Rochelle Q2 Report 2009

1
CO
2
Capture by Aqueous Absorption
Summary of 2nd Quarterly Progress Reports 2009
Supported by the Luminant Carbon Management Program
and the
Industrial Associates Program for CO
2
by Gary T. Rochelle
Department of Chemical Engineering
The University of Texas at Austin
August 1, 2009
Introduction
This research program is focused on the technical obstacles to the deployment of CO
2
capture
and sequestration from flue gas by alkanolamine absorption/stripping and on integrating the
design of the capture process with the aquifer storage/enhanced oil recovery process. The
objective is to develop and demonstrate evolutionary improvements to monoethanolamine
(MEA) absorption/stripping for CO
2
capture from coal-fired flue gas. The Luminant Carbon
Management Program and the Industrial Associates Program for CO
2
Capture by Aqueous
Absorption support 15 graduate students. These students have prepared detailed quarterly
progress reports for the period April 1, 2009 to June 30, 2009. We have also attached
Powerpoint presentations by Freeman, and Nguyen and posters by Chen and Plaza presented at
the Trondheim conference in June 2009. The presentation made at Trondheim by Gary Rochelle
will be sent out separately.
Conclusions
CO
2
solubility and absorption rate for three amine solvents, AEP, EDA, and MDEA/PZ, were
measured with WWC in this quarter and compared to those amines tested previously. 6 m AEP
has a capacity close to 7.7 m HEP but less than 8 m PZ. 12 m EDA and 7 m MDEA/2 m PZ
both have a capacity similar to 8 m PZ. In terms of absorption rate at 5 kPa, MDEA/PZ
performs even a little better than PZ, while EDA is not an attractive solvent. 6m AEP absorbs
CO
2
at a rate less than half of the rate of 8 m PZ. Heat of CO
2
absorption of 12 m EDA is about
80 kJ/mol at average operational CO
2
loading, close to that of 7 MEA. AEP has a similar heat of
absorption as seen for 7.7 HEP.
We have successfully modified the PZ model developed by Hilliard in Aspen Plus
to give a
better fit of recent VLE, heat capacity, and heat of absorption data by adjusting the PZ dielectric
constant along with other parameters.
The calculation of the theoretical difference between the heats of absorption at absorber and
stripper temperatures verifies that the heat of absorption at 40 C should only be approximately
3.2 kJ/mol CO
2
lower than the value at 120 C. This calculation raises concerns about the
accuracy of the available data for heat of absorption.
1
2
Thermal degradation rates for 7 m MDEA are similar to those for 7 m MDEA/2 m PZ,
suggesting that PZ plays a secondary role in the overall degradation process in 7 m MDEA/2 m
PZ.
Oxidation of PZ with low gas flow provides poor water balance and repeatability. This
experimental method needs enhancement to provide adequate quantification of a system that
oxidizes slowly, producing few degradation products.
Mass spectrometry has given the masses of numerous products from numerous masses the
thermal degradation of PZ. These have not been positively identified, but possible candidates
include N-formyl PZ, N-(hydroxymethyl) PZ, 1-methyl PZ, and N,N-diformyl EDA.
Initial results on PZ solutions show that thermally degraded PZ corrodes 316 stainless steel less
than 7 m MEA. After 18 weeks at 150 C, the iron and nickel concentration in an 8 m PZ
solution were 0.9 and 0.7 mM, respectively. For a 7 m MEA experiment at 135 C, the final
concentrations of iron and nickel were 13.7 and 4.2 mM, respectively, after four weeks.
Concentrated, aqueous PZ solutions oxidized 3 to 5 times slower than 7 m MEA in systems with
iron, copper, or stainless steel metals (chromium, nickel, and iron). The thermal degradation rate
in concentrated PZ systems is 23 to 70 times less than 7 m MEA systems.
Thermal loss of 8 m EDA (0.4 loading) is about 35% at 135
o
C after 4 weeks, but only 10% at
100
o
C after 16 weeks. EDA urea is the most concentrated product, representing only 20% of the
lost EDA.
Oxidation of EDA at 55
o
C with 2%/98% CO
2
/O
2
consumed 10% after 168 hours with 1mM
Fe
2+
. However EDA loss was negligible after 168 hours with 100 mM inhibitor A.
Foaminess of 8 m EDA (0.4 loading) is comparable to 7 m MEA (0.4 loading) with or without
formaldehyde.
With an equilibrium CO
2
partial pressures range of 500 to 5000 P at 40
o
C, the lean loading of 8
or 12 m EDA is greater than 0.4 and the rich loading is less than 0.5, giving a capacity with 12 m
EDA of 0.78 moles CO
2
/kg EDA + H
2
O.
Amine partial pressure of 12 m EDA is comparable with 7 m MEA and 8 m PZ at operating
conditions.
CO
2
absorption rate in 12 m EDA is about 50% of that of 7 m MEA at rich conditions.
The viscosity of 8 m EDA is a little less than 8 m PZ solution and comparable with other amines.
The diffusion coefficient of total dissolved CO
2
in 7 to 13 m MEA and 2 to 8 m PZ varies with
the 0.72 power of the viscosity.
The CO
2
absorption rate in MEA showed no significant effect of temperature or amine
concentration on k
g
.
Most PZ experiments did not show an effect of temperature or amine concentration on k
g
.
Some data points have reduced k
g
values but the reduction seems to correlate with the
possibility of these conditions being limited by diffusion of reactants and products near the gas-
liquid interface.
PZ absorbs CO
2
2 to 3 times faster than MEA over the applicable CO
2
capture range for coal-
fired power plants.
2
3
The pseudo first order expression for k
g
was modified to raise the amine activity to the 2
nd

power and include activity coefficients of MEA and CO
2
in loaded MEA solutions. The rigorous
expression was shown to completely account for the negligible deviations in k
g
with changing
temperature and amine concentration.
The K
g
calculated from the pilot plant results concentrated piperazine are consistent with the
values of k
g
obtain by Dugas in the wetted wall column. The pilot plants rates for 8 m Pz are 4
times faster than 7 m MEA and 2 times faster than 9 m MEA.

Total pressure was measured in CO
2
-loaded solutions at 100 to 150
o
C with the following results:
7.4 to 25.8 bar (147
o
C), 8 m PZ, 0.465 loading
1.1 to 7.2 bar, 5 m PZ, 0.293 loading
1.1 to 6.7 bar, 7 m MEA, 0.316 loading
3.3 to 15.9 bar, 7 m MEA, 0.479 loading

The data for PZ at 40 to 190 C are represented well by the empirical relationships:
2
2
1
ln 38.4 ( 102, 000 / ) 20.6 13, 200 3.23
CO
P J mol
RT T
= + + +
ln
( 102, 000 13, 200 )( / )
1
( )
P
H R R J mol
T
= = +

The data for MEA at 40 to 160
o
C are represented well by the empirical relationships:
2
2
1
ln 44.2 ( 116, 000 / ) 29.7 11, 600 17.3
CO
P J mol
RT T
= + + +
ln
( 116, 000 11, 600 )( / )
1
( )
P
H R R J mol
T
= = +

Dynamic absorber modeling gives steady state response time of 4 to 5 minutes for alternative
strategies of load reduction.
In 7 m MEA with 0.1 Fe
++
/50 mM A, the NH
3
production rate increased from 1 to 3 mM/hr as T
increased from 54 to 64
o
C.

EDA oxidizes at 1 mM/hr ammonia with 0.1 mM Fe. 5 mM Cu produces roughly 4 times the
degradation rate as 0.11 mM Fe.
11 m MEA oxidizes to ammonia at a 2 times higher rate than 7 m MEA.
DGA, AMP, and PZ do not oxidize to significant quantities of ammonia or heat stable salts.
Formate reacts with MEA at 135
o
C to form n-formyl-ethanolamine in less than 24 hours.
1. Wetted Wall Column Rate Measurements p. 13
by Xi Chen
The CO
2
solubility and adsorption/desorption rate were measured in the wetted wall column for
6 m 1-(2-Aminoethyl)piperazine (AEP), 12 m Ethylenediamine (EDA), and 7 m MDEA/2 m PZ
blend with varied CO
2
loading (mol CO
2
/mol alkalinity). VLE models of CO
2
were regressed
from experimental data to calculate CO
2
capacity and enthalpy of CO
2
absorption (H
abs
). The
3
4
liquid film mass transfer coefficients (kg) and CO
2
partial pressures (P*) obtained are compared
to those of 8 m piperazine (PZ) and 7 m monoethanolamine (MEA) as well as other amines
studied in the previous quarter.
2. Influence of Liquid Properties on Effective Mass Transfer Area of
Structured Packing p. 27
by Robert Tsai
(also supported by the Separations Research Program)
High viscosity tests (approximately 10 cP) were completed with Sulzer Mellapak 250X
(M250X). In addition, two new Sulzer structured packings (Mellapak 125Y (M125Y) and
MellapakPlus (MP252Y)) were evaluated and compared with Mellapak 250Y (M250Y).
Dry and pre-loading pressure drops for the packings were ordered: M250X (0.4) < M125Y (0.5)
< MP252Y (0.7) < M250Y. Pre-loading liquid hold-ups were ranked as follows: M125Y (0.6) <
M250X (0.9) ~ MP252Y (0.9) < M250Y. (The numbers in parentheses represent the values
relative to M250Y.) Increasing the solution viscosity had little effect on pre-loading pressure
drops but significantly reduced the capacities of the packings. Higher hold-ups were also
associated with the more viscous solutions.
The mass transfer areas of M250Y, M250X, and MP252Y were equivalent, ranging from
roughly 0.651.1 on a fractional basis (a
e
/a
p
) over liquid loads from 130 gpm/ft
2
. M125Y
exhibited higher fractional areas (0.71.2) than M250Y. For all of the packings, a reduction in
surface tension (30 dynes/cm) increased the mass transfer area by 10%. A ten-fold viscosity
enhancement had no appreciable impact on the area. Both findings were consistent with
previous results.
The mass transfer area database was updated, and the current global (a
e
/a
p
) correlation, able to
represent the entire database within limits of 15%, is displayed below:
( )( ) [ ]
116 . 0
3
1
L L
p
e
334 . 1

= Fr We
a
a

A global pre-loading liquid hold-up model was developed for structured packing. The
correlation (shown below) is accurate within approximately 25% with respect to the
experimental data.
( )( ) [ ]
72 . 0
3
2
P L L
84 . 21

= Ga Re h
A basic Excel model was created to evaluate the economics of an amine scrubber (absorber) as a
function of gas throughput and column configuration. The minimum cost was calculated to be
$57/tonne CO
2
for absorber capacities in the 100500 MW range.
3. Modeling Stripper Performance for CO
2
Removal p. 49
by David Van Wagener
Since Hilliard developed thermodynamic models for various amine solvents, additional
experimental data have been collected at new conditions. The data primarily of interest have
been for concentrated piperazine (PZ). The Hilliard model performed well for low
concentrations, 0.9 m5 m, but 8 m PZ will be used in future simulations. VLE data collected by
Dugas as well as heat capacity data collected by Nguyen for concentrated piperazine were
incorporated into previous parameter regression files. Additionally, heat of absorption data were
4
5
collected by Freeman. The parameters to be regressed were reconsidered, and more focus was
put on the heat capacity parameters of the dominant species at relevant loadings. This quarter
the dielectric constant of PZ was also included. The original value used by Hilliard was for
piperidine, a molecule similar to PZ with one amine group instead of two. Including the
dielectric constant in the regression greatly improved the fit, and the new value of the dielectric
constant is between that of piperidine and MEA. The theoretical difference between the heat of
absorption for 8 m MEA at 40 C and 120 C was calculated using an energy balance, and the
difference was found to be approximately 3.2 kJ/mol CO
2
. This difference is much smaller than
what is observed in the experimental data, so the data collection for heat of absorption should be
reevaluated.
4. Solvent Management of MDEA/Piperazine p. 61
by Fred Closmann
(also supported by the Process Science & Technology Center)
A thermal degradation experiment was conducted on 7 m MDEA in the second quarter. The
compounds dimethylaminoethanol (DMAE), diethanolamine (DEA), N,N-dimethyl ethanamine
(DMEA), dimethyl piperazine, 1-(2-hydroxyethyl)-4-methylpiperazine (HMP), and
triethanolamine (TEA) were tentatively identified in degraded solvent samples through ion
chromatography (IC) and IC-mass spectrometry (IC-MS) methods. We calculate an activation
energy for the degradation of MDEA of approximately 47 kJ/gmol, and rates of degradation of
MDEA of 66 and 112 mmolal/day were calculated at 150 C and loadings of 0.1 and 0.2 moles
CO
2
/mole alkalinity, respectively. Proposed degradation mechanisms for the MDEA loss
include the protonation of MDEA, the formation of an MDEA-carbamate, and the
disproportionation of MDEA. All three mechanisms are followed by subsequent reactions and
could involve PZ when present in a solvent blend. However, because the MDEA loss rate is
similar to rates calculated for MDEA in 7 m MDEA/2 m PZ, we believe the role of PZ in the
degradation of the solvent is of lesser importance.
The construction of the integrated solvent cycling/degradation apparatus is near completion, and
basic temperature metrics for amine cycling have been met. Modifications to the system to
achieve accurate temperature measurement are ongoing. Amine cycling degradation
experiments will be completed in 3
rd
quarter 2009.
5. Solvent Management of Concentrated Piperazine p. 71
by Stephanie Freeman
The Cation IC-Mass Spectrometer was used to identify numerous masses of unidentified
degradation products in 8 m PZ thermal degradation experiments. Although not yet positively
identified, possible candidates include N-formyl PZ, N-(hydroxymethyl) PZ, 1-methyl PZ, and
others. Further work is needed to provide positive identification.
Quantification of heavy metals in solution has been used to analyze the corrosion of 316 stainless
steel. Initial results show that thermally degraded PZ solutions contain less metal than 7 m MEA
solutions. After 18 weeks at 150 C, the iron and nickel concentration in an 8 m PZ solution
were 0.9 and 0.7 mM, respectively. For a 7 m MEA experiment at 135 C, the final
concentrations of iron and nickel were 13.7 and 4.2 mM, respectively, after four weeks. Further
work is needed to understand if the amines themselves or their degradation products are
responsible for this corrosion.
5
6
in concentrated PZ systems is 23 to 70 times less than 7 m MEA systems.
6. Ethylenediamine as a solvent for CO
2
capture p. 97
by Shan Zhou
Thermal and oxidative degradation products of 8 m EDA (ethylenediamine) solution were
measured by cation IC, anion IC, HPLC, and TIC. Foaming was measured in samples from
oxidative degradation. Vapor liquid equilibrium and amine volatility of 8 m and 12 m EDA
solution were measured using hot gas FTIR. CO
2
solubility and absorption/desorption rate were
measured in the wetted wall column. Viscosity of 8 m and 12 m EDA was measured at different
temperatures.
About 35% of the EDA was lost after 4 weeks at 135
o
C. Only about 10% of EDA degraded
after 16 weeks at 100
o
C. EDA urea was the most concentrated product from thermal
degradation, representing about 20% of the degraded EDA.
10% of the EDA was oxidized after 168 hours with 1 mM Fe
2+
at 55
o
C. With 100 mM inhibitor
A the oxidation of EDA was insignificant. DETA (diethylenetriamine) and formate were the
most concentrated in the quantified oxidative degradation products of EDA.
The foaminess and foam break time of EDA solution are close to that of 7 m MEA solution,
much better than that of 8 m PZ.
Although EDA is volatile at low CO
2
loading, 8 m and 12 m EDA solutions have amine partial
pressure comparable to 7 m MEA and 8 m PZ in expected lean loading. However, the CO
2
flux
normalized by partial pressure driving force of 12 m EDA solution was lower than that of 7 m
MEA at the same conditions. VLE models were regressed with experiment data from the wetted
wall column. CO
2
2
absorption were calculated. The H
absorption
of
EDA is similar to MEA at operating conditions, higher than most of the other amine systems.
7. Rate Measurements for MEA and PZ p. 124
by Ross E. Dugas
This draft chapter on experimental results from the dissertation includes diaphragm cell
experiments to evaluate diffusion coefficient behavior in 713 m MEA and 28 m PZ. Measured
diffusion coefficients varied to the 0.72 power of the viscosity.
This chapter also includes equilibrium CO
2
partial pressure and rate data for 713 m MEA, 212
m PZ and 7 m MEA/2 m PZ experiments in the wetted wall column at 40, 60, 80, and 100 C.
All of the CO
2
partial pressure measurements matched literature sources very well with the
exception of a few data points near 0.5 loading in MEA solutions. Rate data for MEA and PZ
generally showed no significant effect of temperature or amine concentration on k
g
. Some PZ
data points did show reduced k
g
values but the reduction seems to correlate with the possibility
of these conditions being limited by diffusion of reactants and products near the gas-liquid
interface. Over the applicable CO
2
capture range for coal-fired power plants, PZ was shown to
absorb CO
2
23 times faster than MEA.
The pseudo first order expression for k
g
was modified to a more rigorous form including activity
coefficients. Activity coefficients of MEA and CO
2
in loaded MEA solutions were correlated
6
7
from the literature. The wetted wall column data suggested the amine activity to have a 2
nd
order
dependency. Incorporating these changes into the pseudo first order expression completely
explains the negligible deviations in k
g
with changing temperature and amine concentration.
8. CO
2
Absorption Modeling Using Aqueous Amines p. 152
by Jorge M. Plaza
A CO
2
absorber model for the 8 m PZ solvent is under development. It uses a modified version
of the Hilliard (2008) thermodynamic representation by Van Wagener, version 02/06/09
(Rochelle et al., 2009)). It includes a reduced reaction set based on the more relevant species
present at the expected operating loading (0.30.4 mol CO
2
/mol alkalinity). Kinetics will be
based on Cullinane (2005) and regressed to include data generated by Dugas (Rochelle et al.,
2008a). Initial work accesses proper representation of solvent physical properties. Density and
viscosity were regressed to match experimental data generated by Freeman (Rochelle et al.,
2009; Rochelle et al., 2008b). The activity coefficient of CO
2
was also examined and compared
to values found in Cullinane (2005) as a function of amine concentration and loading.
Density and viscosity match the experimental data with a deviation maximum of 10%. The
model represents correctly the expected change in the activity coefficient of CO
2
with amine
concentration and its values are consistent with the available low concentration experimental
data. However, the effect of loading is not represented correctly. This issue needs to be
addressed. Kinetics will be implemented once this issue is resolved.
The overall gas-side mass transfer coefficient (K
g
) was calculated for the November 2008 PZ
pilot plant data and compared to k
g
data by Dugas (Rochelle et al., 2008a) and previous pilot

plant campaigns. Results show a higher K
g
for 5 m PZ.
9. Total Pressure Measurement of CO
2
Loaded Aqueous Amines at
High T and P p. 165

by Qing Xu, Martin Metzner
In this quarter a series of total pressure measurements were conducted with CO
2
loaded
monoethanolamine (MEA) or piperazine (PZ) at 100 to 160
o
C (for MEA) or 190 C (for PZ). A
500 mL 316 SS autoclave was used as the equilibrium cell. The total pressure of 8 m PZ with
0.465 CO
2
loading varied from 7.4 to 25.8 bar at 100 to 147 C. At 100 to 150 C, for 5 m PZ
with 0.293 CO
2
loading P
t
varied from 1.1 to 7.2 bar; for 7 m MEA with 0.316 CO
2
loading P
t
is
from 1.1 to 6.7 bar; for 7 m MEA with 0.479 CO
2
loading P
t
is from 3.3 to 15.9 bar.
The partial pressure of CO
2
at each experimental condition was estimated by subtracting partial
pressures of water and amines. The calculated results for 7 m MEA match well with the work by
Jou et al. (1995). The regression based on data from 40 to 190 C gives empirical models for
CO
2
partial pressure over loaded aqueous PZ and MEA respectively and the models predict the
data well. Heat of absorption for CO
2
loaded aqueous PZ and MEA was calculated from these
empirical models.
For PZ:
2
2
1
ln 38.4 ( 102, 000 / ) 20.6 13, 200 3.23
CO
P J mol
RT T
= + + + (1)
7
8
ln
( 102, 000 13, 200 )( / )
1
( )
P
H R R J mol
T
= = +
(2)
For MEA:
2
2
1
ln 44.2 ( 116, 000 / ) 29.7 11, 600 17.3
CO
P J mol
RT T
= + + + (3)
ln
( 116, 000 11, 600 )( / )
1
( )
P
H R R J mol
T
= = +
(4)
10. Volatility and heat capacity of amine alternatives p. 187
by Bich-Thu Nguyen
Amine volatility is a crucial screening criterion which affects fugitive emission and necessitates
appropriate water wash design. At a lean CO
2
partial pressure of 500 Pa at 40 C, the ranking of
amine volatility is: 7 m MDEA/2 m PZ (7/6 ppm) < 12 m EDA (9 ppm) < 8 m PZ (14 ppm) < 7
m MEA (28 ppm) < 5 m AMP (112 ppm). The less volatile amines appear to have higher heat of
amine solution than the more volatile amines. There is no apparent correlation between volatility
and the amine heat of desorption.
11. Oxidative Degradation of MEA p. 203
by Alex Voice
Oxidative degradation experiments in the high gas flow (HGF) apparatus continued this quarter.
An experiment was conducted to determine the sensitivity of the oxidative degradation rate of
7 m MEA under kinetically limited conditions in the high gas flow apparatus on various
operating conditions. Temperature was by far the most important parameter, showing a 100
500% change in ammonia production for a 10 C temperature change.
Amine screening for oxidative degradation rate was also conducted in the HGF. Ethylene
diamine (EDA, 8 m) degraded at a rate of 1 mM/hr for 0.11 mM Fe as evidenced by NH
3

production. Addition of 1 mM Fe produced a burst of 0.15 mM of NH
3
, although the steady
state rate remained unchanged. 4.8 m AMP and 17.7 m DGA did not degrade to produce NH
3
in
significant quantities. AMP did not produce significant quantities of heat stable salts. The
presence of other degradation products, including NO
2
and formaldehyde, had a low signal to
noise ratio in the FTIR spectrum, and will require verification. 8 m piperazine did not produce
NH
3
or heat stable salts in the HGF. 11 m MEA produced ammonia at a rate of 4.1 mM/hr, as
compared with 1.8 mM/hr for previous 7 m experiments conducted by Sexton.
Formic acid and 7 m MEA at 0.4 loading reacted rapidly to produce hydroxyl-ethyl-formamide
at 135 C. Formate and formamide concentrations were stable after 24 hours (the first sample
taken). The formate to formamide ratio was 1:1 in a solution which initially contained only
formic acid. In the solution loaded with formic acid and formaldehyde, the ratio was 2:1.
Formate was also detected in the solution containing only formaldehyde at a concentration of
22 mM, well above the ~1 mM formate detected in the neat solution.
The metal dissolution rate (metal concentration divided by time) in thermal degradation
experiments was found to be a clean Arrhenius function of T.
8
9
12. Dynamic Operation of CO
2
Capture p. 216
by Sepideh Ziaii
The dynamic model of the absorber created in ACM has been run at the design and operating
conditions of a pilot plant run with 9 m MEA. The steady state results give 58.8% CO
2
removal,
which is 1.7% less than a reconciled value (59.9%). The temperature profile is not completely
matched with experimental data because of inaccurate calculation of heat of absorption for 9 m
MEA.
In addition, the dynamic model of the absorber has been run for two stripper ratio control
strategies in flexible CO
2
capture. The CO
2
removal, rich loading, and temperature profiles were
calculated for each strategy and overall dynamic and steady state behavior were compared.
13. Electric Grid Level Implications of Flexible CO
2
Capture Operation
p. 223
by Stuart Cohen
A flexible carbon dioxide (CO
2
) capture system with large-scale solvent storage allows
continuous high CO
2
removal from flue gas while power output is increased by turning stripping
and CO
2
compression systems to partial- or zero-load. In such a system, the basic tradeoff is that
longer solvent storage times provide more opportunity to increase plant output when electricity
prices are high, but at the expense of additional energy cost to regenerate stored solvent when
electricity prices are low as well as additional capital costs of solvent inventory, storage tanks,
and larger stripping, compression, and auxiliary equipment.
For storage times of 112 hours using monoethanolamine (MEA) solvent, 10 million kg MEA,
10100 m
3
storage capacity, and $10-$100 million additional capital costs are required. Storage
tanks are a relatively insignificant capital cost relative to that of solvent inventory and larger
stripping/compression equipment. A preliminary optimization study finds that solvent storage of
a few hours or more is attractive when the electricity market experiences large differences
between high and low electricity prices, even if only in the form of infrequent price spikes.
Lower capital costs can improve the economics of a solvent storage system, but capital cost
reductions are insufficient to promote installation of a solvent storage system without a favorable
electricity price distribution. However, 2008 annual average electricity prices in the Electric
Reliability Council of Texas (ERCOT) along with base case parameters and the preliminary
modeling methodology do not provide the conditions required for solvent storage to be desirable.
14. Modeling Absorber/Stripper Performance with MDEA/PZ p. 237
by Peter Frailie
The goal of this study is to evaluate the performance of an absorber/stripper operation that
utilizes the MDEA/PZ blended amine system. Due to the complexity of this system the model
will be developed in several smaller, more manageable parts that can later be combined to form
the final model. The first section that will be developed is an MDEA/PZ model based on
thermodynamic data, which must initially be developed as separate MDEA and PZ models.
Once the MDEA/PZ model has been completed it must be incorporated into separate absorber
and stripper models similar to those developed by Van Wagener and Plaza. Those models can
then be combined to form the final MDEA/PZ absorber/stripper model. This study is currently
in the process of developing the MDEA/PZ model based on thermodynamic data. Over the next
9
10
three months the thermodynamic model should be completed and work should have begun on the
absorber and stripper models.
15. Measurement of Packing Liquid Phase Film Mass Transfer
Coefficient p. 244
by Chao Wang
Packings are widely used in distillation, stripping and scrubbing processes because of their
relatively low pressure drop, good mass transfer efficiency, and ease of installation. Packings
are also are also being investigated for the post combustion carbon capture process for these
reasons. Research continues to focus on development of high performance packing, especially
on minimizing pressure drop, maximizing mass transfer efficiency, and minimizing costs. The
design of packed absorbers for carbon dioxide capture will require the reliable measurement and
accurate prediction of the effective area, gas and liquid film mass transfer coefficient. A variety
of experimental methods for measuring effective area, gas and liquid film mass transfer
coefficient k
L
a have been reported. Consistent measurements of these important design
parameters will begin this summer.
Absorption of CO
2
with NaOH is applied to measure the effective area of packings.
Atmospheric carbon dioxide in air is used as gas phase and 0.1 M NaOH is used as liquid phase.
This is a liquid phase controlled mass transfer system so the liquid phase mass transfer
coefficient k
l
or often referred to as k
g
can be assumed as the overall mass transfer coefficient

K
G
. In the proposed summer work, the gas flow rate set points are 180, 300, 450 ACFM and
liquid flow rate set points are 1, 2.5, 5, 7.5, 10, 15 and 25 gpm/ft
2
(same operating conditions
explored by Robert Tsai). The effective area can then be calculated by the equationl:
RT Zk
y
y
u
RT ZK
y
y
u
a
g
out CO
in CO
G
G
out CO
in CO
G
e
'
2
2
2
2
) ln( ) ln(
= .
Absorption of SO
2
with NaOH is applied to measure the gas phase mass transfer coefficient.
Sulfur dioxide (SO
2
), blended with ambient air at a composition of approximately 80 ppm, will
be absorbed by 1 M NaOH solution. The reaction is instantaneous and the mass transfer process
is controlled by the gas phase. Thus the overall mass transfer coefficient K
G
can be replaced by
gas phase mass transfer coefficient k
G
. This experiment can be combined with the effective area
experiment as long as the gas and liquid flow rates are set at the same level. The gas phase mass
transfer coefficient can be calculated by the equation:
e
out SO
in SO
G
G
ZRTa
y
y
u
k
) ln(
2
2
=
Desorption of toluene in water with air is applied to measure the liquid phase mass transfer
coefficient. Ambient air is used to strip toluene from water. As a result of the high Henrys
constant, the mass transfer resistance is controlled by the liquid phase. The gas flow rates and
liquid flow rates are set at the same value with the effective area measurement to make the 3
experiments consistent. The liquid phase mass transfer coefficient can be calculated by the
equation:
) / ln(
2 1 LA LA
e
L
L
c c
Za
u
k =
10
11
16. Pilot Plant Testing of Advanced Process Concepts using
Concentrated Piperazine p. 254
by Eric Chen
Pilot plant testing of 8 m piperazine in a two-stage heated flash is planned for Fall 2009.
Substantial modifications to the existing pilot plant at SRP will be needed. Process flow
diagrams (PFD) and piping and instrument diagrams (P&ID) have been developed for the new
process. Preliminary specifications for the high pressure pump, cross-exchanger, and steam
heaters have been developed and are in the process of being formally designed and quoted by
various vendors.

Attachments

1. Degradation of Concentrated PZ in CO
2
capture p. 262
by Stephanie A. Freeman, et al.

2. Amine Volatility p. 281
by Thu Nguyen, et al.

3. Foaming Behavior of Piperazine Aqueous Solutions for CO
2
Capture
(poster) p. 298
by Xi Chen, et al.

4. Absorber Intercooling in CO
2
Absorption (poster) p. 299
by Jorge M. Plaza, et al.

11

12
1
Wetted Wall Column Rate Measurements

Quarterly Report for April 1 June 30, 2009
by Xi Chen
and the
2
July 1, 2009
Abstract
The CO
2
solubility and adsorption/desorption rate were measured in the wetted wall column for
6 m 1-(2-Aminoethyl)piperazine (AEP), 12 m Ethylenediamine (EDA) and 7 m MDEA/2 m PZ
blend with varied CO
2
loading (mol CO
2
/mol alkalinity). VLE models of CO
2
were regressed
from experimental data to calculate CO
2
2
absorption (H
abs
). The
liquid film mass transfer coefficients (kg) and CO
2
partial pressures (P*) obtained are compared
to those of 8 m piperazine (PZ) and 7 m monoethanolamine (MEA) as well as other amines
studied in the previous quarter. The capacity of AEP is about 15% less than that of PZ. EDA
and MDEA/PZ both have a similar capacity as PZ. The rate of CO
2
absorption for AEP is less
than half of that for PZ. kg for EDA is only about one third of kg for PZ. MDEA/PZ absorbs
CO
2
at a slightly higher rate than PZ at rich loading, about twice as fast as MEA. EDA has a
much higher H
abs
(80 kJ/mol) than AEP (72 kJ/mol), PZ (70 kJ/mol), or MDEA/PZ (67 kJ/mol)
Introduction
More amine solvents were tested with the wetted wall column (WWC) in this quarter. The
names and chemical structures of amines that have been studied so far are shown in Table 1.
AEP is a derivative of piperazine. With the additional amino group, AEP is expected to have a
greater CO
2
capacity than PZ while still having a high absorption rate. EDA is known for its fast
kinetic rate of reaction with CO
2
at zero or low loading. MDEA/PZ has been reported to have a
good CO
2
capacity and reasonably high reaction rate (Bishnoi, 2000). More measurements have
been done to complete the study on MDEA/PZ.
Table 1: Name and chemical structure of the amines screened in this work
Name Chemical structure
N-(2-hydroxyethyl)piperazine
(HEP)
N
N H
OH

1-(2-Aminoethyl)piperazine
(AEP)
N
N H
NH
2

13
2
2-amino-2-methyl-1-propanol
(AMP)
O H
NH
2
C H
3
CH
3

2-piperidineethanol
(2-PE)
N
H
OH

Ethylenediamine
(EDA)
NH
2
N H
2

Methyldiethanolamine (MDEA)
/Piperazine (PZ)
O H
N
OH
CH
3

NH N H

Experimental Methods
Experimental apparatus, procedure, and analytical methods have been described in previous
reports and will not be repeated here. AEP (99%, Acros), EDA (Labgrade, Fisher), MDEA
(95%99%, Huntsman), and PZ (99%, Alfa Aesar) were used without further purification.
Viscosity measurements were also done for all the amine solutions. The method has been
described by Stephanie Freeman (Rochelle, 2008a). Viscosity under shear rate from 100 s
-1
to
1000 s
-1
was measured and the average value will be reported.
Results and Discussion
CO
2
partial pressure
Equilibrium CO
2
partial pressure is plotted against CO
2
loading for each amine, as shown in
Figures 1, 2, and 3. The data is also tabulated in Table 2. A semi-empirical model was regressed
from solubility data and plotted in solid lines in each figure:
2
/ / ln + + + + = e T d c T b a P (1)
14
3
0
20
40
60
80
100
0.001
0.01
0.1
1
10
100
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
H
e
a
t
o
f
a
b
s
o
r
p
t
i
o
n
(
k
J
/
m
o
l
)
P
*
(
k
P
a
)
CO
2
Loading(mol/molalkalinity)
40C
60C
80C
100C

Figure 1: CO
2
solubility and heat absorption as a function of CO
2
loading for 6 m AEP.
The CO
2
solubility data for 6 m AEP are shown in Figure 1 and Table 2. The gap between lines
for different temperatures indicates the value of heat of absorption. As CO
2
loading increases,
the lines get closer to each other, in line with a decrease in heat of absorption (the orange line in
the figure).
15
4
0.001
0.01
0.1
1
10
100
0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55
P
*
(
k
P
a
)
CO
2
40C
60C
80C
100C
MEA@100C(Hilliard
&Dugas)
MEA@40C

Figure 2: CO
2
solubility for 12 m EDA, compared with data for MEA obtained by Hilliard
and Dugas (Rochelle, 2008b).
2
in 12 m EDA is compared to that of MEA in Figure 2. As CO
2

loading is less than 0.45, CO
2
has a higher solubility in EDA than in MEA. At loadings higher
than 0.45, the trend is reversed.
16
5
0.001
0.01
0.1
1
10
100
0 0.05 0.1 0.15 0.2 0.25
P
*
(
k
P
a
)
CO
2
40C
60C
80C
100C
PZ@100C
PZ@40C(Hilliard&Dugas)

Figure 3: CO
2
solubility for 7 m MDEA/2 m PZ, compared with data for PZ (dashed line)
obtained by Hilliard and Dugas (Rochelle, 2008b).
As shown in Figure 3, MDEA/PZ has a lower CO
2
solubility than PZ. The highest loading for
MDEA/PZ has not been established. More experiments will be done to verify the solubility data.
Table 2: CO
2
solubility and absorption/desorption rates for AEP, EDA, and MDEA/PZ.
Amine/ Conc Temp CO
2
Loading P
*
CO2

kg
(m) (C) (mol/mol alk) (kPa)
(10
7
mol/sPam
2
)
AEP/6 40 0.10 0.01 472.0
AEP/6 40 0.20 0.06 30.3
AEP/6 40 0.29 1.79 5.7
AEP/6 40 0.36 24.95 0.8
AEP/6 60 0.10 0.06 55.7
AEP/6 60 0.20 0.54 21.9
AEP/6 60 0.30 8.62 4.8
AEP/6 80 0.10 0.36 65.2
AEP/6 80 0.20 3.23 29.4
AEP/6 80 0.30 40.50 3.5
AEP/6 100 0.10 2.00 56.2
AEP/6 100 0.20 12.87 20.5
EDA/12 40 0.36 0.03 26.0
EDA/12 40 0.43 0.19 10.3
17
6
EDA/12 40 0.49 4.03 1.7
EDA/12 60 0.22 0.01 N/A
EDA/12 60 0.29 0.03 112.0
EDA/12 60 0.37 0.20 20.4
EDA/12 60 0.43 1.82 7.6
EDA/12 60 0.49 23.76 1.4
EDA/12 80 0.22 0.05 N/A
EDA/12 80 0.29 0.24 56.1
EDA/12 80 0.35 1.52 16.7
EDA/12 80 0.43 9.62
7.7
EDA/12 100 0.22 0.22
N/A
EDA/12 100 0.29 1.64 50.0
EDA/12 100 0.35 7.13 19.9
EDA/12 100 0.43 41.62 5.2
MDEA/7+PZ/2 40 0.09 0.19 16.5
MDEA/7+PZ/2 40 0.14 0.95 10.3
MDEA/7+PZ/2 40 0.19 3.55 6.2
MDEA/7+PZ/2 60 0.09 1.25 16.8
MDEA/7+PZ/2 60 0.14 4.41 9.8
MDEA/7+PZ/2 60 0.19 15.60 6.1
MDEA/7+PZ/2 80 0.03 1.27 27.6
MDEA/7+PZ/2 80 0.09 5.62 12.3
MDEA/7+PZ/2 80 0.14 17.64 6.8
MDEA/7+PZ/2 100 0.03 5.21 16.3
MDEA/7+PZ/2 100 0.09 19.78 7.6

CO
2
capacity
The solubility model enables calculation of CO
2
capacity for 5 kPa rich solution of each amine,
as shown in Figure 4. Since there is some uncertainty about CO
2
loading for MDEA/PZ
solubility data, no calculation will be done for MDEA/PZ at this time. At the same lean CO
2

partial pressure of 0.5 kPa, the sequence for CO
2
capacity from high to low is: 8 m 2-PE > 4.8 m
AMP > 8 m PZ > 7.7 m HEP 6 m AEP 12 m EDA > 7 m MEA. As the lean CO
2
partial
pressure increases or decreases, this trend holds since the curves remains relatively parallel with
each other.

