1 CO 2 Capture by Aqueous Absorption Summary of Second Quarterly Progress Reports 2013 by Gary T. Rochelle Supported by the Texas Carbon Management Program and the Industrial Associates Program for CO 2 Capture by Aqueous Absorption McKetta Department of Chemical Engineering The University of Texas at Austin July 31, 2013 Introduction This research program is focused on the technical obstacles to the deployment of CO 2 capture from flue gas by alkanolamine absorption/stripping. The objective is to develop and demonstrate evolutionary improvements to monoethanolamine (MEA) absorption/stripping for CO 2 capture from gas-fired and coal-fired flue gas. The Texas Carbon Management Program and the Industrial Associates Program for CO 2 Capture by Aqueous Absorption support 15 graduate students. Most of these students have prepared detailed quarterly progress reports for the period April 1, 2013 to June 30, 2013. Conclusions Thermodynamics and Rates 5 m 2MPZ has viscosity of 3.7–6 cP at the CO 2 loading range of 0–0.5 mol/mol alkalinity. The k g ’ avg at typical coal conditions of 3.4 m MDEA/9.8 m MEA is 3.9 Х10 -7 mol/Pa∙s∙m 2 , which is slightly lower than 7 m MEA. The blend has lean/rich loadings that are lower than 7 m MEA. Despite its high alkalinity concentration, the blend has only a moderate capacity at 0.58 mol/kg solvent. At P CO2 * = 1.5 kPa, the -H abs is about 73 kJ/mol, similar to MEA. At 40 °C the heat of absorption of 6 m AEP is 60 to 90 kJ/mol CO 2 at operation loading range (0.27–0.33) and the heat of absorption of 5 m PZ/2 m AEP is 60 to 80 kJ/mol CO 2 at operation loading range (0.3– 0.38). Pulsed field gradient (PFG) spin echo (SE) - H 1 NMR was used to measure the self-diffusion coefficient of unloaded aqueous AEP solutions. The diffusivity of AEP solutions is inversely proportional to its viscosity with a power of 0.6. Modeling With NETL financials, an of 1 is a reasonable factor for expressing purchased equipment cost in annualized $/yr. Over 80% of the purchased equipment cost is represented by the (1) absorber, (2) cross exchanger, (3) reboiler, and (4) compressor. 1 1
171
Embed
CO2 Capture by Aqueous Absorption - University of Texas at ...rochelle.che.utexas.edu/files/2015/06/Rochelle... · Capture by Aqueous Absorption ... (PFG) spin echo (SE) - H1 NMR
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
CO2 Capture by Aqueous Absorption
Summary of Second Quarterly Progress Reports 2013
by Gary T. Rochelle
Supported by the Texas Carbon Management Program
and the
Industrial Associates Program for CO2 Capture by Aqueous Absorption
McKetta Department of Chemical Engineering
The University of Texas at Austin
July 31, 2013
Introduction
This research program is focused on the technical obstacles to the deployment of CO2 capture
from flue gas by alkanolamine absorption/stripping. The objective is to develop and demonstrate
evolutionary improvements to monoethanolamine (MEA) absorption/stripping for CO2 capture
from gas-fired and coal-fired flue gas. The Texas Carbon Management Program and the
Industrial Associates Program for CO2 Capture by Aqueous Absorption support 15 graduate
students. Most of these students have prepared detailed quarterly progress reports for the period
April 1, 2013 to June 30, 2013.
Conclusions
Thermodynamics and Rates
5 m 2MPZ has viscosity of 3.7–6 cP at the CO2 loading range of 0–0.5 mol/mol alkalinity.
The kg’avg at typical coal conditions of 3.4 m MDEA/9.8 m MEA is 3.9 Х10-7
mol/Pa∙s∙m2, which
is slightly lower than 7 m MEA. The blend has lean/rich loadings that are lower than 7 m MEA.
Despite its high alkalinity concentration, the blend has only a moderate capacity at 0.58 mol/kg
solvent. At PCO2* = 1.5 kPa, the -Habs is about 73 kJ/mol, similar to MEA. At 40 °C the heat of
absorption of 6 m AEP is 60 to 90 kJ/mol CO2 at operation loading range (0.27–0.33) and the
heat of absorption of 5 m PZ/2 m AEP is 60 to 80 kJ/mol CO2 at operation loading range (0.3–
0.38).
Pulsed field gradient (PFG) spin echo (SE) - H1 NMR was used to measure the self-diffusion
coefficient of unloaded aqueous AEP solutions. The diffusivity of AEP solutions is inversely
proportional to its viscosity with a power of 0.6.
Modeling
With NETL financials, an of 1 is a reasonable factor for expressing purchased equipment cost
in annualized $/yr.
Over 80% of the purchased equipment cost is represented by the (1) absorber, (2) cross
exchanger, (3) reboiler, and (4) compressor.
1 1
2
OPEX is a relatively constant function of lean loading in intercooled configurations.
The dependence of absorber CAPEX on lean loading is less significant than would be expected.
This is due to the constant inlet vapor flow rate and constant height of SO2 polisher, direct
contact cooler, and water wash.
The CAPEX of the cross exchanger exhibited the greatest dependence on lean loading,
suggesting that the optimum lean loading will largely depend on heat exchanger pricing and
optimization.
The Hanley and Chen model predicts minimal combined contributions to packing reduction from
the effect of liquid rate and packing selection in the recycle intercooling section on the wetted
area available for mass transfer.
The driving force effects of the liquid recycle intercooling design (which includes an
intercooling benefit and penalty for back-mixing) initially show increasing benefits with recycle
rate. The packing reduction from driving force effects reaches a maximum of 12% at a recycle
of 2 L/G.
The model predicts that the benefits of the recycle intercooling system are dominated by the
contribution of reduced liquid-side mass transfer resistance due to the increased liquid rate in the
recycle. The reduction in packing from the mass transfer coefficient contribution is as high as
40% (of the overall 48% reduction) for the highest recycle rate (8 L/G).
With the flash stripper using a warm rich bypass and rich exchanger bypass, 9 m MEA uses 1.5
to 3 kJ/mol less work with stripping at 135 oC rather than 120
oC. The convective steam heater
should make higher temperature feasible with acceptable thermal degradation.
With the flash stripper using a warm rich bypass and rich exchanger bypass, 5 m PZ gives the
same performance as 8 m PZ at a lean loading of 0.26 (assuming a an exchanger LMTD of 5 oC.
Including vapor hold-up may be important to accurately simulate transient behavior, and the
separator vessel model has been updated to include vapor hold-up in the overall material
inventory.
The main process control objectives for amine scrubbing are disturbance rejection, set point
tracking, satisfying constraints, and stable operation with process intensification.
Because of the significant material and energy recycle, amine scrubbing is expected to exhibit
multiple time scale behavior, suggesting the need for a hierarchical controller design.
kL/kLa correlations in the literature all show that the liquid side mass transfer coefficient is
proportional to the square root of diffusivity, but with little or no experimental basis.
Few kL/kLa correlations in the literature have discussed the indirect influence of viscosity on the
liquid side mass transfer coefficient via the effect of viscosity of the diffusion coefficient.
A few investigations have systematically varied viscosity and found that kL/kLa, depends on
viscosity to the -0.5 power.
Solvent Management
At 150 oC and an initial concentration of 7 m tertiary amine/2 m PZ and CO2 loading of 0.1,
initial rates of thermal degradation and activation energy for the tertiary amines are: TEA (1.2
Figure 10: Aspen Plus® model predictions of heat of absorption for 5 m PZ/2 m AEP using
Gibbs-Helmholtz (points) and calorimetric (lines) calculations
Conclusions
1. The thermodynamic model developed for PZ-AEP-H2O-CO2 in Aspen Plus® last quarter
was corrected for speciation prediction by qualitative H1 and C
13 NMR measurement.
2. At 40 °C the heat of absorption of 6 m AEP is about 60–90 kJ/mol CO2 at operation
loading range (0.27–0.33) and the heat of absorption of 5 m PZ/2 m AEP is about 60-80 kJ/mol CO2 at operation loading range (0.3–0.38).
3. Pulsed field gradient (PFG) spin echo (SE) - H1 NMR was used to measure the self-
diffusion coefficient of unloaded aqueous AEP solutions.
4. The diffusivity of AEP solutions was found to be inversely proportional to its viscosity
with a power of 0.6.
Future Work
1. Diffusivity of other common amines will be measured by PFG - SE - H1 NMR.
2. A master’s thesis will be finished next quarter.
3. Some piperidine derivatives will be screened for CO2 capture.
4. A kinetic model for PZ-AEP-H2O-CO2 system will be established based on the
thermodynamic model.
37 37
13
References
Chen X. Carbon dioxide thermodynamics, kinetics, and mass transfer in aqueous piperazine
derivatives and other amines. The University of Texas at Austin. Ph.D. Dissertation. 2011.
Hilliard MD. A Predictive Thermodynamic Model for an Aqueous Blend of Potassium
Carbonate, Piperazine, and Monoethanolamine for Carbon Dioxide Capture from Flue Gas.
The University of Texas at Austin. Ph.D. Dissertation. 2008.
Rochelle GT, et al. "CO2 Capture by Aqueous Absorption, Third Quarterly Progress Report
2012." Luminant Carbon Management Program. The University of Texas at Austin. 2012.
38 38
1
Process Economics for 8 m PZ
Quarterly Report for April 1 – June 30, 2013
by Peter Frailie
McKetta Department of Chemical Engineering
Texas Carbon Management Program
The University of Texas at Austin
July 31, 2013
Abstract
The goal of this study is to evaluate the performance of an absorber/stripper operation that
utilizes MDEA/PZ. Before analyzing unit operations and process configurations,
thermodynamic, hydraulic, and kinetic properties for the blended amine must be satisfactorily
regressed in Aspen Plus®
. The approach used in this study is first to construct separate MDEA
and PZ models that can later be reconciled via cross parameters to model MDEA/PZ accurately.
During the past quarter a base-case absorption/stripping process was designed and evaluated
using concentrated PZ. A techno-economic analysis was also performed to determine the effects
of process modifications on the ultimate cost of CO2 capture. Emphasis was placed on the
relative contributions of major process units to the final cost. All results were generated using
the Independence model. The goal for the next quarter is to finish the techno-economic analysis,
which will complete the work for this project.
Introduction
The removal of CO2 from process gases using alkanolamine absorption/stripping has been
extensively studied for several solvents and solvent blends. An advantage of using blends is that
the addition of certain solvents can enhance the overall performance of the CO2 removal system.
A disadvantage of using blends is that they are very complex compared to a single solvent, thus
making them much more difficult to model.
This study will focus on a blended amine solvent containing piperazine (PZ) and
methyldiethanolamine (MDEA). Previous studies have shown that this particular blend has the
potential to combine the high capacity of MDEA with the attractive kinetics of PZ (Bishnoi,
2000). These studies have supplied a rudimentary Aspen Plus®-based model for an absorber
with MDEA/PZ. This previous work recommended that more kinetic and thermodynamic data
should be acquired for the MDEA/PZ blend before the model can be significantly improved.
Three researchers in the Rochelle lab have been acquiring these data, which are being
incorporated into the model. One of the major goals of this study will be to improve the supplied
Aspen Plus® absorber model with up-to-date thermodynamic and kinetic data. Another major
goal will be to make improvements to the MDEA and PZ thermodynamic models, which should
simplify the construction of the blended amine model.
39 39
2
Methods
During the past quarter the techno-economic model was completed and used to evaluate the cost
effectiveness of the base-case absorption/stripping process described in the previous report.
Calculating cost of CO2 capture
The purchased equipment cost (PEC) for each process unit was calculated using the pricing
methods described in the 2013 Q1 Report and Aspen Plus® results. Over 80% of the PEC can be
attributed to four process units: (1) absorber, (2) reboiler, (3) main cross exchanger, and (4)
compressor. Other process units such as the stripper, pumps, trim cooler, and blower
cumulatively account for a significant cost, but they do not represent a significant opportunity for
design improvement and cost reduction.
In order to compare the effects of process conditions on CAPEX and OPEX, both expenses must
be expressed in dollars per metric ton of CO2 captured. The PEC was converted to these units
using Equation 1.
(1)
In Equation 1 converts the PEC to a total capital requirement (TCR) and annualizes the cost.
Literature values for range from as low as 3 to as high as 10, depending on the process unit in
question. This study will assume a constant value of 5 for for the entire process. The
annualizing factor, , takes into account return on investment (10%), taxes (35% of return on
investment), depreciation (3–10%, depending on plant lifetime), and maintenance (2–3%). If it
is assumed that the plant lifetime is 20 years and the value of the plant at the end of year 20 is $0,
is about 0.2. When multiplied together and yield a factor of 1, meaning that the CAPEX
contribution to the cost of CO2 capture can be calculated by dividing the PEC by the total metric
tons of CO2 captured per year by the process. The number of metric tons of CO2 captured per
year was calculated using the flue gas of a 550 MWe power plant and assuming a 90% capture
rate and 85% annual capacity.
The OPEX was expressed in dollars per metric ton of CO2 captured by calculating the
opportunity cost associated with operating each process unit. Every kWh used to overcome
pressure drop in columns and exchangers, operate the multi-stage compressor, and heat the
reboiler is a kWh of electricity that could not be sold to the market. The cost of electricity
(COE) was initially assumed to be $0.10 per kWh, and the sensitivity of the final solution to this
assumption was tested. The pump work was directly converted to kWh using Aspen Plus®
predictions. Reboiler duty was first converted from a heat duty to an equivalent work in the units
of kJ per mole of CO2 captured using Equation 2.
reboilersn
i i
kiieq
KT
TKTQCOmolkJW
1
sin2
5
575.0/ (2)
In Equation 2, Qi is the reboiler duty, Ti is the reboiler temperature, Tsink is 313K, 0.75 is the
steam turbine efficiency, and 5K is the temperature approach on the reboiler. This is the same
equation commonly used in this study to calculate the equivalent work of the reboiler.
yearpercapturedMTTotal
PECCOMT
2/$
40 40
3
Compressor work was calculated as a function of inlet CO2 pressure according to Equations 3
and 4.
atmP
atmPCOmolkJW in
in
comps 5.4096.4148
log572.4/ 2
(3)
atmP
atmPCOmolkJW in
in
comps 5.4181.2148
log023.4/ 2
(4)
For all cases the CO2 was compressed to 150 bar.
Results and Discussion
Table 1 reports the total cost of CO2 capture for several lean loadings, liquid rates, and absorber
configurations for 8 m PZ.
Table 1: Total cost of CO2 capture in USD per metric ton captured for several lean
loadings, liquid rates, and absorber configurations for 8 m PZ.
Lean Loading
(mol CO2/ mol alk)
1.1 x L/Gmin 1.2 x L/Gmin 1.3 x L/Gmin
IO IC PA IC IO IC PA IC IO IC PA IC
0.2 N/A 43.56 44.41 44.62 45.59 N/A
0.23 44.22 43.70 44.90 44.78 46.22 46.00
0.26 45.67 44.71 46.85 45.94 47.95 46.90
0.29 47.24 46.85 49.13 48.35 50.44 49.40
0.32 50.64 50.61 52.00 52.14 55.00 53.68
0.35 60.04 N/A 60.82 60.73 64.92 N/A
It should be noted that the prices in Table 1 do not include the cost of transportation, storage, and
monitoring (TS&M). A typical cost of TS&M is around $15 per metric ton of CO2. The
intercooling configurations are abbreviated IO IC (in-and-out intercooling) and PA IC (pump-
around intercooling), and they are described in detail in the 2013 Q1 Report (Rochelle et al.,
2013). For all liquid rates and intercooling configurations the cost of CO2 capture increases as
lean loading increases. For most liquid rates and lean loadings the pump-around intercooling is
slightly less expensive than in-and-out intercooling, though the difference in cost is never greater
than $1.32 per metric ton of CO2. As the liquid rate increases the cost of capture increases for all
lean loadings and configurations.
