in the pilot facility at the Institute of Combustion and Power Plant Technology at the University of Stuttgart (IFK) with dedicated one week campaigns. For each week the pilot plant was operated continuously in 24-hour mode with the start-up beginning Sunday and continuous operation following from Monday till Friday. Fig. 4 shows a simplified flow scheme of the pilot plant configuration and system boundaries for checking material balance closure. Several steady state operating points were achieved during each testing week and samples were regularly extracted to allow tracking of composition, particle size, and sorbent

activity. All flows entering and exiting the balance blocks were recorded and the accuracy of gas and solid flow measurements verified. The accuracy of gas analyzers on inlet and outlet flue gas streams and the CO2 product stream were checked every 24 hours with calibration gas. All solids extracted from the plant: fines from cyclones and bag filters, purge sorbent, bed ash, samples and associated purge material were recorded. Blue basins shown in Fig. 4 represent the collection locations of solids periodically extracted from the pilot facility.

— • — • — - Carbonator Balance

— • — • — - Calciner Balance

— • — • — - Overall Balance

Bottom Drain

Bottom Drain

Fig. 4. Pilot plant flow diagram.

3.1. Operational stability

Achieving steady operating conditions, especially with the circulating sorbent bulk, is fundamental to mass balance closure and subsequent process modelling activities. Considering the IFK pilot facility, with a sorbent inventory of approximately 150 kg (122 of which is kg CaO, in the form of CaO, CaCO3 or CaSO4) and a feed rate of limestone 20 kg/h (11 kg/h CaO) the mean residence time of sorbent in the pilot plant (time required to turn-over the system inventory) is 11 hours. With well mixed reactors, stable operation for three residence times is required to purge the system of the initial inventory to a residual value of approximately 5%. Stable operation for only one residence time leads to purging only ~60 % of the original inventory, which may not be representative of the operating conditions. With the specific purpose to achieve a stable sorbent composition at representative operating conditions, the pilot plant was operated for approximately three residence times without changing operating conditions.

During the 30 hour pilot testing period all on-line measured parameters where checked for stability using the coefficient of variation (CV), which is the ratio of the standard deviation to the mean for a particular process parameter. The CV represents the dispersion of the parameter for a given set of data around the mean; larger values are associated with process instability while smaller values are indicative of more precise measurements and stable operating periods. Fig. 5 provides an indication of the operational stability of the pilot plant for a 33 hour period. Average CV refers to the summation of all individual parameter CVs divided by the number of parameters considered. The 2% stability criterion is arbitrarily defined based on measured results and represents the most stable operational periods recorded from the IFK pilot facility.

01 15 m

JU a E ra 10

a E ra 10

o8 m <u a E ra 10

I IVI IT ■"■"■ ■"■ i

T-tfNm^LDtDr^OOCnO^-ltNm^LntDr^

Hours Stabil Operation

Fig. 5. Operational stability plot.

In addition to stability analysis of on-line measurements, samples were extracted over the 33 hour period to track changes in sorbent composition and, through post processing, confirm stable operation. It is worth noting that during the first half of the 33 hour window a systematic change inX SO3 was observed, while in the last third of the testing window a systematic change in sorbent composition or activity was not detectable. Table 1 is a summary of the sorbent compositions shown at the time of extraction from Fig. 5. Values indicated in units of % mol. refer to Ca containing materials only. Samples 2b and 4b denote split samples which were analyzed at independent Alstom laboratories in Vaxjo, Sweden and Chattanooga, Tennessee, USA.

Table 2. Circulating sorbent composition.

CaSO4 %wt. CaCO3 %wt. CaO %wt. Resid. C Ash* %wt. %wt. CaSO4 %mol. CaCO3 %mol. CaO %mol.

Sample 1 3.8 14.2 71.3 0.0 10.7 1.9 9.8 88.3

Sample 2 4.3 14.2 71.8 0.1 9.7 2.1 9.7 88.1

Sample 2b 4.3 14.4 67.2 0.1 14.1 2.3 10.5 87.2

Sample 3 4.3 11.7 76.6 0.4 7.1 2.1 7.7 90.2

Sample 4 3.8 11.7 74.5 0.2 9.8 1.9 7.9 90.2

Sample 4b 4.4 13.2 70.9 0.1 11.4 2.3 9.2 88.5

Average 4.1 13.2 72.0 0.1 10.5 2.1 9.1 88.8

SD 0.2 1.3 3.2 0.1 2.3 0.2 1.1 1.2

95% conf. 0.5 2.5 6.5 0.3 4.6 0.3 2.2 2.4

* ash is used to represent all residual, inactive materials and includes MgO

From the calculated standard deviations (SD) the largest variability in the last third of the testing window appears to be associated with sorbent ash content. Upon closer investigation, a significant portion of this variability is linked to the results obtained from the different laboratories. The source of analysis variation is currently under investigation in an effort to reduce this component of uncertainty.

3.2. Material balance closure

Instrument verification and mass balance reconciliation techniques were applied to identify and correct for gross errors. Material balance closure was obtained with ±10 % of the measured values after the correction of gross errors. One of the largest uncertainties associated with material balance closure was related to the definition of the coal composition (sample and analysis).