18
7
0.1
1.0
0.005 0.05 0.5 5
C
O
2
c
a
p
a
c
i
t
y
(
m
o
l
/
k
g
(
w
a
t
e
r
+
a
m
i
n
e
)
)
LeanPartialPressureofCO
2
(kPa)
7m MEA
6mAEP 8mPZ
8m2PE
4.8mAMP
12mEDA
7.7mHEP

Figure 4: CO
2
capacity as a function of lean CO
2
partial pressure at 40 C.

Heat of CO
2
absorption
Heat of absorption was calculated from the model by applying the Gibbs-Duhem equation:
) / 1 (
) (ln
T d
P d
R H
abs
=
(2)
Heat of absorption obtained for all the amines was compared in Figure 5. For each amine, H
abs

was only plotted within the CO
2
loading range corresponding to 0.5 kPa to 5 kPa CO
2
partial
pressure. Apparently H
abs
of AEP is only slightly higher than that of PZ. EDA has a
comparable H
abs
to MEA.

19
8
65
69
73
77
81
85
0.1 0.2 0.3 0.4 0.5 0.6 0.7
E
n
t
h
a
l
p
y
o
f
C
O
2
a
b
s
o
r
p
t
i
o
n
(
k
J
/
m
o
l
)
CO
2
7.7mHEP
8m2PE
4.8mAMP
8mPZ
7mMEA
6mAEP
12mEDA

Figure 5: Enthalpy of CO
2
absorption vs. CO
2
loading for different amines.
Absorption/Desorption rates
In Figure 6, absorption/desorption rates for 6 m AEP at 40, 60, 80, and 100 C are compared
with those of 7 m MEA and 8 m PZ at 40 C. The rate for AEP is less than for PZ but higher
than MEA as CO
2
partial pressure is less than 5 kPa. However, it drops below MEA at P*
CO2

higher than 5 kPa. Temperature does not have a significant effect on the absorption rate of AEP.
20
9
5E-08
5E-07
5E-06
0.005 0.05 0.5 5
k
g
'

(
m
o
l
/
s
.
P
a
.
m
2
)
P*
CO2
@ 40C (kPa)
7mMEA@40C
8mPZ@40C
100C
80C
60C
40C

Figure 6: CO
2
mass transfer rate for 6 m AEP, compared with 7 m MEA and 8 m PZ.
21
10
1E-07
1E-06
0.01 0.1 1 10
k
g
'

(
m
o
l
/
s
.
P
a
.
m
2
)
P*
CO2
@ 40C (kPa)
7mMEA@40C
8mPZ@40C
100C
80C
60C
40C

Figure 7: CO
2
mass transfer rate for 12 m EDA, compared with 7 m MEA and 8 m PZ.
As shown in Figure 7, kg for EDA is close to that for MEA at 40 C and low P*
CO2
. Absorption
with EDA becomes slower than MEA as P*
CO2
is above 0.2 kPa.

22
11
1E-07
1E-06
0.01 0.1 1 10
k
g
'

(
m
o
l
/
s
.
P
a
.
m
2
)
P*
CO2
@ 40C (kPa)
7mMEA@40C
8mPZ@40C
100C
80C
60C
40C

Figure 8: CO
2
mass transfer rate for (7 m MDEA/2 m PZ), compared with 7 m MEA and
8 m PZ.
The 7 m MDEA /2 m PZ blend performed pretty well at the CO
2
loading we studied. As shown
in Figure 8, at 40 C, kg for the blend has almost the same value as PZ at rich loading with
P*
CO2
around 3.5 kPa, which is desirable. At leaner loading, MDEA/PZ has a slower absorption
rate than PZ, but still about twice as fast as MEA. Also the rates of the blend at 40 C and 60 C
are identical.
23
12
5E-08
5E-07
5E-06
0.05 0.5 5 50
k
g
'

(
m
o
l
/
s
.
P
a
.
m
2
)
P*
CO2
@ 40C (kPa)
8mPZ
7mMEA
7.7mHEP
6mAEP
8m2PE
12mEDA
4.8mAMP
7mMDEA/2PZ

Figure 9: Comparison of CO
2
mass transfer at 40 C for all the amines studied so far.
The curves of kg vs. CO
2
partial pressure at 40 C for all the amines are put together in Figure
9. By interpolation or extrapolation from the current kg data, kg at 5 kPa CO
2
partial pressure
was obtained for each amine solvent and compared. (The values of kg at 5 kPa are also given in
Table 3.) The sequence in absorption rate from high to low is: 7 m MDEA/2 m PZ > 8 m PZ >
7 m MEA > 7.7 m HEP > 6 m AEP > 8 m 2-PE > 4.8 AMP > 12 m EDA. Apparently in terms
of absorption rate, EDA has the worst performance at rich loading. In contrast, MDEA/PZ is an
efficient solvent, even better than PZ.
24
13
0
5
10
15
20
25
30
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
V
i
s
c
o
s
i
t
y
@
4
0
C
(
c
P
)
CO
2
loading(mol/molalka)
7.7mHEP
8 m2PE
4.8mAMP
6 mAEP
12 mEDA

Figure 10: Viscosity of loaded amine solutions vs. CO
2
loading at 40 C.
As shown in Figure 10, viscosity of amine solutions increases with loading. 8 m 2-PE and 6 m
AEP have a viscosity higher than 20 cP at rich loading. A relatively high viscosity was also
found for 7.7 m HEP at high CO
2
loading. Reduction in concentration of these amines might be
considered to lower operation cost.
Table 3: Summary table for all the tested amines
CO
2
Capacity kg @PCO
2
=5kPa
H
abs
@PCO
2

=1.5kPa
Amine
Conc.
(m)
(mol/kg
(water+amine))
(10
7
mol/sPam
2
) (kJ/mol)
MDEA/PZ 7/2 0.71
*
5.7 67
PZ 8 0.79 5.3 70
MEA 7 0.47 3.1 82
HEP 7.7 0.68 2.9 69
AEP 6 0.66 2.3 72
25
14
2-PE 8 1.23 2 73
AMP 4.8 0.96 1.7 73
EDA 12 0.78 1.6 80
*: Capacity for MDEA/PZ is obtained from Bishnois data (Bishnoi, 2000).

Conclusions
CO
2
solubility and absorption rates for three amine solvents, AEP, EDA, and MDEA/PZ, were
measured with WWC in this quarter and compared to those amines tested previously. 6 m AEP
has a capacity close to 7.7 m HEP but less than 8 m PZ. 12 m EDA and 7 m MDEA/2 m PZ
both have a capacity similar to 8 m PZ. In terms of absorption rate, at 5 kPa, MDEA/PZ
performs even a little better than PZ, while EDA is not an attractive solvent. 6 m AEP absorbs
CO
2
at a rate less than half of the rate of 8 m PZ. Heat of CO
2
absorption of 12 m EDA is about
80 kJ/mol at average operational CO
2
loading, close to that of 7 m MEA. AEP has a heat of
absorption similar to 7.7 HEP.
Future Work
Diglycolamine (DGA
), N-Methyl-1,3-Propanediamine, and piperazine derivatives will be tested

with the WWC in the following quarter.
References
Bishnoi S. Carbon dioxide absorption and solution equilibrium in piperazine activated
methyldiethanolamine Department of Chemical Engineering. The University of Texas at
Austin. Ph.D. Dissertation. 2000.
Rochelle GT et al. "CO
2
Capture by Aqueous Absorption, First Quarterly Progress Report 2008."
Luminant Carbon Management Program. The University of Texas at Austin. 2008a.
2
Capture by Aqueous Absorption, Third Quarterly Progress Report
2008." Luminant Carbon Management Program. The University of Texas at Austin. 2008b.

26
1
Influence of Viscosity and Surface Tension on the Effective
Mass Transfer Area of Structured Packing

by Robert Tsai
and the
2
July 14, 2009
Abstract
High viscosity tests (approximately 10 cP) were completed with Sulzer Mellapak 250X
(M250X). In addition, two new Sulzer structured packings (Mellapak 125Y (M125Y) and
MellapakPlus (MP252Y)) were evaluated and compared with Mellapak 250Y (M250Y).
Dry and pre-loading pressure drops for the packings were ordered: M250X (0.4) < M125Y (0.5)
< MP252Y (0.7) < M250Y. Pre-loading liquid hold-ups were ranked as follows: M125Y (0.6) <
M250X (0.9) ~ MP252Y (0.9) < M250Y. (The numbers in parentheses represent the values
relative to M250Y.) Increasing the solution viscosity had little effect on pre-loading pressure
drops but significantly reduced the capacities of the packings. Higher hold-ups were also
associated with the more viscous solutions.
The mass transfer areas of M250Y, M250X, and MP252Y were equivalent, ranging from
roughly 0.651.1 on a fractional basis (a
e
/a
p
) over liquid loads from 130 gpm/ft
2
. M125Y
exhibited higher fractional areas (0.71.2) than M250Y. For all of the packings, a reduction in
surface tension (30 dynes/cm) increased the mass transfer area by 10%. A ten-fold viscosity
enhancement had no appreciable impact on the area. Both findings were consistent with
previous results.
The mass transfer area database was updated, and the current global (a
e
/a
p
) correlation, able to
represent the entire database within limits of 15%, is displayed below:
( )( ) | |
116 . 0
3
1
L L
p
e
334 . 1

= Fr We
a
a

A global pre-loading liquid hold-up model was developed for structured packing. The
correlation (shown below) is accurate within approximately 25% with respect to the
experimental data.
( )( ) | |
72 . 0
3
2
P L L
84 . 21

= Ga Re h
27
2
A basic Excel model was created to evaluate the economics of an amine scrubber (absorber) as a
function of gas throughput and column configuration. The minimum cost was calculated to be
$57/tonne CO
2
for absorber capacities in the 100500 MW range.
Introduction
Packing is commonly used in industrial processes to provide efficient gas-liquid contacting. One
important application for which packed columns are being considered is treating flue gas for CO
2

capture. The conventional method consists of an aqueous amine solvent such as
monoethanolamine (MEA) contacting the gas, resulting in the absorption of CO
2
(Kohl and
Nielsen, 1997). The enriched solvent is sent to a stripper for regeneration and is then recycled
back to the absorber. Gas-liquid contact in both the absorber and stripper is enhanced through
the use of packing.
Reliable mass transfer models are important for design and analysis of these systems. A critical
factor involved in modeling is the prediction of the effective area of packing (a
e
), which can be
considered as the total gas-liquid interfacial area that is actively available for mass transfer. The
current research effort is focused on this parameter. The effective area is especially critical for
CO
2
capture by amine absorption, because the CO
2
absorption rate typically becomes
independent of conventional mass transfer coefficients (k
G
or k
L
0
) but remains directly
proportional to the area. Thus, it is highly desirable to have an accurate area model.
Numerous packing area correlations have been presented in the literature, but none has been
shown to be predictive over a wide range of conditions. The Rocha-Bravo-Fair (Rocha et al.,
1996) and Billet-Schultes (Billet and Schultes, 1993) models, two of the more widely used
correlations for structured packing, seem to be notably poor in their predictions involving
aqueous systems (Tsai et al., 2008b). Wang et al. (2005) performed a comprehensive review of
the available models. The various correlations predict different and sometimes even
contradictory effects of liquid viscosity and surface tension, properties that would be expected to
fundamentally influence the wetted area of packing. It is evident that their role is not well
understood, and there is a definite need for work in this subject matter.
Limited understanding of the fluid mechanics and mass transfer phenomena in packed columns
has been noted, and the need for experiments over a broader range of conditions has been
identified (Wang et al., 2005). The Separations Research Program (SRP) at the University of
Texas at Austin has the capability of measuring packing mass transfer areas. Measurements are
performed by absorbing CO
2
from air with 0.1 M NaOH in a 427 mm (16.8 in) ID column.
However, it is potentially inaccurate to extend these results to other fluids of interest, such as
amine solvents, due to variations in viscosity and surface tension.
The goal of this research is to ultimately develop an improved effective area model for structured
packing. The general objectives are to:
Develop a fundamental understanding of the fluid mechanics associated with
structured packing operation;
Determine suitable chemical reagents to modify the surface tension and viscosity of
the aqueous caustic solutions employed to make packing area measurements, and
characterize potential impacts of such additives on the CO
2
-NaOH reaction kinetics;
28
3
Expand the SRP database by measuring the hydraulic performance and mass transfer
areas of various structured packings over a range of liquid viscosities and surface
tensions;
Combine the data and theory into a semi-empirical model that captures the features of
the tested systems and adequately represents effective area as a function of viscosity,
surface tension, and liquid load.
Packed Column
The packed column had an outside diameter of 460 mm (18 in), inside diameter of 427 mm (16.8
in), and a 3 m (10 ft) packed height. Other sources may be consulted for details regarding the
apparatus and experimental protocol (Tsai et al., 2008a; Rochelle et al., 2008b).
Goniometer
The goniometer (ram-hart Inc., Model #100-00) included an adjustable stage, a syringe support
arm, a computer-linked camera for live image display, and a light source (Rochelle et al., 2006).
This system was used in conjunction with FTA32 Video 2.0 software (developed by First Ten
Angstroms, Inc.) to make surface tension measurements via the pendant drop method.
Rheometer
The Physica MCR 300 rheometer (Anton Paar, USA) employed for viscosity measurements was
first described in the Q4 2006 report (Rochelle et al., 2007). The apparatus was equipped with a
cone-plate spindle (CP 50-1). Temperature was regulated (0.1 C) with a Peltier unit (TEK
150P-C) and a Julabo F25 water bath unit (for counter-cooling). Measurement profiles consisted
of a logarithmically increased or decreased shear rate (100500 s
-1
), with 10 data points recorded
at 15-second intervals. Viscosity was determined from a plot of shear stress (measured) vs. shear
rate.
Materials
A nonionic surfactant, Tergitol
TM
NP-7 (Dow), was used to reduce the surface tension of
solutions. POLYOX WSR N750 (Dow) essentially, poly(ethylene oxide) with a molecular
weight of 300,000 was employed as a viscosity enhancer. Dow Corning
Q2-3183A antifoam
was used for foam suppression, in quantities typically ranging from 50100 ppm
w/v
.
Results and Discussion
Mellapak 250X (M250X) Mass Transfer
Baseline (0.1 M NaOH) and low surface tension (30 dynes/cm) mass transfer area tests with
M250X were presented in the previous quarterly report (Rochelle et al., 2009a). In this quarter, a
high viscosity (11 cP, 40 dynes/cm) experiment was conducted to complete the M250X
characterization. Figure 1 displays the M250X results, together with M250Y data under similar
circumstances. The data points at a given liquid load have been averaged for clarity.
Viscosity (up to approximately 15 cP) has previously been concluded to have no appreciable
impact on the mass transfer area (Tsai et al., 2008b). The new results with M250X affirmed this
29
4
assessment. (The data are close enough that it does not matter if the viscous data may have been
influenced by a small boost (< 10%) from the reduced surface tension (40 dynes/cm); the
conclusion should remain the same.) It appears that on the basis of mass transfer area, M250Y
and M250X are essentially indistinguishable, regardless of physical properties.
0.6
0.7
0.8
0.9
1
1.1
1.2
0 5 10 15 20 25 30
F
r
a
c
t
i
o
n
a
l

a
r
e
a
,

a
e
/
a
p
Liquid load (gpm/ft
2
)
M250Y - Baseline
14 cP, 40 dynes/cm
M250X - Baseline
11 cP, 40 dynes/cm

Figure 1: M250Y and M250X (ap = 250 m2/m3) mass transfer area data.
Mellapak 250X (M250X) Hydraulics
Pressure drop data for M250Y and M250X at a liquid load of 10 gpm/ft
2
are shown in Figure 2.
The results have been normalized by equation 1, a simple power law expression obtained from a
regression of all of the M250Y dry pressure drop data.

856 . 1 M250Y dry,
309 . 0
Z
F
P
=
(1)
Analogous results to M250Y were obtained. That is, while the ten-fold viscosity increase did
noticeably reduce capacity, its impact on pre-loading pressure drops was rather marginal relative
to the effect of irrigation. Thus, it seems that the inherent interaction of gas and liquid is similar
for the two packings, despite their different inclination angles (45 vs. 60).
Hold-up data for M250Y and M250X are presented in Figure 3. The results are displayed on a
relative basis, where each measured fractional hold-up (h
L
) has been normalized by a baseline
value (calculated from an average of the M250Y hold-up(s) at the corresponding liquid load).
The data were somewhat scattered at the low-end liquid loads (< 5 gpm/ft
2
), likely attributable to
accuracy constraints on our volumetric hold-up measurement method. The overall trend, though,
was indicative of lower hold-up in M250X by about 1020% under both baseline (water) and
high viscosity conditions.

30
5
0.1
1
0.5 1 1.5 2 2.5 3 3.5 4 4.5
P

/

P
d
r
y
,

M
2
5
0
Y
F-factor (Pa)
0.5
M250Y - Baseline
14 cP, 45 dynes/cm
M250X - Baseline
14 cP, 45 dynes/cm

Figure 2: M250Y and M250X pressure drop data at liquid load of 10 gpm/ft
2
.
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
2.2
2.4
0 5 10 15 20 25 30 35
R
e
l
a
t
i
v
e

h
L
Liquid load (gpm/ft
2
)
M250Y - Baseline
14 cP, 45 dynes/cm
M250X - Baseline
14 cP, 45 dynes/cm

Figure 3: M250Y and M250X hold-up data. F-factor was low (0.7 Pa
0.5
) to ensure data
were within the pre-loading region.
31
6
Mellapak 125Y (M125Y) Mass Transfer
The primary rationale for testing M125Y (a
p
= 125 m
2
/m
3
) was to expand the lower boundary of
the packing database (previously Mellapak 2Y (M2Y), a
p
= 205 m
2
/m
3
) and observe the behavior
of our global model (Tsai et al., 2008b) at this limit. The effective area data (again, averaged at
each liquid load) for M250Y and M125Y are compared in Figure 4.
0.6
0.7
0.8
0.9
1
1.1
1.2
1.3
0 5 10 15 20 25 30
F
r
a
c
t
i
o
n
a
l

a
r
e
a
,

a
e
/
a
p
Liquid load (gpm/ft
2
)
M250Y - Baseline
30 dynes/cm
M125Y - Baseline
30 dynes/cm

Figure 4: M250Y (a
p
= 250 m
2
/m
3
) and M125Y (a
p
= 125 m
2
/m
3
) mass transfer area data.
M250Y and M2Y were previously found to exhibit approximately the same fractional areas,
which led to speculation that a maximum efficiency had been attained that could not be
surpassed, regardless of packing coarseness (Rochelle et al., 2009b). However, the measured
M125Y fractional areas consistently exceeded the M250Y areas by 10% for both baseline and
low surface tension (30 dynes/cm) experiments, approaching a value of 1.3 at the high-end loads.
It was considered that with such a low specific area packing, the wall mass transfer area could be
becoming significant; for M125Y, a fully wetted, constantly renewing column wall would
account for 7.5% of the packing area, versus only 3.7% for M250Y. Even if this were assumed,
though, there would still be a noticeable difference between M125Y and M250Y.
It would seem that as structured packings become coarser, they behave more like random
packings in the sense that the mass transfer area starts to exceed the specific packing area.
Henriques de Brito et al. (1994) speculated that there should be a greater tendency for liquid flow
instabilities such as rippling or formation of satellite droplets in low a
p
packings, due to longer
film running lengths. While either of these mechanisms could explain our results, it is believed
that the former is more plausible, given that both M250Y and M125Y were equivalently
impacted by a reduction in surface tension. (The droplet theory would suggest more of an
amplifying effect for M125Y.)
32
7
Mellapak 125Y (M125Y) Hydraulics
Dry pressure drop data for M250Y and M125Y are shown in Figure 5, and irrigated pressure
drops (10 gpm/ft
2
) are presented in Figure 6. The results have been normalized by equation 1.
0.4
0.5
0.6
0.7
0.8
0.9
1
1.1
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
P

/

P
d
r
y
,

M
2
5
0
Y
F-factor (Pa)
0.5
M250Y
M125Y

Figure 5: M250Y and M125Y dry pressure drop data.
0.1
1
0.5 1 1.5 2 2.5 3 3.5 4 4.5
P

/

P
d
r
y
,

M
2
5
0
Y
F-factor (Pa)
0.5
M250Y
M125Y

Figure 6: M250Y and M125Y pressure drop data at liquid load of 10 gpm/ft
2
.
33
8
Pressure drop generally scales with specific area, so it was no surprise that the M125Y values
(dry and pre-loading) were lower than the M250Y values by a factor of two. The capacity of
M125Y (defined in terms of the F-factor at loading) was 15% higher as well. Interestingly,
pressure drops for M250X, as shown in Figure 2 and in Rochelle et al. (2009a), were on par with
(actually slightly lower than) the M125Y measurements, meaning that making the flow channels
steeper (45 to 60) has about the same effect as halving the density of packing in the column.
Hold-up data (water) for M250Y and M125Y are displayed in Figure 7. As in Figure 3, the
results have been interpreted on a relative basis.
0.4
0.5
0.6
0.7
0.8
0.9
1
1.1
1.2
1.3
0 5 10 15 20 25 30 35
R
e
l
a
t
i
v
e

h
L
Liquid load (gpm/ft
2
)
M250Y
M125Y

Figure 7: M250Y and M125Y hold-up data. F-factor was low (0.7 Pa
0.5
) to ensure data
The hold-ups for M125Y were around 60% those of M250Y on average, which, assuming the
experimental accuracy (or lack thereof) is not being over-interpreted, would indicate that hold-
up/specific area do not scale as neatly as pressure drop and specific area. A similar observation
was made with M250Y and M2Y, where the two packings were measured to have basically
equivalent hold-ups despite the latter having 20% less area (Rochelle et al., 2009b).
MellapakPlus 252Y (MP252Y) Mass Transfer
MellapakPlus 252Y (MP252Y) is essentially M250Y with a minor modification; the interface
between packing elements, often referred to as the joint and cited as a problem-spot for liquid
accumulation (Green et al., 2007), has been smoothed in order to increase capacity. This is
accomplished by bending the sheets at the top and bottom of each element from the standard 45
inclination to a vertical (90) orientation. With our test packing, this modification occurred at
the top/bottom 0.5-in of each element (8.25 in). The wetted perimeter (L
p
), as defined in Tsai et
al. (2008b) and Rochelle et al. (2009b), was measured to be the same as for M250Y, and the
specific area was assumed to be equivalent as well (250 m
2
/m
3
). (It should be noted that Alix
34
9
and Raynal (2008) listed slightly different channel dimensions for MP252Y than our values, but
even if their numbers were used, the calculated L
p
of MP252Y still would very closely match
(within 3%) that of M250Y.)
The mass transfer area results (averaged at each liquid load) for M250Y and MP252Y are shown
in Figure 8. The two packings were indistinguishable under comparable conditions. This result
would suggest that the joint does not tangibly contribute to the mass transfer area a somewhat
surprising conclusion. It is worth noting that the majority of data in these experiments are
collected at fairly low pressure drops, far from the loading region, where one would not expect
there to be a great deal of gas-liquid turbulence or mixing at the joints. This could explain the
similarities in performance for M250Y and MP252Y. It is possible that the two packings only
deviate (in terms of mass transfer area) near flooding, where M250Y might be anticipated to
exhibit greater mass transfer (at the expense of pressure drop) on account of its more abrupt
element-to-element transition.
0.6
0.7
0.8
0.9
1
1.1
1.2
0 5 10 15 20 25 30
F
r
a
c
t
i
o
n
a
l

a
r
e
a
,

a
e
/
a
p
Liquid load (gpm/ft
2
)
M250Y - Baseline
30 dynes/cm
14 cP, 40 dynes/cm
MP252Y - Baseline
30 dynes/cm
9 cP, 40 dynes/cm

Figure 8: M250Y and MP252Y (a
p
= 250 m
2
/m
3
) mass transfer area data.
MellapakPlus 252Y (MP252Y) Hydraulics
Dry pressure drop data for M250Y and MP252Y are shown in Figure 9, and irrigated pressure
drops (10 gpm/ft
2
) are presented in Figure 10. The results have been normalized by equation 1.
35
10
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
1
1.05
1.1
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
P

/

P
d
r
y
,

M
2
5
0
Y
F-factor (Pa)
0.5
M250Y
MP252Y

Figure 9: M250Y and MP252Y dry pressure drop data.
0.1
1
0.5 1 1.5 2 2.5 3 3.5 4 4.5
P

/

P
d
r
y
,

M
2
5
0
Y
F-factor (Pa)
0.5
M250Y - Baseline
14 cP, 45 dynes/cm
MP252Y - Baseline
12 cP, 45 dynes/cm

Figure 10: M250Y and MP252Y pressure drop data at liquid load of 10 gpm/ft
2
.
The MP252Y pressure drops (dry and pre-loading) were around 70% those of M250Y not quite
as low as M250X but still striking, considering the relatively small modification to the overall
packing geometry. Given that all three 250-series packings (M250Y, M250X, and MP252Y) had
36
11
identical mass transfer areas and yet displayed very different absolute pressure drop behavior,
it would appear that the channel configuration has a drastically larger effect on the vapor flow
than on the liquid. Table 1 summarizes some of the hydraulic data at 5, 10, and 15 gpm/ft
2
.
Table 1: Comparison of capacities for M250Y, M250X, and MP252Y.
F
load
(water)
(Pa)
0.5

F
load
(10x cP)
(Pa)
0.5

P
load
/ P
dry, M250Y

5 gpm/ft
2

M250Y

3.2

2.8

1.4
M250X 3.9 3.2 0.6
MP252Y 4.3 3.9 1.05
10 gpm/ft
2

M250Y
M250X
MP252Y

2.8
3.3
3.7

1.8
3.1
2.9

1.8
0.75
1.15
15 gpm/ft
2

M250Y
M250X
MP252Y

2.7
3.1
3.2

1.2
2.4
1.8

2
0.9
1.4

First, it should be noted that the loading point F-factors (F
load
) provided in Table 1 are only
approximate values, since an exact loading point was not necessarily easy to define in the data.
With that said, several interesting trends are evident. The baseline tests (water) appeared to
show a capacity trend in the order M250Y < M250X < MP252Y. Thus, it would seem there is
an interesting trade-off between M250X and MP252Y, with the former yielding lower absolute
pressure drops but the latter offering slightly more resistance to flooding. The high viscosity
results confounded this conclusion somewhat, since M250X actually appeared to have the most
capacity at 10 and 15 gpm/ft
2
. However, it is suspected that this anomaly may have been
attributable to foaming an erratic phenomenon, especially near flooding and therefore should
not be overly scrutinized. The difference in capacities between the packings also decreased with
increasing liquid load. This occurred presumably because, as liquid began to fill the void spaces
in the packings, they became equally prone to liquid shearing and entrainment. Finally, as has
been mentioned before, the effect of an enhanced liquid viscosity was a decrease in capacity,
with the impact generally becoming more significant at higher liquid loads.
Hold-up data for M250Y and MP252Y are displayed in Figure 11. Similar to the M250X results,
the MP252Y hold-ups were clearly smaller in the moderate liquid load range (515 gpm/ft
2
) but
more closely resembled the M250Y values at higher loads. The x-ray images of Green et al.
(2007) revealed local maxima at the joints for M250Y, with hold-ups being 25 times greater
than in the packing bulk. These spikes would be expected to be less significant for MP252Y due
to the smoother transition between elements, but this does not mean the overall hold-up would
have to be much different. After all, the vast majority of MP252Y is still identical to M250Y.
Consequently, it seemed quite logical for the MP252Y hold-ups to only be slightly lower (10%)
on average. It is not known why the high viscosity tests exaggerated this disparity so much more
than the baseline experiments. It is possible that because the absolute hold-ups were larger, they
were essentially cleaner (i.e. more representative of the M250Y/MP252Y difference).
37
12
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
2.2
2.4
0 5 10 15 20 25 30 35
R
e
l
a
t
i
v
e

h
L
Liquid load (gpm/ft
2
)
M250Y - Baseline
14 cP, 45 dynes/cm
MP252Y - Baseline
12 cP, 45 dynes/cm

Figure 11: M250Y and MP252Y hold-up data. F-factor was low (0.7 Pa
0.5
) to ensure data
Mass Transfer Area Database
Figure 12 shows the entire structured packing mass transfer area database and the global model
(equation 2). The relevant geometric dimensions for each packing are listed in Table 2.
( )( ) | |
116 . 0
3
4
p
3
1
L
116 . 0
3
1
L L
p
e
334 . 1 334 . 1
(
(
|
|
\
|
|
\
|
= =

L
Q
g Fr We
a
a
(2)
The new 250-series data sets (M250X and MP252Y) align with the M250Y data, which makes
sense, since these packings are identical from the viewpoint of equation 2; the model has no
angle dependence or joint-related factor. The M125Y results are the most striking feature of the
updated plot. Whereas one perhaps would have expected the fractional areas to taper off, the
data instead follow the trend of the preceding points. This, of course, raises questions about an
upper fractional limit for structured packings. The model would predict fractional area to
continue increasing for even coarser packings, but it seems natural to believe that there would
have to be some kind of threshold if not for Mellapak 64Y, then for M32Y or M16Y, etc.
This uncertainty aside, it is important to emphasize what we did discover. The lower database
limit was extended all the way down to 125 m
2
/m
3
, and the data were still found to adhere to the
(We
L
)(Fr
L
)
-1/3
dependence a fact that gives the model even greater credibility. Overall, we
have a correlation capable of representing the mass transfer areas of a broad range of structured
packings not only in terms of specific area (125500 m
2
/m
3
) but also accounting for factors
like texture, angle, etc. with very acceptable accuracy ( 15%).
38
13
While the model in its current form is satisfactory, further improvements can obviously be made.
The two issues that need to be addressed both relate to the high a
p
packings (e.g. M500Y), which
are more strongly affected by surface tension and also have a more distinct liquid load asymptote
compared to other packings. One manner of capturing these effects could be to introduce a third
dimensionless group to the model. It has always been hypothesized that capillary phenomena,
such as liquid pooling and bridging between packing sheets, become significant for the high a
p

packings. As such, it does not necessarily make sense to use the Nusselt film thickness as the
characteristic dimension, as was done for the Weber and Froude numbers in equation 2 (Tsai et
al., 2008b), since the film is essentially unbounded. To capture the effects induced by narrow,
constricted packing sheets, it would seem to be more logical to use a geometric parameter for
instance, the sheet spacing as the characteristic dimension. Some work has been done in this
regard, but so far there is nothing noteworthy to present.
Table 2: Structured packing parameters.
Packing Specific area,
a
p
(m
2
/m
3
)
Channel side,
S (mm)
Channel base,
B (mm)
Crimp height,
h (mm)
Source(s)
Mellapak 250Y
(M250Y)
250 17 24.1 11.9 Petre et al. (2003)

Mellapak 500Y
(M500Y)
500 8.1 9.6 6.53 Aroonwilas (2001)

Mellapak 250X
(M250X)

250 17 24.1 11.9 Direct measurement
MellapakPlus 252Y
(MP252Y)

250 17 24.1 11.9 Direct measurement
Mellapak 250Y (smooth)
(M250YS)

250 (?) 17 24.1 11.9 Direct measurement
Mellapak 125Y
(M125Y)

125 37 55 24.8 Direct measurement
Mellapak 2Y
(M2Y)
205 21.5 33 13.8 Sulzer contact
Direct measurement

Flexipac 1Y
(F1Y)
410 9 12.7 6.4 Koch contact
Petre et al. (2003)

Prototype 500
(P500)

500

8.1

9.6

6.53

Assumed same as
M500Y

39
14
0.4
0.5
0.6
0.7
0.8
0.9
1
1.1
1.2
1.3
0.0001 0.001 0.01 0.1
F
r
a
c
t
i
o
n
a
l

a
r
e
a
,

a
e
/
a
p
(We
L
)(Fr
L
)
-1/3
M250Y - Baseline
30 dynes/cm
4 cP, 60 dynes/cm
14 cP, 45 dynes/cm
M500Y - Baseline
30 dynes/cm
4 cP, 45 dynes/cm
10 cP, 40 dynes/cm
M250X - Baseline
30 dynes/cm
11 cP, 40 dynes/cm
MP252Y - Baseline
30 dynes/cm
9 cP, 40 dynes/cm
M250YS - Baseline
30 dynes/cm
M125Y - Baseline
30 dynes/cm
M2Y - Baseline
F1Y - Baseline
6 cP, 65 dynes/cm
P500 - Baseline
30 dynes/cm
-15%
+15%
Equation 2
Figure 12: Structured packing mass transfer area database, compared with global model
(equation 2).
Tsai et al. (2008b) demonstrated the models of Rocha-Bravo-Fair (Rocha et al., 1996) and Billet-
Schultes (Billet and Schultes, 1993) to be especially poor in their handling of aqueous systems
and speculated that this could be due to their heavy reliance on distillation data, which generally
consist of very low surface tension systems. An important objective of this entire research
project was to address this shortcoming and establish a model suitable for aqueous solvents such
as 7 m MEA. Ideally, though, this model would be more universal and be capable of bridging
the apparent gap between hydrocarbon and aqueous systems. To test this, equation 2 was
evaluated under distillation conditions, as reported by Olujic et al. (2000). The SRP
investigation (cyclohexane/n-heptane) consisted of four test pressures and liquid loads ranging
from 120 gpm/ft
2
. The relevant physical parameters are shown in Table 3.
Table 3: Physical properties of the cyclohexane/n-heptane system (avg. at column bottom).
Pressure
(bar)
Liquid density,
L

(kg/m
3
)
Liquid viscosity,
L

(cP)
Surface tension,
(dynes/cm)
Temperature
(C)
0.33 657 0.43 17 61
1.03 625 0.30 14 97
1.66 609 0.23 12 114
4.14 561 0.16 8 154

Figure 13 compares the predictions from equation 2 for the four conditions in Table 3 and for
water (
L
= 1000 kg/m
3
,
L
= 1 cP, = 72 dynes/cm). At the most extreme condition (P = 4.14
bar, = 8 dynes/cm), the model predicts 20% greater mass transfer areas than water for M250Y.
40
15
This is something of an extrapolation data for M250Y at 30 dynes/cm have shown 10% higher
areas at most but at the very least the predicted fractional areas are still reasonable, ranging
from 0.73 to 1.17. In other words, the model does not go berserk, like Rocha-Bravo-Fair does
for water, and therefore it indeed appears to be capable of handling distillation-type systems.
Figure 14 compares equation 2 with the Rocha-Bravo-Fair and Billet-Schultes correlations at the
distillation condition of 4.14 bar. It is interesting to note that the two literature models converge
at this limit, which is perhaps indicative of their development with common data sources. Also
worth pointing out is the fact that the models are actually not too far off from equation 2 in this
case, particularly at moderate liquid loads (1020 gpm/ft
2
). Thus, while we do not endorse the
use of Rocha-Bravo-Fair or Billet-Schultes for the analysis of aqueous systems, they may
actually be acceptable when applied toward the distillation-type systems from which they were
developed.

Figure 13: Predicted M250Y mass transfer areas from equation 2 for cyclohexane/n-
heptane and water.
0.6
0.7
0.8
0.9
1
1.1
1.2
0 5 10 15 20 25
F
r
a
c
t
i
o
n
a
l

a
r
e
a
,

a
e
/
a
p
Liquid load (gpm/ft
2
)
Cyclohexane/n-Heptane: P = 0.33 bar
P = 1.03 bar
P = 1.66 bar
P = 4.14 bar
Water
41
16

Figure 14: Predicted M250Y mass transfer areas from various models for cyclohexane/n-
heptane system at 4.14 bar.
Hold-up Database
While the liquid hold-up model of Suess and Spiegel (1992) has been found to be fairly
reasonable when compared against our data (Rochelle et al., 2008a), it is certainly not without
questionable aspects. For example, the correlation is strictly empirical but is based on a limited
databank (M250Y, M250X, and M500Y). Furthermore, it is systematically overpredictive at low
liquid loads and underpredictive at high ones a feature confirmed by Brunazzi et al. (1995).
Finally, Suess and Spiegel reported no effect of corrugation angle (45 or 60) on hold-up,
whereas our results seemed to show a slight decrease for the steeper configuration. For these
reasons, an attempt was made to develop a new global hold-up model based on our own
hydraulic measurements.
As a first pass, experimental hold-ups were plotted as a direct function of the Nusselt film
thickness (equation 3), since the two parameters should be relatable in some respect. This
approach was unsuccessful in collapsing the entire database.
3
p L
L
L
L L
Nusselt
sin
3
sin
3
L
Q
g g
u

= = (3)
Next, a dimensional analysis approach was taken. Numerous dimensionless group combinations
were examined, but these efforts all failed. Shetty and Cerro (1997) proposed an expression with
a dependence on the Reynolds and Galileo numbers (equation 4), but this was not found to be
effective either.