Figure 1 shows the minimum liquid flowrate in kg per second, CAPEX, and OPEX as a function
of lean loading for the configuration with a liquid flowrate that is 1.2 times the minimum and
pump-around intercooling. As the lean loading increases from 0.2 to 0.35 the minimum liquid
flowrate increases 236%. This is primarily due to the relatively small change in the rich loading
across the lean loading range. The rich loading is set by the temperature of the liquid at the
bottom of the absorber. For all cases this temperature is relatively close to the inlet gas
temperature (313K) because (1) at a low L/G the temperature bulge will be in the top of the
column, and (2) intercooling effectively removes heat from the entire column (Plaza, 2011). If
the rich loading is relatively constant and the lean loading is changing significantly, the
minimum liquid flowrate will also change significantly.
41 41
4
Figure 1: Minimum liquid flowrate (red line), CAPEX (dotted black line), and OPEX (solid
black line) for 8 m PZ assuming a liquid flowrate that is 1.2 times the minimum and pump-
around intercooling.
Figure 2: Equivalent work as a function of lean loading for 8 m PZ with a liquid flowrate
equal to 1.2 times the minimum and pump-around intercooling.
0
1000
2000
3000
4000
5000
$0
$5
$10
$15
$20
$25
$30
$35
$40
$45
0.2 0.23 0.26 0.29 0.32 0.35
Min
imu
m L
iqu
id F
low
rate
(k
g/s
)
$/M
T C
O2
Lean Loading (mol CO2/mol alkalinity)
CAPEX
26
27
28
29
30
31
32
33
0.2 0.23 0.26 0.29 0.32 0.35
Eq
uiv
ale
nt
Wo
rk (
kJ
/mo
l C
O2)
Lean Loading (mol CO2/mol alkalinity)
OPEX
Minimum
Liquid
Flowrate
42 42
5
Figure 1 also shows that the OPEX is a relatively weak function of the lean loading. Figure 2
shows the equivalent work in kJ per mole of CO2 as a function of lean loading for this
configuration. Even though there is a very clear minimum equivalent work around a lean
loading of 0.3 moles of CO2 per mole of alkalinity, the difference between the minimum
equivalent work and the maximum equivalent work in this range is only 6.5%. When compared
to the 58% increase in CAPEX over this lean loading range the variance in OPEX appears
insignificant.
Figure 3 shows the contribution to the CAPEX of each of the major process units for 8 m PZ
with a liquid flowrate equal to 1.2 times the minimum and pump-around intercooling.
Figure 3: Contribution to CAPEX of the absorber (blue), cross exchangers (red),
compressor (green), reboiler (orange), and all other process units (black) as a function of
lean loading for 8 m PZ with a liquid flowrate equal to 1.2 times the minimum and pump-
around intercooling.
The biggest contributor to CAPEX is the absorber, which increases 30% over this lean loading
range. While this is a significant increase, it is not as significant as one might expect given that
the total packing area increases 202% over the same range. This relatively small increase in
CAPEX is due to (1) the constant vapor flow rate, and (2) the constant height assigned to the SO2
polisher, direct contact cooler, and water wash sections. The diameter of the column is primarily
determined by the vapor flowrate, which is constant. Even though the liquid flowrate increases
236% over the loading range, the cross sectional area only increases by 17%. The cross sectional
area determines the cost of distributors, chimney trays, and the shell. If the diameter is constant,
these costs are constant. Also, the heights of the SO2 polisher, direct contact cooler, and water
wash sections are constant, and they account for about half of the total packing area. Even
$0
$2
$4
$6
$8
$10
$12
$14
$16
0.2 0.23 0.26 0.29 0.32 0.35
CA
PE
X (
$/M
T C
O2)
Lean Loading (mol CO2/mol alkalinity)
Absorber
Cross Exchangers
Reboiler
Compressor
Other
43 43
6
though it changes significantly, the total packing area for CO2 capture contributes relatively little
to the overall price of the absorber.
The reboiler price is relatively constant over the entire range of lean loadings. This may be
attributed to the relatively small change in reboiler duty, which is the parameter used to price the
process unit. The reboiler duty may be expressed as the sum of three heat requirements: (1)
steam losses out of the top of the stripper, (2) sensible heat to account for the hot side
temperature approach on the main cross exchanger, and (3) the latent heat of reaction to strip the
CO2 (Van Wagener, 2011). The bypass streams are meant to reduce steam losses, which are
most significant at low lean loadings. The latent heat of reaction decreases as lean loading
increases. Even though the same amount of CO2 is being removed across all cases, the heat of
desorption increases as loading decreases. The average heat of desorption will be greater for
cases with lower lean loadings, assuming a relatively constant rich loading. The hot-side
temperature approach is relatively constant, but the total liquid flowrate increases as lean loading
increases. This will increase the sensible heat requirement. These opposing effects result in a
relatively constant reboiler price.
The compressor price decreases as lean loading increases because the inlet pressure to the
compressor is increasing. What is most surprising about the compressor CAPEX is how little it
decreases (18%). Because they must accommodate a larger vapor volume, the most expensive
stages on a compressor train are the low pressure stages. Increasing the lean loading will
increase the inlet pressure to the compressor, which would eliminate the need for a few of the
lower pressure stages. The elimination of relatively expensive stages should have a greater
effect on the CAPEX of the compressor.
The major process unit showing the greatest dependence on lean loading is the cross exchanger,
which increases 375% in CAPEX over the given loading range. This is primarily due to the
increase in liquid load. A greater volume of liquid will require a greater number of channels to
avoid an excessively large pressure drop.
There are opportunities to improve the CAPEX estimations of all major process units. All of the
prices are based on a limited set of manufacturer quotes. Updating these quotes would improve
the accuracy of the predictions. The cost of the compressor should be a stronger function of inlet
CO2 pressure to reflect the economic advantage of eliminating the low pressure, high volume
stages. Cross exchanger pricing and optimization will be a major priority in the future. It
represents the most significant opportunity for optimizing the CAPEX, and the location of the
optimum solution will be very sensitive to the accuracy of the cross exchanger economics.
Conclusions
An of 1 is a reasonable factor for expressing PEC in dollars per metric ton of CO2
captured.
Over 80% of the PEC is represented by the (1) absorber, (2) cross exchanger, (3) reboiler, and
(4) compressor.
OPEX is a relatively constant function of lean loading in intercooled configurations.
The dependence of absorber CAPEX on lean loading is less significant than would be
predicted. This is due to the constant inlet vapor flow rate and constant height of SO2
polisher, direct contact cooler, and water wash.
44 44
7
The CAPEX of the cross exchanger exhibited the greatest dependence on lean loading,
suggesting that the optimum lean loading will largely depend on heat exchanger pricing and
optimization.
More equipment price quotes must be obtained for all process units to increase confidence in
CAPEX estimates.
Future Work
The remainder of this study will focus on process design and optimization. The set of amines
tested in the techno-economic analysis will be expanded to include the MDEA/PZ blends. This
will complete the work for this project.
References
Bishnoi S. Carbon Dioxide Absorption and Solution Equilibrium in Piperazine Activated
Methyldiethanolamine. The University of Texas at Austin. Ph.D. Dissertation. 2000.
Plaza JM. Modeling of Carbon Dioxide Absorption using Aqueous Monoethanolamine,
Piperazine and Promoted Potassium Carbonate. The University of Texas at Austin. Ph.D.
Dissertation. 2011.
Rochelle GT et al. “CO2 Capture by Aqueous Absorption, First Quarterly Progress Report 2013.”
Texas Carbon Management Program. The University of Texas at Austin. 2013.
Van Wagener DH. Stripper Modeling for CO2 Removal Using Monoethanolamine and
Piperazine Solvents. The University of Texas at Austin. Ph.D. Dissertation. 2011.
45 45
1
Novel Absorber Intercooling Configurations
Quarterly Report for April 1 – June 30, 2013
by Darshan Sachde
Supported by the Texas Carbon Management Program
McKetta Department of Chemical Engineering
The University of Texas at Austin
July 31, 2013
Abstract
Intercooling comparison study results for capture from natural gas combined cycle and coal-fired
power plant applications were presented in the first quarterly report of 2013. These results
focused on the reduction in total packing requirement and potential energy benefits (as measured
by rich loading leaving the absorber) attributed to the new recycle intercooling configuration..
The predicted benefits of the recycle intercooling design were evaluated in further detail.
Specifically, the in-and-out intercooling results and recycle intercooling results for the natural
gas application were compared at a constant amine feed rate or, equivalently, a constant rich
loading (lean loading and CO2 removal were fixed as part of the evaluation). The constant rich
loading case allowed for a direct comparison of the packing requirement for each design. The
predicted reduction in packing (up to 46% reduction in the natural gas application) with recycle
intercooling was then separated into components based on the effect of liquid rate and type of
packing used in the recycle section. These variables (in conjunction with objective function used
to minimize total packing area by distributing the packing in the three column sections) impact
the mass transfer parameters in the absorber model (wetted area and mass transfer resistance) as
well as driving forces in the column (intercooling effect to remove equilibrium limitations and
back-mixing effect due to solvent recycle). The analysis found that the current mass transfer
models used in the evaluation leads to reduced mass transfer resistance as a function of liquid
rate as the dominant contributor to the benefits of the recycle design. This mass transfer
resistance effect leads to an increasing proportion of the total packing to be allocated in the
middle, recycle portion of the column as the recycle rate increased. The optimization result has
the undesired effect of increasing the portion of the column that is well-mixed, and, therefore,
reduces the average driving forces in the column. This serves to offset some of the benefit that
intercooling provides by reducing equilibrium constraints for the driving force.
Introduction
During the past quarter, the recycle intercooling evaluation conducted for CO2 capture from coal
and natural gas power plant flue gas with 8 m PZ was evaluated in further detail to determine the
source of model predicted benefits of the recycle intercooling configuration. Tables 1 and 2
provide an overview of the flue gas conditions and intercooling comparison design parameters
used in the previous analyses.
46 46
2
Table 1: Flue Gas Conditions
Gas Conditions
NGCC Coal
Gas Feed Rate (kg-mol/hr)
114,000 74,000
Gas Feed Rate (kg/hr)
3,230,000 2,140,000
Temperature (°C)
106 57
Pressure (MPa)
0.1 0.1
Composition (Mole %)
NGCC Coal
CO2 4.0% 13.5%
H2O 8.7% 15.2%
N2 74.3% 68.1%
O2 12.1% 2.4%
Data from NETL Case 13 (NGCC) and Case (NETL, 2010)
Table 2: Intercooling Comparison Study: Equipment and Process Design Summary
Equipment and Process Design Parameters
Flue Gas Source NGCC Coal Steel
CO2 Removal 90% 90% 90%
Lean Loading (mols
CO2/mols alkalinity) 0.25 0.297 TBD
Flue Gas CO2
Concentration (mol %) 4% 14% >20%
Recycle Rate
(Lrecycle/G) 0.5–8 2–32 TBD
Maximum Approach to
Flooding 70% 70% 70%
Packing
(No Solvent Recycle in
Section)**
MP 250X MP 250X MP
250X
Packing
(Solvent Recycle in
Section)**
0.5 LRecycle/G: MP 250X 2 LRecycle/G: MP 2X
TBD 1 LRecycle/G: MP 250X 4 LRecycle/G: MP 2X
2 LRecycle/G: MP 2X 8 LRecycle/G: MP 125X
3 LRecycle/G: MP 170X 12 LRecycle/G: MP 64X
47 47
3
5 LRecycle/G: MP 125X 20 LRecycle/G: MP 64X
8 LRecycle/G: MP 64X 32 LRecycle/G: MP 64X
**Coarse Packing required to meet flooding criteria in packing section with solvent recycle.
Packing type varied to approximate identical flooding profiles in each case.
Table 2 includes the range of recycle rates evaluated in this study (defined in the table as liquid
rate in the recycle relative to the overall gas rate in the column). The recycle rates used in the
analysis of the coal system were selected to reflect similar ratios of solvent recycle to solvent
feed rates, as used in the natural gas case.
The packing choices, as discussed in Q4 2012, reflect the need to minimize pressure drop in the
solvent recycle section by use of a coarse packing; as the recycle rate increased, progressively
coarser structured packing was used to maintain relatively constant column diameters and
superficial velocities in the sections outside the solvent recycle section. The implications of
packing choices and mass transfer behavior will be discussed further as part of the detailed
evaluation of the predicted benefits of the recycle design.
Figures 1 and 2 are PFDs for the intercooling configurations considered in the study and reflect
the packing configuration described in the preceding paragraph.
Figure 1: Absorber PFD for In-and-Out Intercooling. Two equal sections of packing (MP
250X) are used with liquid draw-off, cooling, and return between the packed sections. The
solvent is cooled to 40 °C before returning to the column.
48 48
4
Figure 2: Absorber PFD for Recycle Intercooling with Bypass. Three packing sections are
used, with the packing height of each section optimized for each design case to minimize
total packing area. MP-250X is used in the top and bottom section and various coarse
structured packing is used in the middle (recycle section) to maintain 70% max approach
to flood. Solvent is drawn off the bottom of the middle section and cooled to 40 °C. A
portion of the solvent is sent directly to the bottom section of the column (equal to the
nominal liquid feed rate of the column) while the remaining liquid is recycled to the top of
the middle section.
The recycle design is expected to cool the gas more effectively (particularly important in the
NGCC cases with high gas rates or relatively low L/G ratios). In addition, the recycle is
expected to provide benefit due to a high liquid rate per wetted perimeter. The drawback of the
recycle design is the mixing of the solvent on the recycle section. The driving forces of the
column are reduced by mixing a richer solvent with lean solvent entering the middle section of
the column. In addition, the temperature leaving the recycle section will be higher than 40 °C,
diminishing the benefits of intercooling realized in the bottom section of the column. The bypass
design was developed to address the latter shortcoming of the simple recycle design. By splitting
a portion of the recycled solvent after cooling and sending it to the bottom section directly, the
solvent entering the bottom of the column is 40 °C, and the full benefit of intercooling in the
bottom section should be achieved. The subsequent analysis attempts to de-couple the various
effects of the recycle intercooling design to attribute predicted benefits to the specific mechanism
included in the rigorous rate-based absorber model.
Recycle Intercooling with Bypass
Max L/G
49 49
5
Benefits of Recycle Intercooling: Natural Gas Combined Cycle
The intercooling configuration comparison for natural gas combined cycle flue gas source
(Rochelle et al., 2012) highlighted the potential benefits of the recycle intercooling design when
compared to the previously developed in-and-out configuration. Figure 3 presents the results of
the analysis conducted at a constant rich loading (to represent equivalent energy performance for
each design).
Figure 3: Intercooling configuration comparison at constant rich loading (solvent rate) in
terms of total metal packing area (to be differentiated from total wetted area which is a
function of fluid properties and fluid dynamics): recycle with bypass and in-and-out
intercooling. For cases without recycle, the maximum L/G corresponds to the nominal feed
L/G. For the recycle cases, this corresponds to the L/G in the recycle section (feed L +
recycle L). Recycle intercooling simulated at a series of recycle solvent flow rates (and
corresponding max L/G).
The results in Figure 3 highlight potential packing (capital cost) savings associated with the
recycle intercooling design. Table 3 provides the packing distribution by section for the cases
presented in Figure 3.
0
100
200
300
400
500
600
700
800
0 2 4 6 8 10
Tota
l Met
al P
acki
ng
Are
a (1
00
0 m
2)
Max L/G (mol/mol)
0.5 LRecycle/G
In and Out IC
8 LRecycle/G
1
2 3 5
Conditions NGCC (4.1% CO2)
LLDG = 0.25 mols CO2/mols alk. CO2 Removal = 90%
RLDG = 0.365
46% Packing Reduction
50 50
6
Table 3: Packing Distribution by height and metal packing area for NGCC application
with simple recycle intercooling. All cases at lean loading = 0.25, constant rich loading =
0.365, 90% removal. MP-250X in top and bottom sections of column in all cases.