To properly close the material balances the extraction of all solids from the pilot facility were monitored closely. Fig. 6 illustrates the semi-continuous extraction of solids. Stream numbers in Fig 6. correspond to those shown on the process flow scheme in Fig. 4, for example, stream 220 and 320 represent the cumulative solids lost from the Carbonator and Calciner primary cyclones. Fines losses from the Calciner system were found to be larger than fines losses from the Carbonator system. The difference in fines losses may be due to fine fractions in the feed or produced during the initial calcination. The fines losses are observed to remain very stable over the steady state sections of testing since the fluidization velocity in the Carbonator is held constant.

3/17/2014 12:00

3/18/2014 12:00

3/19/2014 12:00

3/20/2014 12:00

3/21/2014 12:00

Fig. 6. Solids extraction logging from pilot plant operation.

3.3. Sorbent attrition

Particle size distributions (PSDs) for samples taken over the course of the 33 hour run were measured with a coulter counter particle size analyzer. The transition of D10, D50, and D90 of the circulating sorbent extracted at both the Carbonator and Calciner loop seals is indicated in Fig. 7. The samples taken during the first hours of testing indicate a smaller PSD than those taken during the remainder of the test. One explanation of the transition of the PSD in the initial stages of operation is linked to the transition from air-fired to oxy-fired conditions in the Calciner. It is postulated that during air fired calcination large partial pressure differences enhance the generation of fines through an intensified breakage mechanism acting during the virgin calcination of the make-up sorbent, which represents a large portion of the circulating sorbent during the beginning stages of testing. After an extended period of oxy-fired operation the distributions in the Carbonator and Calciner systems stabilize, confirming the steady operating conditions. Differences between the steady state PSD in the Carbonator and Calciner system is attributed to a tendency of the Carbonator to accumulate larger particles due to lower fluidization velocities.

«U 500

(U 300

il 100

♦ Carbonator System D10

■ Carbonator System D50 4 Carbonator System D90

♦ Calciner System D10

■ Calciner System D50 4 Calciner System D90 O Calciner Purge D10 □ Calciner Purge D50 A Calciner Purge D90

15 20 25

Run time [h]

Fig. 7. Particle size development.

For each stream of solids exiting the plant, measured PSDs were multiplied by the stream mass flow rates in order to fabricate an attrition characteristic of the Baseline sorbent. The synthesized attrition characteristic for Baseline sorbent is shown in Fig. 8. Stream compositions were considered in the weighting of PSDs irrespective of the size fraction (the composition of different size fractions in one sample were not measured). As such, the finest particles in the synthesized attrition characteristic are likely associated with fly ash and the coarse material associated with bed ash.

Particle Diameter [^m]

Fig. 8. Particle size development.

3.4. Attrition model comparison

Sorbent attrition has been studied by several groups using a calcination/ carbonation loop at the pilot scale [9,10,11] while results vary widely with the source of sorbent materials, a simple model [10] using a linear expression between the attrition rate and the square of the excess gas velocity, (U - Umf) describes results sufficiently well:

-Ka(u-Umf}t + 1 (2)

d =--(3)

Where d0 is the initial average particle size, dt is the size after the cumulative attrition time and Ka is a sorbent specific attrition rate constant. Since, superficial gas velocity was not varied during our testing we compare only the measured attrition with that of other sorbent materials based on Eq. (2) and (3). Fig. 9 shows the size reduction measured in the IFK pilot and compares the result to calculated values based on the measured excess velocity (U-Umf=4.9 m/s) and published values for Ka. The measured attrition rate fits well into the range measured in [11].

Fig. 9. Size reduction after average residence time, 11h.

3.5. Pilot plant confirmation of sorbent deactivation

Extracted samples should represent the population of particles circulating between the reactors for a given steady-state. To avoid sampling stagnant inventory, samples were extracted from loop seals, having characteristically small system volumes and high throughput rates. Extracted samples were analyzed via TGA to confirm conversion and characterize sorbent activity. According to [12] considering the sorbent make-up and circulation rates, FO and FR, and assuming a well-mixed system Eq. (4) can be used to calculate the fractions of sorbent for each successive cycle. Combining Eq. (1) and Eq. (4) allows calculation of the average activity of the population of particles in

Eq. (5).

N-1 R_

(Fo + Fr У

rN = O 1 \ N (4)

XAVE - XrNXN (5)

The calculated XAVE, which is based on fitting parameters determined by TGA, was compared with the measured XAVE of sorbent extracted from the pilot plant. The change in sample mass during the first calcination, after extraction from the carbonation loop seal, yielded a mass fraction of 12.5 wt.% CaCO3, which matched the chemical analysis results, as shown in Table 2. Each measurement point, for the two size fractions: 150ц - 250ц and 300ц -500ц, in Fig. 10 represents the average activity, XAVE, of a sample having a distribution of cycle numbers. Fig. 10 also indicates the results of two model predictions, both accounting for sulfur related deactivation but using different fitting parameters in Eq. (1). By adjusting deactivation parameters, the corrected model consisting of Eq. (1), (4) and (5), was able to describe the measured deactivation of XAVE over an additional 46 cycles. The gap between model predictions appears to be a result of different calcination conditions (programed with the TGA and achieved in the pilot plant). It is likely that lower partial pressures of CO2 and H2O during calcination in the TGA do not capture a deactivation component resulting in the pilot plant with oxy-fired calcination using recycled flue gas.