3
1
Ga
3
1
Re L
096 . 6

= N N h (4)
0.3
0.5
0.7
0.9
1.1
1.3
0 5 10 15 20 25
F
r
a
c
t
i
o
n
a
l

a
r
e
a
,

a
e
/
a
p
Liquid load (gpm/ft
2
)
Equation 2
Rocha-Bravo-Fair
Billet-Schultes
42
17
In the Mass Transfer Area Database section, it was postulated that an additional characteristic
dimension aside from the Nusselt film thickness might be required to fully capture the fluid
mechanics within structured packing. A phenomenon like liquid accumulation within channel
recesses, for instance, might be better suited by a geometric parameter basis. It was decided to
test this idea on equation 4. Surprisingly, when the characteristic length of the Galileo number
(ratio of gravitational-to-viscous forces) was defined by the packing channel side dimension (S),
the hold-up results became somewhat unified. (This modified number was henceforth denoted as
Ga
p
). An Excel power law regression was subsequently performed to determine if a better fit
than the 1/3 and -1/3 exponents could be obtained. Equation 5 was the result of this analysis.
( )( ) | |
72 . 0
3
2
P L L
84 . 21

= Ga Re h (5)
The experimental hold-up data are plotted alongside equation 5 in Figure 15. The global model
is able to represent the majority of points within 25% and has less than half the mean squared
error of the Suess and Spiegel model. An even better fit could be obtained by omitting some of
the low-end liquid load data in future analyses, since they exhibit an unreasonably dramatic (and
therefore suspect) drop-off in hold-ups.
The theoretical significance of equation 5 is not entirely clear, given the somewhat arbitrary
manner in which it was derived. Nevertheless, at worst, it can be considered as an empirical
hold-up model that may be a better (or at least more universal) option than the correlation of
Suess and Spiegel.
0.1
1
10
0.00001 0.0001 0.001
F
r
a
c
t
i
o
n
a
l

h
o
l
d
-
u
p

(
%
)
(Re
L
)(Ga
P
)
-2/3
M250Y - Baseline
30 dynes/cm
6 cP, 60 dynes/cm
14 cP, 45 dynes/cm
M500Y - Baseline
30 dynes/cm
5 cP, 50 dynes/cm
10 cP, 45 dynes/cm
M250X - Baseline
14 cP, 45 dynes/cm
MP252Y - Baseline
12 cP, 45 dynes/cm
M250YS - Baseline
M125Y - Baseline
M2Y - Baseline
P500 - Baseline
+25%
-25%
Equation 5

Figure 15: Structured packing hold-up database, compared with global model (equation 5).
Absorber Economic Model
There are many factors that deserve consideration in the design of the large absorbers planned
for post-combustion CO
2
capture. The selection of packing and the treatment of liquid
43
18
distribution, as well as how to design the column itself (i.e. short/fat vs. tall/skinny), are all
important issues. Much research is devoted to the maximization of certain variables (e.g.
effective area) or minimization of others (e.g. pressure drop), but in the end, the final decision
will rely on economics. Consequently, work was begun on a basic economic model to
investigate this optimization problem.
An Excel program was set up to analyze the interaction of gas rate and column configuration.
Basically, the SOLVER function was utilized to vary superficial gas velocity and calculate a
minimum cost for a given throughput (0.153 MMCFM or roughly 501000 MW). Five major
cost components were identified: the column body (including the shell and auxiliaries such as
manholes, ladders, etc.), packing, pressure drop, blower, and pump. Estimations were derived
from various sources (Peters and Timmerhaus, 1991; Stichlmair et al., 1989; Rochelle et al.,
2005). Several key assumptions were: factors of 4 and 0.4 for the installed and annualized costs
(necessary for conversion to an annualized basis), a conversion factor of 0.6 between 1990 and
present dollars, a pressure drop ceiling of 80% approach to flood (defined at 1025 Pa/m), and a
blower operation cost of $50/MWh. The basis of the model was a 7 m MEA system (unloaded)
with 90% CO
2
removal. A required area of 0.1 m
2
/mol CO
2
-hr was used based on several Aspen
Plus
simulation scenarios run by Plaza. The column was designed as square, and no constraints
were placed on its size.
Table 4 shows the results from a few cases run with Mellapak 250Y. The total minimum cost
was calculated to be around $57/tonne CO
2
, which seemed reasonable, and was found to
decrease with greater throughput. The packing and column dominated the total cost (90%) in
every instance. Interestingly, the column benefitted from economies of scale whereas the
packing did not; it always accounted for roughly $3.30/tonne CO
2
. The cost associated with
pressure drop was smaller than expected, although it did become more significant as the gas load
increased. Over the 100 to 500 MW range, the gas velocity and packing height remained
relatively constant, and the only parameter that really changed was the column side, increasing
from 10 to 20 m.
Table 4: Economic model results for M250Y.
Gas rate
(MW)
Minimum cost
($/tonne CO
2
)
Gas velocity
(m/s)
Column side
(m)
Packing height
(m)
100 $7.35
Packing: 45%
Column: 45%
P: 2.6%
1.53 9.6 10.5
250 $5.71
Packing: 58%
Column: 33%
P: 3.4%
1.56 15.1 10.6
500 $5.11 1.72 20.3 11.6

Packing: 64%
Column: 25%
P: 5.3%

44
19
Conclusions
M250X, MP252Y, and M125Y were characterized and contrasted with the standard M250Y.
The packing geometry greatly impacted the vapor side of the gas-liquid contacting process;
pressure drop scaled directly with specific area (M125Y) and was significantly reduced by
relatively minor modifications to the flow channel angles (M250X and MP252Y). The liquid
side, on the other hand, appeared to be primarily influenced by the specific area. Despite their
varying configurations, all three 250-series packings (M250Y, M250X, and MP252Y) displayed
similar hold-ups, with both M250X and MP252Y being only marginally lower (10%) on average
than M250Y. Furthermore, the 250-series packings had essentially the same mass transfer area.
M125Y exhibited much lower hold-ups (60%) relative to M250Y. Its fractional area was higher
(10%) than M250Y as well, indicating a tendency toward random packing-like behavior (i.e.
effective areas well in excess of the specific area) with increasing structured packing coarseness.
A ten-fold increase in solution viscosity did not appreciably affect pressure drops but did
increase liquid hold-ups and decrease the capacities of the packings. Mass transfer experiments,
performed via absorption of CO
2
into 0.1 M NaOH, showed no perceivable effect of viscosity on
the packing effective area. A reduction in surface tension (30 dynes/cm) resulted in a 10%
increase in area. All of these findings were consistent with our past conclusions.
The mass transfer area database (updated to include the new M250X, MP252Y, and M125Y data
sets) continued to be represented well (15%) by the correlation that was regressed as a function
of (We
L
)(Fr
L
)
-1/3
, further validating the proposed form of our model. The model was checked
under distillation conditions (cyclohexane/n-heptane) and appeared to be capable of handling this
type of systems as well.
The hold-up database collapsed when plotted as a function of (Re
L
)(Ga
p
)
-2/3
. The regressed
model offered an overall improved fit of our data (25%) compared to the Suess and Spiegel
correlation.
Optimum cost estimates of $57/tonne CO
2
for a 100500 MW absorber were obtained from the
economic analysis. It is believed that a good economic model foundation was established, but
obviously, much more work will be necessary to obtain results that we would feel comfortable
reporting in an official setting.
Future Work
Primary goals for the near future include the refinement of the global mass transfer model and
more rigorous analysis of the hydraulic data chiefly, the pressure drop results. The economic
model will also be improved, per the advice of industrial consultants. Journal publications that
cover these topics will hopefully be completed later on in the year.
Nomenclature
a
e
= effective area of packing, m
2
/m
3

a
p
= specific (geometric) area of packing, m
2
/m
3

B = packing channel base, m
F = gas flow factor, (m/s)(kg/m
3
)
0.5
or Pa
0.5

g = gravitational constant; 9.81 m/s
2

45
20
h = packing crimp height, m
h
L
= (total) liquid hold-up, m
3
/m
3

k
G
= gas-side mass transfer coefficient, kmol/(m
2
Pas)
k
L
0
= physical liquid-side mass transfer coefficient, m/s
L
p
= wetted perimeter in cross-sectional slice of packing, m
P = pressure drop, Pa
Q = volumetric flow rate, m
3
/s
S = packing channel side, m
u = velocity, m/s
Z = packed height, m
Greek Symbols
= corrugation angle (with respect to the horizontal), deg
= characteristic length, m
= viscosity, kg/(m-s)
= density, kg/m
3

= surface tension, N/m
Subscripts
G = gas phase
L = liquid phase
Dimensionless Groups
a
f
= fractional area of packing, a
e
/a
p

Fr = Froude number,
g
u
2

Ga or N
Ga
= Galileo number,
2
2 3
g

N
Re
= Reynolds number as defined by Shetty and Cerro (1997),
p
L
Q
4

Re = Reynolds number,
u

We = Weber number,

2
u

46
21
References
Alix P, Raynal L. "Liquid Distribution and Liquid Hold-up in Modern High Capacity Packings".
Chem Eng Res Des. 2008;86:585591.
Aroonwilas A. Mass-Transfer with Chemical Reaction in Structured Packing for CO
2
Absorption
Process. University of Regina. Ph.D. Thesis. 2001.
Billet R, Schultes M. "Predicting Mass Transfer in Packed Columns". Chem Eng Technol.
1993;16(1):19.
Brunazzi E et al. "Interfacial Area of Mellapak Packing: Absorption of 1,1,1-Trichloroethane by
Genosorb 300". Chem Eng Technol. 1995;18(4):248255.
Green CW et al. "Novel Application of X-ray Computed Tomography: Determination of
Gas/Liquid Contact Area and Liquid Holdup in Structured Packing". Ind Eng Chem Res.
2007;46(17):57345753.
Henriques de Brito M et al. "Effective Mass-Transfer Area in a Pilot Plant Column Equipped
with Structured Packings and with Ceramic Rings". Ind Eng Chem Res. 1994;33(3):647
656.
Kohl A, Nielsen R. Gas Purification. Houston, Gulf Publishing Co.: 1997.
Olujic Z et al. "Influence of Corrugation Geometry on the Performance of Structured Packings:
An Experimental Study". Chem Eng Process. 2000;39(4):335342.
Peters MS, Timmerhaus KD. Plant Design and Economics for Chemical Engineers. New York,
McGraw-Hill, Inc.: 1991.
Petre CF et al. "Pressure Drop through Structured Packings: Breakdown into the Contributing
Mechanisms by CFD Modeling". Chem Eng Sci. 2003;58(1):163177.
Rocha JA et al. "Distillation Columns Containing Structured Packings: A Comprehensive Model
for Their Performance. 2. Mass-Transfer Model". Ind Eng Chem Res. 1996;35(5):1660
1667.
Rochelle GT et al. "Integrating MEA Regeneration with CO
2
Compression and Peaking to
Reduce CO
2
Capture Costs. Final Report for Trimeric Corp. Subcontract of DOE contract
DE-FG02-04ER84111." June, 2005.
2
2006." Luminant Carbon Management Program. The University of Texas at Austin. 2006.
2
Capture by Aqueous Absorption, Fourth Quarterly Progress Report
2
Luminant Carbon Management Program. The University of Texas at Austin. 2008.
2
2
47
22
2
Shetty S, Cerro RL. "Fundamental Liquid Flow Correlations for the Computation of Design
Parameters for Ordered Packings". Ind Eng Chem Res. 1997;36(3):771783.
Stichlmair J et al. General Model for Prediction of Pressure Drop and Capacity of Countercurrent
Gas/Liquid Packed Columns". Gas Sep Purif. 1989;3:19-28.
Suess P, Spiegel L. "Hold-up of Mellapak Structured Packings". Chem Eng Process.
1992;31(2):119124.
Tsai RE et al. "Influence of Surface Tension on Effective Packing Area". Ind Eng Chem Res.
2008;47(4):12531260.
Tsai RE et al. "Influence of Viscosity and Surface Tension on the Effective Mass Transfer Area
of Structured Packing". GHGT-9, Washington D.C. 2008.
Wang GQ et al. "Review of Mass-Transfer Correlations for Packed Columns". Ind Eng Chem
Res. 2005;44(23):87158729.

48
1

Modeling Stripper Performance for CO
2
Removal with Amine
Solvents

by David Van Wagener
and the
2
July 1, 2009
Abstract
Since Hilliard developed thermodynamic models for various amine solvents, additional
experimental data has been collected at new conditions. The data primarily of interest have been
for concentrated piperazine (PZ). The Hilliard model performed well for low concentrations,
0.9 m5 m, but 8 m PZ will be used in future simulations. VLE data collected by Dugas as well
as heat capacity data collected by Nguyen for PZ was incorporated into previous parameter
regression files. Additionally, heat of absorption data was collected by Freeman. The
parameters to be regressed were reconsidered, and more focus was put on the heat capacity
parameters of the dominant species at relevant loadings. This quarter the dielectric constant of
PZ was also included. The original value used by Hilliard was for piperidine, a molecule similar
to PZ with one amine group instead of two. Including the dielectric constant in the regression
greatly improved the fit, and the new value of the dielectric constant is between that of piperidine
and MEA. The theoretical difference between the heat of absorption for 8 m MEA at 40 C and
120 C was calculated using an energy balance, and the difference was found to be
approximately 3.2 kJ/mol CO
2
. This difference is much smaller than what is observed in the
experimental data, so the data collection for heat of absorption should be reevaluated.
Introduction
Piperazine (PZ) is of interest as a solvent for CO
2
capture because it has significantly higher
capacity than monoethanolamine (MEA), the baseline and industry standard. A PZ molecule has
two amine groups, which leads to this increased capacity. High solvent capacity leads to less
solvent being circulated between the absorber and stripper, so the stripper reboiler duty decreases
because the sensible heat input for the solvent is lowered. The CO
2
absorption rate for PZ is
enhanced over MEA as well, also possibly due to the two amine groups per molecule. As an
added benefit, PZ has no detectable thermal degradation up to 150 C. Many explored stripper
configurations operate more efficiently at high temperatures, so it is expected that PZ will
perform better than MEA (Freeman et al., 2008).
49
2

Previously, a thermodynamic model was developed for PZ (Hilliard, 2008), and it was used to
simulate a simple stripper with the accompanying rich and lean pumps, cross heat exchanger, and
multi-stage compressor. The simulations produced results with few convergence errors;
however, the behavior while varying the lean loading specification was unexpected. Typically
the calculated equivalent work of the stripper has a single distinct optimum lean loading
(Oyenekan, 2007). Conversely, the PZ simulation demonstrated both a local and global optimum
(Rochelle et al., 2008). The local optimum was at a lean loading of 0.30, an expected value
based upon the measured VLE at absorber conditions. The global optimum was at a lean loading
of 0.15, and the temperature profile was very hot; reaching temperatures over 120 C. A
suggested source for this unusual behavior was the accuracy of the model predictions of
thermodynamic values for the solvent. The predictions that seemed particularly questionable
were the solvent heat capacity and heat of absorption of CO
2
. Prior work generated heat capacity
data for 8 m PZ by using trends in lower concentration piperazine solvents. The model
parameters were regressed to better predict heat capacity. However, the fit was not improved.
The heat capacity data used for the regression were extrapolated from values at lower PZ
concentrations. Though not an ideal source of data, these values were the only ones available at
the time.
More recently, heat capacity data for 8 m PZ was collected by Nguyen. These data were used to
improve the heat capacity predictions in a new regression incorporating A-E heat capacity
parameters. The fit was much improved because the regression only focused on heat capacity
data. The work this quarter continues with the perfection of the PZ model for the use in
simulating absorption/stripping for CO
2
capture.
Methods and Results
The final PZ model from the last quarter was evaluated for overall accuracy of predictions. The
predictions of CO
2
solubility and heat capacity fit the laboratory data very well, and the
speciation followed expected trends. However, the heat of absorption predictions did not follow
typical heat of absorption trends and were lower than collected experimental data on average.
The data collected by Freeman and Aspen Plus
predictions from the PZ model corrected for

heat capacity are shown below in Figure 1.
The heat of absorption predictions decreased with increasing temperature, so the 150 C
predictions were extremely low, dropping as far as 3040 kJ/mol CO
2
. This model will be used
to develop both absorber and stripper simulations. The predictions between 40 C70 C are the
most important for the absorber simulations, but the stripper will be modeled up to 150 C due to
the expected improvement in performance at high temperature with relatively minimal thermal
degradation. In the previous quarter, correlations for PZ were used to calculate the approximate
energy consumption for 2- and 3-stage flash configurations, operating isothermally at 150 C.
Since there are no sensible heat effects between flashes when operating isothermally, only the
heat of absorption matters for this section. It is essential to achieve relatively accurate heat of
absorption predictions in Aspen Plus
to accompany the accurate VLE and heat capacity. For

this reason, further regressions were performed to obtain accurate VLE, heat capacity, heat of
absorption, and speciation.
50
3

Figure 1: Heat of Absorption for 8 m PZ. Data from Freeman (points) and predictions in
Aspen Plus
for heat capacity corrected model (lines).

A substantial database of properties for piperazine has been collected by other contributors,
shown below in Table 1.
Table 1: Available Data Sets for Piperazine Regression
Data Type Temperature (C) PZ Concentration (m) CO
2
Loading
VLE 40-180 2-12 0.148-0.424
Heat Capacity 40-120 3.6-12 0.159-0.400
Volatility 40-60 3.6-5 0.148-0.415
Heat of
Absorption
40-120 2.42-8 0.041-0.488
In addition to a wide range of data sets, there are many parameters that can be varied. These
parameters included:
Ionic heat capacities
Enthalpy of formation for ionic species
Gibbs free energy of formation for ionic species
Tau interaction parameters
51
4

Both the ionic heat capacities and tau interaction parameters are temperature dependent, so each
has multiple parameters for each component. The most dominant species in loaded solutions
were determined to be:
HPZ
+

HPZCOO
PZCOO
-

The regressions using the DRS package in Aspen Plus
were not straightforward. They often

failed to converge, and even completed regressions yielded many flash errors when analyzing the
fit of the new model. What was first planned to be a simple regression activity to match the
available high concentration PZ data turned into a search for regression conditions which yielded
a converging regression, had few to no flash errors after completing, and adequately fit the data
available for 8 m PZ. After many combinations of the regressed parameters and included data
sets, a run with adequate results and no flash errors was eventually stumbled upon. The solution
started with the original Hilliard regression and added the following data to the appropriate data
sets:
VLE: 8 m PZ, 0.230.40 loadings, 40 C100 C
Heat Capacity: 8 m PZ, 0.210.40 loadings, 40 C150 C
Including these data in new data sets did not yield good results. An assortment of parameters for
the dominant species were chosen; see the regressed values below in Table 2. As alluded to
above, this regression converged and had no flash errors when analyzing the fit for 8 m PZ. The
behavior of the VLE, heat capacity, and heat of absorption are shown below in Figures 24. The
most apparent misfit is the VLE, with significant inaccuracies at very low and very high
loadings.
Table 2: Regressed Parameter Values - 1
Parameter Component i Component j Value (SI units)
1 G
aq
form
PZCOO
-
-205423628 70058650
2 H
aq
form
PZCOO
-
-482049819 1313460810
3 C
p
-A PZCOO
-
-6128880 209089229
4 C
p
-B PZCOO
-
21341 565138
5 C
p
-C PZCOO
-
-0.140 20.332
6 G
aq
form
HPZCOO -274324954 403173
7 H
aq
form
HPZCOO -510325070 3313114
8 C
p
-A HPZCOO 1597232 2951470
9 C
p
-B HPZCOO -8765 16428
10 C
p
-C HPZCOO 14.2 22.7
11 C
p
-A
PZH
+

5085443 32344691
12 C
p
-B
PZH+
-14469 174471
13 C
p
-C
PZH+
2.98 232.48
14
m,ca
-A H
2
O (PZH
+
,PZCOO
-
) 7.37 45.07
15
m,ca
-B H
2
O (PZH
+
,PZCOO
-
) 1682 17945
16
ca,m
-A (PZH
+
,PZCOO
-
) H
2
O -181 76
52
5

17
ca,m
-B (PZH
+
,PZCOO
-
) H
2
O 71192 31106
18
m,ca
-A H
2
O (PZH
+
,HCO3
-
) 9.04 118055.40
19
m,ca
-B H
2
O (PZH
+
,HCO3
-
) -538 20327596
20
m,ca
-C H
2
O (PZH
+
,HCO3
-
) -0.978 35923.919
21
ca,m
-A (PZH
+
,HCO3
-
) H
2
O -77.4 104442.7
22
ca,m
-B (PZH
+
,HCO3
-
) H
2
O 186 874129
23
ca,m
-C (PZH
+
,HCO3
-
) H
2
O -484 1744760
24
m,ca
-A PZ (PZH
+
,PZCOO
-
) 3.96 7.92
25
ca,m
-A (PZH
+
,PZCOO
-
) PZ -11.3 2.9

Figure 2: CO
2
Solubility Predictions for Regression 1. Points - Experimental Data. Lines -
Model Predictions.

53
6

Figure 3: Heat Capacity Predictions for Regression 1. Points - Experimental Data. Lines -
Model Predictions.

Figure 4: Heat of Absorption Predictions for Regression 1. Points - Experimental Data.
Lines - Model Predictions.

54
7

This model converged and produced no flash errors in the model analysis, indicating that it could
be a stable thermodynamic model to be used for simulations. However, the VLE was very
inaccurate. The spacing between temperatures appeared roughly correct, but the slope of the
Aspen Plus

predictions was too shallow. The dielectric constant is a variable parameter that
mostly affects the slope of the VLE. The magnitude of this constant is relatable to a component's
ability to stabilize in an ionic solution. The value of this parameter had been specified by
Hilliard to be equal to that of piperidine because the dielectric constant is not available for
piperazine in the literature. Since the value of the dielectric constant for piperazine is uncertain,
it was varied to attempt to better match all sets of data. The regressed constants are displayed in
Table 3, and the predictions are shown in Figures 57.
Table 3: Regressed Parameter Values - 2
1 G
aq
form
PZCOO
-
-219389572 609502.754
2 H
aq
form
PZCOO
-
-479875330 6857582.96
3 C
p
-A PZCOO
-
-2331715 2081603.71
4 C
p
-B PZCOO
-
9202 6131
5 C
p
-C PZCOO
-
-0.0162 0.1439
6 G
aq
form
HPZCOO -273197501 420092
7 H
aq
form
HPZCOO -516582958 2735043.35
8 C
p
-A HPZCOO -14434 67471
9 C
p
-B HPZCOO 1003 245
10 C
p
-C HPZCOO 0.000417 0.19825299
11 C
p
-A
PZH
+

1435790 535810
12 C
p
-B
PZH+
-4323 1462
13 C
p
-C
PZH+
0.0288 0.2337
14
r
-A PZ 23.5 85.0
15
r
-B PZ 6.20 1034.55
16
m,ca
-A H
2
O (PZH
+
,PZCOO
-
) -18.028 2094.437
17
m,ca
-B H
2
O (PZH
+
,PZCOO
-
) -5737.179 704096.224
18
ca,m
-A (PZH
+
,PZCOO
-
) H
2
O -99.4 70.6
19
ca,m
-B (PZH
+
,PZCOO
-
) H
2
O 47711 32337
20
m,ca
-A H
2
O (PZH
+
,HCO3
-
) 18.6 0.5
21
m,ca
-B H
2
O (PZH
+
,HCO3
-
) 8.94 144.90
22
m,ca
-C H
2
O (PZH
+
,HCO3
-
) -3.40 6.87
23
ca,m
-A (PZH
+
,HCO3
-
) H
2
O -7.44 0.40
24
ca,m
-B (PZH
+
,HCO3
-
) H
2
O 11.5 130.9
25
ca,m
-C (PZH
+
,HCO3
-
) H
2
O 0.831 2.980
26
m,ca
-A PZ (PZH
+
,PZCOO
-
) 54.9 9109.8
27
ca,m
-A (PZH
+
,PZCOO
-
) PZ -4.73 0.71

55
8

Figure 5: CO
2
Solubility Predictions for Regression 2. Points - Experimental Data. Lines -
Model Predictions.

Figure 6: Heat Capacity Predictions for Regression 2. Points - Experimental Data. Lines -
Model Predictions.

56
9

Figure 7: Heat of Absorption Predictions for Regression 2. Points - Experimental Data.
Lines - Model Predictions.

The fit is reasonable overall. The VLE fits very well, displaying little discrepancy between the
Aspen Plus
predictions and the experimental data. The vertical spacing for each 20 C
increment should be equal, so the 160 C predictions may be too low. The heat of absorption
follows adequate trends. Based on the VLE, the heat of absorption is expected to be about 6570
kJ/mol CO
2
, but the predictions in the relevant loading range of 0.30.4 display a range of 6080
kJ/mol CO
2
. However, the predictions are well behaved and are tighter packed than in previous
models. The heat capacity predictions are also very close to the data, but start to show deviation
at 150 C.
Since the piperazine dielectric constant was varied with free bounds, its final value was analyzed.
Figure 8 compares the final regressed dielectric constant as a function of temperature to those
provided by Hilliard (2008) and Bishnoi (2002). In Bishnoi's work, the dielectric constant was
assumed to be equal to that of MEA, though the value used is lower than the current documented
literature value, which is listed in the CRC handbook as 31.94 at a temperature of 20 C. Figure
8 shows that the regressed value in the current work approaches the value used by Bishnoi, but is
significantly different than the value used by Hilliard. Without direct measurement of the exact
dielectric constant of PZ, it cannot be determined whether the regressed value is accurate, but
most importantly it allows the Aspen Plus
predictions of solvent properties to closely follow

experimental data.

57
10

Figure 8: PZ Dielectric Constant Used by Various Models
The accuracy of collected heat of absorption data has also been questioned recently. The data
are collected from an apparatus at NTNU in Trondheim, Norway, and it is very consistent
between runs. However, the data can be scattered and may be too high. As stated before, the
expected heat of absorption for 8 m PZ is approximately 65 kJ/mol CO
2
, given by the Gibbs-
Helmholtz relation with the VLE. On the other hand, the data collected for 8 m PZ, shown in
Figure 1 are scattered between 80120 kJ/mol CO
2
before dropping off at a loading of 0.45. The
spread of this data is also worrisome.
The cycle used for this calculation is shown in Figure 9. It is not meant to simulate an actual
absorption/stripping process, but merely quantify the theoretical difference between the heat of
absorption at absorber and stripper temperatures. Below are several equations which were used:
an energy balance rearranged to calculate the heat of absorption difference, and 3 equations to
calculate the heat capacity of the solvent.

1

2

3

4

58
11

The heat capacity correlation used for calculating the heat duty for cooling the lean and heating
the rich streams was developed by Nguyen. Shown above, it calculates the heat capacity for
concentrated piperazine solvents as a function of temperature and PZ, CO
2
, and water mass
fraction. Not shown, the water heat capacity was calculated using the DIPPR correlation for
liquid water. This overall heat capacity was integrated over the range of the absorber and
stripper temperatures. The ideal gas heat capacity for CO
2
from DIPPR was used to determine
the sensible heat required to cool the desorbed CO
2
back to the absorption temperature. The
specifications and result of the calculation is shown below in Table 4.

Figure 9: Energy Balance on CO
2
Absorption/Desorption

Table 4: Calculation of Heat of Absorption Increase with Temperature
Calculationspecifications Heatinput/outputs
Richloading 0.4 Richsensibleheat 568.64 kJ
Leanloading 0.3 Leansensibleheat 568.71 kJ
CO
2
absorbed 1 mol CO
2
cooling 3.18 kJ
Solvent 8 mPZ
Coldside 40 C CalculatedDifference
Hotside 120 C dH
abs120
dH
abs40
3.25 kJ

This calculation demonstrated that the difference between the heats of absorption in 8 m PZ at
40 C and 120 C should be approximately 3.25 kJ/mol CO
2
. This result revealed uncertainty in
the collected heat of absorption data. Not only are the values consistently higher than expected,
but they should be more tightly packed as well.
Cool CO
2
to
40C
59
12

Conclusions
The PZ model developed by Hilliard in Aspen Plus
was regressed to better fit recent VLE, heat

capacity, and heat of absorption data. The key to a successful model seems to be fixing the PZ
dielectric constant. Trials with many combinations of regressed data and parameters were
evaluated, but none produced an accurate model with converging flashes. Including the
dielectric constant in the regression yielded a promising model, and the final value for the
dielectric constant was realistic. In the next quarter, the model will be further evaluated to
potentially adjust the 150 C160 C VLE and volatility predictions.
The calculation of the theoretical difference between the heats of absorption at absorber and
stripper temperatures verifies that the heat of absorption at 40 C should only be approximately
3.2 kJ/mol CO
2
lower than the value at 120 C. This calculation raises concerns about the
accuracy of the apparatus for heat of absorption data collection.

Future Work
The PZ model in Aspen Plus
will be further evaluated to determine if it is ready for use in

simulations. The volatility and high temperature VLE will be addressed. If the dielectric
constant of 8 m PZ or pure PZ is measured, it can be compared against the value obtained by the
regression.
Once the thermodynamic model is completed, configurations with varying levels of complexity
will be analyzed with an end goal of correlating configuration complexity with performance.
The analysis will begin with simple flowsheets like a single flash, a simple stripper, and a
stripper with a preflash. The analysis will continue through more complex options like a double
matrix.

References
Bishnoi S, Rochelle GT. Thermodynamics of Piperazine/Methyldiethanolamine/Water/Carbon
Dioxide. Ind Eng & Chem Res. 2002:604612.
Freeman SA, Dugas RE, Van Wagener DH, Nguyen T, Rochelle GT. CO
2
capture with
concentrated, aqueous piperazine. GHGT-9. Washington, DC: Elsevier, 2008.
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, Austin. Ph.D. Dissertation. 2008.
Oyenekan, Babatunde. Modeling of Strippers for CO2 Capture by Aqueous Amines. Ph.D.
Dissertation, The University of Texas, Austin. Ph.D. Dissertation. 2007.
Rochelle GT et al. CO
2
2008. Luminant Carbon Management Program. The University of Texas, Austin. 2008.
60
1
Solvent Management of MDEA/PZ

by Fred Closmann
the
2
and the
Process, Science, & Technology Center
July 10, 2009
Abstract
A thermal degradation experiment was conducted on 7 m MDEA in the second quarter. The
compounds dimethylaminoethanol (DMAE), diethanolamine (DEA), N,N-dimethyl ethanamine
(DMEA), dimethyl piperazine, 1-(2-hydroxyethyl)-4-methylpiperazine (HMP), and
triethanolamine (TEA) were tentatively identified in degraded solvent samples through ion
chromatography (IC) and IC-mass spectrometry (IC-MS) methods. We calculate an activation
energy for the degradation of MDEA of approximately 47 kJ/gmol, and rates of degradation of
MDEA of 66 and 112 mmolal/day were calculated at 150 C and loadings of 0.1 and 0.2 moles
CO
2
/mole alkalinity, respectively. Proposed degradation mechanisms for the MDEA loss
include the protonation of MDEA, the formation of an MDEA-carbamate, and the
disproportionation of MDEA. All three mechanisms are followed by subsequent reactions and
could involve PZ when present in a solvent blend. However, because the MDEA loss rate is
similar to rates calculated for MDEA in 7 m MDEA/2 m PZ, we believe the role of PZ in the
degradation of the solvent is of lesser importance.
The construction of the integrated solvent cycling/degradation apparatus is near completion, and
basic temperature metrics for amine cycling have been met. Modifications to the system to
achieve accurate temperature measurement are ongoing. Amine cycling degradation
experiments will be completed in 3
rd
quarter 2009.
Introduction
During the 2
nd
Quarter 2009, a thermal degradation experiment was performed on 7 m MDEA
for comparison to similar experiments utilizing 7 m MDEA/2 m PZ. Cation chromatography
was used to determine the loss rate of MDEA, and to identify breakdown products through
comparison to standards. Mass spectrometry (MS) coupled with IC was used to confirm the
identity of breakdown byproducts. Degradation of samples was performed at 120, 135, and
150 C, and loadings of 0.1 and 0.2 moles CO
2
/mole alkalinity. The degradation rate of MDEA
61
2
at 135 C was 53 to 55 mmolal/day at loadings of 0.1 and 0.2 moles CO
2
/mole alkalinity, and 66
and 112 mmolal/day at loadings of 0.1 and 0.2 moles CO
2
/moles alkalinity, respectively, at
150 C.
An additional thermal degradation experiment was initiated using 7 m MDEA/2 m PZ treated
with H
2
SO
4
in sufficient quantity to neutralize 10% of alkalinity. The solution was not loaded
with CO
2
. All sample cylinders failed and the experiment was stopped pending fabrication of a
new set of sample cylinders.
Thermal Experiments
Thermal No. 10
A single thermal experiment was initiated in March, with final samples to be pulled in July 2009.
MDEA degradation rates of 39, 55, and 66 mmolal/day were measured for a loading of 0.1 moles
CO
2
/mole alkalinity at 120, 135, and 150 C, respectively (Table 1). Tables 2, 3, and 4 include
the raw data for MDEA concentration by time and sample cylinder number. MDEA degradation
rates of 53 and 112 mmolal/day were measured for a loading of 0.2 moles CO
2
/mole alkalinity at
135 and 150 C, respectively (Figure 1). Insufficient data points were obtained at 120 C and a
loading of 0.2 due to sample cylinder failure. Previous thermal degradation studies (Thermals
No. 7 and 8) indicated that the rate of loss of MDEA in MDEA/PZ blends may decline after the
PZ is degraded, but this newly reported data suggests that at loadings in the range of 0.1 to 0.2
moles CO
2
/mole alkalinity, the MDEA loss rate in 7 m MDEA is equivalent to that reported in
7 m MDEA/2 m PZ. Figure 2 presents the data from degradation experiments with 7 m
MDEA/2 m PZ and the new MDEA degradation data (large blue diamonds) from Thermal No.
10 at a loading of 0.2 moles CO
2
/mole alkalinity and 150 C. The degradation rate in the MDEA
alone experiment is similar to that for MDEA in the blend, suggesting that the MDEA
degradation mechanism will occur independently of the presence or concentration of PZ.
The compounds dimethylaminoethanol (DMAE)(89.1), diethanolamine (DEA)(105.1), N,N-
dimethyl ethanamine (DMEA)(73.1), dimethyl piperazine (114.2), and 1-(2-hydroxyethyl)-4-
methylpiperazine (HMP)(144.2) were all tentatively identified in degraded 7 m MDEA samples
through a combination of IC standard comparison and quantification of mass using MS. We
have made tentative identification of triethalolamine (TEA)(149.1) in a sample degraded at
150 C for 21 days with MS, which corroborates the findings of Dow Chemical (Bedell, 2009);
Bedell reports that at 180 C, MDEA disproportionates to form TEA and DMEA. In degraded
samples of 7 m MDEA, we have observed both TEA and DMEA. Note that Bedell reports that
disproportionation of MDEA does not occur when the same experiment is conducted with
nitrogen in the sample container headspace.
Other key peaks appearing in degraded MDEA samples include those appearing after MDEA in
IC-MS analysis with masses of 133.1, 103.1, 176.1, 144.1, and 146.1. The peak with mass of
103.1 is possibly 2-(ethylmethylamino)ethanol; a standard for this peak has been ordered and
will be analyzed using cation chromatography to compare peak retention times. Compound
identification was compared to the work of Chakma & Meisen (1997), Dawodu & Meisen
(1996), and Bedell. In that body of literature, the presence of DEA, DMAE, and TEA were
reported in degraded MDEA samples.
62
3
Degradation mechanisms have been postulated for 7 m MDEA/2 m PZ that are consistent with
degradation mechanisms in MDEA alone. A possible mechanism postulated by Dr. Grant
Willson is the protonation of MDEA which then reacts with a strong base (PZ or other amine) at
the nitrogen to form a product (methyl-PZ or hydroxyethyl PZ) plus DEA (Figure 3). One other
mechanism that has been suggested (Rochelle) is the initial formation of DEA-carbamate upon
loading either the blend or MDEA alone, which can then polymerize to form the oxazolidone
ring (similar to MEA mechanism). This step is followed by PZ attacking the carbon adjacent to
the ester-oxygen and the formation of a PZ compound with a molecular weight of 187.1 which
has been observed in IC-MS analyses of degraded blended solvent samples. This same
mechanism could also result in a compound with a mass of 173.1 which has also been observed.
In the absence of PZ, this mechanism may lead to the formation of a polymeric diamine with a
mass of 206.1 (Figure 4). This mass was observed in at least two degraded MDEA samples
(Bombs No. 41 and 44) at a retention time of approximately 35 minutes which is consistent with
the appearance of other diamines using the current IC-MS elution program. This latter
mechanism also supports the finding that PZ plays little or no role in the degradation of MDEA.