Recycle L/G Packing Type in
Recycle Section
% of Total Height % of Total Metal
Packing Area
Top Mid Bottom Top Mid Bottom
0.5 MP-250X 37% 9% 54% 37% 9% 54%
1 MP-250X 36% 16% 48% 36% 16% 48%
2 MP-2X 33% 30% 37% 35% 26% 39%
3 MP-170X 28% 41% 31% 32% 32% 35%
5 MP-125X 28% 56% 16% 40% 37% 23%
8 MP-64X 21% 71% 7% 46% 38% 16%
As expected, the column becomes taller in the recycle section due to the use of progressively
coarser packing; this has consequences for pumping costs, but is not necessarily reflective of
mass transfer effects. However, the total metal packing area (i.e., the physical surface area of the
packing) in the middle section of the column also increases. As evidenced from the optimization
objective function in Equation 1, the most efficient allocation of packing (from a mass transfer
perspective) is progressively weighted to the recycle section with increasing recycle flow rate.
This trend persists even though increased mass transfer area in a well-mixed recycle section
reduces the average driving forces in the column.
[ ] (1)
where:
ap = Specific area of packing (m2/m
3);
hsection = Specific area of packing (m2/m
3);
ACross-Section= Cross-sectional area of column (m2).
The preceding table and figure (and similar results for the analysis for coal applications) clearly
indicate that the recycle configuration improves the mass transfer performance of the column at a
given set of operating conditions when compared to in-and-out intercooling and that these
51 51
7
benefits are correlated with the increasing recycle rate. However, as noted, the benefits of the
recycle design may be attributable to mass transfer enhancement associated with high liquid load
in the recycle section, the use of coarse packing with the high liquid loads, and the enhancement
of intercooling due to cooling of the gas in the recycle section. These benefits would also be
expected to be offset to some degree by the back-mixing effect of recycling the solvent over an
increasingly large portion of the column. These effects are influenced by modeling choices (e.g.,
packing selection, mass transfer model selection, column discretization, etc.). Therefore, it is
critical to understand the source of the predicted benefits to determine if they represent
physically realizable benefits in real systems and to allow sensitivity analyses of modeling
variables and choices that influence the benefits associated with the recycle design.
Quantifying Recycle Intercooling Benefits
The expected effects of recycle intercooling compared to the in-and-out configuration can be
generally described as follows:
Cooling of liquid (and gas to facilitate heat transfer) to address equilibrium constraints;
Reduced average driving forces in the recycle section of the column due to back-
mixing/recycle of the liquid;
Additional mass transfer area generated by high liquid load per wetted perimeter;
Reduced liquid-side mass transfer resistance from enhanced surface to bulk mixing due to
turbulence generated by high liquid rate per wetted perimeter.
The effects can be categorized as driving force effects (intercooling and back-mixing) and mass
transfer enhancement (mass transfer coefficients and interfacial area). As will be discussed, the
contribution of driving force effects to changes in packing requirement (or, alternatively, rich
loading if the analysis was conducted to find energy benefits of the recycle design) are difficult
to quantify and isolate. Therefore, the methodology employed in this work focused on isolating
the mass transfer effects (more readily quantified via the empirical mass transfer models
implemented in Aspen Plus®) and attributing the remaining unexplained changes in packing
requirement to the driving force effects.
The development of the recycle intercooling design (specifically developed for natural gas
applications with low L/G ratios) was for the primary purpose of improved intercooling
performance due to cooling of the gas in addition to the liquid via the recycle design. The mass
transfer benefits were expected to be a secondary benefit as a new degree of freedom (recycle
rate) allowed for the high liquid rate per perimeter in the middle section of the column.
The mechanism for intercooling benefits for PZ systems are well understood and were discussed
in detail in previous work (Plaza, 2011) and thus will not be discussed in depth here. Similarly,
the effect of back-mixing has been studied in detail for countercurrent gas-liquid contacting
systems, most often in the context of hydraulic effects (e.g., entrainment) causing undesirable
axial mixing of the solvent. By recycling rich solvent from a lower portion of the column and
mixing it with lean solvent towards the top of the column, the average driving force in the
recycle section is diminished compared to the standard countercurrent design. As more of the
mass transfer area is contained in the recycle section (as in Table 3), the column approaches a
well-mixed system and the benefits of countercurrent contacting are lost. This is one of the
important limitations in the design and optimization of the recycle configuration.
The mass transfer effects will be discussed in more detail in the following subsections.
52 52
8
Interfacial Area
The effective area of packing is generically defined in Equation 2:
(2)
where:
ae = Effective mass transfer area of packing (m2/m
3);
ap = Specific area of packing (m2/m
3);
A, a,b,c,d = Regression constants;
ρL, ρV = Liquid and vapor density;
uL, uv = Liquid and vapor velocity;
μL, μV = Liquid and vapor viscosity;
L = Characteristic length/dimension of packing.
The dependence of effective area on the variables of the correlation is determined by the
regressed exponents, and thus can vary widely depending on the model selected. In the case of
the recycle intercooling design, the parameters of greatest relevance are the liquid velocity,
specific area of packing, and characteristic dimension of the packing. The liquid velocity is
important due to a large range of liquid rates tested in the recycle analysis while the specific area
and characteristic dimension are representative of the packing, a variable in the middle section of
the column.
For the analysis conducted in this report, the interfacial area model developed by Hanley and
Chen (built into Aspen Plus®) was used (Hanley et al., 2012).
(3)
In this model, the characteristic length is defined as a hydraulic diameter:
(4)
where:
ε = Void Fraction of packing.
53 53
9
Therefore, the Hanley and Chen effective area model can be characterized by the dependence of
the predicted fractional area (the ratio of the specific effective or wetted area to the specific
surface area of the packing) on uL and ap (the void fraction is close to 1 in most cases):
The strong inverse dependence on the specific area of the packing in the Hanley and Chen model
indicates that a coarse packing will generate more mass transfer area per unit of physical surface
area than a fine packing. This effect has been observed in other work and it has been
hypothesized that it arises from the development of ripples, flow instabilities, and droplets in
coarse structured packing (see Tsai, 2010 or Henriques de Brito, 1994 for examples and
discussion). The implication for the recycle intercooling designs is that use of progressively
coarser packing with the increasing recycle rate imparts a secondary benefit of higher effective
area generated per real packing area used. The model developed by Tsai, while also indicating
diminishing returns with finer structured packing, did not have as strong a dependence as the
Hanley and Chen model (fractional area ~ap-0.15
) (Tsai, 2010). Therefore, the selection of an
interfacial area model will influence the optimization when using a design with mixed coarse and
fine packing as in this work. Furthermore, the Hanley model is specific to structured packing –
the potential use of a random or hybrid packing in the recycle section has not been fully
examined and may yet change the effect of the recycle design on the interfacial area for mass
transfer.
The fractional area also displays an inverse dependence on the liquid velocity as in the Hanley
and Chen model. The Tsai area model, in contrast, showed a weak positive dependence on the
liquid rate or velocity (fractional area ~uL0.155
). Part of the explanation for this seemingly
contradictory prediction of the physical behavior in the packing arises from the difficulty of
isolating contributions of a specific physical property or condition that appears in multiple
dimensionless groups (i.e., is related to multiple mechanisms in the fluid dynamics) in the
empirical models developed for packing. In this case, the liquid velocity appears in 3 of the
dimensionless groups (Re, We, and Fr for the liquid) in the generic dimensionless model form
presented in Equation 3. The Hanley and Chen model regression predicts that the We and
Froude number dependencies on velocity cancel, leaving only a liquid Reynolds number
contribution for the velocity. In contrast, Tsai’s regression found no significant effect in the
liquid Reynolds number contribution, but found contributions from both the We and Fr numbers.
Unfortunately, this makes direct physical interpretation of the model dependency difficult; while
properties such as surface tension were varied to develop a We number dependence in the Tsai
model, this also implies some dependence on liquid velocity. Ultimately, in the work by Tsai,
the dimensionless group separation is discarded and the model is recast in terms of physical
properties and a term representing liquid load per perimeter of packing. Tsai concludes this last
term (which contains the liquid velocity and packing geometry contribution) is the most
significant predictor of mass transfer area. Each of the contributions to Tsai’s model can be
explained by experimental data supporting his conclusions even if a simple physical mechanism
is not necessarily evident. The Hanley and Chen model is regressed on a large database
spanning a range of physical systems (hydrocarbon distillation to amine scrubbing systems), and
thus each of the contributions simply reflects a statistical fit and care should be taken in arriving
at any physical conclusions about the variable dependency.
54 54
10
Mass Transfer Coefficients
The liquid side physical mass transfer coefficient will be considered here since the solvent
recycle will most directly impact the liquid side coefficient. As with the interfacial area, the
liquid side mass transfer coefficient can be represented via dimensionless groups:
(5)
where:
kL = Liquid side physical mass transfer coefficient;
A, a, b = Regression constants;
D = Binary diffusion coefficient.
As before, the dependence of the liquid side mass transfer coefficient on the liquid velocity,
packing specific area, and packing characteristic length can be determined from the assigned
exponents in a given correlation. The Hanley and Chen model used in this work follows.
(
) (6)
For the Hanley and Chen model, using the same definition of hydraulic diameter (Equation 4) for
the characteristic length as before, the following results:
The Hanley and Chen model predicts no dependence on the packing specific area, a somewhat
counterintuitive result. Many of the same mechanisms that generate additional mass transfer
area (surface instabilities, turbulence, etc.) in coarse packing compared to fine packing may also
enhance mass transfer by continually replenishing the surface participating in mass transfer. In
contrast, the mass transfer coefficient shows a strong dependence on the liquid velocity. The
high liquid rate per perimeter realized in the recycle design generates turbulence in the liquid
film, reducing the physical mass transfer resistance. The packing selection in the recycle section
will have a strong influence on the liquid velocity since the diameter of the column (set by
flooding constraints) is often fixed by the recycle section. Fine packing results in a larger
diameter, lower superficial velocity, and diminished mass transfer benefits of a high liquid rate.
The effective or wetted mass transfer is directly proportional to mass transfer and, therefore, the
packing required to achieve the specified removal. The effect of the recycle rate and packing
selection is explicitly seen in the preceding section and can lead to direct calculation of changes
in packing requirement. In the case of the physical liquid side mass transfer coefficient,
however, the coefficient represents only part of the overall mass transfer resistance.
(7)
where:
KG = Overall mass transfer coefficient (gas phase basis);
kG = Gas side physical mass transfer coefficient;
kG” = Coefficient representing mass transfer enhancement due to reaction;
55 55
11
As seen in Equation 7, changes in the kL in isolation are not indicative of overall mass transfer
performance; rather, the absolute value of kL relative to the other components in the mass
transfer resistance is important and determines the sensitivity of mass transfer resistance to the
recycle rate and packing selection. In particular, at absorber conditions, the reaction resistance is
expected to be significant (and is often assumed dominant) and equilibrium constraints (with
changing column temperature) can be important in the context of the mass transfer resistance.
Therefore, the prediction of improved mass transfer performance from effects related to the
liquid side physical mass transfer coefficient are dependent not only on the model representation
for the coefficient itself, but also the underlying thermodynamic and kinetic model used in the
simulation over the range of conditions in the absorber. The effect of the mass transfer
coefficient on the packing requirement cannot be calculated directly, but must be developed
indirectly from the model.
Selecting Mass Transfer Models
The preceding discussion highlights the challenge in modeling packing when specific parameters
or physical properties are manipulated to change the performance of the system. In this work,
the use of the liquid rate and packing selection in the recycle section are expected to provide
benefits based on a high-level view of the physics in the system (primarily enhanced turbulence).
However, the use of generalized packing models may not adequately predict the dependence on
the specific variables being manipulated by the modeler or, at the least, do not have the degree of
certainty required to make conclusive statements about the final column design and performance
as a function of the variable(s) of interest. In particular, models that reflect a statistical best-fit of
a range of packing type, operating conditions, and physical properties are only suited for general
prediction of mass transfer behavior by interpolation within the conditions from which the model
was regressed. Therefore, for the type of analysis conducted here, the ideal approach would be
to use a mass transfer model that is specific to the packing selected, includes variation of the
parameters of interest over the range relevant for the modeling, and assigns a dependence to each
parameter individually rather than as part of multiple dimensionless group contributions. While
this diminishes the physical significance and generality of the model, it is a much better suited
approach to modeling the effect of specific variables on mass transfer predictions for a particular
packing. In the absence of such a specific correlation, a general correlation must include a range
of plausible dependency on each variable (e.g., a range on the liquid velocity exponent) to allow
sensitivity analysis to test the robustness of conclusions derived from the parameter
manipulation.
Methodology
The overall approach of isolating the contribution to the change in packing requirement of each
of the effects of the recycle intercooling discussed in the preceding section is summarized as
follows:
OVERALL PACKING REDUCTION
- CONTRIBUTION FROM EFFECTIVE AREA DEPENDENCE ON LIQUID RATE - CONTRIBUTION FROM EFFECTIVE AREA DEPENDENCE ON PACKING SELECTION (SPECIFIC AREA) - CONTRIBUTION FROM MASS TRANSFER COEFFICIENT DEPENDENCE ON LIQUID RATE - CONTRIBUTION FROM MASS TRANSFER COEFFICIENT DEPENDENCE ON PACKING SELECTION
(SPECIFIC AREA)
56 56
12
INTERCOOLING BENEFIT – BACK-MIXIXNG PENALTY (DRIVING FORCE EFFECTS)
Each of the mass transfer contributions is quantified and removed from the overall packing
reduction reported (see Figure 3) leaving a remainder which should reflect the driving force
effects. While this does not lead to a direct quantification of the intercooling benefit, the
intercooling benefit is coupled with the important deterioration of performance expected with the
recycle design and therefore reflects the fundamental driving force trade-off central to the
recycle concept.
The calculation of the individual mass transfer contributions was performed as follows:
1) MODELING STEP: The diameter of the column in the recycle section in each design
case was increased until the superficial liquid velocity matched the values outside of the
recycle section. This modeling approach allows the same amount of liquid to be recycled
through the intercooling section (same maximum L/G as before) providing the same
intercooling benefit as the nominal design, but removes the benefits associated with the
high superficial velocity in the recycle section (see preceding discussion of the effect of
superficial velocity on wetted area and mass transfer coefficients). A new total packing
requirement was calculated from the model; the difference between the packing required
with the larger diameter (lower superficial velocity in the recycle) and the original design
reflects the packing reduction associated with the high superficial velocity in the
recycle section. However, area and mass transfer benefits are still coupled.
2) CALCULATION STEP: Calculate the effect of the liquid superficial velocity on the
wetted area directly from the previously presented correlations. Using the information
from step 1, the change in the packing requirement due to the effect of superficial
liquid velocity on the wetted area and on the mass transfer resistance has now been
isolated independently.
3) CALCULATION STEP: Calculate the change in total packing requirement due to
the use of a coarse packing in the middle section of the recycle design compared to
fine packing throughout the in-and-out intercooling design. This can be calculated
directly from the preceding correlations.
4) The Hanley and Chen model does not predict a dependence of the mass transfer
coefficient on packing type (specific area), so no calculation is needed in this analysis.
The mass transfer enhancements (area and mass transfer resistance related) are now
independently isolated and the remainder of the reported packing reduction is the driving force
effects (intercooling and back-mixing).
This method has potential shortcomings related to the first, modeling step. The mass transfer in
the absorber model is coupled with the energy balance in two ways. First, the location and local
rates of CO2 transfer in the column will determine where heat is generated in the liquid and
creates temperature bulges – the removal or reduction of the bulges is the intercooling benefit
that this method is attempting to deduce. Secondly, the heat transfer in the column is modeled
by Chilton-Colburn analogy to the mass transfer model. Thus, the convective heat transfer
coefficient is proportional to the mass transfer coefficient predicted by the previously discussed
empirical models. Heat transfer coefficients are therefore a function of the liquid (and gas) rate
57 57
13
as well (as would be predicted by any convective heat transfer model). When the superficial
velocity is reduced in step 1 (the gas and liquid velocities change with the diameter), the heat
transfer is potentially impacted as well and may influence the intercooling performance when it
was assumed to be constant. However, the recycle section is modeled as a well-mixed section
and thus heat transfer limitations are likely unimportant. In addition, these effects were assumed
to be higher order effects compared to the primary intercooling mechanisms of a high liquid to
gas rate (gas cooling effect) and cold solvent sent to the bottom of the column.