From Fig. 10, the larger particles are slightly less reactive. Irregularities in the measuredXAVE are again the result of flue gas switching failures.

0 5 10 15 20 25 30 35 40 45

Cycle number (after extraction)

Fig. 10. Extracted sorbent activity.

3.6. Conclusions

Defining the transition point from the fast to the slow reaction regime as a percentage of the maximum rate allows selection of short carbonation times and avoids false sorbent reactivation.

Sulfur adsorption decreases CO2 activity. Although, the sulfation pattern varies largely between sorbents, the deactivation is always disproportionately high for low cycle numbers and disproportionately lower for larger cycle numbers. CaO utilization predictions of an in-house model are capable of describing the sulfation of Baseline and Verdal sorbents.

Ample time to develop sorbent and ash composition is required to ensure the measured activity is representative of the given operating conditions. Three mean residence times are required to effectively purge sorbent from a well-mixed pilot facility, reducing the initial inventory to a residual of 5 %.

Material balance closure is vital in the interpretation of pilot data, instrument verification and mass balance reconciliation techniques can be applied to identify and correct for gross errors. Application of these methods allowed closure of material balances, limiting the largest corrections to within ±10 % of the measured values. One of the largest uncertainties was associated with obtaining a representative coal sample. Closure of material balances requires the tracking of plant inventory and extracted solids and can form the basis for attrition studies under representative process conditions.

Based on the synthesized attrition characteristic, the implied attrition constant Ka is comparable to those suggested in the literature. Further development of a size reduction model should decouple fines production from calcination and from mechanical attrition, from experimental work like [13]. Such a attrition model would aid in scaling performance to larger vessels having lower surface to volume ratios.

The calculated XAVE based on a representative TGA program is significantly larger than that of extracted sorbent from the pilot plant. By adjusting deactivation parameters, the corrected model was able to describe the measured deactivation over an additional 46 cycles. The gap between model predictions appears to be a result of different calcination conditions (programed with the TGA and achieved in the pilot plant). The additional deactivation is thought to be due to higher partial pressures of CO2 and H2O during oxy-fired calcination with recycled flue gas in the pilot plant.

References

[1] Bhatia S.K., Perlmutter D.D. (1983) Effect of the Product Layer on the Kinetics of the CO2-Lime Reaction, AIChE Journal Vol. 29, No.1, pp. 79-86

[2] Alvarez D., Abanades J.C. (2005) Determination of the critical product layer thickness in the reaction of CaO with CO2, Ind. Eng. Chem. Res. Vol. 44, pp. 5608-5615

[3] Grasa G.S., Abanades J.C. (2006) CO2 Capture Capacity of CaO in Long Series of Carbonation/Calcination Cycles, Ind. Eng. Chem. Res. Vol. 45, pp. 561-567

[4] Lysikov A. I., et al. (2007) Change of CO2 carrying capacity of CaO in isothermal recarbonation-decomposition cycles, Ind. Eng. Chem. Res. Vol. 46, pp. 4633-4638

[5] Arias B., Abanades J.C., Grasa G.S. (2011) An analysis of the effect of carbonation conditions on CaO deactivation curves, Chemical Engineering Journal, 167, pp. 255-261

[6] Coppola A., Scala F., et al. (2013) Performance of Natural Sorbents during Calcium Looping Cycles: A Comparison between Fluidized Bed and Thermo-Gravimetric Tests, Energy Fuels Vol. 27, pp. 6048-6054

[7] Martinez I., et al. (2012) Kinetics of Calcination of Partially Carbonated Particles in a Ca-Looping System for CO2 Capture, Energy Fuels Vol. 26, pp. 1432-1440

[8] Grasa G.S., et al. (2008) Sulfation of CaO Particles in a Carbonation/Calcination Loop to Capture CO2, Ind. Eng. Chem. Res. Vol. 47, pp. 1630-1635

[9] Jia, L. et al. (2007) Attrition of Calcining Limestones in Circulating Fluidized-Bed Systems, Ind. Eng. Chem. Res. 2007, Vol. 46, 5199-5209

[10] Chen Z., et al. (2008) Limestone particle attrition and size distribution in a small circulating fluidized bed, Fuel. Vol. 87, pp. 1360-1371

[11] Gonzalez B., et al. (2010) Sorbent attrition in a carbonation/calcination pilot plant for capturing CO2 from flue gases, Fuel. Vol. 89, pp. 2918-2924

[12] Abanades, J.C. (2002) The maximum capture efficiency of CO2 using a carbonation/ calcination cycle of CaO/CaCO3, Chemical Engineering Journal Vol 90, pp. 303-306

[13] Coppola, A. Montagnaro, F., Attrition of Limestone During Fluidized Bed Calcium Looping Cycles for CO2 capture, Combust. Sci. Technol., Vol. 184, pp 929-941