Figure 1: Thermal No. 10, 7 m MDEA/2 m PZ, 120, 135, and 150 C, = 0.1 and 0.2

63
4
Table 1: MDEA and PZ Thermal Degradation Rates
Solvent Temp
(C)
Duration
(Days)
MDEADegRate
(mmolal/day)
PZDegRate
(mmolal/day)
DiamineAppearance
Rate(mmolal/day)
=0.1 =0.2 =0.1 =0.2 =0.1 =0.2
7mMDEA 100 63 66 1852 NA NA NA NA
120 63 0.311 3116 NA NA NA NA
7mMDEA 120 45 39 NA NA NA NA NA
7mMDEA 135 69 55 53 NA NA NA NA
7mMDEA 150 69 66 112 NA NA NA NA
7mMDEA/2m
PZ
100 54 313 194 24 61 12 22
120 54 1111 720 73 95 22 52
7mMDEA/2m
PZw/1mMFe
2+
100 42 NA 313 NA 25 NA 23
120 49 NA 1828 NA 1110 NA 123
7mMDEA/2m
PZ
135 42 98 3015 313 442 204 166
7mMDEA/2m
PZ*
150 28 857 6621 7920 5925 NA NA
*Experimentalloadingswere0.1and0.26.

Table 2: Thermal No. 10 MDEA Degradation Data (120 C)
Temperature=120C,Loading=0.1
Time
(Days) BombNo.
MDEAPk
Area
MDEA
Conc.(m)
0 0.812 7.00
3 32 0.802 6.42
7 1 0.677 4.74
21 30 0.784 5.77
35 31 0.89 6.25
45 33 0.737 5.27

Time
(Days) BombNo.
MDEAPk
Area
MDEA
Conc.(m)
0 0.899 7.00
3 38 0.78 5.64
7 136 0.774 5.46

64
5
Time
(Days) BombNo.
MDEAPk
Area
MDEA
Conc.(m)
0 0.812 7.00
3 4 0.769 6.17
7 34 0.737 5.38
21 49 0.757 5.71
35 57 0.705 5.46

Time
(Days) BombNo.
MDEAPk
Area
MDEA
Conc.(m)
0 0.899 7.00
3 53 0.796 5.75
7 55 0.606 3.92
7 50 0.573 4.03
21 58 0.697 5.08
35 46 0.632 4.14
45 50 0.616 4.45
69 48 0.607 4.85
Time
(Days) BombNo.
MDEAPk
Area
MDEA
Conc.(m)
0 0.812 7.00
3 68 0.757 5.85
3 47 0.775 5.88
21 61 0.703 5.63
35 65 0.689 6.08
69 132 0.231 1.85

Time
(Days) BombNo.
MDEAPk
Area
MDEA
Conc.(m)
0 0.899 7.00
3 59 0.74 5.41
7 41 0.681 4.46
21 44 1.671 11.69
35 69 0.171 1.09
69 70 0.066 0.52

65
6

Figure 2: 7 m MDEA Degradation Comparison to 7 m MDEA/2 m PZ

MDEA/PZ Degradation Protonation Pathway
MDEA MDEAH+
DEA
CO
2
protonation
polymerization
+ H
2
O
DEA-carb
+
Methyl-PZ
PZ

Figure 3: MDEA Protonation Pathway
66
7
Degradation Pathway Formation of DEA-Carb
from MDEA
+
MDEA
CO
2
polymerization
+ H
2
O
DEAcarbamate
+CO
2
+
MDEA

Figure 4: DEA-Carbamate Formation from MDEA
Cycling/Integrated Solvent Degradation Apparatus
During the second quarter of 2009, we continued the construction of the Integrated Solvent
Cycling/Degradation Apparatus (Figure 5). The system consists of an oxidative reactor which is
a hybrid of the Rochelle group low gas flow reactor; the reactor includes a liquid hold-up section
below the stirred section to allow entrained air bubbles to migrate into the stirred section above,
and to mimic hold-up in an absorber unit. At a liquid flow rate of 100 ml/min, the residence time
in the hold-up section is approximately three minutes. Amines are pumped through the system
by a Cole-Parmer positive displacement pump at a nominal rate of 100 to 150 ml/min to an off-
the-shelf heat exchanger manufactured by Exergy operated in a countercurrent flow pattern with
hot amine from the thermal reactor, and into two oil bath preheaters to gain heat, and into a
jacketed thermal reactor to achieve a temperature of 120+ C. Hot amines are returned to the
heat exchanger to give up heat to the amine pumped from the oxidative reactor (countercurrent
flow) and then pumped to an additional heat exchanger (Exergy) for trim cooling purposes. The
trim cooler is provided with tap water on the outer tube side which exits the heat exchanger to
the fume hood drain. Finally, the cooled amine passes through a back-pressure valve used to
control pressure and prevent flashing of solvent components (i.e., water, CO
2
, amine) in the
heated section of the system.
To date, the system has been operated to meet the basic performance goals of 55 C in the
oxidative reactor and 120+ C in the thermal reactor, simultaneously, when loaded with water. A
larger heat exchanger was acquired from Exergy during June after testing of the system indicated
that insufficient heat transfer was taking place in the existing heat exchanger. The heat
exchanger design (tube-in-tube) is the same for the small and large heat exchangers, but the
67
8
length and overall heat transfer area is 4 times greater for the larger heat exchanger. The smaller
heat exchanger is currently being used as the trim cooler.
A solvent consisting of MDEA and PZ in excess of 7 molal MDEA and 2 molal PZ with a
loading of approximately 0.1 moles CO
2
/mole alkalinity was charged into the system. The
system reached a stable operating condition whereby the oxidative reactor temperature was
approximately 50 C and the thermal reactor temperature was 114 C simultaneously. The
temperature at the inlet to the first preheater was 78 C, providing a heat exchanger hot-side
temperature approach of >35 C. Two heat exchanger flow path configurations have been
attempted: (1) the hot liquid leaving the thermal reactor is passed through the outer tube of the
cross exchanger, and (2) the hot liquid leaving the thermal reactor is passed through the inner
tube of the cross exchanger. The latter configuration was recommended by the manufacturer as
being more efficient and is the present configuration for the system. System performance is
similar for both configurations.
The installation of an additional bimetal analog thermometer in-line after the trim cooler yielded
temperature measurements of approximately 30 C, whereas a thermometer placed in the
oxidative reactor read 55+ C, suggesting that the in-line bi-metal analog thermometers
purchased from Arthur Fluids do not give accurate readings of bulk fluid temperature. The
critical and immediate task for system troubleshooting entails the installation of K-type bi-metal
thermocouples/thermowells for in-line temperature measurement on an electronic readout device
for measurement of the fluid temperature at the outlet of the thermal reactor; it is currently
believed that the temperature of the fluid leaving the thermal reactor is above 125 C when
cycling amines. We will acquire a minimum of two accurate temperature measurement
thermowells for this purpose in the immediate term to confirm that our stated temperature goals
of 55 C in the oxidative reactor and 120+ C in the thermal reactor are being simultaneously
met.
Finally, an important finding of the shake-out tests has been that for 7 m MDEA/2 m PZ loaded
to approximately 0.1 moles CO
2
/mole alkalinity, the hot section of the cycling process will need
to be maintained at a minimum pressure of 78 psig to prevent flashing. Initial degradation
studies will likely focus on 7 m MEA at lean loadings (<0.3 moles CO
2
/mole alkalinity). The
total operating pressure of this amine is unknown at this time.
68
9
HeatX
Oxidative
Reactor
PreHeater
No.1
PreHeater
No.2
Thermal
Reactor
Cooling
water
Tosink
Trim
Cooler
Reactor
Holdup
55C
T2
T1
Mixer
Cycling/IntegratedDegradationApparatus
TemperaturePerformance(MDEA/PZ):
T1 89C
T2 109C(*)
*Withwater,T2~125C
P1
PositiveDisplacement
Pump

Figure 5: Integrated Solvent Degradation/Cycling Apparatus
Conclusions
The latest experimental work indicates that : (1) MDEA rates of 66 and 112 mmolal/day were
calculated at 150 C and loadings of 0.1 and 0.2 moles CO
2
/mole alkalinity, respectively; (2)
degradation rates for degraded 7 m MDEA are similar to those for MDEA in 7 m MDEA/2 m
PZ, suggesting that PZ plays a secondary role in the overall degradation process in 7 m
MDEA/2 m PZ; (3) the activation energy for MDEA degradation was kJ/gmol assuming an
Arrhenius relationship; and (4) the construction of the integrated solvent cycling/degradation
apparatus is complete, and we are currently completing shake-out testing including installation of
improved temperature sensing equipment.
Future Work
Future work will focus on achieving accurate measurement of bulk liquid temperatures for the
integrated degradation/solvent cycling apparatus through installation of K-type bi-metal
thermowells/thermocouples. A thermowell will be installed at the outlet of the thermal reactor as
this fitting will accommodate the passage of a probe into the reactor. It is currently believed that
the system achieves the basic goals of 55 C in the oxidative reactor and 120+ C at the outlet of
the thermal reactor. Confirmation of these temperatures is essential to the operation of the
system. Once final shake-out is complete, each experiment will be conducted for a period of up
69
10
to three weeks, requiring some level of automation. Automation of the system will be achieved
through the installation and integration of temperature probes and a pressure gauge with PicoLog
software for continuous capture of data. Planned experiments for the integrated degradation
system include 7 m MEA and 7 m MDEA/2 m PZ.
During the third quarter 2009, we will initiate a new acid treatment experiment similar to the
H
2
SO
4
experiment stopped in the second quarter due to sample cylinder failure. This experiment
will be conducted with 7 m MDEA/2 m PZ. Based on a recommendation by Dr. Willson, this
experiment will utilize toluene sulfonic acid to neutralize 10% of solvent alkalinity. We will also
use new sample cylinders to eliminate failures in the experiment.
Experimental work with HPLC, IC, and IC-MS will continue in order to identify products in
degraded 7 m MDEA and 7 m MDEA/2 m PZ. We will continue to use our knowledge of the
degradation products to derive mechanisms for the degradation processes that take place in these
solvents. In particular, we will make an effort to understand whether disproportionation,
protonation, and/or the formation of carbamates are the leading mechanisms for MDEA loss.
Once the MDEA degradation process is understood, we will use this knowledge to determine the
role of PZ in the degradation of 7 m MDEA/2 m PZ. As in the second quarter, future efforts will
include confirming the presence of the DEA-based urea compound with HPLC. Analytical
methods developed for the batch thermal and oxidative degradation experiments over the last
year will be utilized to analyze samples collected from the cycling apparatus.

References
Bedell S. "Autoxidation of methyldiethanolamine - Lab studies and implications for flue gas
carbon capture". Dow Chemical Company, Trondheim CCS Conference, June 17, 2009.
Chakma A, Meisen A. "Methyl-Diethanolamine Degradation - Mechanism and Kinetics". Can J
Chem Eng. 1997:75.
Dawodu OF, Meisen A. "Degradation of Alkanolamine Blends by Carbon Dioxide". Can J Chem
Eng. 1996:74.

70
1
Oxidation and Thermal Degradation of Concentrated
Piperazine

by Stephanie A. Freeman
and the
2
July 1, 2009
Abstract
The Cation IC-Mass Spectrometer was used to identify numerous masses of unidentified
degradation products in 8 m PZ thermal degradation experiments. Although not yet positively
identified, possible candidates include N-formyl PZ, N-(hydroxymethyl) PZ, 1-methyl PZ, and
others. Further work is needed to provide positive identification.
Quantification of heavy metals in solution has been used to analyze the corrosion of 316 stainless
steel. Initial results show that thermally degraded PZ solutions contain less metal than 7 m MEA
solutions. After 18 weeks at 150 C, the iron and nickel concentrations in an 8 m PZ solution
were 0.9 and 0.7 mM, respectively. For a 7 m MEA experiment at 135 C, the final
concentrations of iron and nickel were 13.7 and 4.2 mM, respectively, after four weeks. Further
work is needed to understand if the amines themselves or their degradation products are
in concentrated PZ is 23 to 70 times less than 7 m MEA.
Introduction
Concentrated aqueous piperazine (PZ) is being investigated as a possible alternative to 30 wt %
(or 7 m) MEA in absorber/stripper systems to remove CO
2
from coal-fired power plant flue gas.
Aqueous PZ has been given a proprietary name of ROC20 for 10 m PZ and ROC16 for 8 m PZ.
Previous reports include the proprietary name, while the concentration of PZ will be explicitly
used in this document.
Preliminary investigations of PZ have shown numerous advantages over 7 m MEA systems
(Freeman, 2008a). PZ solutions have less oxidative and thermal degradation, as previously
shown at concentrations of 5 and 8 m PZ (see previous quarterly reports). The kinetics of CO
2

absorption are faster in concentrated PZ (Cullinane and Rochelle, 2006; Dugas, 2008). The
capacity of concentrated PZ is greater than that of MEA while the heat of absorption and
volatilities are comparable (Freeman, 2008).
71
2
This quarter was focused on completing the blank oxidation experiments started last quarter.
The oxidation experiments were conducted in an attempt to quantify the volatility of PZ and
other mechanical issues with the low gas flow apparatus. Work began in earnest this quarter on
quantifying metals in solution and using the mass spectrometer to identify degradation products.
A long-term thermal degradation experiment was planned and begun.

Analytical Methods
Total Inorganic Carbon Analysis (TIC)
Quantification of CO
2
loading was performed using a total inorganic carbon analyzer. In this
method, a sample is acidified with 30 wt % H
3
PO
4
to release the CO
2
present in solution
(Hilliard, 2008). The CO
2
is carried in the nitrogen carrier gas stream to the detector. PicoLog
software was used to record the peaks produced from each sample. A calibration curve was
prepared at the end of each analysis using a TIC standard mixture of K
2
CO
3
and KHCO
3
. The
TIC method quantifies the CO
2
, CO
3
-2
, and HCO
3
-
present in solution. These species are in
equilibrium in the series of reactions shown below.
CO
3
2

+ 2H
+
HCO
3
+ H
+
H
2
CO
3
CO
2
+ H
2
O
Acidification of the sample shifts the equilibrium toward CO
2
which bubbles out of solution and
is detected in the analyzer.
Acid pH Titration
Titration with 0.2 N H
2
SO
4
was used to determine the concentration of amines in experimental
samples. The automated Titrando apparatus (Metrohm AG, Herisau, Switzerland) was used for
this method. A known mass of sample was diluted with water and the autotitration method was
then used. The Titrando titrates the sample with acid while monitoring the pH. The equivalence
points are recorded. The equivalence point around a pH of 3.9 corresponds to basic amine
species in solution (Hilliard, 2008). The test is not sensitive to the type of amine, so if PZ has
degraded to ethylenediamine (EDA), the titration test will detect the sum of contributions from
the species.
Anion IC
The anion IC was used to determine the concentration of glycolate, acetate, formate, chloride,
nitrite, sulfate, oxalate, and nitrate in experimental samples. A Dionex ICS-3000 instrument
with AS15 IonPac column, 4-mm Anionic Self-Regenerating Suppressor (ASRS), carbonate
removal device (CRD), and carbonate removal from eluent generation was used as previously
described by Andrew Sexton using a linear KOH eluent concentration (Sexton, 2008). No major
modifications have been made to the method in this quarter.
Cation IC
The cation IC was used to determine the concentration of PZ and ethylenediamine (EDA) in
experimental samples. A Dionex ICS-2500 instrument with CS17 IonPac column with 4-mm
Cationic Self-Regenerating Suppressor (CSRS) was used as previously described by Andrew
Sexton with a linear increase of methanesulfonic acid (MSA) concentration in the eluent (Sexton,
2008). No major modifications have been made to the method in this quarter.
72
3
NaOH Treatment for Amides
An analytical test for the formation of amides was developed by Andrew Sexton and has been
included in the results shown here. Experimental samples were treated with 5 N NaOH (in equal
gravimetric amounts) and allowed to sit overnight. The anion IC analytical method was then
used to quantify increases in the concentrations of analytes as compared to the original samples
(Sexton, 2008). In most cases, the main increases were in the production of formate and oxalate
following NaOH treatment.
The addition of strong base reverses the amide formation reaction that has occurred during the
experiment. As an example, the formation of N-formylPZ is shown in Eqn. 1 below:

The addition of NaOH hydrolyzes the bond between the amine group and the carbon of the
formyl group to reverse the reaction. In this way, the free formate created from reversing this
reaction can be used to identify the formate bound as N-formylPZ. The same process can be
used to identify the oxalate amine of PZ.
Atomic Absorption Spectrophotometry (AA)
The concentration of individual heavy metals in solution has been measured using a Perkin
Elmer 1100B flame atomic absorption spectrophotometer (AA). The AA uses a lamp producing
a specific wavelength of light for each metal of interest. The AA also uses a flame fueled by
acetylene with air added at either a rich or lean ratio. For iron and nickel, a lean (blue) flame is
used that requires 2.5 L/min of acetylene with 8 L/min of air. Chromium requires a rich (yellow)
flame with a higher acetylene to air ratio. Samples analyzed for chromium also require the
addition of 2% ammonium chloride (NH
4
Cl) to suppress interference from iron and nickel. The
AA atomizes the liquid sample and mixes it with fuel and air that is then sent to the flame. In the
flame, the elements of interest are reduced to unexcited atoms which absorb the light at the
characteristic wavelength of that metal. The absorbance of light is measured on the opposite end
of the lamp to determine the amount of light absorbed by the analyte of interest. This is
correlated to concentration using a set of known standards and a third order polynomial
calibration curve.
Cation IC Mass Spectrometry (MS)
Mass spectrometry (MS) with cation IC is used to help identify unknown peaks on cation IC
chromatography or other unknown cations in solution. A thermal TSQ MS is attached to a
Dionex ICS-2500 IC with an IonPac CS17 analytical column (4 x 250 mm), IonPac CG17 guard
column (4 x 50 mm), Dionex AS-25 autosampler, and 4-mm CSRS is used as described above.
The main modification is that the suppressor does not run in regeneration mode, the outlet
analyte stream is sent to the MS and an additional water stream is added into the suppressor.
After separation on the cation IC column, the sample stream enters the MS, which uses
electrospray ionization (ESI) to ionize the molecules in solution. The ions then enter the mass

+

(1)
(PZ) (Formate) (N-formylPZ)
73
4
analyzer of the MS which sorts ions based on their mass to charge ratio (M/Z) using an electric
field. The standard conditions used are a range of 50 to 300 M/Z. There is a 2.2 minute delay in
the resolution of the cation IC chromatogram and the MS spectrum that is taken into account
while analyzing data.
Results
Oxidative Degradation
Continuing work on the oxidation of PZ has been completed this quarter. Samples from OE5,
OE5B, OE6 and OE6B were reanalyzed this quarter due to unexpected trends in the data. The
revised data is presented. Also, most of this quarter focused on trying to understand the
limitations of the low gas flow apparatus through four blank experiments. Finally, all the past
PZ oxidation data was reanalyzed in preparation for the 5
th
Trondheim Conference on Carbon
Capture and Sequestration (TCCCS) in June.
Reanalysis of OE5, OE5B, OE6, and OE6B Samples
Samples from OE5, OE5B, OE6, and OE6B were reanalyzed this quarter because of some
discrepancies in the trends of the initial data.
For OE5/OE5B, the concentration profiles reported in the 4
th
quarter of 2008, shown below in
Figure 1, had some unusual results (Rochelle, 2009b). The main issues were in the PZ, post
NaOH PZ, and ethylenediamine (EDA) profiles. All the OE5 and OE5B samples were re-run on
the cation IC to reevaluate the PZ and EDA concentrations before and after NaOH treatment.
The new concentration profiles are shown in Figure 2 below, with similar axes ranges for better
comparison. The three profiles mentioned demonstrate the improved analysis. No EDA was
found in OE5B, which matched the lack of EDA in OE5. It was suspect that the initial analysis
shown EDA in the final set of samples, suggesting that the removal of the solution from reactor
and the foaming test initiated the creation of EDA. The PZ concentration profile is improved in
terms of trends but no shows a larger discrepancy at 70 hours, the point in between the two
experiments. Finally, the post NaOH PZ profile is slightly improved. The original data showed
an unexplained dip in the concentration during OE5, which is now removed. The anion IC was
not rerun as the profiles for formate, oxalate, acetate, and nitrite were relatively clear in their
trends.
For OE6/OE6B, the concentration profiles reported in the 4
th
quarter of 2008, shown below in
Figure 3, also had some unusual results (Rochelle, 2009b). The main issues were in the post
NaOH formate, PZ, and post NaOH PZ profiles. All the OE6 and OE6B samples were re-run on
the cation IC to reevaluate the PZ concentrations before and after NaOH treatment. The formate
results were reported incorrectly, so a re-analysis on the anion IC was not necessary.
The original reporting of the concentration profiles is shown in Figure 3 with the new, updated
profiles shown in Figure 4.
The post NaOH formate was reported incorrectly in the 4
th
quarter of 2008. A miscalculation
made the formate appear to reach upwards of 15 mM while the actual concentration is closer to
3.0 mM. The profiles for PZ and post NaOH PZ improved with the new analysis. For PZ, the
concentrations for OE6B (after 70 hrs) are more smooth while for post NaOH PZ the
concentrations for OE6 (before 70 hrs) show a more reasonable curve.
74
5

Figure 1: Concentration Profiles for OE5 and OE5B (8 m PZ, 1 mM Fe
2+
, =0.3, 55C)
reported Q4 2008. OE5 ended and OE5B started at 70 hours.

Figure 2: Reanalyzed Concentration Profiles for OE5 and OE5B (8 m PZ, 1 mM Fe
2+
,
=0.3, 55C). OE5 ended and OE5B started at 70 hours.

75
6

Figure 3: Concentration Profiles for OE6 and OE6B (8 m PZ, 1 mM Fe
2+
, 100 mM
Inhibitor A, =0.3, 55C) reported Q4 2008. OE5 ended and OE5B started at 70 hours.

Figure 4: Reanalyzed Concentration Profiles for OE6 and OE6B (8 m PZ, 1 mM Fe
2+
, 100
mM Inhibitor A, =0.3, 55C). OE5 ended and OE5B started at 70 hours.

76
7
Limitations of the Low Gas Flow Apparatus
Previous experiments with PZ in the system showed losses of PZ in the range of 10 to 20% while
the production of measureable degradation products is very low. This apparent disconnect
between PZ loss and the production of products is what is currently being investigated.
Last quarter, the first experiment in a set of four was reported. This quarter, the remaining three
experiments were performed and the results of all are reported here. The first two, OE9 and
OE10, looked at PZ degradation in the presence of a N
2
/CO
2
gas stream without any catalysts or
additives. The second two, OE11 and OE12, looked at PZ loss in the presence of an O
2
/CO
2
gas
stream without catalysts or additives.
Samples obtained were analyzed for PZ concentration (sulfuric acid titration), CO
2
concentration
(TIC), and degradation products (anion and cation IC). The results from PZ titration (orange line
with filled circles) and IC analysis for each experiment are shown in Figures 5 through 8.

Figure 5: Concentration Profiles for OE9 (8 m PZ, 55 C, a=0.3, 100 mL/min N
2
/CO
2
)
77
8

Figure 6: Concentration Profiles for OE10 (8 m PZ, 55 C, a=0.3, 100 mL/min N
2
/CO
2
)

Figure 7: Concentration Profiles for OE11 (8 m PZ, 55 C, a=0.3, 100 mL/min O
2
/CO
2
)

78
9

Figure 8: Concentration Profiles for OE12 (8 m PZ, 55 C, a=0.3, 100 mL/min O
2
/CO
2
)

Overall, very few degradation products were formed in any experiment while the PZ
concentration decreased. There are points that appear erroneous in each graph. For OE9 (Figure
5), the apparent spike in EDA concentration at a time of 150 hours does not seem correct. The
post-NaOH treatment sample did not display this increase in EDA concentration and this is
assumed to be an artifact of something that went wrong in the cation IC at that point.
For OE10 (Figure 6), there is a definite increase in the concentration of PZ and the detected
degradation products. This increase was caused by the plug in the rubber stopper on the top of
the reactor being left off after sampling at 150 hours. This open hole allowed a large amount of
water to evaporate, concentrating all the species in solution. For further analysis, the experiment
will be cut off at 250 hours when compared to OE9.
For OE11 (Figure 7) and OE12 (Figure 8), the curves have unexpected increases in the total
formate concentration but these issues are due to the inaccuracy of the experimental method and
will not be adjusted.
Last quarter, a method for analyzing the data in terms of loss from the vapor phase was
introduced (Rochelle, 2009a). In this method, the known partial pressure of water and PZ at
55 C is used along with the known PZ and CO
2
concentrations to estimate the loss of mass from
the reactor during the experiment. A mass balance was created for each sampling point based on
knowing the initial total mass in the reactor, the sampling mass, the mass of water added each
time, and the final total mass of the reactor. The estimated mass lost from the reactor is shown
for each of the four oxidation experiments in Figures 9 and 10.
79
10

Figure 9: Estimated Total Mass Lost During OE9 and OE10 (8 m PZ, 55 C, a=0.3, 100
mL/min N
2
/CO
2
)

Figure 10: Estimated Total Mass Lost During OE11 and OE12 (8 m PZ, 55 C, a=0.3, 100
mL/min O
2
/CO
2
)
80
11
The estimated mass lost from each reactor appears to be a very inconsistent and diverse
calculation. There are a few assumptions in the estimate that may have led the values towards an
inconsistent answer. The large loss of water from the reactor during OE10 at 250 hours is
captured, however, in an estimated loss of 75 g. Although the data points are scattered, the
overall average of their values may still prove useful. To get an average value, the mass loss
recorded at each sampling data point was divided by the hours between it and the previous point.
This was done to normalize the time in between sampling and these values were the hourly loss
of mass in grams per hour. Then, and average and standard deviation was calculated and
reported in Table 1 as the estimated mass loss.
To compare this value with the other set of data that is known, the water balance loss was
calculated as well. The water balance loss assesses the differences in the overall mass balance of
the reactor by knowing the initial mass, how much water was added at each sampling point and
the final mass. This value gives an estimate of how much mass loss there is hourly in the low
gas flow apparatus operating at 55 C. These values are also shown in Table 1.
Table 1: Comparison of Mass Losses in OE9 through OE12
Expt
Estimated Mass Loss
(g/hr)
Water Balance Loss
(g/hr)
OE9 0.32 0.16 0.30
OE10 0.39 0.54 0.42
OE11 0.23 0.12 0.31
OE12 0.30 0.13 0.32

The estimation procedure predicted losses ranging from 0.23 to 0.39 g/hr with a standard
deviation up to 0.53 gram per hour. The data for OE10 clearly reflects the error in the rubber
stopper that allowed a large amount of water to evaporate at the end of the experiment.
Excluding this experiment, the range of 0.23 to 0.32 gram per hour is seen in the remaining three
experiments. The water balance loss values are in better agreement, with between 0.30 and 0.32
gram per hour expected, excluding OE10. Although this data does not help explain the losses of
PZ seen in these experiments, it does indicate the proper amount of water to be added to the
reactors to replace the water that evaporates. Before, a rule of thumb of 15 grams every two days
was used without any specific data to back this up. It was close, as these new values indicate
14.4 grams of water every 48 hours is the best estimate to maintain the level in the reactor.
A secondary goal of these four experiments was to determine the repeatability of a low gas flow
experiment. As OE10 and OE12 were repeats of the same experiment as OE9 and OE11, a
comparison of the results provides an idea of how repeatable the low gas flow procedure is. The
concentration profile of PZ and the main degradation products (formate, total formate, and EDA)
for OE9 and OE10 is shown in Figure 11. The same is shown for OE11 and OE12 in Figure 12.

81
12

Figure 11: Comparison of OE9 and OE10 Concentration Profiles (Closed Points = OE9,
Open Points = OE10)

Figure 12: Comparison of OE11 and OE12 Concentration Profiles (Closed Points = OE11,
Open Points = OE12)

82
13
Although the repeatability of the low gas flow experiment is only average based on the
concentration profile, it is not unexpected given the limitations of the apparatus. The
concentration of degradation products is so low that most of the results are within the detection
limit of the equipment being used, generally understood to be near 5 mM for our samples. The
production of EDA in the four experiments is suspect. In OE9 there is very little detectable EDA
while none was found in OE12. OE10 had a large spike of EDA in the middle of the experiment
while OE11 showed a steady production of EDA during the experiment. The detection of EDA
using the cation IC is difficult for PZ-based experiments because a high dilution (10000X) is
needed to reduce the PZ peak to reasonable levels. This dilutes the EDA concentration making
detection difficult. EDA and PZ have nearly identical response factors on the cation IC, but the
small concentration makes EDA more problematic. Also, there is a bump in the baseline exactly
where EDA elutes, adding to the difficulty.
The PZ concentration, on the other hand, should be easily measured in the Cation IC and the
scattered results indicate the propagation of error in the low gas flow experiment that is not well
understood. There are numerous opportunities for error while running the experiment and a few
while analyzing the samples. These results show that more care is needed in operating the
reactors so that large errors are not introduced due to human error or water balance issues.
Reanalysis of Past Oxidative Degradation Data
In preparation for the TCCCS in Trondheim, Norway, all of the PZ oxidation data collected so
far was revisited. The presentation for the conference focused on comparing oxidation of PZ
with MEA, and the effect of Inhibitor A on oxidation. All of the following figures have the same
axes, the percent of the initial amine against the time in hours. All of the scales in the figures are
the same to allow direct comparison. The improvement in each case is demonstrated going from
the red data series to the blue data series on each figure.
The difference in the metal-catalyzed oxidation of PZ and MEA is demonstrated in the following
three figures. In each, an 8 to 10 m PZ solution is compared to a 7 m MEA solution with the
sample metal catalyst. All experiments were performed using the low gas flow apparatus at
55 C with agitation at 1400 rpm.
The effect of iron, copper, and stainless steel metals are shown in Figures 13, 14, and 15,
respectively, below.
In all three cases, the oxidation of PZ is less than that of a comparable 7 m MEA experiment.
This improvement results in approximately 3 to 5 times less amine loss. The effect is especially
pronounced in the case of stainless steel metals where there is essentially no loss of PZ up until
275 hours.
Inhibitor A has been shown to effectively reduce the oxidation of MEA in the presence of metals
such as iron and copper (Goff, 2005; Sexton, 2008). The effect of Inhibitor A on iron-catalyzed
degradation of PZ is shown in Figure 16 below. The effect of Inhibitor A on copper-catalyzed
degradation of PZ is shown in Figure 17.

83
14

Figure 13: Loss of Amine for Iron-Catalyzed Degradation for PZ and MEA

Figure 14: Loss of Amine for Copper-Catalyzed Degradation for PZ and MEA
84
15

Figure 15: Loss of Amine for Stainless Steel-Catalyzed Degradation for PZ and MEA

Figure 16: Loss of PZ for Iron-Catalyzed Degradation with and without Inhibitor A
85
16

Figure 17: Loss of PZ for Copper-Catalyzed Degradation with and without Inhibitor A

Thermal Degradation
Thermal degradation data for concentrated PZ was also reanalyzed. A new comparison was
created to illustrate the difference in degradation rates of PZ and MEA. This new comparison is
shown in Figure 18. In this figure, the amine loss rate as a percent per week is plotted against the
inverse temperature in Kelvin. The labels of degrees Celsius were added as a reference for the
reader.

86
17

Figure 18: Comparison of Thermal Degradation Rates for Concentrated PZ and 7 m MEA
Solutions.
In this comparison, the logarithmic rate of thermal degradation is linearly related to the inverse
temperature, indicating an Arrhenius relationship between temperature and amine loss. Using an
Arrhenius relationship (Equation 2), the activation energy can be calculated from the slope of
each linear trendline. In Equation 2, the rate is represented by the rate constant k which is
directly proportional to the pre-exponential factor, A. the rate is also directly proportional to the
quantity of the exponent of the activation energy, E
A
, divided by the universal gas constant, R,
and the temperature, T. Taking the natural logarithm of each side gives a form directly related to
the trendlines in Figure 18 (Equation 3). The activation energies shown in Figure 18 were
calculated in this way.

(2)

(3)
The activation energy of concentrated PZ and MEA is nearly the same, both around 130 kJ per
mole. From the data at 135 and 150 C, the degradation rate of concentrated PZ is 23 to 70 times
less than that of 7 m MEA.
87
18
Additional thermal degradation data from screening work of Davis and new data for 8 m EDA
from Shan Zhou were also analyzed (Davis, 2009; Rochelle, 2009a). The thermal degradation
rates for 7 m (AMP), 7 m diglycolamine (DGA), 7 m hydroxyethylpiperazine (HEP), 7 m
methyldiethanolamine (MDEA)/2 m PZ, 8 m EDA, and 7 m diethylenetriamine (DETA) were
compared to the PZ and MEA data in Figure 19. These experiments were all performed at
135 C and a loading of 0.4 mole CO
2
per mole alkalinity. As Davis found, DETA has a faster
thermal degradation rate than 7 m MEA while the other amines fall between PZ and MEA in
terms of degradation. The additional EDA data for 100 and 120 C from Zhou show that the
thermal degradation of EDA is similar to MEA systems, although the activation energy is
slightly lower at 97 kJ per mole.

Figure 19: Comparison of Thermal Degradation Rates for Screened Amines Compared to
Concentrated PZ and 7 m MEA Solutions

Trace Metals Analysis
Analysis for the concentration of trace metals in solution using AA was started this quarter. This
work investigates the leaching of metals into solution in thermal degradation experiments as a
way to quantify and understand corrosion in our stainless steel cylinders. Iron, nickel, and
chromium will be the target species first as they are the primary components of stainless steel.
The first samples analyzed were TE9, a thermal degradation experiment at 150 C for 18 weeks.
The iron and nickel concentrations were measured and are shown in Figure 20 below. Iron is
quickly dissolved into solution, with the first sample showing a concentration only slightly lower
than that at 18 weeks. The iron concentration overall does not increase significantly, remaining
around 0.9 to 1.0 mM. Nickel, on the other hand, appears to increase slightly throughout the
experiment, reaching about 0.7 mM.
88
19

Figure 20: Iron and Nickel Concentration for TE9

Alex Voice measured the iron and nickel concentrations for one of Davis experiments on 7 m
MEA (Davis, 2009). This experiment at 135 C and a loading of 0.4 mole CO
2
per mole
alkalinity had an amine loss of 6.7%/week. The results of the metals concentration analysis are
shown alongside the previously reported PZ data in Figure 21. The PZ loss for that experiment
was 0.4%/week.

Figure 21: Comparison of Iron and Nickel Concentrations for PZ and MEA
89
20
The concentration of iron and nickel is significantly higher for MEA than PZ and appear to be
constantly increasing. The concentration did not reach a level value within the four weeks of the
experiment. The reason for this difference is not yet clear. One explanation is that the MEA
itself is more corrosive than PZ and causes higher metal concentrations in solution. Another
possible explanation is that the MEA is degrading much faster than the PZ and the degradation
products are responsible for initiating and continuing corrosion. Both the apparent corrosion rate
(as read through the metals concentration directly) and the degradation rate are higher for MEA
and the two effects cannot be separated at this point.
Mass Spectrometry (MS) for Identification of Unknown Compounds
The MS-Cation IC was used this quarter to attempt to identify unknown degradation compounds.
The cation IC was used to separate the samples to allow compounds to be separated before
entering the MS. Three samples were run this quarter and the results analyzed.
First, the final sample from TE9 was analyzed. TE9 was a thermal degradation experiment with
8 m PZ solution at a loading of 0.3 mole CO
2
per mole alkalinity at 150 C for 18 weeks. This
experiment lost 6.3% of the initial PZ, a loss of 0.4%/week. The results from the MS analysis
are listed in Table 2. The retention time on the cation IC column is listed in the first column and
the mass-to-charge ratio (M/Z) in the second column. When analyzing the MS data, any
significant peak of a single mass was recorded, even if a peak on the cation IC did not appear. In
this way, masses of molecules that may have been separated by the column but do not register
with the conductivity meter because they are neutral are also seen. The third column lists any
species that were positively identified, which is only PZ at this point. The final column lists
proposed degradation products based solely on the predicted molecular weight from the M/Z
ratio.
Table 2: MS Results for the Final Sample of TE9
Retention Time
(min)
M/Z Ratio
Species
Identified
Possible Species??
13.1 120.1 - MDEA
15.2 115.1 -
N-formyl PZ, 2,5-piperazinone, N-ethyl PZ,
dimethyl PZ
15.2 229.2 - -
30.5 87.1 PZ -
30.5 105.1 - -
30.5 187.1 - -
32.5 101.1 - N-methyl PZ, 2-piperazinone
32.5 117.0 - N-(Hydroxymethyl) PZ, N,N-diformyl EDA
33.2 115.1 -
N-formyl PZ, 2,5-piperazinone, N-ethyl PZ,
N,N-dimethyl PZ
33.2 113.1 - Triethylenediamine
34.6 130.1 - PZ-carbamate
35.6 199.1 - 1,1'-carbonylbis PZ
35.6 197.1 - -
35.6 200.1 - -
90
21
37.8 197.1 - -
39.9 117.1 - N-(Hydroxymethyl) PZ

The results show a variety of different masses exist in this sample but few are identified yet. The
MS-Cation IC chromatogram is shown in Figure 22. The bottom panel in the figure is the cation
IC chromatogram, as would be expected from cation IC run alone. The top panel is the total MS
signal that has been adjusted to match the position on the cation IC chromatogram.
The mass of 119.1 g/mol (M/Z of 120.1) at the beginning of the sample is most likely
contamination of methyldiethanolamine (MDEA) from the other users of the Cation IC-MS
machine. A mass of 116.1 (M/Z of 117.1) was found at multiple points on the chromatogram
which was possibly an N-(Hydroxymethyl) PZ. This molecule has been previously proposed as
a product of formaldehyde reacting with PZ and a precursor of the formation of N-formylPZ
(Freeman, 2009). The remaining masses found have not been positively identified but appear to
be PZ-based degradation products in very small amounts. Some are expected, although the
mechanism is not clear, such as 1-methyl PZ, dimethyl PZ, N,N-diformyl EDA, or 1,1-
carbonylbis PZ. Others, such as triethylenediamine, are not very likely, but are listed since the
molecular mass matches. The chemical structures of the proposed degradation products are
shown below in Figure 23 for reference.