Also, the change in diameter affects the gas velocities which would have a similar impact on the
gas-side physical mass transfer coefficient. While the gas-side resistance in the absorber has
been previously assumed to be unimportant, this has not been explicitly proven in this analysis
and so the method used in step 1 may exaggerate the liquid-side resistance contribution by
lumping changes to the gas-side resistance into the liquid-side term. A sensitivity analysis of the
gas-side mass transfer coefficient is needed for the analysis to support the idea that it is
insignificant.
Results
Figure 4 presents the different components of the packing reduction compared to the baseline in-
and-out intercooling design for the constant rich loading (constant liquid rate) analysis
represented previously in Figure 3.
-0.3% -1% 0%
2%
6% 6% 6%
12% 10%
6%
1% 4%
10%
17%
22%
33%
40%
10%
15%
28%
32%
41%
48%
-10%
0%
10%
20%
30%
40%
50%
60%
0.5(MP-250X)
1(MP-250X)
2(MP-2X)
3(MP-170X)
5(MP-125X)
8(MP-64X)
Pac
kin
g R
ed
uct
ion
(vs
.In
-An
d-O
ut
IC)
Recycle L/G (mol/mol)
Area Effects
Intercooling - Back-Mixing
kL ~ f(Liquid Rate)
TOTAL REDUCTION
58 58
14
Figure 4: Isolated contributions to overall packing reduction for recycle intercooling design
compared to in-and-out intercooling design. Recycle rates from 0.5 to 8 L/G are presented
for the natural gas combined cycle application. All cases with following specifications:
The area effects (first bar in each group reading from left to right) include the combined effect of
liquid rate and packing specific area on the wetted area (or fractional area); the two were
combined since the overall contribution to the change in packing requirement is small in all
cases. The area effects are negative (or increase the packing requirement compared to in-and-out
intercooling) at the lowest recycle rates because the Hanley and Chen model predicts a negative
dependence on liquid velocity (as discussed before); as the packing is changed to progressively
coarser packing, however, the fractional area dependence on the packing geometry (negative
dependence on specific area of the packing) begins to become more important, and the area
effects ultimately contribute to a reduction in the packing requirement. In general, however, it is
clear that the Hanley and Chen model predicts minimal effect on wetted area of the packing
due to the conditions in the recycle.
The combined intercooling and back-mixing effect of the recycle (driving force effects) are
shown in the second bar from the left in each case. This trend shows an initially increasing
benefit from intercooling at the lowest recycle rates to a maximum of 12% packing reduction at
the 2 L/G recycle rate; beyond this recycle rate, the benefit diminishes and appears negligible at
the highest recycle rates. There are two explanations for this trend. First, as the recycle rate is
increased, the marginal benefit of intercooling diminishes – the temperature in the system is not
affected by an incremental increase in liquid rate beyond some point. This would explain a
flattening in intercooling benefits with recycle rate; the decline is explained by the back-mixing
contribution. As illustrated in Table 3, as the recycle rate increased, more of the packing was
allocated to the well-mixed recycle section of the column by the optimization. Therefore, the
average driving forces in the column are progressively diminished with recycle rates and
undercut the benefit from intercooling.
Finally, the last contribution to the overall packing reduction, is the effect of the liquid rate on
the mass transfer resistance (as determined indirectly from the modeling step (step 1) in the
procedure described before). The third bar from the left for each case reflects this contribution.
Clearly, this is the dominant source of the packing reduction predicted by the model. This result
indicates that the physical liquid-side mass transfer coefficient (as predicted by the Hanley and
Chen model) is of a magnitude where it is a significant contributor to the overall mass transfer
resistance; when coupled with the strong dependence on liquid rate predicted by the model, the
mass transfer resistance contribution becomes very important in the recycle configuration.
This highlights the importance of isolating these effects. If the intercooling benefit is the most
likely to be important in real systems or the model is most likely to accurately predict
intercooling benefits, the current design is less than optimal – the strong dependence on mass
transfer resistance contributions led to a large amount of packing in the middle of the column,
reducing average driving forces and wiping out the intercooling benefits. If the mass transfer
benefits are not realized at this magnitude in a real system, the design will lead to poor
performance compared to the cost of the recycle system (in particular, the pumping costs over
the large recycle section). Therefore, future work needs to determine which of the effects
59 59
15
isolated in this initial analysis are in fact significant and what a reasonable range of sensitivity is
for each parameter. This may come from literature review or evaluation of data collected in this
research group for specific conditions and types of packing.
Conclusions
A method was developed to isolate the different contributions to the packing reduction predicted
for the recycle intercooling configuration compared to the in-and-out configuration in an effort
provide a detailed understanding of the recycle design. The results of this evaluation are
summarized as follows:
The Hanley and Chen model predicts minimal contributions to packing reduction from
the effect of liquid rate and packing selection in the recycle section on the wetted area
available for mass transfer. The two effects (liquid rate and packing geometry)
effectively cancel over the range of conditions tested.
The driving force effects of the liquid recycle design (which includes an intercooling
benefit and penalty for back-mixing) initially show increasing benefits with recycle rate
(the marginal benefit of improved intercooling outweighs the back-mixing impact of the
recycle). The packing reduction from driving force effects reaches a maximum of 12% at
a recycle of 2L/G. At higher recycle rates (> 2L/G), the back-mixing effect becomes more important while
the incremental benefit of intercooling diminishes. This indicates that the current design
(increasing amount of packing in the recycle section with recycle rate) does not maximize
benefits from the intercooling effect. In addition, higher recycle rates may not provide
enough benefit from intercooling in general to justify increased pumping costs. The model predicts that the benefits of the recycle system are dominated by the
contribution of reduced liquid-side mass transfer resistance due to the increased liquid
rate in the recycle. The reduction packing from the mass transfer coefficient contribution
is as high as 40% (of the overall 48% reduction) for the highest recycle rate (8 L/G).
This effect drives the optimization to minimize total packing area by selectively putting
more packing in the middle (recycle) section of the column as the liquid rate is increased.
This strong dependence requires further investigation and should be subject to thorough
sensitivity analysis to determine if the benefits are realizable in a real system.
Process Safety Considerations
A critical component of large-scale deployment of amine based capture systems will be
management of solvent losses due to process failures or upsets. In the scope of the absorber, this
includes flooding of the absorber column which would lead to the potential venting of large
amounts of solvent. Several operating circumstances and components of absorber design should
be considered to prevent flooding of the column during operation:
1) Flooding Under Standard Operating Conditions: Flooding during normal operation is
primarily prevented during the equipment design process. Absorber designs typically
include a safety margin to prevent operation in the flooding region (70% to 80% of
flooding velocity). However, the flooding limit (and associated safety margin) is based on
pressure drop correlations specific to the packing type, operating conditions, and fluid
properties used in experimental development of the correlations. Therefore, an important
aspect of absorber design for novel solvents and configurations (e.g., recycle
60 60
16
intercooling) is testing and development of correlations relevant to amine scrubbing for
power plant applications. This includes experimental efforts such as air-water packing
experiments and pilot scale operations such as those conducted within our group at the
Pickle Research Center.
2) Flooding Due to Equipment Failure: This category includes scenarios such as the failure
of equipment downstream of the absorber. For example, shutdown or failure of pumps
downstream of the absorber could lead to accumulation of liquid in the absorber and and
eventually flooding of the column. Prevention measures would include level sensors in
the absorber sump to trigger shutdown of feed flows, pump sparing or pumping to
emergency solvent storage, and shutdown associated with pressure drop measurements in
the absorber.
3) Flooding during start-up/shut-down: This scenario is particularly relevant to capture
applications for power plants where flexible operation of the scrubbing system or power
plant conditions may require frequent cycling of the amine scrubbing system. During
ramping/shutdown of the capture system, gas and liquid flow rates to the absorber must
be carefully controlled in tandem; high gas velocities (relative to the liquid rate) can be
realized in the absorber that would not be observed in normal operations. Pressure drop
sensors, liquid level control sensors, and flow meters for the gas and liquid can be used to
provide data to design a control strategy to avoid flooding conditions during ramping.
Future Work
The intercooling evaluation performed for coal and natural gas applications will be considered in
further detail using the new information from the isolation of benefits. The mass transfer model
dependencies on packing selection and liquid rate will be evaluated from a range of literature
models and experimental data in our group to provide means for a sensitivity analysis. Alternate
methods to isolate benefits of the recycle design will be considered. For example, the prediction
of a minimum liquid rate for different configurations (or different recycle rates) can serve as a
proxy for the intercooling benefits realized by a design (equilibrium pinches removed). The
findings will be extended to selection of packing to optimize performance in the recycle section
and development of new absorber configuration concepts.
References
Hanley B, Chen CC. “New Mass-Transfer Correlations for Packed Towers.” AICHE J.
2012;58:132–152.
Henriques de Brito M, von Stockar U, Menendez Bangerter A, Bomio P, Laso M. “Effective
Mass-Transfer Area in a Pilot Plant Column Equipped with Structured Packings and with
Ceramic Rings.” Ind Eng Chem Res. 1994;33:647–656.
Plaza JM. Modeling of Carbon Dioxide Absorption using Aqueous Monoethanolamine,
Piperazine, and Promoted Potassium Carbonate. The University of Texas at Austin. Ph.D.
Dissertation. 2011.
Rochelle GT et al. "CO2 Capture by Aqueous Absorption, Fourth Quarterly Progress Report
2012." Luminant Carbon Management Program. The University of Texas at Austin. 2013.
Tsai RE. Mass Transfer Area of Structured Packing. The University of Texas at Austin. Ph.D.
Dissertation. 2010.
61 61
1
Modeling and Optimization of
Advanced Stripper Configurations
Quarterly Report for April 1 – June 30, 2013
by Yu-Jeng Lin
Supported by the Texas Carbon Management Program
McKetta Department of Chemical Engineering
The University of Texas at Austin
July 31, 2013
Abstract
In this work, advanced stripper configurations have been modeled and optimized using Aspen
Plus®. Equivalent work is used as an indicator of energy performance as well as heat duty. A
rich exchanger bypass strategy that recovers stripping steam heat by using a cross exchanger has
been proposed. To get better energy performance, this strategy is applied to advanced stripper
configurations. The next pilot plant configuration, flash stripper with warm rich bypass and rich
exchanger bypass, has offered an 8.4% energy improvement for 8 m PZ and 4.4% for 9 m MEA.
One objective of this work is to demonstrate the flexibility with different operating temperature
of the flash stripper with warm rich bypass and rich exchanger bypass. Since existing power
plants may have different pressure levels of steam extracted from the crossover pipe between the
low and intermediate pressure turbines, the stripper needs to adapt to different operating
temperatures. By equivalent work analysis, the flexibility of this configuration has been
demonstrated in the operating range from 120–150 oC for 8 m PZ and 120–135
oC for 9 m MEA.
When comparing two different regeneration temperatures, the higher operating temperature has
greater improvement at lower lean loading but is less efficient at higher loading.
Another objective is to investigate energy performance using 5 m PZ. 5 m PZ has lower
viscosity, which leads to a higher heat transfer coefficient. Even though 5 m PZ has lower CO2
capacity which increases the sensible heat, but less heat exchanger area is required to attain the
same temperature approach. A preliminary comparison of 5 m and 8 m PZ has been done using
the same 5 oC LMTD specifications in the cross exchanger.
Introduction
In post-combustion CO2 capture, steam usage for lean solvent regeneration in the stripper and
CO2 compression work are the main contributions to the energy requirement. Implementing CO2
capture incurs a 20–30% penalty on electricity output for a typical coal-fired power plant
(Rochelle, 2009). Alternative stripper configurations could improve energy efficiency
significantly compared to a simple stripper.
62 62
2
Several previous studies have been done to improve equivalent work by introducing alternative
many of these evaluations neglect cost-centers by ignoring compression cost or are flawed by being
based on only equilibrium reactions. The economic heuristic to be developed would overcome both of
these flaws.
There have been prior efforts to develop consistent costing methodology. Each effort offers its own
set of guidelines for costing the process (contingency, interest during construction, etc), however they
all conflict on the values assigned (DOE, 2005; IEAGHG, 2013; NETL, 2013). This is due to different base
years and locations for which the models were developed. These studies will be consulted for guidance
in costing equipment and checking predicted costs.
Methods and Preliminary Results
Rigorous Thermodynamic and Kinetic Models
85 85
15
Figure 2 Sequential thermodynamic regression schema showing systems regressed and order of
regression for the diamine system of 2MPZ/PZ.
As detailed in the objectives, a large set of data must be regressed to create a robust model that not
only represents the regressed data but that also extrapolates well. The eNRTL model is chosen for the
thermodynamics due to its integration with Aspen Plus®. The model is constructed through sequential
regression (P. Frailie, Plaza, Van Wagener, & Rochelle, 2011), starting with the amine-water system then
moving to the amine-water-CO2 system. If the goal is a diamine blend, then two additional steps are
needed, namely amine x-amine y-water, and then amine x-amine y-water-CO2, as illustrated in Figure 2.
The expected overall number of parameters for a single amine-water-CO2 model is 20‒30, while for a
diamine system around 60 parameters are needed. The primary parameters are the eNRTL cross
parameters (τij) (X. Chen, 2011; P. Frailie et al., 2011). A preliminary thermodynamic result for 8 m
2MPZ is shown in Figure 3.
86 86
16
Figure 3 CO2 solubility of 8 m 2MPZ. Curves are the model, while filled points are WWC and open points
total pressure (X. Chen, 2011; Xu, 2011).
The thermodynamic and kinetic models are intimately linked through speciation which is governed
by reaction equilibrium constants. All of the models mentioned in previous section used a polynomial
expression to calculate these constants. This resulted in using two sets of fG ,
fH , and
PC values
to calculate thermodynamic properties and equilibrium constants. If these two sets differ, there is a
thermodynamic inconsistency, which is usually manifested in unusual heat capacity behavior. For this
reason, the equilibrium constants are calculated using Equation 4 to ensure that the same set of fG ,
fH , and
PC values is used throughout the model.
dTR
CdT
R
C
TRT
H
RT
HG
RT
GK
T
T
P
T
T
Peq
00
1ln 0
0
00
(4)
Assured of consistent thermodynamics, the kinetic model may proceed. The goal of kinetic
modeling is to match wetted wall column (WWC) data by varying the reaction pre-exponential (ko),
activation energy (EA), and the diffusivity of products and reactants (α and β of Equation 6) (X. Chen,
2011). As viscosity is involved in the calculation of diffusivity, it is regressed first. Then, the density is
regressed to allow for converting from volumetric flow rates to mass flow rates. Once the hydraulics are
done, the reaction parameters are regressed.
1.E+00
1.E+01
1.E+02
1.E+03
1.E+04
1.E+05
1.E+06
1.E+07
0 0.1 0.2 0.3 0.4 0.5
P*CO2
(Pa)
Loading (mol CO2/mol alk)
100 °C
20 °C 40 °C
80 °C
60 °C
120 °C
140 °C
160 °C
87 87
17
Figure 4 Kinetic modeling flow sheet.
As with all regressions, fewer parameters give a more stable result, and so to minimize the number
of parameters, the fewest reactions possible are used to match the experimental flux of CO2. To further
reduce the number of parameters, only one or two reactions are chosen to be regressed, while the
others are ratioed using a Brønsted plot (Cullinane, 2005). All reversible reactions are represented using
a power law, shown in Equation 5, with separate forward and reverse reactions.
(5)
(6)
where k0 is the reaction pre-exponential, EA is the activation energy, R is the universal gas constant, and
Tref is the reference temperature. Do, β, and α are adjustable parameters.
At higher temperatures, the diffusivity becomes more limiting, and so the parameters of Equation 6
are regressed. Once everything has been regressed, the model is checked. If needed, another iteration
is performed, as sketched in Figure 4. Once the thermodynamic data is well matched, and predicted
fluxes are within 20% of experimental fluxes, process simulation can take place.
Generic Amine Model
88 88
18
Figure 5 CO2 solubility of 4 m 2MPZ/4 m PZ. Curves are the model; filled points WWC (X. Chen,
2011); open points total pressure (Xu, 2011).