Figure 22: Cation IC-MS Result for Final Sample of TE9; Top panel - Total MS Signal,
Bottom Panel - Cation IC Chromatogram
91
22

or

N-formyl PZ
(114.1 g/mol)
2,5-piperazinone
(114.0 g/mol)
N-ethyl PZ (or 2-ethyl PZ)
(114.2 g/mol)

or

or

N,N-dimethyl PZ (or 2,5-
dimethyl PZ)
(114.2 g/mol)
N-methyl PZ (or 2-methyl
PZ)
(100.2 g/mol)
2-piperazinone
(100.0 g/mol)

N-(Hydroxymethyl) PZ
(116.1 g/mol)
N,N-diformyl EDA
(116.0 g/mol)
Triethylenediamine
(112.2 g/mol)

PZ carbamate
(129.1 g/mol)
1,1'-carbonylbisPZ
(198.1 g/mol)
Figure 23: Chemical Structures of Possible Degradation Products
92
23

The final samples of OE9 and OE10 were also analyzed by Cation IC-MS. These two
experiments were low gas flow oxidation experiments using 8 m PZ with a loading of 0.3 mole
per mole alkalinity at 55 C for 14 days. The gas phase of the experiment was 98% N
2
/2% CO
2

as the goal of the experiment was to assess the volatility losses of the apparatus under standard
operating conditions. These two experiments were duplicates but yielded slightly different
results for the liquid phase analysis, as shown in prior sections of this report.
The Cation IC-MS results are shown in Tables 3 and 4 and Figures 24 and 25 below. The OE9
sample contained eleven different masses but only PZ was positively identified. The early peaks
of 61.1 and 119.1 g/mol (M/Z of 62.1 and 120.1) are most likely monoethanolamine (MEA) and
MDEA, respectively, contamination from other users of the MS machine. OE10 had only two
strong masses in the sample, one of which was clearly PZ. The other strong mass of 116.1 g/mol
(M/Z of 117.1) may be N-(Hydroxymethyl) PZ as discussed for the TE9 sample.

Table 3: MS Results for the Final Sample of OE9
Retention Time
(min)
M/Z Ratio
Species
Identified
Possible Species??
10.6 62.1 - Carbamic Acid, MEA
11.4 90.1 - -
13.2 120.1 - MDEA
15.6 134.1 - -
30.8 87.1 PZ -
30.8 105.1 - -
30.8 187.1 - -
32.6 101.1 - N-methyl PZ
32.6 99.1 - -
35.5 204.1 - -
35.5 202.1 - -

Table 4: MS Results for the Final Sample of OE10
Retention Time
(min)
M/Z Ratio
Species
Identified
Possible Species??
30.6 87.1 PZ -
34.0 117.1 - N-(Hydroxymethyl) PZ

93
24

Figure 24: Cation IC-MS Result for Final Sample of OE9; Top panel - Total MS Signal,

Figure 25: Cation IC-MS Result for Final Sample of OE10; Top panel - Total MS Signal,
Discussion
The MS work started this quarter demonstrated the utility of the machine to elucidate the soup of
degradation products present in heavily degraded PZ samples. Very few of the degradation
products of PZ have been identified so far as this MS work has just begun. There is a lot of work
to be done, especially with thermally degraded samples, to identify the major products.
Unfortunately, despite the presence of a variety of masses detected by the MS, there are no or
very few significant, unidentified peaks on the cation IC chromatogram. This leads to the
assumption that either the PZ is not degraded and is being lost through physical means
(entrainment, evaporation, volatility, water balance issues, etc) or it is degrading to neutral
molecules. Either way, more work is needed to identify and quantify the degradation products of
PZ.
94
25
Conclusions
The current low gas flow experiment is imprecise in terms of water balance and repeatability.
The low gas flow was first proven on systems that heavily degrade whereas PZ may require an
experimental apparatus with higher sophistication to allow the quantification of a system that
oxidizes slowly, producing few degradation products.
The MS has already proved to be a useful tool for identifying unknown compounds. There are
numerous masses identified in the thermal degradation of PZ that have yet to be positively
identified. Possible candidates include N-formyl PZ, N-(hydroxymethyl) PZ, 1-methyl PZ,
N,N-diformyl EDA, and others. Further work is needed to provide positive identification.
Quantification of heavy metals in solution has been shown to be an important tool for analyzing
corrosion at a basic level. Initial results on PZ show that thermally degraded PZ corrodes 316
stainless steel less than 7 m MEA. After 18 weeks at 150 C, the iron and nickel concentrations
in an 8 m PZ solution were 0.9 and 0.7 mM, respectively. For a 7 m MEA experiment at 135 C,
the final concentrations of iron and nickel were 13.7 and 4.2 mM, respectively, after four weeks.
Further work is needed to understand if the amines themselves or their degradation products are
in concentrated PZ is 23 to 70 times less than 7 m MEA.
Future Work
Thermal degradation of PZ has been analyzed with only a small number of experiments to date.
Multiple long-term thermal degradation studies have been started this quarter and will finish next
year. These experiments will be continuously monitored as samples are taken periodically. A
summary of the experiments started this quarter is shown in Table 5. Already, there has been
some failure of cylinders for various reasons, so the experiments may not finish exactly as
planned. Next quarter, an update will be given as to the status of these experiments.
Table 5: Thermal Degradation Experiments Started this Quarter
Expt Solution
Loading
(mol CO
2
/mol alk)
Temperature
(C)
Duration
(weeks)
Expected
End Date
TE10 8 m PZ 0.3 135 72 Oct 2010
TE11 8 m PZ 0.4 135 72 Oct 2010
TE12 8 m PZ 0.3 175 12 Aug 2009
TE13 8 m PZ 0.4 175 12 Aug 2009
TE14 8 m PZ 0.3 150 30 Jan 2010
TE15 8 m PZ 0.4 150 30 Jan 2010

Additional thermal degradation experiments are planned to begin in the next two quarters. These
experiments will look at the effect of 100 mM Inhibitor A and PZ concentration on PZ thermal
degradation. The second experiment will look at 5 and 12 m PZ.
95
26
Corrosion studies using the AA to measure the concentration of heavy metals in solution will
continue next quarter. Samples from the MEA thermal degradation work of Davis are being
analyzed for iron, nickel, and chromium (Davis, 2009). The samples from the PZ pilot plant
campaign from last fall will be analyzed as well as other PZ thermal degradation samples. The
goal of this work is to understand the corrosion characteristics of PZ in terms of temperature,
loading, and amine concentration.

References
Cullinane JT, Rochelle GT. "Kinetics of carbon dioxide absorption into aqueous potassium
carbonate and piperazine." Ind & Engr Chem Res. 2006;45(8):25312545.
Davis J. Thermal Degradation of Aqueous Amines Used for Carbon Dioxide Capture. The
University of Texas at Austin. Ph.D. Dissertation. 2009.
Dugas R. Absorption and desorption rates of carbon dioxide with monoethanolamine and
piperazine. GHGT-9. Washington, DC. 2008.
Freeman SA. Degradation of Concentrated Piperazine Used for Carbon Dioxide Capture. The
University of Texas at Austin. Research Proposal. 2009.
Freeman SA, Dugas R, Van Wagener D, Nguyen T, Rochelle GT. "Carbon dioxide capture with
concentrated, aqueous piperazine." IJGGC. Accepted for publication, 2008.
Goff GS. Oxidative Degradation of Aqueous Monoethanolamine in CO
2
Capture Processes: Iron
and Copper Catalysts, Inhibition, and O
2
Mass Transfer. The University of Texas at
Austin. Ph.D. Dissertation. 2005.
Carbonate, Piperazine, and Monoethanolamine for Carbon Dioxide Capture from Flue
Gas. The University of Texas at Austin. Ph.D. Dissertation. 2008.
Rochelle GT. "CO
2
Luminant Carbon Management Program. The University of Texas at Austin. 2009a.
Rochelle GT. "CO
2
Capture by Aqueous Absorption, Fourth Quarterly Progress Report 2008."
Luminant Carbon Management Program. The University of Texas at Austin. 2009b.
Sexton A. Amine Oxidation in CO
2
Capture Processes. The University of Texas at Austin. Ph.D.
Dissertation. 2008.

96
1

Ethylenediamine as a Solvent for CO
2
Capture

Quarterly Report for January 15 June 30, 2009
by Shan Zhou
and the
2
July 7, 2009
Abstract
Thermal and oxidative degradation products of 8 m EDA (Ethylenediamine) were measured by
cation IC, anion IC, HPLC, and TIC. Foaming was measured in samples from oxidative
degradation. Vapor liquid equilibrium and amine volatility of 8 m and 12 m EDA were
measured using hot gas FTIR. CO
2
solubility and absorption/desorption rates were measured in
the wetted-wall column. Viscosity of 8 m and 12 m EDA was measured at different
temperatures.
About 35% of the EDA was lost after 4 weeks at 135
o
C. Only about 10% EDA degraded after
16 weeks at 100
o
C. EDA urea was the most concentrated product from thermal degradation,
representing about 20% of the degraded EDA.
10% of the EDA was oxidized after 168 hours with 1mM Fe
2+
at 55
o
C. With 100 mM inhibitor
A the oxidation of EDA was insignificant. DETA (Diethylenetriamine) and formate were the
most concentrated in the quantified oxidative degradation products of EDA.
The foaminess coefficient and foam break time of EDA are close to that of 7 m MEA, and much
better than those of 8 m PZ.
Although EDA is volatile at low CO
2
loading, 8 m and 12 m EDA have amine partial pressure
comparable to 7 m MEA and 8 m PZ in expected lean loading. However, the CO
2
flux
normalized by partial pressure driving force of 12 m EDA solution was lower than that of 7 m
MEA at the same conditions. VLE models for 12 m EDA were regressed with experiment data
from the wetted-wall column and FTIR. CO
2
2
absorption were
calculated. The H
absorption
of CO
2
in EDA is similar to MEA at operating conditions, higher than
most of the other amine systems.
Introduction
Ethylenediamine (EDA) is a primary diamine, which is reported to react rapidly with CO
2
at low
CO
2
loading. Table 1 summarizes measurements in the literature of the reaction kinetics of CO
2

and aqueous EDA. Most of this work focuses on unloaded solution. The kinetics are all
reported as a second-order reaction between aqueous EDA and CO
2
. The last row is reaction
97
2
between CO
2
and MEA. Sharma (1965) used the same experimental method for EDA and MEA.
EDA has a higher reaction constant than MEA according to Sharma.
Table 1: Literature on the Reaction between CO
2
and Aqueous EDA
Authors
k
2
adjusted
to 303K
(m
3
mol
-1
s
-1
)
Method
Experiment
T(K)
C
EDA
(mol l
-1
)
Activation
Energy
(kJ/mol)
Jensen and Christensen
(1955)*
22.3
Competitive
reaction
291 0.1** -
Sharma (1965)* 21.6 Liquid jet 298 - -
Weiland and Trass (1971)* 138 Liquid jet 298.5 0.1~0.9 -
Sada et al.(1977)* 25.0 Liquid jet 298 0.4~1.2 -
Hikita et al.(1977) 17.9 Rapid mixing 278.5~317 0.007~0.03 53.6
Li et al. (2007) 17.7 Stopped-flow 298~313 0.03~0.07 50.56
MEA: Sharma(1965) 10.6 Liquid jet 298 - -
*The rate constant values were calculated from measurements at the other temperatures by Sada et al.
(1985). The activation energy of the reaction was 53.6 kJ mol
-1
according to Hikita et al. (1977).
**The author did not offer concentration units expressly.
Jensen and Christensen (1955) introduced 15% CO
2
and 85% atmospheric air into aqueous
solution containing both 0.1 M EDA and 0.2 M NaOH at 18 C. The gaseous mixture was
depleted, and the solution was immediately analyzed after being shaken for 2 minutes. Rate
constant for the reaction (1) was calculated.
NHCOOH CH CH N H CO NH CH CH N H
2 2 2 2 2 2 2 2
+
(1)
According to the other authors' calculation, the second-order reaction rate constant should be
22.3 m
3
mol
-1
s
-1
adjusted to 303K.
Sharma (1965) used several types of apparatus to obtain the rate constants for the reaction
between CO
2
and various substances. The jet apparatus was used for 0.05~10
2
m
3
mol
-1
s
-1
. The
second-order rate constant of CO
2
and EDA in this work is 15.1 m
3
mol
-1
s
-1
(10
4.18
lmol
-1
s
-1
) at
298K. However, the solvent concentration and CO
2
partial pressure were not mentioned in the
paper.
Weiland and Trass (1971) solved the following equilibrium equations using a method due to
Rosenbrock (1960). The dicarbamate form was calculated as monocarbamate.
( ) ( )
+
+ + OH NH CH N H O H NH CH N H
3 2 2 2 2 2 2 2 2
(2)
( ) ( )
+
+
+
+ + OH NH CH N H O H NH CH N H
3 2 2 3 2 3 2 2 2
(3)
( ) ( )
+ +
+ H NHCOO CH N H NHCOO CH N H
2 2 2 2 2 3
(4)
( ) ( )
-
3 2 2 2 2 2 2 2 2
HCO NH CH N H O H NHCOO CH N H + +

(5)

+
3 2
HCO OH CO
(6)
O H CO OH HCO
2
2
3 3
+ +

(7)
98
3
Most of the carbamate is in zwitterion form. Weiland and Trass also calculated an extremely
low equilibrium partial pressure of CO
2
.
A laminar liquid jet was used to measure the physical and chemical absorption rates. Chemical
absorption experiments were done using uncarbonated EDA of several concentrations under
vacuum. Exposure time was varied by adjusting the jet velocity. The pseudo-first-order rate
constants found from the slopes of the data had a linear relation with EDA concentration. The
second-order rate constant at 25.5
o
C is 100.4 m
3
mol
-1
s
-1
4% for a 95% confidence interval.
Sada et al. (1977) measured the rates of chemical absorption of carbon dioxide into aqueous
EDA using a laminar liquid jet with and without a surface-active agent (Phensol NP-100 (0.025
%)). The absorption rates were measured volumetrically at 298K and at atmospheric pressure.
The first-order reaction rate constant was derived by comparing the observed absorption rates
under the fast reaction regime with the penetration theory solution. Solvent concentration was
varied from 0.396 mol/l to 1.145 mol/l. Liquid flow rate was maintained constant. CO
2
was
depleted. The value of the rate constant for the second order reaction was derived as 17.5
m
3
mol
-1
s
-1
.
Hikita et al. (1977) used the rapid mixing method to investigate the kinetics of reaction of carbon
dioxide with EDA. The extent of the reaction at any point along the observation tube was
calculated from the observed temperature difference. The concentration of EDA varied from
0.00744 to 0.0284M. The CO
2
concentration was in the range of 0.004460.00742M. The
temperature was varied from 5.5 to 40.4
o
C. The activation energy was found to be 53.6 kJ mol
-1

(12800 cal gmol
-1
). The second-order rate constant was correlated by the empirical equation:
logk
2
=13.49-2799/T (8)
The average deviation of equation 8 is 4.1%.
Li et al. (2007) observed pseudo-first-order rate constant (k
0
) for the reaction between CO
2
and
EDA aqueous solution at 298, 303, 308, 313K. EDA concentration was varied from 26.2 mol/m
3

to 67.6 mol/m
3
. The standard model SF-51 stopped-flow equipment manufactured by Hi-Tech
Scientific, Ltd. (UK) was used. The second-order rate constant can be expressed by equation 9.

=
T
6080
exp 10 7 . 8
9
2
k
(9)
or logk
2
=9.9395-2651/T.
Most of the previous reports of the second-order rate constant agreed with each other except
Weiland and Trass (1971). The second-order rate constant was observed for fresh EDA solution.
EDA aqueous solution with CO
2
loaded has not been investigated. High EDA concentration
with CO
2
loaded solution which is close to industrial application needs to be studied.
A few papers discuss stability of EDA with CO
2
. Mulvaney and Evans (1948) introduced a
method of producing imidazolidone from EDA and CO
2
directly. At 200~230
o
C, 400~900 psi,
conversion from EDA to imidazolidone can reach 80%. This process relies on high temperature
and high pressure and may play an important role in EDA thermal degradation.
Sexton (2008) and Davis (2009) have reported a few degradation experiments with 3.5 m EDA.
This current work focused on a broader investigation of more concentrated EDA solution. Both
unloaded and loaded 8 m EDA solutions were used in thermal degradation at different
99
4
temperatures. 1 mM Fe
2+
was added into the low gas flow oxidative degradation system as a
catalyst and the effect of inhibitor A was investigated. Oxidative degradation was also measured
with high gas flow. VLE data were measured with both the hot gas FTIR and the wetted-wall
column. CO
2
capacity and heat of CO
2
absorption was calculated from the VLE model and
compared with 7 m MEA and 8 m PZ. Viscosity of partial loaded EDA aqueous solution was
measured and compared with PZ.
Thermal degradation experiments were performed as detailed in the Davis dissertation (2009).
EDA was sealed in 10 mL stainless steel bombs. All the bombs were put in a forced convection
oven at 100
o
C to 135
o
C for different times.
The low gas flow apparatus described by Sexton (2008) was used to measure oxidative
degradation. 350 mL solution was kept in a glass vessel and maintained at 55
o
C. A gas flow of
100 mL/min of 2% CO
2
/98% O
2
was introduced at the top of the gas liquid interface, while the
solution was agitated at 1400 RPM. 1 mM Fe
2+
was added as a catalyst for both oxidation
experiments.
The high gas flow apparatus was also described detailed by Sexton (2008). Solution volume was
approximate 350 mL. A gas flow of 5 L/min of 2% CO
2
/15% O
2
/83%N
2
was bubbled into the
solution. The agitator in the solution was operated at 1400 RPM at most time. Fe
2+
, Cu
2+
and
inhibitor A was added during the experiment.
The hot gas FTIR was used to measure the vapor-liquid equilibrium. The detailed experiment
method and procedure have been described in Hilliard (2008).
The wetted-wall column was used for CO
2
rate measurement. Details of the apparatus have been
described by Bishnoi & Rochelle (2000).
A foaming experiment was performed using an adapted ASTM D892 method. Detailed
apparatus and experiment procedure have been described in previous reports. (Rochelle et al.
2008). Foaminess (F) is defined in equation 10:
G
V V
G
V
F
t
g
0
= =
(10)
where V
g
is the total steady volume (m
3
) of gas trapped in the liquid, V
0
is the original liquid
volume (m
3
), Vt is the total steady volume (m
3
) of content in the cylinder during foaming, G is
the gas flow rate (m/s). The break time of foam was measured as the period required for the
foam to break completely after the gas flow was discontinued.
Analytical Methods
Cation IC: A Dionex ICS-2500 Ion Chromatography System with CS17 IonPac column with
CSRS 4 mm self-regenerating suppressor was used to determine concentrations of EDA and the
cationic species in degraded solutions. The system and analytical method are the same as
described by Davis (2009). All the samples were diluted about 10000 times for cation analysis.
Anion IC: A Dionex ICS-3000 Dual RFIC Ion Chromatography System with AS15 IonPac
column and ASRS 4 mm self-regenerating suppressor were used to quantify glycolate, acetate,
formate, nitrite, oxalate, and nitrate in the degraded solutions. The system and analytical method
are the same as described by Sexton (2008). The samples were diluted 50 times to get signals
100
5
from the detector.
HPLC: HPLC analysis of nonionic species was performed using the PL-ELS 2100 evaporative
light scattering detector. The analytical method is the same as described in Sexton (2008).
Degraded solutions were diluted 10 to 50 times to protect the detector from high signal.
TIC: The concentration of carbon dioxide in the solution was measured using the Total
Inorganic Carbon Analyzer. CO
2
loading of solutions can be calculated from the result of TIC.
Acid pH Titration: Metrohm 835 Titrando with an 801 stirrer was used to determine the total
concentration of amines in the solution. The analytical method is the same as described by
Freeman.
NaOH Treatment for Amides: Oxidative degradation samples were treated with 5 N NaOH
and then analyzed using anion IC. The increase of carboxylic acid concentration between pre-
and post-NaOH treatment samples is taken to be amide. This analytical method has been
described by Sexton (2008).
Viscosity measurement: Viscosity of solutions was measured using one Physica MCR 300
cone and plate rheometer. The measurement method has been described by Freeman (Rochelle
et al., 2008) in a previous report, with the slight modification that at 80
o
C the viscosity was
measured for 5 seconds rather than 10 seconds.
Material
Table 2: Chemical Reagent Specifications
Reagent CAS# Supplier Molecular Weight Assay % Lot #
Ethylenediamine
1
107-15-3 Strem Chemicals 60.10 99
A3566128,
A4879029(v)
Ethylenediamine
2
107-15-3 Fisher Chemical 60.10 100.2 064516
2-Imidazolidone 120-93-4 Sigma-Aldrich 86.09 96 07806DH
1 Used in most of the experiments and measurements except Wetted Wall Column.
2 Used in Wetted Wall Column.
Results
All the raw experiment data compiled from liquid phase analysis for thermal and oxidative
degradation experiments and raw data from the other experiments and measurements are listed
below.
Table 3: Thermal degradation of 8 m EDA at 100
o
C with 0.4 mols CO
2
/2 mols EDA
(Analyzed in May, 2009)
Experiment Time (weeks) 2 4 6 8** 10 13 16
Ethylenediamine Urea (mM)* 1.95 5.49 12.10 19.63 18.85 26..33 32.03
DETA (mM) 0 0 0 0 0 0 0
Formate (mM) 1.38 2.16 3.01 4.52 22 6.66 6.43
Imidazolidone (mM) 0 0 0 0.03 0.02 0.12 0.14
EDA (M) 4.2 4.2 4.1 4.1 4.1 4.1 4.0
* Calculated from EDA standard curve
**This point is from analysis in March
101
6
o
C with 0.4 mols CO
2
/2 mols EDA
Experiment Time (weeks) 1 2 3 4 5 6 7 8
Ethylenediamine Urea (mM)* 17.63 37.21 44.62 56.65 64.77 74.84 81.07 93.58
DETA (mM) 0 0 2.39 2.37 1.17 2.35 2.52 2.38
Formate (mM) 5.09 13.23 10.97 29.05 26.09 23.25 25.85 35.04
Oxalate (mM) 0.01 0 0.01 0.01 0.01 0 0.01 0
Acetate (mM) 0 0 0 0 0 0 4.07 0
Glycolate (mM) 0 0 0 0.14 0 0.15 0.21 0.26
Nitrate (mM) 0.06 0.02 0.04 0.17 0.11 0.01 0.14 0.02
Nitrite (mM) 0.03 0.01 0.01 0.06 0.05 0.01 0.07 0.02
Sulfate (mM) 0.09 0.12 0.15 0.17 0.16 0.13 0.17 0.11
Chloride (mM) 0.24 0.05 0.31 0.38 0.23 0.05 0.22 0.04
Imidazolidone (mM) 4.42 4.21 4.26 4.42 4.62 4.63 5.08 5.78
EDA (M) 4.2 4.0 3.9 3.9 3.8 3.7 3.5 3.5
RT=3.7 min on Cation (mM)** 15.87 19.14 19.28 19.12 23.63 23.78 28.00 31.29
** Calculated from MEA standard curve
o
C with 0.4 mols CO
2
/2 mols EDA
Experiment Time (weeks) 1 2 3 4
Ethylenediamine Urea (mM)* 66.16 98.01 109.27 113.74
DETA (mM) 2.42 3.64 4.75 4.86
Formate (mM) 11.65 35.66 42.36 48.93
Oxalate (mM) 0 0 0.01 0.02
Acetate (mM) 0 0 0 0
Glycolate (mM) 0 0 0.08 0.43
Nitrate (mM) 0.03 0.04 0.23 0.10
Nitrite (mM) 0.01 0.01 0.02 0.02
Sulfate (mM) 0.07 0.12 0.15 0.11
Chloride (mM) 0.13 0.05 0.60 0.05
Imidazolidone (mM) 7.39 14.59 17.83 22.64
EDA (M) 3.7 3.2 3.1 2.9
RT=3.7 min on Cation (mM)** 29.37 46.56 62.37 76.17
o
C with 0.2 mols CO
2
/2 mols EDA
Experiment Time (weeks) 2 4 8
Ethylenediamine Urea (mM)* 3.33 5.76 10.46
DETA (mM) 0 0 0
Formate (mM) 20.74 22.49 27.82
Oxalate (mM) 0.03 0.11 0.07
Acetate (mM) 0 0 0
Glycolate (mM) 0 0 0
Nitrate (mM) 0 0 0
102
7
Nitrite (mM) 0 0 0
Sulfate (mM) 0.11 0.10 0.18
Chloride (mM) 0.05 0.02 0.04
Imidazolidone (mM) 0 0 0
EDA (M) 4.8 4.8 4.7
RT=3.7 min on Cation (mM)** 10.92 11.19 10.68
Table 7: Thermal degradation of 8m EDA at 120
o
C with 0.2 mols CO
2
/2 mols EDA
Experiment Time (weeks) 1 2 3 5 8
Ethylenediamine Urea (mM)* 8.76 18.70 27.14 40.33 67.01
DETA (mM) 0 0 0 1.68 2.65
Formate (mM) 23.46 23.90 26.64 30.77 53.38
Oxalate (mM) 0.04 0.03 0.06 0.04 0.07
Acetate (mM) 0 0 0 0 0
Glycolate (mM) 0 0 0 0 0
Nitrate (mM) 0 0 0 0 0.75
Nitrite (mM) 0 0 0 0 0
Sulfate (mM) 0.14 0.11 0.17 0.22 0.36
Chloride (mM) 0.06 0.02 0.04 0.10 0.40
Imidazolidone (mM) 0 0 0 0 0
EDA (M) 4.8 4.8 4.7 4.7 4.7
RT=3.7 min on Cation (mM)** 11.28 13.24 13.19 14.89 17.03
o
C with 0.2 mols CO
2
/2 mols EDA
Experiment Time (weeks) 1 2 3 4 6 8
Ethylenediamine Urea (mM)* 39.70 79.30 101.83 124.01 151.50 162.72
DETA (mM) 1.95 4.18 4.06 4.85 5.93 7.06
Formate (mM) 30.94 45.48 50.44 61.26 68.87 82.44
Oxalate (mM) 0.04 0.04 0.04 0.04 0.02 0
Acetate (mM) 0 0 0 0 0 0
Glycolate (mM) 0 0 0 0 0 0
Nitrate (mM) 0 0 0 0 0 0
Nitrite (mM) 0 0 0 0 0 0
Sulfate (mM) 0.13 0.12 0.09 0.12 0.15 0.16
Chloride (mM) 0.03 0.03 0.03 0.04 0.03 0.09
Imidazolidone (mM) 0 0 7.46 7.73 8.72 8.70
EDA (M) 4.7 4.6 4.5 4.4 4.2 4.2
RT=3.7 min on Cation (mM)** 17.75 22.39 27.45 33.26 39.60 46.38
103
8
o
C with 0.4 mols CO
2
/2 mols EDA (02/09)
Experiment Time (weeks) 1 2 3 4 5 6 7 8
Ethylenediamine Urea (mM)* 107.62 202.78 202.78 215.03 244.41 248.93 254.32 242.06
DETA (mM) 3.46 4.25 5.90 7.07 7.29 9.37 9.34 10.82
Formate (mM) 38.02 57.88 58.89 70.27 75.41 67.53 68.46 81.06
Oxalate (mM) 0.00 0.02 0.02 0.02 0.03 0.02 0.06 0.03
Acetate (mM) 0 0 0 0 0 0 0 0
Glycolate (mM) 0.07 0.18 0.27 0.43 0.43 0.45 0.49 0.48
Nitrate (mM) 0 0 0 0 0 0 0 0
Nitrite (mM) 0 0 0 0 0 0 0 0
Sulfate (mM) 0.06 0.09 0.11 0.13 0.11 0.11 0.19 0.11
Chloride (mM) 0.03 0.04 0.04 0.06 0.03 0.04 0.03 0.02
Imidazolidone (mM) 7.61 10.06 14.35 18.42 22.97 17.01 46.48 30.34
EDA (M) 4.0 3.8 3.5 3.3 3.2 3.0 2.9 2.9
RT=4.0 min on Cation (mM)** 0 0 32.02 40.46 42.02 46.42 47.63 45.60
RT=4.5 min on Cation (mM)** 0 0 0 25.52 32.33 24.19 25.15 41.66
Table 10: Oxidative degradation of 8 m EDA at 55 C with 1 mM Fe
2+
Experiment Time (hours) 0 24 48 73 96 120 144 168
DETA (mM) 0 5.67 10.78 18.18 21.48 26.04 29.39 33.41
Formate (mM) 0.33 2.08 4.08 5.22 6.43 7.83 8.25 10.53
Formamide (mM) 3.50 10.34 19.01 22.64 26.49 26.45 34.51 40.43
Oxalate (mM) 0.01 0 0.01 0.01 0.02 0.03 0.05 0.08
Nitrite (mM) 0 0.60 0.93 1.27 1.62 2.07 3.11 3.96
Sulfate (mM) 0.91 0.90 0.88 0.88 0.85 0.84 0.82 0.90
Chloride (mM) 0 0.04 0.05 0.04 0.04 0.04 0.04 0.04
EDA (M) 4.3 4.2 4.2 4.1 4.0 3.8 3.8 3.8
RT=3.7 min on Cation (mM)* 11.55 41.23 48.84 53.54 64.22 75.40 70.84 69.60
* Calculated from MEA standard curve
Table 11: Oxidative degradation of 8 m EDA at 55
o
C with 1 mM Fe
2+
/100 mM A
Experiment Time (hours) 0 24 48 72 96 120 144 168
DETA (mM) 0 2.53 1.90 2.05 1.94 1.97 2.38 2.12
Formate (mM) 0.16 1.23 1.74 2.57 3.48 4.13 5.03 6.25
Formamide (mM) 2.53 10.58 12.45 13.35 12.80 15.10 15.88 14.54
Oxalate (mM) 0 0.06 0.01 0 0.01 0.01 0.01 0.01
Nitrite (mM) 0 0.08 0.06 0.08 0.09 0.09 0.08 0.10
Sulfate (mM) 1.01 0.93 1.00 0.89 0.96 0.88 0.88 0.97
Chloride (mM) 0.02 0.03 0.05 0.05 0.07 0.06 0.10 0.04
EDA (M) 4.2 4.3 4.3 4.1 4.3 4.1 4.1 4.1
RT=4.5 min on Cation (mM)* 0 2.45 2.89 3.06 3.92 3.54 3.77 3.99
* Calculated from MEA standard curve
104
9
Table 12: High gas flow oxidation of 8 m EDA at 55 C with multiple additives
Experiment date 0406
Experiment Time (hours) 137
Concentration at the end(mM) Average rate(mM/hr)
EDA loss 440 3.21
NH
3
yield 470 3.43
CO yield 8 0.06
CH
4
yield 14 0.10
NO
2
yield 10 0.07
CH
3
COH yield 17 0.12
Volatile EDA 28 0.20
Formate 10.49 0.077
Nitrite 1.59 0.012
Oxalate 0.091 0.0007
Nitrate 1.60 0.012
Sulfate 4.34 0.032
Chloride 0.04 0.00029
DETA 22.57 0.16
Formamide* 31.37 0.23
RT=3.5 min on Cation ** 106.33 0.78
RT=4.5 min on Cation** 21.6 0.16
* Formate standard curve does not cover the formate concentration in the samples. It is calculated from lower
concentration's standard curve.
**Calculated from MEA standard curve
Table 13: Vapor-Liquid Equilibrium and Volatility Data of EDA Solution from FTIR
Solution Addition F (10
-3
m
2
s) t(s)
8m EDA LGF 1mM Fe 31 14
8m EDA LGF 1mM Fe/100 mM A 10 6
8m EDA HGF 1mM Fe/5mM Cu100 mM A 23 9
8m EDA HGF 1mM Fe/5mM Cu 26 9
8m EDA 0.4 ldg - 27 7
8m EDA 0.4 ldg 0.01 mM Fe 30 8
8m EDA 0.4 ldg 0.1 mM Fe 31 8
8m EDA 0.4 ldg 0.2 mM Fe 28 8
8m EDA 0.4 ldg 0.3 mM Fe 30 10
8m EDA 0.4 ldg 0.5 mM Fe 29 10
8m EDA 0.4 ldg 1 mM Fe 36 10
8m EDA 0.4 ldg 200 mM CH2O 36 15
Table 14: Vapor-Liquid Equilibrium and Volatility Data of EDA Solution from FTIR
EDA
conc.(m)
T(C) T(K) Loading(mol CO
2
/
2 mols EDA)
p
CO2
(kPa) p
EDA
(kPa)
7.914 40.016 313.166 0 0 2.414E-2
7.914 60.033 333.183 0 0 1.457E-1
7.932 40.016 313.166 0.411 0.249 9.37E-4
7.932 60.006 333.156 0.411 2.695 6.189E-3
7.816 40.014 313.164 0.487 4.544 0.459E-3
7.816 60.025 333.183 0.487 24.054 2.905E-3
105
10
11.566 39.995 313.145 0 0 7.305E-2
11.566 60.000 333.150 0 0 4.312E-1
11.904 40.007 313.157 0.424 0.203 1.412E-3
11.904 59.991 333.141 0.424 2.420 8.637E-3
11.578 40.004 313.154 0.501 20.728 0.160E-3
11.578 60.046 333.196 0.501 33.060 1.199E-3
Table 15: Vapor-Liquid Equilibrium and Rate Data from Wetted Wall Column
Conc.(m) Temp(
C) CO
2
Loading P
CO2
kg'
12 40 0.362 28 2.60E-06
12 40 0.429 190 1.03E-06
12 40 0.486 4031 1.71E-07
12 60 0.219 9.3
12 60 0.29 27 1.12E-05
12 60 0.367 203 2.04E-06
12 60 0.429 1816 7.58E-07
12 60 0.491 23756 1.41E-07
12 80 0.22 49
12 80 0.292 242 5.61E-06
12 80 0.353 1522 1.67E-06
12 80 0.43 9621 7.73E-07
12 100 0.219 220
12 100 0.288 1643 5.00E-06
12 100 0.351 7128 1.99E-06
12 100 0.432 41621 5.19E-07
Table 16: Viscosity of EDA solution
Conc.(m) CO
2
Loading Temp(C) Viscosity(cP)
8 0.2 25 5.24
8 0.2 40 3.47
8 0.2 60 2.06
8 0.2 80 1.49
8 0.3 25 6.35
8 0.3 40 4.21
8 0.3 60 2.66
8 0.3 80 1.91
8 0.4 25 7.88
8 0.4 40 5.42
8 0.4 60 3.33
8 0.4 80 2.39
8 0.5 25 9.78
8 0.5 40 6.72
8 0.5 60 4.16
8 0.5 80 3.17
106
11
12 0.2 25 10.45
12 0.2 40 6.43
12 0.2 60 3.60
12 0.2 80 2.40
12 0.3 25 14.64
12 0.3 40 9.11
12 0.3 60 5.34
12 0.3 80 3.43
12 0.4 25 21.17
12 0.4 40 13.10
12 0.4 60 7.41
12 0.4 80 4.75
12 0.5 25 27.13
12 0.5 40 16.48
12 0.5 60 9.72
12 0.5 80 6.19
Thermal Degradation
Seven experiments were performed for thermal degradation with 8 m EDA loaded with 0.2 or
0.4 moles CO
2
/2 moles EDA at 100135
o
C (Tables 39). Unloaded EDA solution did not
degrade. Table 17 summarizes these results.
Table 17: Summary of thermal degradation results
EDA
concentration (m)
CO
2
Loading Temp (C)
Experiment time
(weeks)
EDA remaining
after 8 weeks (%)
8 0.4 100 16 94
8 0.4 120 8 79
8 0.4 135 4 -
8 0.4 135 8 58
8 0.2 100 8 95
8 0.2 120 8 94
8 0.2 135 8 83
Figures 1 to 3 are typical chromatograms of thermally degraded 8 m EDA. There are several
unknown product peaks. Molecular weights of the products on cation IC were determined
through IC-Mass spectra by Davis. The peak whose retention time is about 14.0 min on cation
IC should be EDA urea (N,N'-bis(2-aminoethyl) urea, shown in Figure 1). However, this species
cannot be quantified accurately because we have no standard sample.
The two peaks whose retention times are 15 min and 25.3 min on the anion IC can also be seen
on the chromatogram of the initial solution with CO
2
loaded but not on that of unloaded solution.
The size of the peaks changes little over 8 weeks. That means they are products of the reaction
between EDA and CO
2
, but not degradation products.
There are two peaks of imidazolidone on the HPLC chromatogram of degraded EDA solution.
Both peaks appear even in the chromatogram of standard imidazolidone. The summation of the
two peak areas is used to quantify imidazolidone in the samples.
107
12
-0.5
0.5
1.5
2.5
3.5
4.5
5.5
6.5
0 5 10 15 20
Time/min
R
e
s
p
o
n
s
e
/
s
*
m
i
n