To create a generic amine model, an existing rigorous Aspen Plus® model is adapted using minimal
experimental data, maybe as few as three CO2 solubility data points. (A full list of basis rigorous models
is provided in the Appendix.) The most similar amine model is selected as a basis, and then CO2 solubility
are fit using the eNRTL cross parameters as well as Gibbs free energy of select amine species. These
data determine capacity and heat of absorption, making their accurate representation critical to
matching process performance. If the thermodynamic model fit is well and the other predicted
properties are reasonable, the kinetics are adjusted to match WWC flux data. This will be done by
adjusting existing reaction parameters.
It is important to keep in mind the limitations of this model. It will not extrapolate well, and many
properties of potential interest will not be regressed but rather simply rely on the basis rigorous model.
However, the benefits of quickly crafting a model to capture gross process performance outweigh these
drawbacks, and these drawbacks can be dealt with by collecting more data for fitting.
As a first step towards this generic model, a preliminary thermodynamic model for 2MPZ/PZ was
created by modifying the pure PZ model using the above procedure. This model matches the CO2
solubility data as shown in Figure 5.
Viscosity Process Effects Viscosity effects on process design will be explored in two ways. One is by comparing a high
viscosity and a low viscosity solvent, such as 7 m MEA and 8 m 2MPZ. The second is by comparing one
solvent at different concentrations, such as 5 m and 8 m 2MPZ. In this exploration, the same process
configuration will be used and its equipment will be sized. Particular attention will be paid to the size of
40 °C
60 °C 80 °C
100 °C 120 °C
140 °C
160 °C
1.E+00
1.E+01
1.E+02
1.E+03
1.E+04
1.E+05
1.E+06
1.E+07
1.E+08
0 0.1 0.2 0.3 0.4 0.5
P*CO2
(Pa)
Loading (mol CO2/mol alk)
CO2 Solubility
89 89
19
the heat exchanger—as increased viscosity decreases heat transfer rates—and the amount of packing in
the absorber—as increased viscosity decreases mass transfer rates. The results of this exploration will
guide future work in this area, but at a minimum one other process configuration will be examined to
determine if viscosity is a deciding factor in process design.
Economic Evaluation The first step will be to determine a representative base case process model that applies to all
solvents of interest. Most likely this will be a simple stripper and an intercooled absorber, which will be
the same configuration used for the viscosity studies. With a base case established, the equipment can
be sized. Most sizing rules are well established, but the heat exchanger has historically given trouble.
For this reason, new methods of sizing the heat exchanger will be trialed.
With this foundation laid, an exploration of different amines and amine concentrations can be done.
This will test the sizing rules and allow for refinement, while at the same time leading to a greater
understanding of capital and operating expense tradeoffs, especially as shown in packing area and
equivalent work. Using this exploration and the spreadsheet model developed by Peter Frailie, an
economic heuristic will be developed to allow for quick estimation of total capital expenditure.
Essentially, the rigorous financial and equipment costing models will be reduced down to one equation
for the total annualized cost (TAC) using purchased equipment cost (PEQ) and energy of the form:
CenergyBPEQATAC (7)
TimelineFall 2013
2MPZ model completion
2MPZ/PZ model completion (paper)
model comparison for generic model
Spring 2014
AMP/PZ model completion (paper)
viscosity sensitivity analysis for 5 m and
8 m 2MPZ
generic model development
Fall 2014
generic model completion (oral
presentation and paper for GHGT-12)
viscosity study comparing a high and a
low viscosity solvent
Spring 2015
economic heuristic development
generic model validation
Fall 2015
economic heuristic completion (paper)
process simulation using generic
models
Spring 2016
dissertation write up
graduation
90 90
20
References
Abu-Zahra, M. R. M., Schneiders, L. H. J., Niederer, J. P. M., Feron, P. H. M., & Versteeg, G. F. (2007). CO2 capture from power plants. International Journal of Greenhouse Gas Control, 1(1), 37–46. doi:10.1016/S1750-5836(06)00007-7
Austgen, D. (1989). A Model of Vapor-Liquid Equilibriua for Acid Gas-Alkanolamine-Water Systems. The University of Texas at Austin.
Bishnoi, S., & Rochelle, G. T. (2000). Absorption of carbon dioxide into aqueous piperazine: reaction kinetics, mass transfer and solubility. Chemical Engineering Science, 55(22), 5531–5543. doi:10.1016/S0009-2509(00)00182-2
Brand, C. V, Rodriguez, J., Galindo, A., Jackson, G., & Adjiman, C. S. (2013). Validation of a process model of CO2 capture in an aqueous solvent , using an implicit molecular based treatment of the reactions, 00, 1–6.
Chen, C.-C., Britt, H. I., Boston, J. F., & Evans, L. B. (1982). Local Composition Model for Excess Gibbs Energy of Electrolyte Systems. AIChE Journal, 28(4), 588–596.
Chen, C.-C., & Song, Y. (2004). Generalized electrolyte-NRTL model for mixed-solvent electrolyte systems. AIChE Journal, 50(8), 1928–1941. doi:10.1002/aic.10151
Chen, X. (2011). Carbon Dioxide Thermodynamics, Kinetics, and Mass Transfer in Aqueous Piperazine Derivatives and Other Amines. The University of Texas at Austin.
Chowdhury, F. A., Okabe, H., Yamada, H., Onoda, M., & Fujioka, Y. (2011). Synthesis and selection of hindered new amine absorbents for CO2 capture. Energy Procedia, 4, 201–208. doi:10.1016/j.egypro.2011.01.042
Cullinane, J. T. (2005). Thermodynamics and Kinetics of Aqueous Piperazine with Potassium Carbonate for Carbon Dioxide Absorption. The University of Texas at Austin.
Desideri, U., & Paolucci, A. (1999). Performance modelling of a carbon dioxide removal system for power plants. Energy Conversion and Management, 40(18), 1899–1915. doi:10.1016/S0196-8904(99)00074-6
DOE. (2005). Carbon Capture and Sequestration Systems Analysis Guidelines (p. 67).
Dugas, R. E. (2009). Carbon Dioxide Absorption , Desorption , and Diffusion in Aqueous Piperazine and Monoethanolamine. The University of Texas at Austin.
Edali, M., Idem, R., & Aboudheir, A. (2010). 1D and 2D absorption-rate/kinetic modeling and simulation of carbon dioxide absorption into mixed aqueous solutions of MDEA and PZ in a laminar jet apparatus. International Journal of Greenhouse Gas Control, 4(2), 143–151. doi:10.1016/j.ijggc.2009.11.005
91 91
21
EPA. (2009). Endangerment and Cause or Contribute Findings for Greenhouse Gases Under Section 202 (a) of the Clean Air Act. earth1.epa.gov. Retrieved from http://earth1.epa.gov/climatechange/Downloads/endangerment/rtc_volume_3.pdf
Ermatchkov, V., & Pe, Ä. (2006). Solubility of Carbon Dioxide in Aqueous Solutions of N-Methyldiethanolamine in the Low Gas Loading Region, 6081–6091.
Frailie, P., Plaza, J., Van Wagener, D., & Rochelle, G. T. (2011). Modeling piperazine thermodynamics. Energy Procedia, 4, 35–42. doi:10.1016/j.egypro.2011.01.020
Hanley, B., & Chen, C. (2012). New mass transfer correlations for packed towers. AIChE journal, 58(1). doi:10.1002/aic
IEAGHG. (2013). Criteria for Technical and Economic Assessment of Plants with Low CO2 Emissions.
Incropera, F. P. (2006). Fundamentals of Heat and Mass Transfer (6th ed.). Wiley.
Intergovernmental Panel on Climate Change (IPCC) (2005). Summary for Policymakers of the Intergovernmental Panel on Climate Change.
International Energy Agency (IEA) (2008). Energy Technology Perspectives.
International Energy Agency (IEA) (2010). Carbon capture and storage. Energy Policy. Retrieved from http://www.sciencedirect.com/science/article/pii/S0301421508004436
International Energy Agency (IEA) (2011). CO2 Emissions from Fuel Combustion 2011 (p. 133). OECD Publishing. doi:10.1787/co2_fuel-2011-en
Kohl, A., & Nielsen, R. (1997). Gas Purification (5th ed., p. 1395). Houston: Gulf Professional Publishing.
Li, H., Li, L., Nguyen, T., Rochelle, G. T., & Chen, J. (2013). Characterization of Piperazine / 2-Aminomethylpropanol for Carbon Dioxide Capture, 00, 1–13.
Li, L., Li, H., Namjoshi, O., Du, Y., & Rochelle, G. T. (2013). Absorption rates and CO2 solubility in new piperazine blends, 00, 1–16.
Li, L., Voice, A. K., Li, H., Namjoshi, O., Nguyen, T., Du, Y., & Rochelle, G. T. (2013). Amine blends using concentrated piperazine. Energy Procedia.
Mac Dowell, N., & Shah, N. (2013). Identification of the cost-optimal degree of CO2 capture: An optimisation study using dynamic process models. International Journal of Greenhouse Gas Control, 13, 44–58. doi:10.1016/j.ijggc.2012.11.029
Mangers, R., & Ponter, A. (1980). Effect of viscosity on liquid film resistance to mass transfer in a packed column. Industrial & Engineering Chemistry …, 530–537. Retrieved from http://pubs.acs.org/doi/abs/10.1021/i260076a005
92 92
22
Mathias, P. M., Reddy, S., Smith, A., & Afshar, K. (2013). A Guide to Evaluate Solvents and Processes for Post-Combustion CO2 Capture, 00, 1–8.
NETL. (2013). Capital Cost Scaling Methodoloy.
Nuchitprasittichai, A., & Cremaschi, S. (2011). Optimization of CO2 capture process with aqueous amines using response surface methodology. Computers & Chemical Engineering, 35(8), 1521–1531. doi:10.1016/j.compchemeng.2011.03.016
Oyenekan, B. (2007). Modeling of Strippers for CO2 Capture by Aqueous Amines. The University of Texas at Austin.
Plaza, J.M., & Rochelle, G. T. (2011). Modeling pilot plant results for CO2 capture by aqueous piperazine. Energy Procedia, 4, 1593–1600. Retrieved from http://www.sciencedirect.com/science/article/pii/S1876610211002268
Plaza, Jorge M. (2011). Modeling of Carbon Dioxide Absorption using Aqueous Monoethanolamine, Piperazine and Promoted Potassium Carbonate. The University of Texas at Austin.
Rochelle, G. T. (2009). Amine scrubbing for CO2 capture. Science (New York, N.Y.), 325(5948), 1652–4. doi:10.1126/science.1176731
Singh, P. (2011). Amine Based Solvent for CO2 Absorption “From Molecular Structure to Process”. University of Twente.
Trimeric Corporation. (2005). Integrating MEA Regeneration with CO2 Compression and Peaking to Reduce CO2 Capture Costs. Power.
Tsai, R. (2010). Mass Transfer Area of Structured Packing. The University of Texas at Austin.
U. S. Energy Information Administration (2012a). Short-term energy outlook (p. 44). Retrieved from http://www.osti.gov/energycitations/product.biblio.jsp?osti_id=6030138
U. S. Energy Information Administration (2012b). AEO2012 Early Release Overview, 2012, 1–13.
Versteeg, G., & Swaaij, W. van. (1988). On the kinetics between CO2 and alkanolamines both in aqueous and non-aqueous solutions—II. Tertiary amines. Chemical engineering science, 43(3), 587–591. Retrieved from http://www.sciencedirect.com/science/article/pii/0009250988870180
Xu, Q. (2011). Thermodynamics of CO2 Loaded Aqueous Amines. University of Texas at Austin.
Zhang, Y., & Chen, C.-C. (2011). Thermodynamic Modeling for CO2 Absorption in Aqueous MDEA Solution with Electrolyte NRTL Model. measurements, 163–175. Retrieved from http://www.aspentech.com/downloads/thermodynamic_modeling_co2.pdf
93 93
23
Zhang, Y., Chen, H., & Chen, C. (2009). Rate-based process modeling study of CO2 capture with aqueous monoethanolamine solution. Industrial & …, 9233–9246. Retrieved from http://pubs.acs.org/doi/abs/10.1021/ie900068k
Zhang, Y., Que, H., & Chen, C.-C. (2011). Thermodynamic modeling for CO2 absorption in aqueous MEA solution with electrolyte NRTL model. Fluid Phase Equilibria, 311, 67–75. doi:10.1016/j.fluid.2011.08.025
Zong, L., & Chen, C.-C. (2011). Thermodynamic modeling of CO2 and H2S solubilities in aqueous DIPA solution, aqueous sulfolane–DIPA solution, and aqueous sulfolane–MDEA solution with electrolyte NRTL model. Fluid Phase Equilibria, 306(2), 190–203. doi:10.1016/j.fluid.2011.04.007
Disclaimer: This presentation was prepared as an account of work sponsored by an agency of
the United States Government. Neither the United States Government nor any agency
thereof, nor any of their employees, makes any warranty, express or implied, or assumes
any legal liability or responsibility for the accuracy, completeness, or usefulness of any
information, apparatus, product, or process disclosed, or represents that its use would not
infringe privately owned rights. Reference herein to any specific commercial product,
process, or service by trade name, trademark, manufacturer, or otherwise does not
necessarily constitute or imply its endorsement, recommendation, or favoring by the United
States Government or any agency thereof. The views and opinions of authors expressed
herein do not necessarily state or reflect those of the United States Government or any
agency thereof.
Appendix
94 94
24
Figure 6 A review of prior kinetic modeling efforts. Reproduction of Table 3.1 (Jorge M.
Plaza, 2011).
95 95
25
Table 1 List of available rigorous models in Aspen Plus® (X. Chen, 2011; Dugas, 2009; H.
Li, Li, Nguyen, Rochelle, & Chen, 2013; L. Li, Li, Namjoshi, Du, & Rochelle, 2013; Jorge M. Plaza,
The fundamental equations for the eNRTL model are presented below. For a thorough explanation see
(Song & Chen, 2009).
PDHexlcexex GGG ,, (8)
PDH
i
lc
ii lnlnln (9)
expG (10)
where Gex is the excess Gibbs free energy, Gex,lc the local contribution to the Gex, Gex,PDH the long-range
contribution to Gex, γ the activity coefficient, α the non-randomness factor (the same as in NRTL), and τ
the binary interaction parameter.
Appendix B: Importing Non-Databank Components into an Existing Aspen Plus® Simulation and Wetted Wall Column Data Regression Procedures
96 96
26
Importing Non-Databank Components into an Existing Aspen Plus® Simulation Purpose: This procedure is useful for merging two existing Aspen thermodynamic models . It was
developed for the creation of 2MPZ/PZ blend model by merging 2MPZ into the Independence
model.
Created: 2013-06-12 by Brent Sherman
Modified: 2013-6-24 by Brent Sherman
It is best to merge the models in as few as sittings as possible. The reason is there are many different
things that need to be changed, and these are scattered throughout the Aspen interface. Failure to
update all of them will lead to cryptic error messages that will lead you in the wrong direction.
Throughout this procedure, I will refer to pulling from a source model. This is the model that has the
components you want to merge into the other model. The other model I’ll call the destination model.
Add the New Components Under ComponentsSpecification in the source model, right click and copy and paste
each component one-by-one into the destination model.
o with each paste, click User Defined in both simulations. I make sure the Aliases
match, and the properties on the Conventional Component Basic Data sheet
(the next page) match, and lastly that the Conventional Component
Additional Data also match.
add the amine and zwitterion to the list of Henry’s components: ComponentsHenry
CompsHC-1
Change the Parameters Now, begin to change all the parameters. In all cases, you will look at the source model and change the
destination model parameters to match. First, change the scalar parameters.
Use PropertiesPure ComponentUSRDEF to change non-temperature dependent
parameters. Watch out as the default unit may conflict with the retrieve parameter unit.
Then start changing the temperature dependent parameters
o If a T-dependent parameter is not showing, select
PropertiesParametersPure Component and click New… .
o Add the amine to THRSWT-1: ParameterPure ComponentTHRSWT-1
The flags here are critical. They determine what property methods are used to
calculate all of the basic thermo properties (critical T and P, compressibility, Cp,
etc)
Similarly, the flags in TRNSWT-1 select the transport submodels used.