Figure 1: Cation chromatogram of thermally degraded EDA, =0.4, T=135
o
C, t=4weeks
-1
1
3
5
7
9
11
13
15
0 5 10 15 20 25 30 35
Time/min
R
e
s
p
o
n
s
e
/
s
*
m
i
n

Figure 2: Anion chromatogram of thermally degraded EDA, =0.4, T=135
o
C, t=4weeks
Formate
RT=29.8
EDA
MW=60
DETA
MW=103
RT=3.7
MW=86/172
RT=4.0
MW=61/14
RT=4.5
MW=88
Background
Background
Product of reaction
between CO
2
and
EDA
EDA urea
MW=146
O
N H
2
NH NH
NH
2
EDA urea
108
13
0
50
100
150
200
250
300
0 5 10 15 20
Time/min
R
e
s
p
o
n
s
e
/
m
V

Figure 3: HPLC chromatogram of thermally degraded EDA, =0.4, T=135 C, t=4weeks
Figure 4 shows EDA as a function of time in the thermal degradation experiment. As in other
amine systems, EDA degrades faster at higher temperatures and with higher CO
2
loading. About
40% of 8 m EDA (0.4 loading) was lost after 4 weeks at 135
o
C, which agrees with Davis (2009),
who screened several amines in thermal degradation experiments. At the same conditions, amine
loss of 8 m EDA (0.4 loading) is similar to 7 m MEA (0.4 loading), which Davis found to be
37%. At 120
o
C after 8 weeks EDA loss was about 20%. Absorbent loss dropped to 10% at
100
o
C after 16 weeks. Ideal stripper temperature for EDA should not be higher than 110
o
C.
Degradation process is slower with 0.2 CO
2
loading. 8 m EDA at 0.2 loading degraded 20%
after 8 weeks at 135
o
C, which is similar to 8 m 0.4 loading at 120
o
C.
50
60
70
80
90
100
0 1 2 3 4 5 6 7 8
Time/week
E
D
A

R
e
m
a
i
n
i
n
g
/
%
100 C
120 C
135 C I
135 C II

Figure 4a: Thermal degradation of partially loaded 8 m EDA; =0.4
EDA and other
amine products
Imidazolidon
N H NH
O
109
14

Figure 4b: Thermal degradation of partially loaded 8 m EDA; =0.2
Table 18 lists carbon mass balance of three thermally degraded samples. Each unknown
molecule is assumed to consume 2 EDA molecules. Imidazolidone, formate, and DETA
consume little EDA. The unknowns from cation IC are more concentrated. As there are no
standards for the unknowns and EDA urea, there could be a significant error in the estimates of
their concentrations. Thermally degraded samples were not treated with NaOH, so there is no
formamide information in table 18.
Table 18: Carbon mass balance of 8 weeks thermally degraded samples
Actual concentration (mM) Equated EDA/Total EDA Loss (%)
Concentration
s
(mM)
135
o
C
0.4 loading
135
o
C
0.2 loading
120
o
C
0.4 loading
135
o
C
0.4 loading
135
o
C
0.2 loading
120
o
C
0.4 loading
EDA urea 328 163 94 32 39 20
Formate 81 82 35 2 5 2
IM 30 9 6 1 1 1
DETA 11 7 2 1 2 1
Unknowns 203 93 67 20 22 14
Imbalance 44 31 63
EDA loss 2055 832 940
EDA and CO
2
are known to produce imidazolidone at higher temperature and higher pressure
(Mulvaney and Evans, 1948). EDA urea may be developed from imidazolidone before the
samples are analyzed. However, imidazolidone cannot be detected in less degraded samples.
The EDA urea is always the most concentrated degradation product. After 8 m 0.2 EDA was
mixed with imidazolidone and left in a 120
o
C oven overnight. Significant EDA urea was
present then. Detailed data are in Table 19.
Table 19: Experiments with Imidazolidone and 8 m 0.2 loading EDA at 120
o
C
Initial Concentration Post reaction concentration
# EDA (M) Imidazolidone (M) EDA(M) EDA urea (M)*
1 4.32 1.39 3.87 0.24
2 4.30 0.64 4.22 0.08
*EDA urea concentration was calculated from EDA standard curve
110
15
Since the main products of EDA thermal degradation are EDA urea and other amines, CO
2
capacity of the solution does not decrease as much as EDA concentration, while the reverse
reaction of EDA urea composition may also happen when CO
2
is stripped.
Oxidative degradation
Low gas flow and high gas flow apparatus were set up for the EDA oxidative degradation
experiment. All the solutions in this section were 8 m EDA. CO
2
was loaded before the
experiment, but loading varied during the experiment because CO
2
gas was introduced into the
reactor. All the oxidative experiments are summarized in Table 20.
Table 20: Summary of oxidative degradation results
Apparatus Additives Experiment time
(hr)
Degradation rate
(mM/hr)
Low gas flow 1 mM Fe
2+
168 3.2
Low gas flow 1 mM Fe
2+
/100 mM A 168 0.1
High gas flow 1 mM Fe
2+
/5 mM Cu
2+
/100 mM A 140 3.2
Inhibitor A has been reported to be an effective degradation inhibitor for many amine systems.
A was investigated with catalyst Fe
2+
together for EDA. The oxidative degradation of 8 m EDA
is fast with 1 m Fe
2+
, and much slower with 100 mM inhibitor A. EDA loss decreases
significantly, as shown in Figure 5.
3.7
3.8
3.9
4
4.1
4.2
4.3
4.4
0 20 40 60 80 100 120 140 160
Time/hour
E
D
A

C
o
n
c
e
n
t
r
a
t
i
o
n
/
M

Figure 5: EDA loss in oxidative degradation, 55
o
C, 98% O
2
/2% CO
2

Hydroxyethyl-formamide, DETA, and formate were the most concentrated oxidation products of
3.5 m EDA in Sextons study (2008). Sexton used the HPLC-UV to detect hydroxyethyl-
formamide, which method was not used in this work. Formate is the most concentrated known
product of anion IC. The increase of formate between pre- and post-NaOH treated samples is
taken to be formamide. Glycolate and acetate were not detected in either of the low gas flow
oxidative degradation experiments. DETA (diethylenetriamine) is the most concentrated known
1 mMFe
2+
/100 mM A
1 mMFe
2+
111
16
product of cation IC. However, monoamines, whose retention times are close to MEA, are
present in greater concentrations. All of the main degradation products in the two experiments
are shown in Figure 6. The monoamines were quantified using the MEA standard curve.
0
20
40
60
80
0 20 40 60 80 100 120 140 160
Time/hour
C
o
n
c
e
n
t
r
a
t
i
o
n
/
m
M

Figure 6: Oxidative degradation products with low gas flow, 8 m EDA, =0.4, 55
o
C, 1 mM
Fe
2+

Formate decreased about 50% when A was added into the solution, while the DETA decreased
about 90%. Inhibitor A can play an important role in protecting EDA from oxidative
degradation.
Fe
2+
, Cu
2+
and inhibitor A were added into the reactor at different times during the high gas flow
oxidative degradation experiment, so that the effect of theses compounds on EDA loss is difficult
to quantify. Generally, there is low NO
x
, but high NH
3
in the degradation products. CO, CH
4
,
and acetaldehyde are also present in the gas phase. The unknown monoamines are the most
concentrated products in the liquid phase following LGF experiment. Formamide, formate, and
DETA are also present. Cu
2+
is a stronger catalyst for EDA oxidative degradation than Fe
2+
,
following the other amines. Inhibitor A can protect EDA from degradation with both Fe
2+
and
Cu
2+
. Detailed data for all kind species found in the HGF experiment are listed in Table 21.
NH
3
production rate in the gas phase is showed in Figure 7. Where there is a red line in the
figure is the time additives were added in to the solution. There are some other rapid changes in
NH
3
rate, which was caused by system changes, e.g. a reduction in N
2
.

All Formamide /A
Rt=3.7 min on Cation
DETA
Rt=3.7 min on Cation /A
All Formamide
DETA /A
112
17
Table 21: High gas flow oxidation of 8 m EDA at 55 C with multiple additives
Species
Concentration
at the end
(mM)
Average rate
(mM/hr)
Species
Concentration
at the end
(mM)
Average rate
(mM/hr)
EDA loss 440 3.21 Nitrate 1.60 0.012
NH
3
yield 470 3.43 Sulfate 4.34 0.032
CO yield 8 0.06 Chloride 0.04 0.00029
CH
4
yield 14 0.10 DETA 22.57 0.16
NO
2
yield 10 0.07 Formamide* 31.37 0.23
CH
3
COH yield 17 0.12
Volatile EDA 28 0.20
RT=3.5 min on
Cation **
106.33 0.78
Formate 10.49 0.077
Nitrite 1.59 0.012
RT=4.5 min on
Cation**
21.6 0.16
Oxalate 0.091 0.0007
* Formate standard curve does not cover the formate concentration in the samples. It is calculated from lower
concentration's standard curve.
**Calculated from MEA standard curve
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
Time (min)
N
H
3

p
o
d
u
c
t
i
o
n

r
a
t
e

(
m
M
/
h
r
)

Figure 7: NH
3
production rate in HGF experiment
Foaming tendency
All of the final samples from oxidative degradation were tested for foaming. Fe
2+
and CH
2
O
(formaldehyde) were added into undegraded, loaded EDA to detect their effect. All the tests
were performed at 40
o
C.
The EDA molecule has a short carbon chain like MEA, and has two amino groups like PZ. 8 m
+0.1
mM Fe
2+

+1 mM Fe
2+

+5 mM
Cu
2+

+50 mM A +50 mM A
113
18
EDA (0.4 loading) foamed like 7 m MEA (0.4 loading) at 40
o
C. At the same conditions, both
foaminess and break time of EDA was much less than that of 8 m PZ (0.3 loading) as compared
in Table 22.

Table 22: Foaming tendency compare at G=2 mm/s, t=40
o
C, V0=400 mL
Samples Foaminess coefficient (10
-3
m
2
s) Break time (s)
7 m 0.4 loading MEA* 21 5
8 m 0.3 loading PZ* 90 30
8 m 0.4 loading EDA 27 7
* Rochelle et al. (2008)
Fe
2+
is one of the main corrosion products in the CO
2
capture process. Fe
2+
was varied from 0.01
mM to 1 mM (Figure 8). With Fe
2+
less than 0.5 mM, EDA foaminess is not sensitive.
Foaminess increased by 35% at 1 mM Fe
2+
, but break time changed little. This behavior is more
similar to that with MEA. MEA foaming tendency was not sensitive to small amounts of Fe
2+
(Rochelle et al., 2008). Formaldehyde was considered to be one of the degradation products and
added into the solution. Foaminess increased by 35% with 200 mM formaldehyde, while break
time doubled. EDA foaminess was not sensitive to degradation.
26
28
30
32
34
36
38
0 0.2 0.4 0.6 0.8 1
Ferrous concentration /mM
F

/
1
0
-
3
m
2
s

Figure 8: Foaminess coefficient as function of [Fe
2+
] at 40
o
C
CO
2
Solubility and Amine Volatility
8 m and 12 m EDA solutions were investigated using an equilibrium experiment with the hot gas
FTIR. VLE data for 12 m EDA were also obtained with CO
2
absorption rate data from the
wetted wall column. The VLE model of 12 m EDA was given as equation 11 by regressing VLE
data from both wetted wall column and FTIR.
Lnp=49.17-16290/T-50.19+15607/T+48.08
2
(11)
where p is CO
2
partial pressure (Pa), T is temperature (K), and is loading (mol CO
2
/equivalent
amine). This is only an estimated model, relative error between experiment data and calculated
value can be 50%.
114
19
Figure 9 shows the CO
2
solubility for 12 m and 8 m EDA with CO
2
loading. The predicted lines
are also present in Figure 9. The experiment points and predicted lines of 12 m EDA seem to
agree with each other well except the two points from FTIR at 0.5 loading. Assume p
*
CO2
at rich
loading is 5 kPa, rich loading of EDA should be around 0.5. According to the absorption
reactions, two amine functions can capture one CO
2
. Loading of 0.5 is a critical value. CO
2

partial pressure on rich EDA solution is very sensitive at lower temperatures as shown in Figure
9.
Assume p
*
CO2
for lean loading is 0.5 kPa, CO
2
capacity (mol CO
2
/kg EDA + Water) can be
derived from equations 1112.
( )
1000 / MW Molarity 1
Molarity loading lean - loading rich
Capacity CO
amine amine
amine
2

+
=
(12)
CO
2
capacity of 12 m EDA is calculated to be 0.78 mol CO
2
/ kg EDA + Water, almost the same
with 8 m PZ and higher than 7 m MEA.

0.001
0.01
0.1
1
10
100
0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55
Loading / mol CO2/equivalent amine
C
O
2

p
a
r
t
i
a
l

p
r
e
s
s
u
r
e

/

k
P
a

Figure 9: CO
2
solubility in 12 m and 8 m EDA
CO
2
partial pressure over 12 m EDA is compared with 8 m PZ in Figure 10. p*
CO2
over 8 m PZ
is much higher than over 12 m EDA at the same loading and same temperature. As CO
2
loading
at operation temperature is close to critical value, CO
2
absorption rate in EDA is measured to be
about one third of PZ.
40 C
100 C
80 C
60 C
Lnp=49.17-16290/T-50.19+15607/T+48.08
2
WWC 12 m
FTIR 8 m
FTIR 12 m
115
20
0.001
0.01
0.1
1
10
100
0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55
C
O
2

p
a
r
t
i
a
l

p
r
e
s
s
u
r
e

/

k
P
a

Figure 10: Comparison of CO
2
solubility between 12 m EDA and 8 m PZ.
*Work with Chen **Dugas (2009)
CO
2
heat absorption can also be estimated from Equations 11 and 13. Heat absorption formula is
given as Equation 14.

(13)

H=(14313-9849)R (14)
where H is CO
2
absorption heat (J / mol CO
2
), R is universal gas constant (8.314 J/mol/K).
H is independent of T in Equation 13. The H value calculated from eq. 14 is considered as the
value at 70
o
C, because the temperature range in the wetted wall column experiment is
40~100
o
C. Figure 11 shows H in EDA from other authors. H in EDA solution should be a
positive function of T, and higher than that of PZ solutions.
( )
R H =
T
p
p
2
CO
/
/ 1
ln
100 C
80 C
60 C
40 C
12 m EDA*
8 m PZ**
116
21
60
70
80
90
100
110
120
130
10 30 50 70 90 110 130
t /C
H

/

k
J
/
m
o
l

C
O
2

Figure 11: CO
2
heat of absorption in EDA
Figure 12 compares the measured values of EDA volatility to those of PZ and MEA. Pure EDA
is volatile and flammable. The EDA partial pressure over unloaded solution is also much higher
than MEA and PZ. However, EDA partial pressure decreases quickly as CO
2
is loaded into the
solution. In the operating range of CO
2
loading, which is about 0.4~0.5 for EDA, amine partial
pressure over 12 m EDA solution is lower than 7 m MEA. As shown above, the heat of CO
2

absorption in EDA is higher than MEA and PZ. The interaction of CO
2
and EDA is strong.
EDA and CO
2
exist as carbamate in the solution. As EDA has two basic groups, there is little
free EDA molecule in the solution when the loading is 0.5 and EDA partial pressure drops
sharply. At the same time, CO
2
partial pressure increases dramatically. At operating loading,
amine partial pressure of 8 m PZ solution is similar to 8 m EDA.
HIKITA,1977
Trass and Weiland, 1971
Rochelle
MEA
PZ
117
22
0.1
1
10
100
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
loading / mol CO2/mol equivalent amine
p
a
m
i
n
e

/

P
a

Figure 12: EDA volatility in partial loaded solutions, 40
o
C.
*Hilliard (2008), **Nguyen (2008), ***Work with Nguyen.
Reaction Rate
The fast reaction rate between EDA and CO
2
is one of the reasons we want to explore EDA as a
solvent for CO
2
capture. Nevertheless, most of the literature is focused on the reaction between
unloaded EDA and CO
2
, which differs from the operating conditions in a typical absorber.
The absorption rate was measured in the wetted wall column. The experimental and data
analysis methods are the same as those used by Ross Dugas and Xi Chen for other amines.
Generally, all experiment runs are at less than 50% gas film control. The overall mass transfer
coefficient K
G
was calculated from the flux and driving force. The gas phase mass transfer
coefficient k
g
was calculated from a correlation (Cullinane, 2005). The liquid film mass transfer
coefficient, k
g
', was calculated using Equation 15.
'
g g G
k
+
k
=
K
1 1 1
(15)
where k
g
' is obtained liquid film mass transfer coefficient, which is a function of both physical
diffusion of reactants and products and reaction kinetics.
Results with12 m EDA are presented in Figure 13. The CO
2
partial pressure was too low to be
measured at low CO
2
loading. CO
2
partial pressure is very sensitive to CO
2
loading in EDA
solution when loading is greater than 0.4. Conformation of carbamate at one side of EDA
weakens the other amino group. The CO
2
absorption rate decreases rapidly with increased
loading. Figure 14 shows the obtained k
g
' of 7 m MEA, 8 m PZ, and 12 m EDA. k
g
' of 12 m
EDA is higher than 7 m MEA when CO
2
partial pressure is low, but lower than 7 m MEA when
the p
*
CO2
is in operation range.
8m PZ
7m MEA
8m EDA
12m EDA
118
23
10 100 1000 10000 100000
p*CO2(Pa)
k
g
'
(
m
o
l
/
s

P
a

m
2
)
7m MEA 40C 8m PZ 40C 12m EDA 40C
7m MEA 60C 8m PZ 60C 12m EDA 60C

Figure 13: Liquid film mass transfer coefficient of 12 m EDA, 7 m MEA, and 8 m PZ
Viscosity
The viscosity of 8 m and 12 m EDA with 0.2 to 0.5 moles CO
2
/2 EDA was measured using an
Anton Parr cone and plate viscometer (Figures 14a d). Temperature varied from 25 C to
80 C. Generally, the measured time for each sample is 100 s. Since evaporation is much faster,
the measuring time for samples at 80 C was 50 s. As with other amines, the viscosity increases
with EDA and CO
2
loading. The logarithm of viscosity has a linear relationship with the
reciprocal of temperature.
1
10
0.1 0.2 0.3 0.4 0.5 0.6
V
i
s
c
o
s
i
t
y

/

c
P
25C
40C
60C
80C

Figure 14a: Viscosity of 8 m EDA solution with loading
1E-7
1E-6
1E-5
1E-4
119
24
1
10
0.0028 0.0029 0.0030 0.0031 0.0032 0.0033 0.0034
1/T / 1/K
V
i
s
c
o
s
i
t
y

/

c
P
0.2 loading
0.3 loading
0.4 loading
0.5 loading

Figure 14b: Viscosity of 8 m EDA solution with temperature
1
10
100
0.1 0.2 0.3 0.4 0.5 0.6
V
i
s
c
o
s
i
t
y

/

c
P
25C
40C
60C
80C

Figure 14c: Viscosity of 12 m EDA solution with loading
120
25
1
10
100
0.0028 0.0029 0.0030 0.0031 0.0032 0.0033 0.0034
1/T / 1/K
V
i
s
c
o
s
i
t
y

/

c
P
0.2 loading
0.3 loading
0.4 loading
0.5 loading

Figure 14d: Viscosity of 12 m EDA solution with temperature
Function groups have an important effect on the viscosity of aqueous amines. Additional groups
may increase viscosity (Hartono, 2009). Higher concentrations of basic groups also increase
viscosity linearly (Freeman, 2008). 12 m EDA solution has a higher viscosity value than 8 m
EDA solution with the same loading because of higher concentrated basic groups. 8 m EDA
solution seems to have lower viscosity than 8 m PZ at the same loading because of the smaller
molecule weight.

Conclusions
Thermal loss of 8 m EDA (0.4 loading) is about 35% at 135
o
C after 4 weeks, but only 10% at
100
o
C after 16 weeks. EDA urea is the most concentrated product, representing only 20% of the
lost EDA.
Oxidation of EDA at 55
o
C with 2%/98% CO
2
/O
2
consumed 10% after 168 hours with 1 mM
Fe
2+
. However EDA loss was negligible after 168 hours with 100 mM inhibitor A.
Foaminess of 8 m EDA (0.4 loading) is comparable to 7 m MEA (0.4 loading) with or without
formaldehyde.
With an equilibrium CO
2
partial pressures range of 500 to 5000 Pa t 40
o
C, the lean loading of 8
or 12 m EDA is greater than 0.4 and the rich loading is less than 0.5, giving a capacity with 12 m
EDA of 0.78 moles CO
2
/kg EDA + H
2
O.
Amine partial pressure of 12 m EDA is comparable with 7 m MEA and 8 m PZ at operating
conditions.
CO
2
absorption rate in 12 m EDA is about 50% that of 7 m MEA at rich conditions.
The viscosity of 8 m EDA is a little less than 8 m PZ and comparable with other amines.
EDA is not a perfect absorbent for CO
2
capture. Slow absorption rate and high amine demand
for equivalent CO
2
is the most disadvantageous.
121
26
Future Work
Density is one important property of partially loaded EDA. The density of EDA at different
loadings will be measured at different temperatures. Further summary and analysis will be the
last part of the solvent evaluation.

References
Bishnoi S, Rochelle GT. Absorption of CO
2
into aqueous piperazine: reaction kinetics, mass
transfer and solubility. Chem Engr Sci. 200;55:5532-5542.
Cullinane JT. Thermodynamics and Kinetics of Aqueous Piperazine with Potassium Carbonate
for CO
2
Absorption. The University of Texas at Austin. Ph.D. Dissertation. 2005.
Davis J. Thermal Degradation of Amines Used in Carbon Dioxide Removal Applications. The
Dugas R. Absorption and desorption rates of carbon dioxide with monoethanolamine and
piperazine. Energy Procedia. 2009;1:11631169.
Freeman S et al. Carbon dioxide capture with concentrated, aqueous piperazine Energy
Procedia. 2009;1:14891496.
Hartono A. Characterization of diethylenetriamine (DETA) as absorbent for Carbon Dioxide.
Norwegian University of Science and Technology. Ph.D. Dissertation. 2009.
Hikita H, Asai S, Ishikawa H, Honda M. The kinetics of reactions of carbon dioxide with
monoisopropanolamine, diglycolamine and ethylenediamine by a rapid mixing method.
Chem Eng J. 1977;14:2730.
Jensen A, Christensen R. Studies on Carbamates XI the Carbamate of Ethylenediamine. Acta
Chem Scand. 1955;9(3):486492.
Li J, Henni A, Tontiwachwuthikul P. Reaction kinetics of CO
2
in aqueous ethylenediamine,
ethyl ethanolamine, and diethyl monoethanolamine solutions in the temperature range of
298313K, using the stopped-flow technique. Ind Eng Chem Res. 2007;46:44264434.
Mulvaney JF, Evans RL. Synthesis of Ethylene Urea (Imidazolidone-2). Ind & Engr Chem.
1948;40:393397.
2
Capture by Aqueous Absorption, Second Quarterly Progress Report
2008. The University of Texas at Austin. 2008.
2
2008. The University of Texas at Austin. 2008.
Rosenbrock HH. Computer Journal. 1960;3:175.
2
Dissertation. 2008.
Sharma MM. Kinetics of reactions of carbonyl sulphide and carbon dioxide with amines and
catalysis by Brnsted bases of the hydrolysis of COS. Trans Faraday Soc. 1965;6:681
688.
122
27
Sada E, Kumazawa H, Butt MA. Absorption of carbon dioxide into aqueous solutions of
ethylenediamine: Effect of interfacial turbulence. Chem Engr J. 1977;13:213217.
Sada E, Kumazawa H, Han ZQ. Kinetics of Reaction between carbon dioxide and
ethylenediamine in non-aqueous solvents. Chem Engr J. 1985;31:109115.
Trass O, Weiland RH. Absorption of carbon dioxide in ethylenediamine solutions II. Pilot Plant
Study of Absorption and Regeneration. Can J Chem Engr. 1971;49:773781.
Weiland RH, Trass O. Absorption of carbon dioxide in ethylenediamine solutions I. Absorption
kinetics and equilibrium. Can J Chem Engr. 1971;49:767772.

123
1
Mass Transfer and CO
2
Partial Pressure Results

(Dissertation Chapter 4)
by Ross Dugas
and the
2
July 10, 2009
This chapter includes the experimental results of the diaphragm cell and the wetted wall column.
The diaphragm cell measures diffusion coefficients in CO
2
loaded MEA and PZ solutions. The
wetted wall column measures CO
2
partial pressure and CO
2
absorption/desorption rate in CO
2

loaded MEA and PZ. The chapter concludes with a section predicting CO
2
reaction rates in
MEA solutions.
4.1 NECESSITY OF EXPERIMENTS
4.1.1 Need for Diaphragm Cell Experiments
Work by Versteeg and Van Swaaij (1988) has shown that the diffusion of N
2
O and CO
2
in
aqueous amines generally follow the viscosity dependence in Equation 4.1. Snijder et al. (1993)
have shown that alkanolamine diffusion in aqueous alkanolamine solutions follow the viscosity
dependence in Equation 4.2.
( ) ( )
Water O N eSolution A O N
D CONSTANT D
8 . 0
2 min
8 . 0
2
= = (4.1)
( ) ( )
Water e A eSolution A e A
D CONSTANT D
6 . 0
min min
6 . 0
min
= = (4.2)
The N
2
O and CO
2
diffusivity relationship in Equation 4.1 was confirmed with MDEA solutions
but resulted in less satisfactory results for AMP (Tomcej and Otto, 1989; Xu, Otto et al., 1991).
If the diffusion relationships are dependent on amines, the relationship in Equation 4.1 may not
directly apply to MEA, PZ, or MEA/PZ. The current work uses a diaphragm cell to measure
diffusion coefficients in MEA and PZ.
4.1.2 Need for the Current Wetted Wall Column Experiments
A significant amount of data is available on rate studies concerning the reaction of CO
2
and
monoethanolamine. Numerous references are compiled in Table 2.1 of the Literature Review.
Almost all of this data was obtained at low MEA concentration in unloaded solutions.
Unfortunately, these data do not allow for the prediction of CO
2
absorption/desorption rates in
concentrated, CO
2
loaded monoethanolamine solutions. Concentrated, CO
2
loaded MEA
solutions are non-ideal solutions which can have significant activity coefficient and ionic
strength effects not seen in the present literature data. Therefore, to predict CO
2

absorption/desorption rates in concentrated, CO
2
loaded MEA solutions, rate experiments with
concentrated, CO
2
loaded MEA solutions must be performed.
124
2
Currently, only Aboudheir (2003) has provided a major data source on the CO
2
reaction rates in
concentrated, loaded MEA. Dang (2003) provides a few more data points for comparison. This
work provides a second major data source of CO
2
reaction rates in concentrated, CO
2
loaded
MEA.
As Table 2.2 of the Literature Review summarized, there is little CO
2
rate data in piperazine
solutions. Of the 4 literature sources, none have been tested near industrial conditions. 1.5 m PZ
was the most concentrated solution studied. None of the CO
2
rate data in piperazine have been
obtained using CO
2
loaded solutions. The current work measures CO
2
rates at high CO
2
loading
in 2, 5, 8, and 12 m PZ. These data should provide a much greater insight into the CO
2
capture
performance of industrial systems.
4.2 AMINE CONCENTRATION BASIS MOLALITY, MOLARITY, AND WT %
Wetted wall column rate experiments were conducted on 7, 9, 11, and 13 m MEA, 2, 5, 8, and
12 m PZ, and 7 m MEA/2 m PZ. A molality basis is convenient in experimentation because it
does not change with the addition of other components and does not require density
measurements. However, many other researchers are accustomed to molarity or amine mass
fraction. Table 4.1 shows the experimental amine concentrations on each basis. Molarity and
mass fraction are presented on a CO
2
-free basis. Calculated molarities use the density at 25 C.
The correlation by Weiland (1998) was used to determine MEA densities. PZ densities were
obtained by extrapolating density measurements by Freeman back to zero loading (Rochelle,
Dugas et al., 2008). A measured density of 1.02 was used for 7 m MEA/2 m PZ.
Table 4.1: Concentration conversions for the wetted wall column experiments
Molality Molarity Mass
m M wt%
7 5.0 30
9 5.9 35
11 6.7 40
13 7.4 44
2 1.7 15
5 3.6 30
8 4.9 41
12 6.2 51
M
E
A
/
P
Z
7 - MEA
2 - PZ
4.5 - MEA
1.3 - PZ
27 - MEA
11 - PZ
M
E
A
P
Z

Molality is defined as mol/kg water while molarity is defined as mol/l solution. Molality and
molarity do not scale linearly.
4.3 DIAPHRAGM CELL RESULTS
Diffusion experiments were carried out in a diaphragm cell for 7, 9, and 13, m MEA and 2, 5,
and 8 m PZ. Table 4.2 summarizes results for each experiment.
The membrane-cell integral diffusion coefficient, D, is a complex concentration and time
averaged value which is somewhat different from the fundamental diffusion coefficient, D. The
fundamental diffusion coefficient is defined with respect to one species. The membrane-cell
integral diffusion coefficient is the effective diffusion coefficient of all of the CO
2
species in
solution. More details are given in Experimental Methods, section 3.1.3 of the dissertation.
125
3
Table 4.2: Diaphragm cell results for monoethanolamine and piperazine solutions
CO
2
Loading Temp Time Visc Approach to Material Balance
(mol/mol
alk
) (C) (h) (cP)
(m
2
/s)
Equilibrium (%) Error (%)
0.25-0.35 236 2.8 2.2E-10 34 7
0.45-0.55 261 3.3 4.7E-10 62 4
0.25-0.35 93 3.8 3.7E-10 19 16
0.44-0.49 138 4.5 3.2E-10 22 25
13m MEA 0.16-0.31 261 5.8 3.8E-10 58 7
0.24-0.32 72 1.7 6.1E-10 24 14
0.35-0.41 146 1.6 5.8E-10 37 26
0.25-0.32 166 5.2 2.5E-10 20 32
0.33-0.39 308 5.4 2.7E-10 48 3
0.25-0.29 237 14.5 1.2E-10 20 27
0.34-0.41 409 16.5 8.9E-11 27 4
30
Solution
7m MEA
9m MEA
2m PZ
5m PZ
8m PZ
D

The viscosity in Table 4.2 is the viscosity of the average loading of the solutions in the two
chambers. For MEA, the viscosity was obtained from correlations produced by Weiland (1998).
For PZ, the viscosity was obtained from a regression by Plaza (Rochelle, Chen et al., 2009) using
viscosity measurements by Freeman (Rochelle, Sexton et al., 2008). Plaza regressed PZ
viscosity to a similar form used by Weiland (1998) and shown in Equation 4.3. In Equation 4.3,
refers to the mass fraction of the amine and refers to the CO
2
loading in mol
CO2
/mol
alk
. In
the correlation by Weiland (1998) refers to the mass percent of the amine. Weiland uses a
CO
2
-free basis for the mass percentage. Plaza incorporates the CO
2
mass into the mass fraction.

+ + + + + +
=
2
2
] 1 ) ( )][ ( ) [(
exp
T
g fT e d c T b a
O H
(4.3)
The obtained constants from the correlation are shown in Table 4.3.
Table 4.3: Viscosity parameters for PZ solutions
Parameter Value
a 487.52
b 1389.31
c 1.58
d 4.50
e 8.73
f 0.0038
g 0.30

Table 4.2 also shows an approach to equilibrium and a material balance for each experiment.
The material balance was calculated by comparing the change in CO
2
loading of the bottom
chamber to the change in CO
2
loading in the top chamber. It does not represent the total amount
of CO
2
which was lost during an experiment. A 25% material balance error could be represented
as the top CO
2
loading changing from 0.20 to 0.215 while the bottom chamber CO
2
loading
changed from 0.30 to 0.28.
The approach to equilibrium is the change in CO
2
loading in a chamber divided by half the
difference in CO
2
loading of the original two solutions. If 0.2 and 0.3 CO
2
loading solutions
reach 0.225 and 0.275 CO
2
loadings by the end of the experiment, then the approach would be
50%. A 100% approach would result in both solutions achieving a 0.25 CO
2
loading.
126
4
12 m PZ was also tested in the diaphragm cell but meaningful results were not obtained. The
Mettler Toledo DE40 density meter was not able to reproducibly analyze the 12 m PZ samples.
The solutions may have been too viscous or may not have been homogeneous. 12 m PZ at 20 C
(the temperature of the density measurement) is about 5060 cP depending on the CO
2
loading
(Rochelle, Sexton et al., 2008).
Diffusion coefficients are typically a function of viscosity. Figure 4.1 plots the diffusion
coefficient and viscosity data in Table 4.2. The diffusion coefficient of 1 m piperazine is shown
for comparison (Sun, Yong et al., 2005).
1x10
-10
1x10
-9
1 10
D
i
f
f
u
s
i
o
n

C
o
e
f
f
i
c
i
e
n
t

(
m
2
/
s
)
Viscosity (cP)
Sun (2005)
1 m PZ
Circles - 7, 9, 13 m MEA
Diamonds - 2, 5, 8 m PZ
m = -0.72
30C

Figure 4.1: Diffusion coefficient-viscosity relationship for MEA and PZ (Sun, Yong et al.,
2005)
The data seem to show a slope of -0.72 although a fair amount of data scatter is apparent. This
0.72 value seems reasonable compared to a 0.8 dependence for N
2
O and a 0.6 dependence for
amines cited by Versteeg and Van Swaaij (1988). The membrane-cell integral diffusion
coefficient cited here most likely refers to the carbon dioxide carrying species since CO
2
loading
changes were measured. In that case the measured diffusion coefficient would most closely
represent the diffusion coefficient of the carbamate species.
The data also compare favorably to the piperazine diffusion coefficient data point measured by
Sun (2005). Extrapolating the trend line in Figure 4.1 to the Sun data point viscosity would
show the trend line slightly under predicting the diffusion coefficient. However, the diffusion
coefficient of a PZ carbamate may be slightly lower than PZ due to the larger size and possibly
more hydrogen bonding on the ionic species.
Overall, the 0.72 dependence the diaphragm cell provides seems reasonable and can be used in
the rate model.
127
5
4.4 WETTED WALL COLUMN RESULTS
4.4.1 Tabulated Wetted Wall Column Data
Tables 4.44.6 provide tabulated k
g
rate data and equilibrium CO
2
partial pressure data. Section
2.3.1 explains why rate data are presented in terms of k
g
rather than rate constants. k
g
is the
liquid film mass transfer coefficient in gas film units, defined by Equation 4.4. Figure 2.1 in the
Literature Review graphically defines kg.

) (
*
, 2 , 2
2 '
b CO i CO
CO
g
P P
N
k
= (4.4)
Each row of the following tables represents the results of 6 experimental inlet CO
2
partial
pressures. Much more detailed data including gas flow rates, pressures, and inlet and outlet CO
2

partial pressures can be found in Appendix A. Appendix A also includes the liquid film physical
mass transfer coefficient, k
l
o
, and the gas film resistance percentage of each experiment.
Experiments were designed to be less than 50% gas film controlled. In some experiments k
l
o

may be limiting such that CO
2
mass transfer is restricted by diffusion limitations in the system.
Table 4.4: CO
2
equilibrium partial pressure and rate data obtained from the wetted wall
column with aqueous MEA
MEA Temp CO
2
Loading P*
CO2
k
g
' MEA Temp CO
2
Loading P*
CO2
k
g
'
m C mol/mol
alk
Pa
mol/s
.
Pa
.
m
2
m C mol/mol
alk
Pa
mol/s
.
Pa
.
m
2
0.252 15.7 3.34E-06 0.261 14.0 3.36E-06
0.351 77 1.40E-06 0.353 67 1.76E-06
0.432 465 7.66E-07 0.428 434 7.14E-07
0.496 4216 3.47E-07 0.461 1509 4.34E-07
0.252 109 2.92E-06 0.261 96 3.35E-06
0.351 660 1.70E-06 0.353 634 1.80E-06
0.432 3434 9.28E-07 0.428 3463 8.71E-07
0.496 16157 3.76E-07 0.461 8171 5.02E-07
0.271 1053 2.85E-06 0.256 860 4.35E-06
0.366 4443 1.87E-06 0.359 3923 1.93E-06
0.271 5297 2.98E-06 0.256 4274 3.72E-06
0.366 19008 1.40E-06 0.359 18657 1.56E-06
0.231 10.4 - 0.252 12.3 3.08E-06
0.324 34 1.86E-06 0.372 84 1.28E-06
0.382 107 1.40E-06 0.435 491 6.96E-07
0.441 417 8.36E-07 0.502 8792 1.62E-07
0.496 5354 3.02E-07 0.252 100 2.98E-06
0.231 61 3.80E-06 0.372 694 1.54E-06
0.324 263 2.44E-06 0.435 3859 7.56E-07
0.382 892 1.47E-06 0.502 29427 1.93E-07
0.441 2862 9.57E-07 0.254 873 4.21E-06
0.496 21249 3.24E-07 0.355 3964 1.85E-06
0.265 979 3.24E-06 0.254 3876 3.66E-06
0.356 4797 1.75E-06 0.355 18406 1.56E-06
0.265 4940 3.40E-06
0.356 21534 1.33E-06
7
9
80
100
40
60
40
60
80
100
11
13
80
100
40
60
40
60
80
100

12 m PZ experiments at 40 C could not be run in the wetted wall column due to the high
viscosity of the solution. A thin liquid film on the surface of the stainless steel rod could not be
maintained. Also 12 m PZ samples with approximately 0.40 CO
2
loading were not tested due to
solubility limitations.
128
6
Table 4.5: CO
2
column with aqueous PZ
PZ Temp CO
2
Loading P*
CO2
k
g
' PZ Temp CO
2
Loading P*
CO2
k
g
'
m C mol/mol
alk
Pa
mol/s
.
Pa
.
m
2
m C mol/mol
alk
Pa
mol/s
.
Pa
.
m
2
0.240 96 3.32E-06 0.231 68 4.27E-06
0.316 499 2.04E-06 0.305 530 1.98E-06
0.352 1305 1.39E-06 0.360 1409 1.14E-06
0.411 7127 5.55E-07 0.404 8153 3.53E-07
0.240 559 3.33E-06 0.231 430 4.41E-06
0.316 2541 2.06E-06 0.305 2407 2.02E-06
0.352 5593 1.38E-06 0.360 7454 9.57E-07
0.411 25378 3.84E-07 0.404 30783 3.20E-07
0.239 2492 3.34E-06 0.253 3255 3.61E-06
0.324 12260 1.32E-06 0.289 9406 1.97E-06
0.239 9569 2.40E-06 0.253 13605 2.18E-06
0.324 39286 9.12E-07 0.289 32033 1.20E-06
0.226 65 4.39E-06 0.231 331 4.19E-06
0.299 346 2.57E-06 0.289 1865 1.85E-06
0.354 1120 1.69E-06 0.354 6791 7.73E-07
0.402 4563 7.93E-07 0.222 2115 4.24E-06
0.226 385 4.75E-06 0.290 9141 1.48E-06
0.299 1814 2.62E-06 0.222 7871 3.78E-06
0.354 5021 1.80E-06 0.290 33652 8.30E-07
0.402 17233 6.59E-07
0.238 2192 4.67E-06
0.321 9699 1.91E-06
0.238 8888 3.52E-06
0.321 36960 1.02E-06
60
2
5
80
100
80
100
40
60
40
40
60
60
8
12
80
100
80
100

Table 4.6: CO
2
column with 7 m MEA/2 m PZ
MEA PZ Temp CO
2
Ldg P*
CO2
k
g
'
m m C mol/mol
alk
Pa
mol/s
.
Pa
.
m
2
0.242 27 3.45E-06
0.333 166 1.96E-06
0.416 1425 8.76E-07
0.477 7418 4.32E-07
0.242 178 4.00E-06
0.333 1256 2.03E-06
0.416 7122 9.08E-07
0.477 33704 3.75E-07
0.242 1138 4.29E-06
0.333 6174 2.12E-06
0.242 4340 4.83E-06
0.333 26571 1.23E-06
7 2
40
60
80
100

4.4.2 Equilibrium CO
2
Partial Pressure
The figures in the following sections graphically represent the data in Tables 4.44.6 along with
applicable literature data.
129
7
4.4.2.1 Monoethanolamine
Figure 4.2 shows wetted wall column obtained CO
2
equilibrium partial pressure values in 7, 9,
11, and 13 m MEA compared to Jou (1995) and Hilliard (2008) values. Hilliard used an
equilibrium cell to measure CO
2
partial pressures with an FTIR (Fourier transform infrared
spectroscopy) analyzer to quantify the CO
2
concentration. Jou also measured the equilibrium
partial pressure with an equilibrium cell.
1
10
100
1000
10000
100000
1000000
0.05 0.15 0.25 0.35 0.45 0.55
CO
2
Loading (mol/mol
alk
)
P
*
C
O
2

(
P
a
)
Hilliard 3.5 m MEA
Hilliard 7 m MEA
Hilliard 11 m MEA
7 m MEA
9 m MEA
11 m MEA
13 m MEA
Jou 7 m MEA
Open Points Hilliard (2008) 3.5, 7, 11 m MEA
Dashes Jou (1995) 7 m MEA
Filled Points Current Work 7, 9, 11, 13 m MEA
100C
80C
60C
40C
1
10
100
1000
10000
100000
1000000
0.05 0.15 0.25 0.35 0.45 0.55
CO
2
Loading (mol/mol
alk
)
P
*
C
O
2

(
P
a
)
Hilliard 3.5 m MEA
Hilliard 7 m MEA
Hilliard 11 m MEA
7 m MEA
9 m MEA
11 m MEA
13 m MEA
Jou 7 m MEA
Open Points Hilliard (2008) 3.5, 7, 11 m MEA
Dashes Jou (1995) 7 m MEA
Filled Points Current Work 7, 9, 11, 13 m MEA
100C
80C
60C
40C

Figure 4.2: Equilibrium CO
2
partial pressure measurements in MEA solutions at 40, 60,
80, and 100 C (Jou, Mather et al., 1995; Hilliard, 2008)
The 3.5, 7, and 11 m MEA data by Hilliard (2008), the 7 m MEA data by Jou (1995), and the
current work at 7, 9, 11, and 13 m MEA agree well at each of the 4 temperatures. The current
data represented by the filled data points seem to show a little deviation from the other data near
0.5 loading. These data seem to show a dependence with respect to amine concentration. At
both 40 and 60 C near 0.5 loading the 13 m data has a higher CO
2
partial pressure than the 11 m
MEA data which is higher than the 7 m MEA data. The 11 m MEA data by Hilliard both at 40
and 60 C show a higher CO
2
partial pressure than 7 or 3.5 m MEA data at high CO
2
loading.
However, the 7 m MEA data from the wetted wall column provide higher CO
2
partial pressure
values than the 7 m MEA data by Hilliard (2008) or Jou (1995).
The effect of amine concentration on the CO
2
partial pressure of the MEA system at high loading
is not necessarily erroneous. Amine concentration should not affect CO
2
equilibrium partial
pressures for carbamate producing systems when compared at equivalent CO
2
loading.
However, amine concentration is extremely important in bicarbonate producing systems. MEA
systems begin producing significant bicarbonate concentrations approaching 0.5 loading. This
difference is based on the stoichiometry of the carbamate and bicarbonate reactions. The
mathematics of the difference are explained in Appendix B.
130
8
The increased CO
2
partial pressure of the higher MEA concentration near 0.5 loading is likely
due to an increased concentration of bicarbonate. At lower CO
2
loading bicarbonate
concentration is insignificant and MEA concentration has no effect on the equilibrium CO
2

partial pressure of the system.
4.4.2.2 Piperazine
Figure 4.3 shows wetted wall column obtained CO
2
equilibrium partial pressure values in 2, 5, 8,
and 12 m PZ compared to Ermatchkov (2006a) and Hilliard (2008). Hilliard used an equilibrium
cell to measure CO
2
partial pressure with an FTIR (Fourier transform infrared spectroscopy)
analyzer to quantify the CO
2
concentration. Ermatchkov measured the equilibrium partial
pressure using headspace gas chromatography (2006b).
10
100
1000
10000
100000
0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45
CO
2
Loading (mol/mol
alk
)
P
*
C
O
2

(
P
a
)
Hilliard 0.9 m PZ
Hilliard 2 m PZ
Hilliard 2.5 m PZ
Hilliard 3.6 m PZ
Hilliard 5 m PZ
8 m PZ
5 m PZ
12 m PZ
2 m PZ
Ermatchkov 1-4 m PZ
Open Points Hilliard (2008) 0.9, 2, 2.5, 3.6, 5 m PZ
Dashes Ermatchkov (2006) 1-4.2 m PZ
Filled Points Current Work 2, 5, 8, 12 m PZ
100C
80C
60C
40C
10
100
1000
10000
100000
0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45
CO
2
Loading (mol/mol
alk
)
P
*
C
O
2

(
P
a
)
Hilliard 0.9 m PZ
Hilliard 2 m PZ
Hilliard 2.5 m PZ
Hilliard 3.6 m PZ
Hilliard 5 m PZ
8 m PZ
5 m PZ
12 m PZ
2 m PZ
Ermatchkov 1-4 m PZ
Open Points Hilliard (2008) 0.9, 2, 2.5, 3.6, 5 m PZ
Dashes Ermatchkov (2006) 1-4.2 m PZ
Filled Points Current Work 2, 5, 8, 12 m PZ
100C
80C
60C
40C

2
partial pressure measurements in PZ solutions at 40, 60, 80,
and 100 C (Ermatchkov, Perez-Salado Kamps et al., 2006a; Hilliard, 2008)
All the data in Figure 4.3 matche very well at 40, 60, and 80 C. Neither Ermatchkov (2006a) or
Hilliard (2008) provide data at 100 C but the 100 C data look reliable based upon the spacing
from the 80 C data and the overlap of the amine concentrations.
Unlike the CO
2
partial pressure measurements in the MEA system, the PZ system does not show
a dependence of amine concentration at high loading. This is because the CO
2
loading is not
high enough to see appreciable quantities of bicarbonate. Since only carbamates are produced,
none of the data show an effect of amine concentration when plotted versus CO
2
loading.
4.4.2.3 7 m MEA/2 m PZ
Very little data for equilibrium CO
2
partial pressure are available for 7 m MEA/2 m PZ. Figure
4.4 includes the current data (filled points) compared against Hilliard (2008) represented as the
131
9
open points. Again, Hilliard used an equilibrium cell to measure CO
2
partial pressure with an
FTIR analyzer to quantify the CO
2
concentration.
1
10
100
1000
10000
100000
0.05 0.15 0.25 0.35 0.45
CO
2
Loading (mol/mol
alk
)
P
C
O
2
*

(
P
a
)
Open Points Hilliard (2008)
Filled Points Current Work
100C
80C
60C
40C
7 m MEA/2 m PZ

2
partial pressure measurements in 7 m MEA/2 m PZ at 40, 60,
80, and 100 C (Hilliard, 2008)
Although there are limited data for 7 m MEA/2 m PZ, the available equilibrium CO
2
partial
pressure data show a very good match despite using two very different experimental apparatuses.
Other MEA/PZ concentrations were not studied due to significant concerns on thermal
degradation discovered by Davis (2009). Davis found that the more reactive PZ will react
preferentially with an oxazolidone intermediate formed by thermally degrading MEA.
Essentially, PZ protects MEA in the thermal degradation of the blended system. PZ in the
absence of MEA will not thermally degrade significantly because there is no pathway to produce
oxazolidone.
4.4.3 CO
2
Capacity
The equilibrium CO
2
partial pressures in Figures 4.24.4, allow for the determination of the CO
2

capacity of the systems. The CO
2
capacity is defined as the difference in the CO
2
concentration
from the rich to the lean amine streams, not the total CO
2
concentration in any particular steam.
The CO
2
capacity leads to the amount of CO
2
that would be removed from the system during a
circulation of the amine solution.
The CO
2
capacity is important because of energy tradeoffs of the sensible heat and the heat of
absorption. Circulating less solvent reduces the sensible heat duty since the stripper must heat all
the solution from the cross-exchanger outlet temperature to the stripper temperature. This
temperature difference is the same as the cross-exchanger temperature approach. However,
circulating too little solvent to achieve a very high CO
2
capacity generally results in a very low
132
10
lean loading or CO
2
partial pressure. Stripping to very low CO
2
partial pressures increases the
stripping steam required per mole of CO
2
and can cause inefficient operation of the stripper. The
optimal operating lean loading and thus CO
2
capacity for a given amine system requires a
significant optimization with a complex model since CO
2
reaction rates will change drastically
with changing CO
2
loading. Since the optimal lean loading and thus CO
2
partial pressure of that
lean loading cannot be easily determined, Figure 4.5 is constructed to compare the CO
2
capacity
of 8 m PZ and 7 m MEA at 40 C for any lean partial pressure. Alternative amine systems allow
for an increase in the CO
2
capacity of the system without requiring the system to strip to lower
CO
2
partial pressures. Figure 4.5 includes CO
2
loading values next to some of the data points.
Since CO
2
capacity relates to the sensible heat of the solution and the total dissolved CO
2
has a
negligible partial heat capacity, CO
2
capacities are calculated on a mol
CO2
/kg (water+amine)
basis. It is not appropriate to include the CO
2
in the weight of the solution since it has an
effective negligible sensible heat. Essentially, a mole of MEA has almost the same heat capacity
as MEA carbamate (Hilliard, 2008). Nguyen has seen the same effect in PZ systems (Rochelle,
Chen et al., 2009).
0
0.5
1
1.5
2
2.5
10 100 1000
C
O
2

C
a
p
a
c
i
t
y

w
i
t
h

a

5

k
P
a

R
i
c
h

S
o
l
n
(
m
o
l

C
O
2
/
k
g
(
w
a
t
e
r
+
a
m
i
n
e
)
)
Lean Partial Pressure (Pa)
8 m PZ
7 m MEA
.36 Ldg
.31
.23
.15 Ldg
.47
.31
.19
40C
.39
.54
.20
.30
.37
.49
11 m MEA

Figure 4.5: Operating CO
2
capacity of 8 m PZ and 7 and 11 m MEA assuming a 5 kPa rich
CO
2
partial pressure at 40 C (7 and 11 m MEA data from Hilliard (2008))
Figure 4.5 assumes a 5 kPa CO
2
partial pressure rich solution. In a coal-fired power plant CO
2

enters the absorber near 12 mole %, or 12 kPa since it is near atmospheric pressure. Therefore,
the assumption of a 5 kPa CO
2
partial pressure rich solution at 40 C represents a 5/12 or a 42%
approach to saturation at the bottom of the absorber if the solution exits at 40 C.
With the stated assumptions detailed above, 8 m PZ exhibits about a 70% greater CO
2
capacity
than 7 m MEA and about a 50% greater CO
2
capacity than 11 m MEA. Some of the increased
133
11
capacity arises from the increased amine strength of the piperazine, 8 m versus 7 m for MEA.
The majority of this increased CO
2
capacity is due to the fact that each mole of piperazine has
two functional nitrogen groups. This allows PZ to react twice in the CO
2
reaction, whereas MEA
can only react once. PZ solutions allow for much greater CO
2
capacities than MEA and thereby
lower required liquid flow rates and the sensible heat input requirement of the reboiler.
4.4.4 CO
2
Reaction Rates
As previously stated in section 2.3.1 (Ch. 2), CO
2
absorption rates should be reported in terms of
k
g
. The definition of k
g
is reiterated in Equation 4.5. k
g
is a the liquid film mass transfer
coefficient converted to gas phase units.

) (
*
, 2 , 2
2 '
b CO i CO
CO
g
P P
N
k
= (4.5)
Obtained k
g
values for each MEA experiment are plotted against the measured equilibrium
partial pressure at the temperature of the experiment in Figure 4.6. Figure 4.6 includes 7, 9, 11,
and 13 m MEA rate data at 40, 60, 80, and 100 C.
1E-07
1E-06
1E-05
10 100 1000 10000 100000
P*
CO2
(Pa)
k
g
'

(
m
o
l
/
s
.
P
a
.
m
2
)
40C 60C 80C 100C
7, 9, 11, 13 m MEA
1E-07
1E-06
1E-05
10 100 1000 10000 100000
P*
CO2
(Pa)
k
g
'

(
m
o
l
/
s
.
P
a
.
m
2
)
40C 60C 80C 100C
7, 9, 11, 13 m MEA

Figure 4.6: CO
2
absorption/desorption rates in MEA solutions at 40, 60, 80, and 100 C
Each shape in Figure 4.6 represents a different MEA concentration but the MEA concentration
does not seem to significantly affect the measured k
g
. This is interesting and was unexpected,
considering k
g
is often represented by the pseudo first order approximation result shown in
Equation 4.6.

2
2 2 '
] [
CO
b CO
g
H
Am k D
k = (4.6)
Equation 4.6 includes a term for the amine concentration in the numerator. For Equation 4.6 to
hold true, other terms in Equation 4.6 must change with concentration to offset the change in the
concentration term. The diffusion coefficient and the Henrys constant are both affected by
134
12
changes in concentration. The Henrys constant shown in Equation 4.6 is not the true
thermodynamic Henrys constant which refers to the solubility in water. The Henrys constant
shown in Equation 4.6 refers to the CO
2
solubility in the solution. It is a function of the amine
concentration, CO
2
loading and temperature (Browning and Weiland, 1994; Hartono, 2009).
The diffusion coefficient of CO
2
will go down slightly with increasing MEA concentration due
to the viscosity effect. The CO
2
solubility decreases (H
CO2
increases) with increasing amine
concentration and this change cancels most of the increasing MEA concentration term. Contrary
to the data, Equation 4.6 does predict an amine concentration effect on k
g
.
Figure 4.6 seems to imply that k
g
in MEA solutions increases with increasing temperature.
However, that assertion is wrong. Rather than each increasing temperature curve having a higher
k
g
, it has a higher CO
2
equilibrium partial pressure. A close look at Figure 4.6 reveals that the
k
g
is almost identical with increasing temperature. The 7 m MEA (circles) data point at 40 C
near 15 Pa has a k
g
of approximately 3.3*10
-6
mol/s
.
Pa
.
m
2
. The lowest loading data points for
7 m MEA at 60, 80, and 100 C each show a k
g
of approximately 3.0*10
-6
mol/s
.
Pa
.
m
2
. Each of
these 4 data points has a similar CO
2
loading and k
g
, verified in Table 4.4.
Since temperature has little effect on the measured k
g
, the temperature dependent terms in
Equation 4.6 must somehow cancel each other. The diffusion coefficient, rate constant, and
Henrys constant are all temperature dependent. The diffusion coefficient will decrease with
increasing temperature due to viscosity changes. The rate constant will increase with increasing
temperature as shown by regressed literature data (Versteeg, Van Dijck et al., 1996). The
solubility of CO
2
and N
2
O in water decreases with increasing temperature (Versteeg and Van
Swaaij, 1988). Like concentration, Equation 4.6 does not predict k
g
to be independent of
temperature as the data indicate.
It would be convenient to show Figure 4.6 in terms of CO
2
loading on the x-axis but the CO
2

loading basis would prohibit the MEA data from being compared to other amines. Different
amines can only be compared on a partial pressure basis since the definition of CO
2
loading is
somewhat arbitrary and each amine has a different CO
2
loading operating range. However, we
can plot the x-axis in terms of the equilibrium CO
2
partial pressure at a given temperature. This
results in two points with the same CO
2
loading being plotted at the same value on the x-axis
regardless of temperature. In this respect it is similar to plotting the x-axis on a CO
2
loading
basis. However, this basis has the advantage that it also allows a fair comparison of the CO
2

reaction rates with different amines since it is on a partial pressure basis. The equilibrium CO
2

partial pressure at 40 C can be viewed as a surrogate for CO
2
loading. The MEA rate data is
plotted versus the equilibrium CO
2
partial pressure at 40 C in Figure 4.7.
135
13
1E-07
1E-06
1E-05
10 100 1000 10000
P*
CO2
@ 40C (Pa)
k
g
'

(
m
o
l
/
s
.
P
a
.
m
2
)
7, 9, 11, 13 m MEA
100C
80C
60C
40C
1E-07
1E-06
1E-05
10 100 1000 10000
P*
CO2
@ 40C (Pa)
k
g
'

(
m
o
l
/
s
.
P
a
.
m
2
)
7, 9, 11, 13 m MEA
100C
80C
60C
40C

Figure 4.7: CO
2
absorption/desorption rates in MEA solutions at 40, 60, 80, and 100 C,
plotted against the 40 C equilibrium CO
2
partial pressure
The MEA data clearly show that the amine concentration and temperature do not affect the k
g
of
the MEA solution. These points fall nicely on each other and make the determination of k
g
for
MEA solutions relatively simple. It is important to note that measured k
g
values drastically
decrease with increasing equilibrium CO
2
partial pressure at 40 C (CO
2
loading). This decrease
of about a factor of 10 in the value of k
g
from approximately 0.25 to 0.50 CO
2
loading is
primarily due to the reduction of free MEA available for reaction.
PZ rate data at 40, 60, 80, and 100 C are compared in Figure 4.8. 12 m PZ data is not included
in the plot since the equilibrium partial pressures of 12 m PZ at 40 C could not be determined
using the wetted wall column. These solutions were too viscous for wetted wall column
operation.
136
14
1E-07
1E-06
1E-05
10 100 1000 10000
P*
CO2
@ 40C (Pa)
k
g
'

(
m
o
l
/
s
.
P
a
.
m
2
)
2, 5, 8 PZ
100C
80C
60C
40C
1E-07
1E-06
1E-05
10 100 1000 10000
P*
CO2
@ 40C (Pa)
k
g
'

(
m
o
l
/
s
.
P
a
.
m
2
)
2, 5, 8 PZ
100C
80C
60C
40C

Figure 4.8: CO
2
absorption/desorption rates in PZ solutions at 40, 60, 80, and 100 C,
plotted against the 40 C equilibrium CO
2
partial pressure
The PZ data do not converge quite as nicely as the MEA data. The measured k
g
values of the
PZ data do not seem to be dependent on the amine concentration. However, there are some
temperature effects. At the lowest CO
2
loading near 70100 Pa, the black 100 C data points
seem to drop below the trend of the other data. At the next higher CO
2
loading near 300500 Pa,
the green 80 C data points drop from the trend while the 100 C data points drop far below the
trend. At the two highest loadings only 40 and 60 C data is available but the red 60 C data
points routine fall slightly below the 40 C data points.
The observed temperature effects in the PZ data suggest that diffusion of products and reactants
may be limiting the reaction of the CO
2
with the amine. At the lowest CO
2
loading, there is
plenty of free amine at the interface and CO
2
fluxes are generally small at the lower
temperatures. Recall, tested CO
2
partial pressures range roughly from 02 times the equilibrium
partial pressure, not the equilibrium partial pressure at 40 C. Therefore fluxes at 100 C are
very high compared to the other temperatures and it is possible that these fluxes, combined with
fast CO
2
reaction rates, are depleting the interface of reactive PZ and PZ carbamate. At the next
highest loading, there is less free PZ carbamate at the interface while CO
2
fluxes are higher due
to the increased equilibrium partial pressure of the solutions. At this loading it seems as though
the 80 C data are now being somewhat restrained by diffusion limitations while 100 C seem
severely hampered by the diffusion of reactants and products near the interface. At the higher
loadings, the PZ carbamate concentration continues to decrease while tested partial pressures
continue to increase, thereby possibly slowing the measured mass transfer coefficients.
137
15
It is important to keep in mind that although the PZ rate data suggest this diffusion limiting
phenomenon, a comprehensive model is required to verify it. Alternatively, the MEA
experiments do not seem to be limited by the diffusion of reactants and products.
It is also important to keep in mind that the proposed diffusion limitation for the PZ experiments
in the wetted wall column may not be seen in industrial columns. The wetted wall column has a
somewhat smaller liquid film physical mass transfer coefficient, k
l
o
, than a typical industrial
column. This is due to the 9.1 cm stainless steel contactor. In a packed industrial column, either
structured or random packing, the mean flow path of the solvent is probably closer to 24 cm.
The more frequent mixing of the solvent will refresh the interface and discourage depletion of
the reactants at the gas-liquid interface.
The MEA, PZ, and the MEA/PZ data are compared in Figure 4.9. MEA is represented by the
empty points. PZ is represented by the filled data points. 7 m MEA/2 m PZ data is shown as Xs.
1E-07
1E-06
1E-05
10 100 1000 10000
P*
CO2
@ 40C (Pa)
k
g
'

(
m
o
l
/
s
.
P
a
.
m
2
)
Filled Points 2, 5, 8 m PZ
Empty Points 7, 9, 11, 13 m MEA
100C
80C
60C
40C
Xs 7 m MEA/2 m PZ
1E-07
1E-06
1E-05
10 100 1000 10000
P*
CO2
@ 40C (Pa)
k
g
'

(
m
o
l
/
s
.
P
a
.
m
2
)
Filled Points 2, 5, 8 m PZ
Empty Points 7, 9, 11, 13 m MEA
100C
80C
60C
40C
Xs 7 m MEA/2 m PZ

Figure 4.9: CO
2
absorption/desorption rates in MEA, PZ, and MEA/PZ solutions at 40, 60,
80, and 100 C, plotted against the 40 C equilibrium CO
2
partial pressure
Most of the PZ data points form a trend line above the MEA data. These data show that k
g
for
PZ is 23 times faster than MEA. This means PZ reacts with CO
2
23 times faster than MEA.
To a first approximation 1/2 to 2/3 less packing in the absorber would be required for PZ
compared to MEA.
The 7 m MEA/2 m PZ rate data generally fall between the MEA and PZ. The CO
2
loading near
200 Pa seems to show some diffusion limitations at the 100 C condition.
138
16
4.4.4.1 Rate Comparisons with Literature
4.4.4.1.1 Monoethanolamine
Rate data obtained in this work are compared to literature values in this section. As previously
stated, there is limited rate data on highly loaded concentrated amines. For a proper comparison
on a k
g
basis, some raw data is required.
Figure 4.10 shows a comparison of 7 m MEA rate data at 40 and 60 C. Aboudheir (2003)
provides rate data obtained from a laminar jet absorber. At each condition a few measurements
were made and these data overlap nicely. Figure 4.10 also includes 4 wetted wall column data
points obtained by Dang (2003). Dang used the same wetted wall column used in this work. A
single 40 C data point from Hartono (2009) is also included in Figure 4.10.
4x10
-7
6x10
-7
8x10
-7
1x10
-6
3x10
-6
0 0.1 0.2 0.3 0.4 0.5
k
g
'

(
m
o
l
/
s
.
P
a
.
m
2
)
CO
2
Loading (mol/mol)
7 m MEA
Circle - Harono (2009)
X's - Aboudheir (2003)
Squares - Dang (2003)
Triangles - Current Work
60C 40C

Figure 4.10: CO
2
reaction rate comparison on a k
g
basis for 7 m MEA at 40 and 60 C
(Aboudheir, Tontiwachwuthikul et al., 2003; Dang and Rochelle, 2003; Hartono, 2009)
The data by Dang fit very nicely with the newly obtained wetted wall column data. The data by
Aboudheir also fit nicely at the two higher CO
2
loadings. The data by Aboudheir (2003) near 0.1
loading show a lower k
g
value than an extrapolation of the wetted wall column data would
predict. However, the unloaded rate data by Hartono (2009) supports these 0.1 CO
2
loading
values and suggests that the liquid film mass transfer coefficient, k
g
, may not change
significantly from 0 to 0.25 CO
2
loading.
No wetted wall column experiments were conducted below 0.2 CO
2
loading. The wetted wall
column cannot accurately obtain rate data in MEA solutions at CO
2
loading much lower than
0.25 because the system becomes dominated by the gas film mass transfer coefficient. Recall the
gas film mass transfer coefficient of the column was originally characterized using unloaded
MEA (Pacheco, 1998).
139
17
The data by Aboudheir (2003) seem to show a fairly reproducible effect of temperature. In each
of the three CO
2
loadings, the 60 C data points exhibit about 50% higher k
g
values. The wetted
wall column data including Dang (2003) and the current work do not clearly show a trend.
Figure 4.7 more clearly shows that there is no discernable temperature effect on the CO
2

absorption/desorption rates in MEA solutions.
Unloaded MEA rate data found in the literature can also be compared to the highly loaded,
highly concentrated MEA rate data presented here. As the Literature Review detailed, there are
numerous literature sources which report rate data in unloaded, relatively dilute MEA solutions.
However, most of these data sources only report obtained rate constants and do not detail values
used for the Henrys constant or the diffusion coefficient. Neither do they include fluxes and
driving forces which allow for the calculation of k
g
.
Laddha and Danckwerts (1981) provide calculated rate constants along with the solubility and
diffusion parameters which allow for the calculation of the measured flux and K
G
. No gas film
mass transfer coefficients were given for the stirred cell experiments. The rate constants
(expressed through Equation 4.7) for the 6 tested amine concentrations ranged from 5.49 to 6.28
m
3
/(mol
.
s) at 25 C. These rate constants compare favorably with 5.99 m
3
/(mol
.
s) value predicted
by a correlation developed from a review of literature data (Versteeg, Van Dijck et al., 1996).
[ ][ ]
2 2 2
CO MEA k r
CO
= (4.7)
Hartono (2009) provides all the important experimental data from his CO
2
absorption into MEA.
This allows for the calculation of K
G
and then k
g
. The rate experiments performed using a
string of discs were determined to be 518% gas film controlled. The calculated k
g
from the
experiments by Hartono (2009) and the calculated K
G
values from Laddha (1981) are shown in
Figure 4.11.
7x10
-7
8x10
-7
9x10
-7
1x10
-6
2x10
-6
3x10
-6
4x10
-6
0 1 2 3 4 5
K
G

o
r

k
g
'

(
m
o
l
/
s
.
P
a
.
m
2
)
MEA Concentration (Molarity)
25C
25C
30C
40C
50C
K
G
- Laddha and Danckwerts (1981)
k
g
' - Hartono (2009)

Figure 4.11: CO
2
reaction rates in unloaded MEA solutions (Laddha and Danckwerts,
1981; Hartono, 2009)
140
18
Figure 4.11 shows the Laddha data at 25 C below the Hartono data at 25 C. This is expected
since the Laddha data does not remove the gas film resistance from the system. Recall, the
liquid film mass transfer coefficient, k
g
, must be larger than the overall mass transfer
coefficient, K
G
. In cases where the gas film mass transfer coefficient, k
g
, is limiting, K
G
can be
significantly lower than k
g
. In a stirred cell experiment with unloaded MEA, it is quite likely
that gas film mass transfer resistance is significant since stirred cells often have this concern.
The Laddha data in Figure 4.11 is not as descriptive of CO
2
rates into MEA as the Hartono data
because gas film resistances due to operating conditions and the geometry of the apparatus
cannot be extracted from the reported data.
Figure 4.11 shows a couple of interesting points. First of all it shows a dependence of k
g
on the
MEA concentration at lower MEA concentrations. At higher MEA concentrations k
g
becomes
independent of concentration, although at different amine concentrations for different
temperatures. This independence of concentration on k
g
is also seen in the current MEA rate
data (Figure 4.7) which was taken at high MEA concentrations.
4.4.4.1.2 Piperazine
Although Table 2.2 of the Literature Review only lists 5 references for CO
2
reaction rates into
aqueous PZ solutions, all provide a fair amount of raw experimental data. Sun (2005), Derks
(2006), Cullinane (2006) and Samanta (2007) include unloaded PZ rate data while Bishnoi
provides CO
2
loaded rate data. All five data sources use low piperazine concentrations.
Derks uses a stirred cell and a semi-continuous gas phase operation. Numerous CO
2
partial
pressures were tested for each amine to determine when the pseudo first order condition applies.
At high CO
2
partial pressures, diffusion in the liquid phase limits CO
2
mass transfer. For 1.0 M
PZ at 40 C, approximately 1.5 kPa CO
2
was the threshold for the onset of the pseudo first order
condition. Inlet CO
2
partial pressures above 1.5 kPa showed a distinct effect of the partial
pressure on the measured K
G
. Below the threshold, the overall mass transfer coefficient is
independent of the inlet partial pressure.
Sun (2005) and Samanta (2007) each measured the absorption into unloaded PZ solutions using
wetted wall columns. However, each uses very high CO
2
partial pressures, typically about 5
kPa. At these high CO
2
partial pressures and amine concentrations below 1 M, CO
2
fluxes into
the liquid phase may be restricted by diffusion. In fact, Sun (2005) tested a few lower CO
2

partial pressures and these data verify that the system is not operating in the pseudo first order
regime at the 5 kPa CO
2
pressure experiments, which comprise most of the data. Although we
cannot extract a meaningful k
g
value from these raw data, they are still valuable data. These
data require a model to properly account for the diffusion limitations in the system.
Cullinane provides all the required data to directly calculate k
g
. At each condition, 5
measurements were made. Obtained k
g
values were shown to range 30% from the mean due
to the high dependence of the gas film mass transfer coefficient. The 1.2 m PZ experiments were
5473% gas film controlled. Only 25 and 60 C experiments were tested. The Cullinane
experiments all use very low CO
2
partial pressures (< 250 Pa) so the pseudo first order condition
should apply.
Figure 4.12 shows a comparison of the obtained 2 m PZ wetted wall column rate data with some
literature obtained values. Figure 4.12 includes an unloaded 1.0 M PZ data point from Derks.
This point is actually the obtained overall mass transfer, K
G
, not the liquid film mass transfer
coefficient, k
g
. Derks does not provide a gas film mass transfer coefficient correlation to
quantify if or how much gas phase resistance limits CO
2
absorption into the solution. For
141
19
purposes of comparison, the K
G
obtained from Derks is plotted alongside the k
g
data and the k
g

model prediction from Cullinane (2005).
6x10
-7
8x10
-7
1x10
-6
3x10
-6
5x10
-6
7x10
-6
0 0.1 0.2 0.3 0.4
k
g
'

(
m
o
l
/
s
.
P
a
.
m
2
)
CO
2
Loading (mol/mol
alk
)
Triangle - Derks (2006) (K
G
)1.0 M PZ
Squares - Bishnoi (2000) 0.06-0.31 m PZ
Line - Cullinane (2005) 1.8 m PZ Model Prediction
X's - Cullinane (2006) 1.2 m PZ (25 and 60C)
Circles - Current Work 2 m PZ
40C
PZ
60C
25C

Figure 4.12: CO
2
reaction rate comparison on a k
g
basis for PZ solutions at 40 C (Bishnoi
and Rochelle, 2000; Cullinane, 2005; Cullinane and Rochelle, 2006; Derks, Kleingeld et al.,
2006)
Figure 4.12 shows a good match of the current 2 m PZ rate data with the 1.8 m PZ model
prediction by Cullinane (2005). The loaded Bishnoi data shows mass transfer coefficients
somewhat below the current data. This is expected due to the very low PZ concentration (0.06 to
0.31 m PZ) in these experiments. Interestingly, these data show the same trend as the 2 m PZ
data. Very low amine concentrations also exhibited a reduced k
g
in MEA solutions (Figure
4.11).
The unloaded data in Figure 4.12 is difficult to analyze. The 25 and 60 C data points by
Cullinane show a significant temperature effect at zero loading. These 1.2 m PZ data points
show a decent fit to the 1.8 m 40 C model prediction. The Derks overall mass transfer
coefficient falls significantly below the other unloaded data, which is not unexpected. This
suggests that the gas film mass transfer coefficient is likely limiting mass transfer into the PZ
solution. The limitation of the gas film mass transfer coefficient is a disadvantage of stirred cell
reactors to measure CO
2
reaction rates of very fast amines.
4.5 PREDICTING K
G
FOR MEA WITH THE PSEUDO FIRST ORDER ASSUMPTION
The pseudo first order conditions are fulfilled when the bulk amine concentration is the same as
the amine concentration at the interface. At this condition, the diffusion of reactants and
products does not hinder CO
2
mass transfer and the mass transfer rate can be determined
analytically by the rate equation. However, an analytical expression to calculate k
g
at highly
concentrated, highly loaded conditions has previously remained elusive and thus required
experimentation to determined CO
2
mass transfer rates.
142
20
Equation 4.8 yields the rate equation typically used for CO
2
reaction with MEA. The rate
expression can be used in a material balance to determine the CO
2
flux. The grouping of terms
in Equation 4.10 can be treated as the liquid film mass transfer coefficient, k
g
.
[ ][ ]
2 2
CO MEA k r
f CO
= (4.8)
0 ] ][ [
] [
2
2
2
2
2
=
CO Am k
x
CO
D
f CO
(4.9)
) (
] [
*
, 2 , 2
2
2
2 b CO i CO
CO
CO
CO
P P
H
Am k D
N = (4.10)

2
2 '
] [
CO
b CO
g
H
Am k D
k = (4.11)
The rate constant in Equation 4.11 is known from the literature data. The amine concentration
can be chosen. The diffusion coefficient and Henrys constant can be estimated to predict k
g
.
This method has not been effective at predicting k
g
in highly concentrated, highly loaded MEA
solutions. This disconnect required a closer look into the assumptions in Equation 4.11.
4.5.1 Activity Coefficients
The rate expression is determined by the activity of the reactants, not the concentration. It
cannot be assumed that activity coefficients are near 1.0 in highly loaded, highly concentrated
MEA solutions. These solutions are highly ionic and should be treated accordingly. Accounting
for activity coefficients in the rate expression (Equation 4.12) leads to a modified k
g
expression
shown in Equation 4.13.
[ ] [ ]
2 2 2
CO MEA k r
CO MEA f CO
= (4.12)

2
2 2 '
] [
CO
CO b MEA CO
g
H
Am k D
k

= (4.13)
Including activity coefficients into the rate expression requires an understanding of what the
activity coefficients of MEA and CO
2
are in these solutions. MEA activity coefficients can be
obtained from amine volatility experiments. CO
2
activity coefficients can be obtained from
Henrys solubility data.
4.5.1.1 Monoethanolamine
Amine volatility data is very scarce but Hilliard (2008) provides a nice sample of 3.5, 7, and
11 m MEA volatility data. These experiments coincide with the CO
2
partial pressure
experiments Hilliard ran in an equilibrium cell. The FTIR analyzer he used can simultaneously
measure gas phase concentrations of multiple components.
The MEA volatility data was treated via the modified Raoults Law in Equation 4.14. DIPPR
database obtained value of 164 and 666 Pa were used for the equilibrium partial pressure of pure
MEA at 40 and 60 C (1979).