97 97
27
Once all the pure component ones are changed, change the Binary Interaction
parameters.
o Henry, NRTL, and VLCLK.
Then do the Electrolyte Pair parameters.
o Cut and paste the GMELCC, GMELCD, and GMELCE values.
Verify Parameters In both simulations, use ToolsRetrieve Parameter Results to see all parameters
for the components of interest.
o Put these into Excel, mark the ones that are different, and start making them match up
o Keep track of what has been changed as Aspen doesn’t always take the first change
(unpredictable and non-reproducible, so be scrupulous and double check)
When you think you are done, retrieve parameter results and check again.
o Error-prone as cutting and pasting can truncate accidentally.
o Easy to lose track of parameters in the sea of numbers.
o Easy to forget to change units.
Check which parameters are actually used by looking at the switches in THRSTW-1 and
TRNSWT-1.
Update Chemistry You must copy over the equilibrium reactions in order for the system to speciate. Don’t forget
to do both GLOBAL and REDUCED.
If you fail to do so and attempt to calculate the properties for components that are not present
(ie you add the amine and try to calculate the related species’ properties), you will see the
following error, “EOS LIQUID VOLUME CALCULATION FAILED TO CONVERGE
AFTER 99 ITERATIONS”.
If your source model has kinetic reactions, copy those as well.
Update Subroutines You must merge each subroutine. This will involve adding in new variables, if switches, and
other code.
o vl2u2.f
o mul2u2.f
o dl0u.f
Verify Thermodynamic Integrity The best way to verify proper behavior is through Property Analysis blocks.
Before copying and pasting the property analysis blocks over, you need to copy and paste the
Prop-Sets.
Now copy and paste over the Analysis blocks.
o gamma
o heat of absorption
o pKa
o density
o speciation
o vapor pressure
98 98
28
When it’s all done, verify that you have achieved the same performance that you previously
had.
Remember that for pKa, you need to use the chemistry that includes protons.
Wetted Wall Column Data Regression Purpose: To match the experimental flux within ±20% by changing rate constants and the effective
diffusion coefficient of reactants and products.
Created: 2012-06-25 by Brent Sherman
Last Updated: 2013-06-27 by Brent Sherman
Prerequisites Subroutines
o area (area.obj)
o diffusivity (dl0u.obj)
o pressure drop (pressuredrop.obj)
o density (vl2u2.obj)
o viscosity (mul2u2.obj)
o bubble point (drusr0.obj)
o mass transfer (masstransfer.obj)
Raw data from the WWC (one .xlsx file for each loading and temp. combo)
An Excel workbook for pre- and post-processing.
Complete thermo model
Complete hydraulic model (vl2u2 and mul2u2 must be current)
Method
Raw Data Pre-processing While processing the data, check the experimental fit. Do any of the points seem to be outliers? Is the
line straight and through the origin? What would cause these things?
Extract the raw experimental data from the Excel files supplied by the experimenter.
o There are three sheets in each file: Sheet1, Gas Film Resistance, and Gas Mixtures.
Regression data is in the first two sheets.
Be careful of the units and diameter scaling, as this will change the flux values.
From the first, take the fluxes, Ptotal, P*CO2, and Qtotal.
From the second, take Qgas (both values), kg, kg’, and Kg
o This data should go in the provided Excel template, which looks like below. There is far
more to the spreadsheet than that shown below.
99 99
29
o All of the data extracted is put into one large table whence many more things are
calculated. Check molecular weights.
o This data is preprocessed and sent to another table for input into Aspen with the proper
units.
o Check the diameter of the WWC and rescale as necessary. The WWC is either 10x or
100x the physical diameter.
Aspen Simulation
Figure 7 Aspen WWC PFD.
Description
In the above figure are the WWC and two flash blocks. The rich and lean flash blocks simply flash the
rich and lean solvents in a bubble point calculation to determine the equilibrium partial pressure of CO2.
The lean solvent is fed in three separate streams to a mixer. The heat of mixing and speciation is then
dealt with using the heater block. The vapor stream contains N2 and CO2. This feed gas and a liquid
100 100
30
water stream go to the saturator, which is a T-P flash block. This process ensures the liquid and vapor
feeds are isothermal with the WWC itself.
The WWC is a three stage RADFRAC column. Its diameter is 10 or 100x the actual WWC diameter, but
the height is the same (9.1 cm). The WWC uses the counter-current flow model with a discretized liquid
film and 5% liquid hold up. It uses a customized mass transfer and interfacial area subroutine, rendering
the choice of packing irrelevant.
There are three calculator blocks used. C-LDGS simply checks the loadings to make sure they are as they
should be. C-FLUX computes the flux and KG, the overall gas side mass transfer coefficient. C-KEQ
calculates the equilibrium constants for each reaction, ratios the rates of reaction, and back-calculates
the reverse rates of reaction. For the sake of this block, all activation energies are set to 0, which
reduces kr to kf/Keq.
Method: Generating a set of ko’s and EA’s
Pick two points at 40 and 60 °C.
o one dominated by the bicarbonate reaction
o one dominated by the carbamate
Note that practically speaking, the carbamate is the most significant reaction
across the whole range as it is the fastest reaction.
Cut and paste all flowrates from Excel into Aspen.
Update the temperature and pressure of the 2MPZ stream. The transfer blocks will propagate
these throughout the model.
In the reaction set, make sure all activation energies are set to 0.
Adjust loading to give the same error in the absorption and desorption points.
o Always re-initialize before each run.
o Run with a fixed set of ko’s and EA’s.
o Start with the experimental loading if you have no better guess.
o Run the simulation and calculate the ratio of the predicted flux to the experimental flux.
o Run until the difference beteen the ratios is less than 1% or you have adjusted the
loading as much as possible.
o Don’t be afraid if the ratio is poor, and the loading is maximally adjusted. This process is
iterative, and the results will improve over time.
Adjusting loading notes
o Make sure the design spec and the KEQ calculator block are disabled, as both change ko
values.
o only adjust 10% of the operational loading range (this works out to 0.01 mol/mol alk. for
2MPZ)
o The idea behind this is that at 0 driving force, there should be zero flux. Thus having the
same over- or under-prediction for both points will ensure the line goes through the
origin.
101 101
31
o Setting the activation energies to zero makes the reaction pre-exponential equal to the
rate of reaction.
o You will always use adjusted loading values for Aspen.
Now, run the regression using the adjusted loading to get k values.
o Activate the design spec and the KEQ calculator block.
o Run it for the desorption points at high loading. Record KEQ‘s and kf.
arbitrary decision to use high loading desorption points
o Input the experimental flux into the design spec as you run each point.
o Set the design spec to vary one of the forward ko’s. Record the result.
o Then rerun varying the other ko’s.
You should be able to perfectly match the flux.
o Then repeat for the second temperature.
o Back calculate kr from KEQ and kf.
o You should now have for each reaction a KEQ, a kf, and a kr.
Calculate the activation energies.
o Use goal seek to solve the following equation by varying EA.
o C
ref
AC k
TTR
Ek 6040
110
o Ratioed reactions will have the same activation energy as the reaction they are ratioed
to.
o It’s a good idea to double check that using the ko and EA calculated, you duplicate the k
at 60C.
o Repeat this procedure for the reverse reactions.
You now have a set of ko’s and EA’s that are thermodynamically consistent. Fix these. Turn off
the design spec and the KEQ calculator block.
From here on out, you will simply be repeating the loading adjustment calculation.
Now, repeat the loading adjustment for all of the data points using the fixed set of ko’s and EA’s.
Adjust the loadings until the under- or over-prediction is less than 1%. However, do not adjust
more than 10% of the operational loading range.
Plot the data vs T and loading. Look for trends indicating systematic bias. Sample plots are
shown below.
Figure 9 8 m 2MPZ kinetic fit.
0.6
1.0
1.4
1.8
Fluxpred/ Fluxexp
Loading (mol/mol alk.)
40 abs 40 des
60 abs 60 des
80 abs 80 des
100 abs 100 des
103 103
33
Figure 10 8 m 2MPZ kinetic fit.
Method: Starting a New Iteration You checked your data, and there is a trend. Or there is no trend, but you simply want a tighter fit. You
can regress reaction rate constants or diffusion parameters.
Changing the reaction rate
o change the ko as you please
o repeat the loading adjustment for the same two data points previously used
o repeat the regression step
o then, fix the ko’s and EA’s and run through all the data points repeating the loading
adjustment
Changing the diffusion parameters
o no need to the re-regress ko’s and EA’s
o simply run back through the data, repeating the loading adjustment
Appendix C: CCSI IAB Meeting April 2013 Posters
0.6
1.0
1.4
1.8
40 60 80 100
Fluxpred/ Fluxexp
T (°C)
40 abs 40 des
60 abs 60 des
80 abs 80 des
100 abs 100 des
104 104
34
105 105
35
106 106
1
Dynamic Modeling and Control of Amine Scrubbing
Quarterly Report for April 1 – June 30, 2013
by Matthew Walters
Supported by the Texas Carbon Management Program
McKetta Department of Chemical Engineering
The University of Texas at Austin
July 31, 2013
Abstract
A dynamic model of solvent regeneration using piperazine is being developed in gPROMS®
. To
improve the quality of the code and reduce the chance of mistakes, all of the thermophysical
properties of the amine solvent have been compiled into one dynamic link library file.
gPROMS® interfaces with this user defined properties package by treating it as a foreign object.
The previously developed model of a separator vessel has been modified to account for vapor
inventory, since this may affect the transient behavior of the system. The goal of creating a
dynamic model is for the development of process control strategies. The major process control
objectives for amine scrubbing have been identified: reject disturbances from the upstream
power plant, track set point changes made in response to grid demand, obey process constraints,
and allow for stable operation with process intensification. A multiple time scale behavior is
demonstrated, which suggests the need for a hierarchical controller design.
Introduction
Dynamic models are useful tools for designing process control strategies, optimizing off-design
steady state conditions, and understanding the system response to input changes or disturbances.
Several models have been developed and validated to predict the steady state operation of amine
scrubbing (see review by Wang et al., 2011). However, there are only a few examples of works
concerned with the dynamics of this process (Kvamsdal et al., 2009; Lawal et al., 2009;
Tobieson et al., 2012; Ziaii, 2012). A major objective of this research is to develop and validate
a detailed equation-based dynamic model of an amine scrubbing plant using piperazine (PZ)
solvent and advanced process configurations. Another objective is to use the model to develop a
process control strategy that satisfies process constraints as well as economic objectives. The
strategy should be effective in the presence of model and parameter uncertainty. This research
strives to address some of the challenges associated with process design and control so that
recommendations may ultimately be made for the operation of carbon capture from commercial
scale power plants.
Thermophysical Properties Package for Piperazine
Process flowsheeting software, such as Aspen Plus® or gPROMS
®, generally provides a
thermophysical properties library which allows the user to perform equilibrium calculations for
chemical compounds available in its database. In the case of amine-H2O-CO2 systems,
107 107
2
properties calculated from the built-in libraries do not provide adequate agreement with
experimental data to accurately simulate an amine scrubbing process. It therefore becomes
necessary to either modify the existing thermophysical properties package (for example see
Frailie et al., 2011) or to create a custom package which is consistent with the experimental
results. In gPROMS®, the best way to include equation-based thermophysical properties for an
amine scrubbing system is through developing an independent properties library for the desired
amine solvent. gPROMS® can then call this library without having to repeatedly include the
same set of equations in the model of each individual unit operation. Additionally, the unit
models can be designed for a generic amine solvent, so future users have the option to create
their own library for a new amine without needing to modify the existing gPROMS® code.
Table 1 lists the properties that should be included in the library, the variables needed to
calculate the property, and the source of this information for piperazine (PZ).
Table 1: Thermophysical properties included in the library.
Thermophysical Property Function of Source for PZ
Xu, 2011
DIPPR
Constant ( = 0 ) Assumed Nonvolatile
Freeman, 2011
Freeman, 2011
---
∫
Frailie et al., 2011
∫
DIPPR
∫
DIPPR
Xu, 2011
DIPPR
To implement an independent library in gPROMS®, the equations used to calculate the properties
in Table 1 were coded in C++ using a template provided by Process Systems Enterprise (PSE).
Equations describing the analytical derivative of each physical property with respect to all input
variables were also required to be included in the code. This is because gPROMS®
interfaces
with a custom built package by treating it as a foreign object and no longer has access to the
equations describing the thermophysical properties. Mathematica® was used to obtain analytical
expressions for the derivatives. Microsoft Visual Studio 2012® was used to build the C++ code
into a dynamic link library (DLL) file. Instructions on how to build the DLL using the templates
provided by PSE is given in the Appendix. Once the DLL is saved in the appropriate directory,
any model in gPROMS® is able to call this user-created library by declaring the DLL filename as
a foreign object. There is no apparent change in simulation speed as a result of including the
thermophysical properties in a foreign object instead of coding the equations within each unit
operation. Figure 1 describes how information is passed between gPROMS® and the foreign
object.
108 108
3
Figure 1: gPROMS® calls the desired thermophysical property (“physprop”) and passes
the current values of the liquid mole fractions and temperature to the foreign object. The
foreign object returns the numerical value of the thermophysical property, along with the
numerical value of its derivative with respect to each calculation input.
Two-Phase Separator Vessel Model
A previously presented model of a separator vessel performing an equilibrium flash calculation
for CO2 desorption assumed that vapor hold-up was negligible (Walters, Dunia, et al., 2012).
However, Kumar & Daoutidis (1999) have shown that models which include the vapor hold-up
can lead to different transient behavior than a model that assumes the vapor hold-up is negligible.
Therefore, the separator vessel model was updated so that vapor hold-up is included in the total
vessel material hold-up. As discussed in the previous section, the model should be formulated so
that it is not amine specific. The new two-phase model is represented in gPROMS® by
Equations 1–19, along with calls for the appropriate thermophysical properties listed in Table 1.
, for i = CO2, H2O, amine (1–3)
(4)
∑ (5)
∑ (6)
∑ (7)
(8)
(9)
, for i = CO2, H2O, amine (10–12)
(13)
(14)
109 109
4
∫
(15)
∫
∫
(16)
, for i = CO2, H2O, amine (17–19)
The specific enthalpy, composition, and flow rate entering the vessel are specified by the inlet
stream conditions, pressure at the inlet and outlets of the tank are set by the flash calculation, and
the effluent flows are specified by the outlet streams. In the case of a heated flash tank, the heat
duty is the energy supplied by steam minus heat loss. When there is a steam heater upstream,
heat loss to the surroundings is the only component of the heat duty.
Overview of Process Control Strategy Development
An effective process control strategy for a post-combustion capture plant would accomplish the
following objectives, which are described in more detail in the proceeding sections:
1. Reject disturbances from the upstream power plant.
2. Track set point changes made in response to grid demand.
3. Obey process and safety constraints.
4. Allow stable operation while minimizing the total solvent inventory and intensifying
process integration.
Disturbance Rejection and Set Point Tracking
The operation of the amine scrubbing plant is highly dependent on the upstream power plant and
upstream pollution control processes (for example flue gas desulfurization). The flow,
composition, and temperature at the absorber inlet are determined by these upstream operations
and are treated as disturbances to the amine scrubbing process. The availability of steam for
solvent regeneration, cooling water temperature, and ambient conditions are also considered
disturbances that affect process outputs. The control strategy developed in this research should
quickly respond to upsets or load changes in the power plant or the other disturbances mentioned
here, which may be either measured or unmeasured. In addition to stabilization, the controller
should minimize energy use when disturbances occur. Manipulated inputs such as bypass flows,
compressor speed, and valve positions should be set to minimize an energy cost function.
In addition to frequent disturbances, it is also expected that set point changes will occur in
response to the electricity grid demand. Cohen (2012) found that operating a carbon capture
process flexibly increases the ability of the plant to provide grid reliability services and improves
grid resiliency at minimum and maximum electricity demand. At minimum demands, flexible
capture helps respond to intermittency in renewable energy generation. At maximum demands,
carbon capture can be ramped down so peak load can be met without installing additional grid
capacity. Therefore, we are interested in a control strategy that is able to move from 100% load
to ~20% load and back quickly and without oscillations. Fast and stable responses to set point
changes will therefore be an essential part of the process control strategy, along with the
regulation objective mentioned above.
Process and Safety Constraints
110 110
5
Varying degrees of constraints exist in the amine scrubbing process. Some constraints
correspond to hard input constraints that cannot physically be violated, for example the
maximum speed of the compressor or the maximum pressure of steam available from the IP/LP
crossover. Soft operational constraints, like the solvent degradation temperature, can be
temporarily violated but an appropriate control design would penalize these violations so the
process conditions quickly return to acceptable values. Finally, hard safety constraints like the
maximum pressure and temperature rating of a heat exchanger, the compressor surge limit, and
pressure limit for pump cavitation must be obeyed at all times because violating them could lead
to a plant shutdown and potentially create serious safety issues. These process constraints must
be taken into consideration for an effective control strategy design.
Multiple Time Scale Behavior
The general trend in the chemical process industry is the development of increasingly integrated
process designs that use extensive material and energy recycling and minimize overall inventory.
With enormous capital and operating costs, amine scrubbing will certainly adhere to this trend.
Figure 2 shows the material flows in a typical amine scrubbing process. Fis denotes the molar
flow of stream i at steady state. Based on the work of Baldea & Daoutidis (2012), the following
observations are made regarding the relative magnitudes of these flows:
i. The nominal flows of the amine solvent streams in the recycle loop, the CO2-rich solvent
flow (FRichs) and the CO2-lean solvent flow (FLean
s), are of comparable magnitude:
ii. The amount of CO2 entering the absorber in the flue gas (FCO2,ins) is of comparable
magnitude to the amount of CO2 exiting the stripper (FCO2,outs):
iii. The flow of the recycled solvent is much greater than the combined CO2 throughput and
the makeup amine and water (FAmines and FH2O
s), which is reflected in a large recycle
number (Rc):
iv. The flow of the purge stream is much smaller than the rate of material entering the
system, reflected in a small purge number (Pu):
These observations may suggest that the overall plant is a singularly perturbed system with the
possible existence of three distinct time scales: a fast time scale at the unit level associated with
the large recycle loop flows, an intermediate time scale at the process level associated with the
small feed and product flows, and a slow time scale of impurity levels associated with the purge
stream. A time scale decomposition will be performed to confirm whether the system is
singularly perturbed. Singularly perturbed systems complicate model-based process control
because the equations are inherently stiff and reduced order modeling is required.
In order to satisfy control objectives while simultaneously minimizing solvent inventory and
maximizing material and energy recycling, the controller design must take into account the
111 111
6
presence of multiple time scales. When systems are clearly separable into multiple time scales
such as amine scrubbing, a hierarchical control strategy is warranted (Scattolini, 2009). At the
fast unit-level time scale, standard PI controllers are usually sufficient to achieve regulatory
objectives such as level control. At the intermediate and slow time scales, a supervisory
controller is needed to achieve overall process objectives, such as the percentage of CO2
removed from the flue gas.
Figure 2: Process material and energy flows.
Literature Review of Amine Scrubbing Process Control
Limited examples exist of process control strategy development for amine scrubbing. Ziaii
(2012) developed a multi-loop cascade controller that showed adequate disturbance rejection and
set point tracking. Panahi & Skogestad (2011) used an economic optimization procedure to
select the best self-optimizing controlled variables. This work was extended (Panahi &
Skogestad, 2012) to include a plantwide control strategy consisting of a regulatory layer with
PID control and a supervisory layer with model predictive control. The dynamic model of the
plant used to test this control scheme contains significant simplifications and assumptions.
Åkesson et al. (2012) developed a nonlinear model predictive controller for the stripper and
demonstrated the controller on a simplified plant model.
Conclusions
A thermophysical properties library has been developed for piperazine which can be
called by any unit operation model developed in gPROMS®.
Including vapor hold-up may be important to accurately simulate transient behavior, and
the separator vessel model has been updated to include vapor hold-up in the overall
material inventory.
The main process control objectives for amine scrubbing are disturbance rejection, set
point tracking, satisfying constraints, and stable operation with process intensification.
Because of the significant material and energy recycle, amine scrubbing is expected to
exhibit multiple time scale behavior, suggesting the need for a hierarchical controller
design.
Future Work
Implement a model for a segment of stripper packing in gPROMS®.
112 112
7
Perform a dynamic validation of the two-stage flash configuration using pilot plant data. Implement a model for a segment of absorber packing in gPROMS
®.
Perform a plantwide model validation using pilot plant data.
Notation
Cp specific heat capacity (kJ/mol∙K)
F molar flowrate (mol/s)
H specific enthalpy (kJ/mol)
M molar hold-up (mol)
mw molecular weight (kg/mol)
P pressure (Pa)
Q heat rate (kW)
R gas constant (J/mol∙K)
T temperature (K)
U internal energy (kJ)
V volume (m3)
x liquid mole fraction
y vapor mole fraction
z bulk fluid mole fraction
Greek
ΔH specific enthalpy of phase change (kJ/mol)
ρ density (kg/m3)
Subscript
des desorption of CO2
i component (CO2, H2O, PZ)
ref reference
sol loaded solution
vap vaporization of H2O
Superscript
in inlet
L liquid effluent
T total hold-up
V vapor effluent
113 113
8
References
Åkesson J, Laird CD, Lavedan G, Prölß K, Tummescheit H, Velut S, Zhu Y. " Nonlinear Model
Predictive Control of a CO2 Post-Combustion Absorption Unit." Chem Eng Tech.
2012:35(3);445–454.
Baldea M, Daoutidis P. Dynamics and Nonlinear Control of Integrated Process Systems. New
York, Cambridge University Press. 2012.
Cohen SM. A Techno-economic Plant- and Grid-Level Assessment of Flexible CO2 Capture. The
University of Texas at Austin. Ph.D. Dissertation. 2012.
Frailie PT, Plaza JP, Van Wagener DH, Rochelle GT. "Modeling piperazine thermodynamics."
Energy Proc. 2011;4:35–42.
Freeman SA. Thermal Degradation and Oxidation of Aqueous Piperazine for Carbon Dioxide Capture. The University of Texas at Austin. Ph.D. Dissertation. 2011.
Kvamsdal HM, Jakobsen JP, Hoff KA. "Dynamic Modeling and Simulation of a CO2 Absorber
for Post-Combustion CO2 Capture." Chemical Engineering and Processing: Process
Intensification. 2009:48(1);135–144.
Kumar A, Daoutidis P. "Modeling, analysis and control of ethylene glycol reactive distillation
column." AIChE J. 1999;45(1):51–68.
Lawal A, Wang M, Stephenson P, Yeung H. "Dynamic Modeling and Simulation of CO2
Chemical Absorption Process for Coal-Fired Power Plants." Computer Aided Chemical
Engineering. 2009:27;1725–1730.
Panahi M, Skogestad S. "Economically Efficient Operation of CO2 Capturing Process PartI:
Self-Optimizing Procedure for Selecting the best Controlled Variables." Chemical
Engineering and Processing: Process Intensification. 2011:50(3);247–253.
Panahi M, Skogestad S. "Economically Efficient Operation of CO2 Capturing Process. Part II.
Design of Control Layer." Chemical Engineering and Processing: Process Intensification.
2012:52;112–124.
Scattolini R. "Architectures for Distributed and Hierarchical Model Predictive Control- A
Review." J Proc Cont. 2009:19;723–731.
Tobieson FA, Hillestad M, Kvamsdal H, Chikukwa A. "A General Column Model in CO2SIM
for Transient Modelling of CO2 Absorption Processes." Energy Proc. 2012:23;129–139.
Walters MS, Dunia RH, Edgar TF, Rochelle GT. "Two-stage flash for CO2 regeneration:
dynamic modeling and pilot plant validation." In: 11th International Conference on
Greenhouse Gas Control Technologies. Kyoto, Japan. 2012.
Xu Q. Thermodynamics of CO2 Loaded Aqueous Amines. The University of Texas at Austin.
Ph.D. Dissertation. 2011.
Ziaii SF. Dynamic Modeling, Optimization, and Control of Monoethanolamine Scrubbing for
CO2 Capture. The University of Texas at Austin. Ph.D. Dissertation. 2012.
114 114
9
Appendix
In Visual Studio 2012®, the following steps should be taken to create a DLL containing the
thermophysical properties of an amine solvent:
1. File → New → Project.
2. Select Win32 Console Application, and change name/location as desired.
3. In the wizard, Next → Select DLL as application type and check empty project → Finish.
4. Right click Source Files → Add → Existing: Add your main .cxx file. This should be a
modified version of foi_demo_cpp.cxx which is provided by PSE. You only need to
change the equations and function names, the shell of the code should not be changed.
3. Increased CO2 loading results in an increased rate of degradation of the tertiary amine
compared to piperazine in PZ-activated tertiary amine solvents.
4. PZ-activated tertiary amine solvents whose tertiary amine has at least one hydroxyethyl
group present lose alkalinity more rapidly than PZ-activated tertiary amine solvents
whose tertiary amine has no hydroxyethyl groups present.
5. The selectivity of DEA over MAE in the degradation of PZ-activated MDEA is greater
than 95%.
References
Freeman SA. Thermal Degradation and Oxidation of Aqueous Piperazine for Carbon Dioxide
Capture. The University of Texas at Austin. Ph.D. Dissertation. 2011.
Rochelle GT et al. "CO2 Capture by Aqueous Absorption, Fourth Quarterly Progress Report
2012." Luminant Carbon Management Program. The University of Texas at Austin. 2013.
134 134
1
Aerosol and Volatile Control in CO2 Capture
Quarterly Report for April 1 – June 30, 2013
by Steven Fulk
Supported by the Texas Carbon Management Program
McKetta Department of Chemical Engineering
The University of Texas at Austin
July 31, 2013
Abstract
A modular Phase Doppler Interferometer (PDI) analyzer was tested at the Post-Combustion
Carbon Capture Center (PC4) on the Pilot Solvent Test Unit (PSTU) at the National Carbon
Capture Center (NCCC) in Wilsonville, Alabama on June 5, 2013. Equipment and technical
support of the analyzer was provided by William Bachalo and Chad Sipperley from Artium
Technologies, Inc. Southern Research Institute (SRI) provided optical access windows, set 140° apart, immediately downstream of the water wash column and a wooden support table for
particle measurement. Carl Landham of SRI coordinated and oversaw the demonstration.
A small, pocket-sized nebulizer was used to demonstrate the efficacy of the transmitter/receiver
system with ad hoc beam expander for a dense particle cloud containing 0.1–20 μm droplets.
The PDI analyzer measured a well-behaved, log-mean particle distribution with a count-mean
diameter of around 5 μm.
The analyzer was then set in place and realigned to measure the droplet size distribution at the
center of the duct exiting the water wash. The PDI analyzer was unable to measure a particle
size distribution due to the high concentration of particles ≤ 1 μm. Though larger particles were
visually present, the dense fog of submicron drops precluded a measureable response above the
background signal.
Artium has suggested that focusing the transmitted lasers to 2–5 μm in diameter (50 μm was
used in this test) would increase signal response at higher particle concentrations; however, the
optical path length will need to be reduced. The Self-Contained PDI system used on the aerosol
growth column would provide ideal conditions for measurement.
The high concentration of submicron particles indicates that coagulation is likely still an
important mechanism for aerosol growth/agglomeration. Furthermore, the contribution of the
aerosol mass to the total mass balance is significant and must be accounted for in simulations.
Particle growth may be inhibited by gas-side mass transfer of amine from the bulk solvent.
Fabrication of the aerosol growth column continued in this quarter. The absorber column, sump,
distributor inlets, and the presaturator vessel have been sent off for fabrication and welding. The
extruded aluminum structural framing and support plates for flanged elements were fabricated
and bolted in place. Power supplies and control/measurement device connections have been
135 135
2
wired. Liquid and gas tubing have been cut and swaged in place where allowable without the
welded pieces for dimensioning.
Introduction
Volatile emissions are a primary concern for CO2 capture plants using amine scrubbers.
Emissions constitute increased economic expense through solvent loss as well as being a source
of potentially hazardous environmental pollutants. Compounds found in treated flue gas include
contaminants from thermal degradation and oxidation as well as combustion byproducts.
Degradation and reaction products have a wide range of toxicity and biodegradation
characteristics which potentially represent unacceptable emissions; as a result, recent work has
focused on estimating volatile losses and assessing their toxicological impact.
Volatile emissions can be reduced through the use of an absorber column using recycled water as
a solvent, called a water wash. Design considerations for water wash systems include liquid
distribution methods to adequately wet packing with small liquid rates, and balancing water in
the absorber/stripper system by adjusting the total volatile concentration in the wash water.
Water wash columns have relatively flat efficiency profiles, meaning the removal efficiency is
not a strong function of either the gas or liquid flow rates or the operating temperature.
Emissions with Aerosols
Recent pilot-plant measurements have shown that normal water wash columns are ineffective at
controlling volatile loss of amine and other pollutants due to the presence of aerosols. In 2011,
MHI presented pilot test results for both KS-1TM
and MEA which showed that emissions were
proportional to inlet SO3 concentration (MHI, 2012). Amine levels out of the wash section were
0.4–23.2 ppmv and 0.8–67.5 ppmv for KS-1TM
and MEA, respectively, for 0–3 ppmv inlet SO3.
Aerosols were visually present at the direct-contact cooler (DCC) and wash outlets. At the
Maasvlakte pilot plant, TNO and SINTEF jointly tested a 30 wt % MEA CO2 capture unit with a
downstream water wash complete with online gas and aerosol phase sampling (TNO, 2012;
SINTEF, 2012). Excessive emissions were observed; aerosols, not physical entrainment, were
responsible for the increase. Lithium and rubidium carbonate (Li2CO3, Rb2CO3) tracers in the
solvent and wash loops verified negligible entrainment. A Brownian demister unit (BDU) was
installed downstream of the wash section which reduced emissions to previously simulated
levels, indicating the bulk of emissions were contained in the droplet phase. Mean droplet
diameters (dDrop) were measured using light extinction coefficients and ranged between 0.76–
7.88 μm at the BDU inlet and 0.2–1.74 μm at the outlet. The quality of the inlet flue gas and the
absolute temperature of the absorber influenced the emission rate. More recently, a baseline
study using MEA at NCCC in Wilsonville, Alabama saw higher amine emissions than expected
(NCCC, 2012). The number of absorber beds (2–3), intercoolers (0–2), and inlet SO3
concentration (1.8 and 3.2 ppmv) were varied as part of a parametric test on emission rate. Their
work concluded that carry-over was proportional to inlet SO3 and also to the concentration of
MEA in the wash water. Emissions were inversely related to absorber temperature. In all
studies, aerosols increased emissions roughly 1–2 orders of magnitude.
It is clear from pilot plant observations and emission studies that removing aerosols is a key part
of reducing possible releases from amine-based CO2 capture plants. The failure of conventional
wash columns and the potential financial impact of particle collectors necessitate fundamental
research to identify more practical means of controlling emissions for large-scale processes.
136 136
3
Understanding interconnectivities of the bulk CO2 removal process operating conditions and
aerosol dynamics can provide the necessary insight required to either design or operate a system
with the intention of suppressing droplet growth; or conversely, to condition aerosols for easier
removal.
Safety
Amine carryover from aerosols represents hazards to plant operation by not only increasing
potential exposure to plant facilitators, but also a risk of upsetting process equipment. Insoluble
amines may precipitate unexpectedly or at a much faster rate than predicted. Clogging of pipes
and process vessels can lead to excess pressure and potential bursts and leakage. Heat tracing
and H2O flushing can resorb precipitated amine in lines.
PDI Test at NCCC
A major goal of this work is to identify the best method for sampling aerosols in a CO2 capture
system. Prior work has shown this task to be difficult, especially when relying on methods more
suitable for either ambient or solid particle sampling. It is expected that with soluble (aqueous)
aerosols that extractive sampling will lead to measurement error due to transmission losses,
evaporation/condensation, and agglomeration. In situ sampling methods can potentially
circumvent these problems; however, most in situ methods, which rely on optical measurement,
have limitations at higher particle concentrations. The PDI system is expected to be the ideal in
situ type measurement for particle sampling since the measurement focuses on the signal of two
coherent crossing beams, rather than an in-and-out type measurement which is more subject to
particle shadowing and intensity losses due to refraction and reflection.
Because of the high cost of a modular PDI system, a cooperative purchasing agreement was
made between UT and Southern Company to procure a PDI system that would work at both the
pilot scale (8” duct) and bench scale (1.5” duct). Pending a successful test at both facilities, the
PDI system would be cost-shared and utilized at both locations. The first test was scheduled at
NCCC to be held in late May or early July of 2013.
NCCC is located at the Gaston Steam Plant in Wilsonville, Alabama next to the Coosa River.
The Gaston Steam Plant consists of 5 units. Primary power production is done in Unit #5 (880
MW). The older Units #1–4 are used during peaking times and scheduled/unscheduled outages
of Unit #5. Flue gas from the boilers is treated with SCR, ESP, and a Chiyoda FGD unit prior to
being sent up the stack. The PC4 test facility takes a slip stream of flue gas off the duct
connecting the Chiyoda scrubber to the Unit #5 stack. A pipe rack runs across the coal delivery
conveyor to a header for the PC4 facility. A fraction of this test gas (0.5 MWe) is used to run the
PSTU pilot plant. Figure 1 shows the stack of Unit #5, the Chiyoda FGD unit, and the draw-off
pipe sent across the facility to the PSTU.
The PDI test was conducted on June 5, 2013. Aerosol size distributions were measured
immediately downstream of the water wash located on the 8th
floor platform. Figure 2 shows the
optical access windows and table setup provided by NCCC. The windows are equipped with N2
flush lines entering the top of the flanged connection to prevent condensation on the windows.
In the event of condensation, liquid can be drained through a small line at the bottom of the
flanged piece. Stainless steel cones were welded onto each flanged piece to protect the windows
from excess condensation and to prevent laser intensity degradation prior to sampling the gas.
137 137
4
These precautionary designs anticipated very dense aerosol plumes. Figure 3 shows the
dimensions of the optical access windows at NCCC.
The CO2 capture system was running a proprietary Chiyoda solvent which had been in service
for approximately 1000 hours in the PSTU. The absorber was operating without intercooling
and a single stage water wash was in use. A thick white fog was visibly present at various
viewports along the absorber column and at the inlet of the water wash. The plume at the sample
location appeared thinner (less opaque) compared to the wash inlet. Additional details of the test
including amine emission levels, CO2 capture rates, and operating conditions were held as
proprietary. Unfortunately, it is unknown how much amine was removed across the demisters in
the wash tower and absorber. Emissions data using an extractive sampling system with an
impinger train was taken on the day of the PDI test, but results are currently unreported.
Figure 1: Flue gas from Unit #5 (880 MW) is drawn off after the Chiyoda SO2 scrubber
(shown in the left picture) and is piped into the PSTU (shown on the right) and other
smaller pilot plants, test skids, and bench-top apparatuses located at PC4. The feed rate to
PSTU is equivalent to a 0.5 MWe scrubbing system (10 tpd CO2 captured).
138 138
5
Figure 2: Optical access windows downstream of the water wash column. The gas flow is
oriented downward. The viewing cones inside of the flanges are flushed with N2 entering
from the top to prevent excessive window condensation. A liquid drain line is available at
the bottom of each flanged window.
The modular PDI system provided by Artium for the June 5 test had an additional beam
expander for the transmitter to increase the beam crossing angle to 28.1° in the vertical plane.
The custom-built beam expander was constructed of adjustable lenses attached to the mounting
rail. Prior to testing on the PSTU, the beams from the transmitter and the receiver had to be
aligned to the designed focal lengths of 350 and 500 mm for the transmitter and receiver,
respectively. A small, pocket-sized nebulizer was used to demonstrate the efficacy of the
transmitter/receiver system with an ad hoc beam expander for a dense particle cloud containing
0.1–20 μm droplets. The PDI analyzer measured a well-behaved, log-mean particle distribution
with a count-mean diameter of around 5 μm. The nebulizer fog was shown to be well within the
measuring capabilities of the PDI.
The analyzer and peripherals were then repacked and lifted by crane to the 6th
floor landing of
the PSTU and hand-carried the rest of the way to the 8th
floor where the water wash and duct
work are located. The equipment was secured on the support with C-clamps and quickly
realigned. Figure 4 shows the PDI setup prior to data collection. The transmitter and beam
expander are located in the foreground and the receiver is in the background. Figure 5 shows a
view of the beam crossing at the center of the duct and the detector output displayed on an
139 139
6
oscilloscope. A closer view of the beam crossing can be seen in Figure 6. The significant haze
surrounding the beams indicates a high concentration of particles scattering light.
Figure 3: Dimensions of the optical access windows on the PSTU. The viewing cones are
designed to accommodate the crossing angles of the transmitter and receiver.
Figure 4: The PDI transmitter (foreground) and receiver (background) mounted to
support rails prior to data collection.
140 140
7
Figure 5: Receiver output to an oscilloscope. The oscilloscope display shows a noisy
baseline for both detectors with no discreet Doppler bursts, indicating a high concentration
of particles ≤ 1 μm.
Figure 6: Close-up view of the transmitted beam crossing. The intersection of the laser
beams is where the gas is sampled for moving particles. 50 μm diameter beams were used
in this test.
141 141
8
Results
The PDI analyzer was unable to measure a particle size distribution inside the water wash outlet
duct due to a high concentration of particles ≤ 1 μm. Droplets and larger aerosols were visibly
present but were not measured by the instrument. The operability of the analyzer was
reconfirmed by measuring the particle size distribution of the nebulizer on the wooden support
structure built around the water wash duct. The test confirmed the same result as previously
measured.
Without emission data from the Chiyoda campaign it is difficult to assess to what extent aerosols
increased amine loss in the absorber and out of the water wash. Though there was a high
concentration of submicron particles, the significant portion of carryover may still reside in
particles ≥ 1 μm in diameter. Fog appeared to increase in opacity moving upwards in the
absorber and into the water wash; however, the fog looked visually thinner at the water wash
outlet. It is possible that significant particle loss occurs in the water wash, but this observation
needs to be confirmed through measurement.
Artium suggested that focusing the transmitted lasers to 2–5 μm in diameter (50 μm was used in
this test) would increase signal response at higher particle concentrations; however, the optical
path length would need to be reduced. The PDI analyzer may be able to measure particle size
distributions on smaller bench-scale systems but would require duct work modification to be
used at the pilot scale. A pilot plant’s exhaust would need to be split with a bypass line
somewhere around 1.5” in diameter to support a smaller and more focused PDI setup.
Conclusions
The modular PDI analyzer with transmitter beam expander was unsuccessful at measuring the
particle size distribution in the outlet duct from the water wash column at the PSTU located at
NCCC in Alabama. A high concentration of aerosols ≤ 1 μm precludes measurement of larger
droplets. However, due to the proprietary nature of the Chiyoda campaign, it is unclear what the
total amine emissions were and to what extent aerosols contributed. Particle collection across
the water wash is also unknown from these measurements.
Aerosols comprise a non-negligible portion of the total emitted amine. Emissions models must
include the mass contained in the aerosolized phase to correctly predict particle growth and,
subsequently, total emissions. The rate of aerosol growth depends on the rate at which amine
can transfer from the bulk liquid, through the bulk gas, and condense on the aerosol. High
concentrations of submicron particles indicate that coagulation may still be a significant
mechanism of aerosol growth throughout the absorber and water wash.
Future Work
Artium has proposed testing the Self-Contained PDI system on a bench-scale absorber with
thinner lasers, roughly 2–5 μm in diameter, to determine if the instrument is capable of
measuring larger particles in the presence of a submicron particle background at high
concentrations. The aerosol growth column represents an ideal test unit. If successful at the
bench scale, the outlet duct of the pilot plant at Pickle Research Campus will be modified for
aerosol measurements during future campaigns.
142 142
9
The aerosol growth column support structure is built and most instrumentation is connected.
Once parts are returned from welding and fabrication, gas and liquid lines can be connected and
controllers will be tested.
Acknowledgements
The authors wish to acknowledge the collaborative effort between NCCC, SRI, and Artium
Technologies undertaken to test the PDI analyzer. NCCC and SRI provided access to their pilot
plant and invested in modifications to their pilot plant to test the PDI analyzer. Artium
Technologies provided test equipment and technical support.
References
Mitsubishi Heavy Industries (MHI). “Amine Emission Control Technology of KM CDR
ProcessTM
.” Presented at the Amine Workshop in Palo Alto, California. August 16, 2011.
National Carbon Capture Center (NCCC). “National Carbon Capture Center: Post Combustion.”
Presented at the 2012 NETL CO2 Capture Technology Meeting. July 10, 2012.
Netherlands Organization for Applied Scientific Research (TNO). “Emission Reducing
Technologies Aerosols.” Presented at UTCCS-1 in Austin, Texas. January 25, 2012.
Rochelle et al. "CO2 Capture by Aqueous Absorption: Fourth Quarterly Progress Report 2012."
Luminant Carbon Management Program, 2013.
SINTEF. “Emission Studies at the Maasvlakte CO2 Capture Pilot Plant.” Presented at UTCCS-1
in Austin, Texas. January 25, 2012.
143 143
1
Amine Degradation in Pilot Plants
Quarterly Report for April 1 – June 30, 2013
by Paul Nielsen
Supported by the Texas Carbon Management Program
McKetta Department of Chemical Engineering
The University of Texas at Austin
July 31, 2013
Abstract
A long-duration pilot plant campaign using PZ was conducted by CSIRO at the Tarong coal-
fired power plant in Australia. After 700 hours of parametric testing, steady state operation was
conducted with stripper sump operating temperatures 120 °C and 155 °C for 420 hours each.
During the 120 °C run, formate and its formamide accumulated at a rate of 0.056 mmol/kg/hr.
This increased to 0.17 mmol/kg/hr after the stripper temperature was raised. The rate of stainless
steel metal ion accumulation due to corrosion also increased significantly from 0.14 to 1.0
μmol/kg/hr when the stripper temperature was raised.
MNPZ in the Tarong solvent reached a steady state of approximately 7 mmol/kg after 7 weeks
with the stripper operating at 120 °C. After raising the stripper temperature to 155 °C the MNPZ
concentration dropped rapidly down towards a new steady state of 2 mmol/kg. Both
observations are in line with what was predicted using the model developed for MNPZ
decomposition.
For PZ cycled from 55 to 150 °C between a thermal and oxidative reactor in the HTCS cycling
apparatus, 75% of the nitrogen loss could be accounted for by the accumulation of ammonia,
formate, FPZ, 2-piperazinol (2-PZOH), ethylenediamine (EDA), volatile loss of PZ, and other
observed degradation products. This is a significant improvement over a previous material
balance done for PZ cycling oxidation in the ISDA, which could only quantify 27% of PZ
decomposition, but which did not measure volatile ammonia loss.
1-methylpiperazine was observed to form from the cycled oxidation of PZ in the HTCS and pilot
plants. This was shown in a bench-scale thermal degradation experiment to be the result of the
reaction and subsequent thermal decomposition of PZ and formaldehyde.
Introduction
Piperazine (PZ) has shown promise as a solvent for carbon dioxide capture, with greater
capacity, absorption rate, and thermal and oxidative stability than the baseline
monoethanolamine (MEA) solvent. However, PZ degradation is not as thoroughly characterized
as MEA, with a significant portion of the mass balance still unidentified (Freeman, 2011).
144 144
2
Tarong long-duration PZ campaign
CSIRO conducted a long-duration pilot plant campaign at the Tarong coal-fired power plant
using 8 m PZ. Liquid samples of lean and rich solvent, wash water, and stripper condensate
were collected weekly and analyzed for degradation product accumulation. A list of samples
received and analyzed is shown below in Table 1. During the first 856 hours of operation,
operating variables such as L/G and stripper operating temperature were varied. After this, the
stripper operating temperature was set to 120 °C and other process variables were held constant
for a long-duration run of 425 hours. The stripper operating temperature was then raised to
155 °C kept constant for another 421 hours.
Table 1: Samples received and analyzed from Tarong pilot plant PZ campaign
Date Hours of operation
11/15/2012 749.1
11/22/2012 781.3
11/29/2012 839.9
12/4/2012 855.6
12/13/2012 927.5 120 °C
12/20/2012 986
1/10/2013 1078.4
1/17/2013 1178.6
1/24/2013 1229.9
2/22/2013 1280.6
2/28/2013 1327.0 155 °C
3/5/2013 1376.6
3/7/2013 1424.6
3/13/2013 1475
3/15/2013 1514.1
3/21/2013 1601.7
3/28/2013 1676
4/3/2013 1701.7
Table 2: Typical flue gas composition at Tarong (average of Line 6 FTIR analysis, 11/15–
1/14)
Water vapor H2O 5.00 vol %
Carbon dioxide CO2 11.90 vol %
Oxygen O2 6.94 vol %
Nitrogen N2 76.16 vol %
Carbon monoxide CO 47.71 ppm
NOx 210.20 ppm
Nitrogen oxide NO 208.92 ppm
Nitrogen dioxide NO2 1.28 ppm
Sulfur dioxide SO2 0.55 ppm
Sulfur trioxide SO3 0.002 ppm
145 145
3
Ammonia NH3 0.49 ppm
Hydrogen chloride HCl 0.32 ppm
Hydrogen fluoride HF 1.79 ppm
Degradation and reclaiming modeling
The Texas Carbon Management Program in coordination with Trimeric and URS has conducted
a review of solvent reclaiming technologies. The project is sponsored by IEAGHG. TCMP’s
contribution to the review is a complete survey and analysis of solvent degradation in order to
determine the composition of both the feed to the reclaimer and the sludge waste products that
must be treated.
Experimental Methods
A full description of the High Temperature Oxidation Reactor (HTOR or HTCS) can be found in
Alex Voice’s recent Ph.D. dissertation (Voice, 2013)
Analytical Methods
All analytical methods used have been discussed in previous quarterly reports (Rochelle et al.,
2013).
Safety: International Shipping of Pilot Plant Samples
Samples from Tarong in Australia were shipped in 30+ mL vials. The vials were sealed with
tape, and were then placed in sealed bags and packed with packing material in cardboard boxes
before being shipped. Upon arrival in Austin, TX, the samples were opened inside a fume hood
to vent any vapor buildup that may have occurred. No samples leaked in transit.
Results and Discussion
Tarong High Temperature PZ Campaign Results
Figure 1 shows the accumulation of formate, stainless steel metal ions (SSM), 2-piperazinol (2-
PZOH), and ethylenediamine (EDA) during the long-duration runs conducted at Tarong. The
rate of formate accumulation increased from 0.056 to 0.166 mmol/kg/hr when the stripper
temperature was raised from 120 °C to 155 °C. The rate of SSM accumulation from corrosion
follows a similar trend. The SSM ions may be catalyzing degradation, the formate accumulation
may be catalyzing corrosion, or there may be a synergistic effect between the two trends. 2-
PZOH and EDA are intermediate degradation products and do not show any long term trends.
146 146
4
Figure 1: Accumulation of total formate, stainless steel metal ions, 2-piperazinol (2-PZOH),
and ethylenediamine (EDA) during the 120 and 155 °C long-duration tests
Figure 2 shows the concentration of MNPZ measured in the solvent, as well as the concentration
predicted by the MNPZ degradation model developed by Fine and described in Equations 1 and
2. The observed accumulation matches the model prediction very closely (Fine et al., 2013).
(1)
( )
(2)
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0
20
40
60
80
100
120
700 900 1100 1300 1500 1700
mm
ol/
kg s
tain
less
ste
el m
etal
ions
(SS
M)
mm
ol/
kg
Operating hours
EDA
2-PZOH
Total formate
SSM ions
120 °C 155 °C
147 147
5
Figure 2: Measured and predicted concentration of MNPZ in lean and rich solvent
samples. Model parameters: NO2 in the flue gas yNO2 = 1.2 ppmv, residence time in the