*
MEA MEA MEA MEA MEA
P x P P y = = (4.14)
Equation 4.14 requires the mole fraction of MEA, which is easy to determine accurately below
0.4 CO
2
loading. Above a 0.45 CO
2
loading bicarbonate concentrations can become significant
while free MEA concentration become very small. At these high CO
2
loadings it is very difficult
to accurately determine the free MEA concentration. Due to this uncertainty, no data above 0.45
CO
2
loading was used in the determination of MEA activity coefficients. Figure 4.13 shows the
calculated MEA activity coefficients from the Hilliard (2008) data.
143
21
y = 1.115x + 0.339
y = 0.679x + 0.294
0.0
0.4
0.8
1.2
1.6
2.0
0.1 0.2 0.3 0.4 0.5
CO
2
Loading (mol/mol)
M
E
A

A
c
t
i
v
i
t
y

C
o
e
f
f
i
c
i
e
n
t
7 m MEA, 40C 7 m MEA, 60C
11 m MEA, 40C 11 m MEA, 60C
3.5 m MEA, 40C 3.5 m MEA, 60C

Figure 4.13: Calculated MEA activity coefficients for 3.5, 7, and 11 m MEA at 40 and 60 C
(Hilliard, 2008).
The large errors in Figure 4.13 seen at high loading are not particularly important. These large
errors are the result of fairly small deviations in either the vapor-liquid equilibrium model or the
CO
2
loading reported by Hilliard.
The Hilliard data show an increasing MEA activity coefficient with increasing CO
2
loading. The
MEA activity coefficient also seems to be a function of temperature, with higher temperatures
having lower activity coefficients. Amine concentration does not seem to be a major factor in
the determination of the activity coefficient. The 3.5, 7 and 11 m MEA data sets each tend to
overlap fairly well. Figure 4.13 includes trend lines for the 40 and 60C data. The equations of
the trend lines reproduced in Equation 4.15 and 4.16 give a pretty good estimation of the MEA
activity coefficient as a function of CO
2
loading.
( ) 339 . 0 115 . 1
2 40 ,
+ = Loading CO P
C MEA
(4.15)
( ) 294 . 0 679 . 0
2 60 ,
+ = Loading CO P
C MEA
(4.16)
4.5.1.2 Carbon Dioxide
The activity of CO
2
in loaded MEA solutions can be obtained from Henrys solubility data.
Unfortunately very little N
2
O solubility data has been reported in concentrated, CO
2
loaded MEA
systems. Browning and Weiland (1994) present 12 N
2
O solubility data points in 10, 20, and 30,
wt % MEA solutions up to 0.4 CO
2
loadings at 25 C. No other N
2
O solubility data varying
amine concentration and CO
2
loading is known. The N
2
O solubility data was regressed into
Equation 4.17. Equation 4.17 includes the MEA concentration in wt %.
) )( ( 7 . 294 ) ( 1115 ) ( 31 . 18 4093
2 2 25 , 2
Ldg CO MEA Ldg CO MEA H
C O N
+ + = (4.17)
Figure 4.14 shows the N
2
O solubility data points from Browning as well as regressed curves for
10, 20, and 30 wt % MEA.
60C
40C
144
22
4000
4500
5000
5500
6000
6500
7000
7500
8000
0 0.1 0.2 0.3 0.4
CO
2
Loading (mol/mol)
H
N
2
O

(
k
P
a
.
L
/
m
o
l
)

Figure 4.14: N
2
O solubility data (Browning and Weiland, 1994) and model (lines) in 10, 20,
and 30 wt % MEA solutions at 25 C.
Figure 4.14 shows that Equation 4.17 does a very good job of predicting the N
2
O solubility as a
function of amine concentration and CO
2
loading. Figure 4.14 also illustrates how drastically the
N
2
O solubility decreases with increasing loading and amine concentration. It is imperative that
both the amine concentration and CO
2
loading are considered in the estimation of the Henrys
constant. The literature suggests these factors have generally been ignored.
Equation 4.17 allows for the calculation of the solubility of CO
2
via the N
2
O analogy. The CO
2

and N
2
O solubility data in water as a function of temperature has been compiled and regressed
(Versteeg and Van Swaaij, 1988). The implementation of the N
2
O analogy assumes that N
2
O
and CO
2
solubility in concentrated, loaded amines has the same temperature dependence as N
2
O
and CO
2
in water. Fortunately this assumption is no longer required for MEA solutions.
Hartono (2009) recently published N
2
O solubility data in loaded 30 wt % (7 m) MEA solutions.
He measured N
2
O solubility from 2587 C for 0, 0.2, 0.4, and 0.5 CO
2
loading solutions.
Figure 4.15 illustrates the N
2
O solubility results for each of the 4 CO
2
loadings.
10 wt% MEA
30 wt% MEA
20 wt% MEA
25C
145
23
8
8.5
9
9.5
10
10.5
0.0027 0.0029 0.0031 0.0033
1/Temp (1/K)
l
n

(
H
N
2
O
)

(
P
a
.
m
3
/
m
o
l
)

Figure 4.15: N
2
O solubility data (points) and trend lines for 0, 0.2, 0.4, and 0.5 CO
2
loaded
7 m MEA solutions (Hartono, 2009)
The natural log of the N
2
O solubility plotted versus inverse temperature yields straight lines for
each of the 4 CO
2
loadings. The slope of the lines corresponds to the temperature behavior of
N
2
O solubility in 7 m MEA. The slopes of the four lines are approximately equal with an
average value of -1905. The N
2
O solubility temperature dependence in loaded MEA solutions
can be added to Equation 4.17 which is valid at 25 C. Equation 4.18 should be valid from 25 to
at least 87 C, the temperature range of the regressed data.

[ ]

+ + =
15 . 298
1905
exp
1905
exp
) )( ( 7 . 294 ) ( 1115 ) ( 31 . 18 4093
2 2 2
T
Ldg CO MEA Ldg CO MEA H
O N
(4.18)
Similar to the N
2
O solubility from Browning (1994), Hartono shows the N
2
O solubility
decreasing with increasing CO
2
loading. Unfortunately the data do not agree completely. Both
Hartono and Browning measure N
2
O solubility at 25 C for 7 m MEA. Figure 4.16 shows the
disagreement in the two data sets.
0.5
CO
2
Loading
0.4
0.2
0
7 m MEA
146
24
4500
5000
5500
6000
6500
7000
7500
8000
0 0.1 0.2 0.3 0.4 0.5
CO
2
Loading (mol/mol)
H
N
2
O

(
k
P
a
.
L
/
m
o
l
)

Figure 4.16: N
2
O solubility in 7 m MEA at 25 C (Browning and Weiland, 1994; Hartono,
2009)
Since these are the only two data sets for N
2
O solubility in loaded MEA solutions, it is not
possible to tell which data set is erroneous. In this work the Browning (1994) data set has been
used to quantify the effects of CO
2
loading and MEA concentration on N
2
O solubility. The
Hartono (2009) data set has been used to quantify the effect of temperature on N
2
O solubility.
Using the N
2
O analogy, the solubility of CO
2
in loaded MEA solutions can be determined using
Equation 4.18 along with the solubility of CO
2
and N
2
O in water. An accurate calculation of the
Henrys constant of CO
2
allows for the determination of the activity coefficient of CO
2
via
Equation 4.19.

O H CO CO CO
H H
2 , 2 2 2
= (4.19)
2 CO
H gives the effective solubility of CO
2
in the solution.
O H CO
H
2 , 2
is the true thermodynamic
Henrys constant which refers to the solubility of CO
2
in water. The activity coefficient of CO
2

varies between 1.3 and 2.5 for 713 m MEA wetted wall column experiments.
5.4.2 Diffusion Coefficient of CO
2

Work by Versteeg and Van Swaaij (1988) has shown that the diffusion of N
2
O and CO
2
in
aqueous amines generally follows the viscosity dependence in Equation 4.20.
( ) ( )
Water O N eSolution A O N
D CONSTANT D
8 . 0
2 min
8 . 0
2
= = (4.20)
The N
2
O and CO
2
diffusivity relationship in Equation 4.20 was confirmed with MDEA solutions
but resulted in less satisfactory results for AMP (Tomcej and Otto, 1989; Xu, Otto et al., 1991).
Diaphragm cell experiments in loaded MEA and PZ solutions yield a viscosity dependence of
0.72 compared to the 0.8 obtained by Versteeg and Van Swaaij (1988) for N
2
O. Although the
0.72 dependence obtained from the diaphragm cell experiments does necessarily represent CO
2

diffusion, or diffusion of any other specific species, the 0.72 dependence was used for
calculation of the diffusion coefficient in the pseudo first order expression.
7 m MEA
Browning
(1994)
Hartono
(2009)
25C
147
25
( )
( )
( )
eSolution A
Water CO
eSolution A CO
D
D
min
72 . 0
72 . 0
2
min 2

= (4.21)
The viscosity of water at the wetted wall column experimental temperatures was found from
Watson (1986). MEA solution viscosity values were obtained from Weiland (1998).
5.4.3 Monoethanolamine Order
With accurate estimations for the activity coefficients of MEA and CO
2
, the MEA concentration
dependence on k
g
can be examined. It was found that the rate data show a 2
nd
order dependence
on the MEA concentration. This 2
nd
order dependence can be satisfied from either the zwitterion
or termolecular mechanism, although the termolecular mechanism is more likely for MEA. The
termolecular mechanism allows for the following base catalysis reaction expression below.
[ ] [ ] ( ) [ ] [ ]
2 2 2 2
CO MEA O H k MEA k r
O H MEA CO
+ = (4.22)
For the 2
nd
order dependence to be observed [ ] MEA k
MEA
must be much greater than [ ] O H k
O H 2 2
.
Cullinane (2005) used Bronsted Theory to relate pK
a
s to rate constants. According to the pK
a
-
rate constant relationship found for piperazine, an amine with the pK
a
of MEA would produce
O H
MEA
k
k
2
ratios near 180, thereby making catalysis by water negligible in almost all the wetted wall
column experimental conditions. In 7 m MEA the base catalysis by water would approach that
by MEA only after more than 95% of the total MEA had been reacted.
In order for the 2
nd
order amine dependence to submit to the zwitterion mechanism,
r
k must be
much greater than [ ]
B k
b
in Equation 4.23 yielding Equation 4.24. This is generally not
accepted as true for MEA (Danckwerts 1979). However, in the previous treatment of literature
data leading to this conclusion, activity coefficients and the Henrys constant of CO
2
were not
considered rigorously. At least at these highly concentrated, highly loaded MEA systems, these
parameters are very important.

+
=
] [
1
] ][ [
2
2
B k k
k
k
CO Am
r
b f
r
f
CO
(4.23)
[ ][ ]
= ] [
2 2
B k CO MEA
k
k
r
b
r
f
CO
(4.24)
5.4.4 Predicting k
g
Results
Instituting the 2
nd
order amine dependence into the pseudo first order condition result yields
Equation 4.25

O H CO co
b MEA CO
CO
CO b MEA CO
g
H
MEA k D
H
MEA k D
k
2 , 2
5 . 0
2
2 2
2
2
2
2 2
2 '
] [ ] [

= = (4.25)
The rate constant in the simple form of the pseudo first order expression (Equation 4.11)
assuming 1
st
order amine kinetics was compiled into Equation 4.26 (Versteeg, Van Dijck et al.,
1996).

s mol
m
T
k

=
3
8
5400
exp 10 4 . 4 (4.26)
148
26
The temperature dependence of this expression did not require readjustment. The pre-
exponential factor was scaled until the calculated k
g
values from Equation 4.26 matched wetted
wall column measured values. Figure 4.17 is a parity plot quantifying the fit of the modified
pseudo first order expression. Figure 4.17 also includes 0.5, 1, 2, 3, 4, and 5 M unloaded MEA
rate data (Hartono, 2009).
1E-07
1E-06
1E-05
1E-07 1E-06 1E-05
Measured k
g
' (mol/s
.
Pa
.
m
2
)
C
a
l
c
u
l
a
t
e
d

k
g
'

(
m
o
l
/
s
.
P
a
.
m
2
)

Figure 4.17: Parity plot of measured k
g
versus calculated k
g
from modified pseudo first
order expression for 7, 9, 11, and 13 m MEA at 40 and 60 C and 0.55 M unloaded MEA
at 40 C (Hartono, 2009)
Figure 4.17 shows an excellent agreement between the wetted wall column measured and
calculated k
g
values from the modified pseudo first order expression in Equation 4.25. After
adjusting the k
g
expression to include 2
nd
order MEA dependence, the amine concentration
dependence was perfectly accounted for by modifying the Henrys constant, the diffusion
coefficient, and the activity coefficients of MEA and CO
2
according the available literature data.
The model only compares 40 and 60 C wetted wall column k
g
data due to the lack of available
MEA volatility data above 60 C. However these data match very well and no unaccounted
temperature effect seems to be present. The higher temperature increases the rate constant,
decreases the MEA activity coefficient, decreases the CO
2
activity coefficient, and increases the
effective Henrys constant. The effects of each of these parameters (obtained from literature
data) essentially cancel in the k
g
expression. The same was seen in the wetted wall column
experiments where the temperature effect did not significantly affect k
g
. Although this
comparison only includes 40 and 60 C, the model will likely do a fair job extrapolating to the 80
and 100 C experimental data.
CO
2
reaction rates with unloaded MEA can also be predicted using the modified pseudo first
order expression. At higher MEA concentration, 25 M, the expression very closely predicted
the experimental k
g
. At the two lower MEA concentrations, 0.5 and 1 M, the expression under
7, 9, 11, 13 m MEA .
40C - Circles..
60C Triangles
Xs 0.5-5 M MEA, 40C
.Hartono (2009)
0.5 M
1 M
2 M
3 M
4 M
5 M
149
27
predicted the experimental k
g
. This underestimation is partially due to the neglected H
2
O
catalysis in the modified pseudo first order expression. According to Equation 4.22, base
catalysis by water can become significant or even dominant at very low amine concentrations. It
is important to remember that the activity coefficient correlations (Equations 4.15, 4.18, and
4.19) are likely less accurate for unloaded, dilute MEA data than the highly concentrated, highly
loaded data.
This analysis was able to quantitatively show how the temperature and concentration dependent
parameters in the modified pseudo first order expression cancel each other to show negligible
temperature and concentration effects in k
g
. It is important to note that all the parameters were
estimated from available literature data and only the implementation of 2
nd
order MEA kinetics
was required to account for the lack of temperature and concentration dependences in the rate
data of the highly loaded, 713 m MEA experiments.
References
(1979). Vapor Pressures and Critical Points of Liquids XIV: Aliphatic Oxygen-Nitrogen
Compounds. Item 79030. London, Engineering Sciences Data.
Aboudheir A, Tontiwachwuthikul P, et al. "Kinetics of the reactive absorption of carbon dioxide
in high CO
2
-loaded, concentrated aqueous monoethanolamine solutions." Chem Engr Sci.
2003;58:51955210.
Bishnoi S, Rochelle GT. "Absorption of CO
2
into Aqueous Piperazine: Reaction Kinetics, Mass
Transfer and Solubility." Chem Engr Sci. 2000;55:5531-5543.
Browning GJ, Weiland RH. "Physical Solubility of CO
2
in Aqueous Alkanolamine via Nitrous
Oxide Analogy." J Chem Eng D. 1994;39:817-822.
Cullinane JT. Thermodynamics and Kinetics of aqueous piperazine with potassium carbonate for
CO
2
absorption. The University of Texas at Austin. Ph.D. Dissertation. 2005;295.
Cullinane JT, Rochelle. "Kinetics of CO
2
Absorption into Aqueous Potassium Carbonate and
Piperazine." I&ECR. 2006;45(8):2531-2545.
Danckwerts PV. "The reaction of CO
2
with ethanolamines." Chem Engr Sci. 1979;34(4):443-
446.
Dang H, Rochelle GT. "CO
2
Absorption Rate and Solubility in MEA/PZ/H
2
O." Sep Sci & Tech.
2003;38(2):337-357.
Davis JD. Thermal Degradation of Aqueous Amines Used for CO
2
Capture. The University of
Texas at Austin. Ph.D. Dissertation. 2009;278.
Derks PWJ, Kleingeld T et al. "Kinetics of Absorption of CO
2
in Aqueous Piperazine Solution."
Chem Engr Sci. 2006;61(20):6837-6854.
Ermatchkov V, Perez-Salado Kamps A et al. "Solubility of CO
2
in Aqueous Solutions of N-
Methyldiethanolaine in the Low Gas Loading Region." Ind Eng Chem Res.
2006b;45:60816091.
Ermatchkov V, Perez-Salado Kamps A et al. (2006a). "Solubility of Carbon Dioxide in Aqueous
Solutions of Piperazine in the Low Gas Loading Region." J Chem Eng D.
2006a;51(5):17881796.
150
28
Hartono A. Characterization of diethylenetriamine (DETA) as absorbent for CO
2
. Trondheim,
Norway, Norwegian University of Science and Technology. Ph.D. Dissertation 2009;243.
Hilliard M. A Predictive Thermodynamic Model for an Aqueous Blend of Potassium Carbonate,
Piperazine, and Monoethanolamine for CO
2
Capture from Flue Gas. The University of
Texas at Austin. Ph.D. Dissertation 2008;1025.
Jou F-Y, Mather AE et al. "The Solubility of CO
2
in a 30 Mass Percent Monoethanolamine
Solution." Can J Chem Engr. 1995;73(1):140147.
Laddha SS, Danckwerts PV. "Reaction of CO
2
with Ethanolamines: Kinetics from Gas
Absorption." Chem Engr Sci. 1981;36:479482.
Pacheco MA. (1998). Mass Transfer, Kinetics and Rate-Based Modeling of Reactive Absorption.
The University of Texas at Austin. Ph.D.: 291.
2
Capture by Aqueous Absorption, First Quarterly Progress Report 2009.
2
2009. Luminant Carbon Management Program. The University of Texas at Austin. 2009.
2
2
Samanta A, Bandyopadhyay SS. "Kinetics and Modeling of CO
2
Absorption into Aqueous
Solutions of Piperazine." Chem Engr Sci. 2007;62(24):73127319.
Snijder ED, te Riele MJM et al. "Diffusion Coefficients of Several Aqueous Alkanolamine
Solutions." J Chem Engr Data. 1993;38(3): 475-480.
Sun W-C, Yong C-B et al. "Kinetics of the Absorption of CO
2
into Mixed Aqueous Solutions of
2-amino-2methyl-1-propanol and Piperazine." Chem Engr Sci. 2005;60(2):503516.
Tomcej RA, Otto FD. "Absorption of CO
2
and Nitrous Oxide into Aqueous Solutions of
Methyldiethanolamine." AIChE J. 1989;35(5):861864.
Versteeg GF, Van Dijck LAJ, et al. "On the Kinetics between CO
2
and Alkanolamines both in
Aqueous and Non-aqueous Solutions. An Overview." Chem Engr Comm. 1996;144:113
158.
Versteeg GF, Van Swaaij WPM. "Solubility and diffusivity of acid gases (CO
2
, nitrous oxide) in
aqueous alkanolamine solutions." J Chem Engr Data. 1988;33(1):2934.
Watson JR,. Sengers JV. "Improved International Formullations for the Viscosity and Thermal
Conductivity of Water Substance." J Phys Chem Ref Data. 1986;15:1291.
Weiland RH, Dingman JC et al. "Density and Viscosity of Some Partially Carbonated Aqueous
Alkanolamine Solutions and Their Blends." J Chem Engr Data. 1998;43(3):378382.
Xu S, Otto FD, et al. "Physical Properties of Aqueous AMP Solutions." J Chem Engr Data.
1991;36(1):7175.

151
1
Modeling CO
2
Absorption Using Aqueous Amines

by Jorge M. Plaza
and the
2
July 1, 2009
Abstract
A CO
2
absorber model for the 8 m PZ solvent is under development. It uses a modified version
of the Hilliard (2008) thermodynamic representation by Van Wagener version 02/06/09
(Rochelle et al., 2009). It includes a reduced reaction set based on the more relevant species
present at the expected operating loading (0.30.4 mol CO
2
/mol alkalinity) Kinetics will be
based on Cullinane (2005) and regressed to include data generated by Dugas (Rochelle et al.,
2008a). Initial work accesses proper representation of solvent physical properties. Density and
viscosity were regressed to match experimental data generated by Freeman (Rochelle et al.,
2008b: Rochelle et al., 2009). The activity coefficient of CO
2
was also examined and compared
to values found in Cullinane (2005) as a function of amine concentration and loading.
Density and viscosity match the experimental data with a deviation of maximum 10%. The
model represents correctly the expected change in the activity coefficient of CO
2
with amine
concentration and its values are consistent with the available low concentration experimental
data. However, the effect of loading is not represented correctly. This issue needs to be
addressed. Kinetics will be implemented once the effect of loading on CO
2
activity coefficients
is resolved.
The overall gasside mass transfer coefficient (K
g
) was calculated for the November 2008 PZ
pilot plant data and compared to k
g
data by Dugas (Rochelle et al., 2008a) and previous pilot

plant campaigns. Results show a higher K
g
for 5m PZ.

PZ Model Development
Thermodynamic Model
Thermodynamics are based on a re-regression of the Hilliard (2008) PZ model by Van Wagener
version 02/06/09 (Rochelle et al., 2009) to include data generated at amine concentration above
5m. The reaction set was reduced to four reactions involving the most significant species present
at the loading range proposed for operation for this solvent (0.30.4 mol CO
2
/mol alkalinity)
152
2
based on speciation by Hilliard (2008). Species that are present in small concentrations have
been omitted from the model (CO
3
=
, H
+
, OH
-
, PZH
2
++
). The final reaction set is as follows:
2 +
2

+
+
(1)
2
+
2

2
+
+
(2)

+
2
+
2

3
+
+
(3)
+
+

+
+
(4)

The vapor-liquid equilibrium (VLE) was verified with the new reaction set (Figure 1) and it
continues to adequately represent the behavior of the partial pressure of CO
2
with loading.

Figure 1: VLE verification for the Hilliard modified by Van Wagner version 02/06/09
(Rochelle et al., 2009) for 8 m PZ. Points are by Dugas (Rochelle et al., 2008a). Lines
generated with the reduced reaction set in Aspen Plus
RateSep
TM
.

10
100
1000
10000
100000
1000000
0.2 0.25 0.3 0.35 0.4 0.45
P
C
O
2
*
(
P
a
)
Loading (mol CO
2
/mol alkalinity)
40
o
C
60
o
C
80
o
C
100
o
C
153
3
Physical Properties Regression
Density and viscosity were fit to data generated in the lab by Freeman (Rochelle et al., 2008b:
Rochelle et al., 2009) for PZ from 5m to 9 m PZ which corresponds to the range of operation of
the pilot plant.
Density: Density data for 5 m, 7 m, 8 m and 9 m PZ and from 0.20 to 0.46 mol CO
2
/mol
alkalinity was regressed using the Clarke model for liquid molar volume for electrolyte solutions.
The liquid molar volume for the solution (V
m
l
) is calculated using the following:

(5)
where:
V
solv
is the liquid molar volume for the solvent;
V
ca
is the electrolyte effective partial molar volume;
x
ca
is the apparent electrolyte mole fraction.
The liquid molar volume for the solvent is calculated using the following expression:

(6)
where:
V
w
*
is the water volume from the steam tables.
x
nws
is the sum of the mole fractions of all non-water solvents.
V
nws
l
is the liquid molar volume for the mixture of all non-water solvents calculated
using the Rackett equation.
The Rackett expression was used for PZ and CO
2
:

1+(1
)
2
7

(7)
where:

0.5
1
2

(8)

(9)

(10)

(11)

(12)
154
4
The electrolyte contribution in the molar volume is determined using the apparent mole fraction
(x
ca
) calculated from the true ionic composition based on the proposed system chemistry.

1 +
(13)
The parameters V
ca
, and A
ca
for equation 13 were regressed for the cation PZH
+
and the other
ionic species along with the values for Z
i
RA
in equation 10 for CO
2
and PZ using Aspen Plus

Data Regression System (DRS). Table 1 shows the resulting values.
Table 1: Regressed parameters for the density of PZ-CO
2
-H
2
O
Parameter Component(s) Value
Z
i
RA
PZ 0.2459
CO
2
0.2799
V
ca

PZH
+
- PZCOO
-
0.2165
PZH
+
- PZ(COO
-
)
2
0.3198
PZH
+
- HCO
3
-
0.1651
A
ca
PZH
+
- PZ(COO
-
)
2
-0.2324
PZH
+
- HCO
3
-
-0.5939

The regressed values adequately represent the change in density with loading at the selected
amine concentrations (Figures 2-5)
Viscosity: Data available for 5 m, 7 m and 9 m PZ was fit using the expression of viscosity as a
function of loading, amine concentration and temperature proposed by Weiland et al. (1998).

=
+ + + + + + 1
2
(14)
where:
s
,
w
are the viscosity of the solution and water, respectively (cP);
is the weight fraction of PZ;
is the loading of the solution (mol CO
2
/mol alkalinity);
T is the temperature (K).

155
5

Figure 2: Density fit for 5 m PZ. Points by Freeman (Rochelle et al., 2009). Lines are fit in
Aspen Plus
.

Aspen Plus
.

1.04
1.06
1.08
1.1
1.12
1.14
0.10 0.20 0.30 0.40 0.50
r
(
g
/
c
m
3
)
Loading (mol CO
2
/mol alkalinity)
40
o
C
60
o
C
20
o
C
1.06
1.08
1.1
1.12
1.14
1.16
0.15 0.25 0.35 0.45
r
(
g
/
c
m
3
)
Loading (mol CO
2
/mol alkalinity)
40
o
C
60
o
C
20
o
C
156
6

Aspen Plus

Aspen Plus
.

Table 2 shows the regressed parameters for this correlation. This expression was selected
because it was not possible to adequately fit the viscosity data with the available correlations in
Aspen Plus
. It was implemented in Aspen Plus
using a FORTRAN subroutine. The obtained

results are presented in Figures 68. The 80
o
C extrapolated line is added to show that the
viscosity correlation is well behaved at the absorber operating temperature range.
1.06
1.08
1.1
1.12
1.14
1.16
1.18
0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50
r
(
g
/
c
m
3
)
Loading (mol CO
2
/mol alkalinity)
20
o
C
40
o
C
60
o
C
1.08
1.1
1.12
1.14
1.16
1.18
1.2
0.15 0.20 0.25 0.30 0.35 0.40 0.45
r
(
g
/
c
m
3
)
Loading (mol CO
2
/mol alkalinity)
20
o
C
40
o
C
60
o
C
157
7
Table 2: Regressed viscosity parameters for PZ.
Parameter a b c d e f g
Value
487.52 1389.31 1.58 4.50 8.73 -0.0038 -0.30

Figure 6: Viscosity results for 5 m PZ. Points by Freeman (Rochelle et al., 2008b). Lines
are fit in Aspen Plus
.

.

1
10
0.20 0.25 0.30 0.35 0.40 0.45
m
(
c
P
)
Loading (mol CO
2
/mol alkalinity)
25
o
C
40
o
C
60
o
C
80
o
C (extrapolated)
2
20
0.20 0.25 0.30 0.35 0.40 0.45
m
(
c
P
)
Loading (mol CO2/mol alkalinity)
25
o
C
40
o
C
60
o
C
80
o
C (extrapolated)
158
8

.
CO
2
Activity Coefficient:
The activity coefficient of CO
2
(
CO2
) was evaluated in Aspen Plus
for variable amine

concentration and CO
2
loading. Model results are compared against experimental points by
Cullinane (2005). Figure 9 shows that the model adequately represents the change in
CO2
with
amine concentration. However, for 8m PZ at various loadings an erratic behavior can be
observed with a maximum around a loading of 0.30 (Figure 10).
2
20
0.20 0.25 0.30 0.35 0.40 0.45
m
(
c
P
)
Loading (mol CO2/mol alkalinity)
25
o
C
40
o
C
60
o
C
80
o
C (extrapolated)
159
9

Figure 9: CO
2
Activity coefficient change for unloaded solutions. VLE by Van Wagener
version 02/06/09 (Rochelle et al., 2009). Points from Cullinane (2005). Lines Aspen Plus

model results.

0.95
1.05
1.15
1.25
1.35
1.45
0 2 4 6 8 10
g
C
O
2
PZ concentration (m)
25
o
C
40
o
C
160
10

Figure 10: Change in CO
2
activity coefficient with loading for 8 m PZ. VLE by Van
Wagener version 02/06/09 (Rochelle et al., 2009). Lines Aspen Plus
model results.
PZ Pilot Plant Analysis
PZ at 5 m, 8 m and 9 m was tested at the Pickle Research Center in November 2008. The
absorber was packed with Mellapak 2X and operated at volume liquid-gas ratios (L/G) from 4.6
x10
-3
to 6.9 x 10
-3
. Inlet CO
2
gas concentration varied from 1012%.
The overall gas-side mass transfer coefficient (K
g
) was calculated using absorber data and the
following expression:

(15)
N
CO2
is the absorbed CO
2
(gmol/s). V
p
is the column packing volume (m
3
) and a
eff
is the
packing effective area calculated using the Tsai et al. (2008) correlation:

= 1.329
1
3
4
3
0.116

(16)
where:
a
p
is the nominal packing area (m
2
/m
3
);
L
is the liquid density (kg/m
3
);
is the surface tension (N/m);
0.0
0.5
1.0
1.5
2.0
2.5
0.1 0.2 0.3 0.4 0.5
g
C
O
2
Loading (mol CO
2
/mol alkalinity)
25
o
C
40
o
C
161
11
Q is the liquid flow (m
3
/s);
L
p
is the wetter perimeter in cross sectional slice of packing (m).
LMPD is the Log Mean Pressure Difference calculated between absorber inlet and outlet
conditions:

=
ln

(17)
P is the difference between the actual CO
2
partial pressure and the CO
2
pressure in equilibrium
with the solvent loading.
The calculated K
g
was plotted against the arithmetic average between inlet and outlet CO
2

equilibrium pressures. Results were plotted in Figures 11 and 12 along with values for k
g
by
Dugas (Rochelle et al., 2008a) and previous pilot plant studies (Cullinane, 2005: Plaza et al.,
2008).

Figure 11. Comparison between K
g
and k
g
data for 5 m PZ. Lines are k

g
data by Dugas
(Rochelle et al., 2008a). Diamond shape points are for 5 m PZ (large diamond is the
average of the runs). Squares are average results for other concentrations tested in the
November 2008 campaign. Circles are data from other campaigns and solvents (Cullinane,
2005: Plaza et al., 2008).
1.E-07
1.E-06
1.E-05
100 1000 10000
K
g

-
k
g
'
(
g
m
o
l
/
P
a
-
m
2
-
s
)
PCO
2
*
(Pa)
5m PZ 60
o
C
5m PZ 40
o
C
5m PZ
9m PZ
8m PZ
7.5m PZ
6.4m K
+
/1.6m PZ
7 MEA
5m K
+
/2.5 m PZ
9m MEA
162
12

Figure 12: Comparison between K
g
and k
g
data for 8 m PZ. Lines are k

g
data by Dugas
(Rochelle et al., 2008a). Diamond shape points are for 8 m PZ (large diamond is the
average of the runs). Squares are average results for other concentrations tested in the
November 2008 campaign. Circles are data from other campaigns and solvents (Cullinane,
2005: Plaza et al., 2008).

Conclusions
PZ model development
The VLE for CO
2
PZ was accurately represented with the new reduced reaction set. Density
and viscosity were also adequately implemented for the PZ model at absorber conditions.
The CO
2
activity coefficients for unloaded solutions at low PZ concentrations are consistent with
the experimental data presented by Cullinane (2005) and behave adequately at higher
concentrations (>2 m). However, experimental data at higher concentrations is required to verify
the accuracy of the extrapolation.
The effect of loading on the CO
2
activity coefficient is not being represented correctly by the
model at the desired PZ concentrations. The presence of a maximum at a mid loading point
needs to be addressed and experimental data is necessary for the regression of the activity
coefficient at the proposed PZ concentrations (> 5 m) and loadings.
PZ Pilot Plant Analysis
The calculated K
g
data from the pilot plant campaign is consistent with the values of k
g
obtained
by Dugas (Rochelle et al., 2008a) in the wetted wall column (Figures 11 and 12).
1.E-07
1.E-06
1.E-05
100 1000 10000
K
g
-
k
g
'
(
g
m
o
l
/
P
a
-
m
2
-
s
)
PCO
2
*
(Pa)
8m PZ 40
o
C
8m PZ 60
o
C
5m PZ
9m PZ
8m PZ
7.5m PZ
5m K
+
/2.5 m PZ
9m MEA
7 MEA
8m PZ 40
o
C
8m PZ 60
o
C
5m PZ
9m PZ
8m PZ
7.5m PZ
5m K
+
/2.5 m PZ
9m MEA
7 MEA
6.4m K
+
/1.6m PZ
163
13
The 5 m PZ run exhibits the highest K
g
values (Figure 12). It is 5 times higher than the data for
7 m MEA, 4 times higher than 5 m K
+
/2.5 m PZ and 20% higher than 8 m PZ.

Future Work
Work will continue to develop a rigorous PZ model for high concentration PZ (> 5 m). Issues
related to the activity coefficient of CO
2
with respect to loading will be addressed by modifying
some of the parameters necessary to calculate
CO2
. The values of
CO2
at high loadings will be
approximated using values by Cullinane (2005) for K
+
/PZ.
Kinetics will be developed using the reduced reaction set, values reported by Cullinane (2005)
and recent data by Dugas (Rochelle et al., 2008a)
Data from the November 2008 campaign will be used to validate the PZ model. Data
reconciliation in Aspen Plus
will be used for this purpose. Additional absorber configurations

including intercooling will be evaluated. The critical L/G (Plaza et al., 2009) for the upcoming
pilot plant campaign will be predicted.
The amine water wash and the flue gas blower will be included in the developed model and the
most recent model for MEA.
References
Cullinane JT. Thermodynamics and Kinetics of Aqueous Piperazine with Potassium Carbonate
for Carbon Dioxide Absorption. The University of Texas at Austin. Ph.D. Dissertation.
2005
Gas. The University of Texas at Austin. Ph.D Dissertation. 2008
Plaza JM et al., "Modeling CO
2
Capture with Aqueous Monoethanolamine" in 9th International
Conference on Greenhouse Gas Control Technologies. Elsevier: Washington D.C. 2008.
Plaza JM et al. "Absorber Intercooling in CO
2
Absorption by Piperazine Promoted Potassium
Carbonate". Submitted to AIChE J. 2009.
2
2008". Luminant Carbon Management Program. The University of Texas at Austin.
2008a
2
2008b
2
Capture by Aqueous Absorption, First Quarterly Progress Report 2009".
Luminant Carbon Management Program. The University of Texas at Austin. 2009
Tsai R et al., "Influence of Viscosity and Surface Tension on the Effective Mass Transfer Area
of Structured Packing" in 9th International Conference on Greenhouse Gas Control
Technologies. Elsevier: Washington D.C. 2008.
Weiland R et al. "Density and Viscosity of Some Partially Carbonated Aqueous Alkanolamine
Solutions and Their Blends". J. Chem. Eng. Data. 1998. 43: p. 378-382.

164
1
Total P of CO
2
Loaded Aqueous Amines at High T

by Qing Xu, Martin Metzner
and the
2
July 7, 2009
Abstract
In this quarter a series of total pressure measurements were conducted with CO
2
loaded
monoethanolamine (MEA) or piperazine (PZ) at 100 to 160
o
C (for MEA) or 190 C (for PZ).
A 500 mL 316 SS autoclave was used as the equilibrium cell. The total pressure of 8 m PZ
with 0.465 CO
2
loading varied from 7.4 to 25.8 bar at 100 to 147 C. At 100 to 150 C, for
5 m PZ with 0.293 CO
2
loading P
t
varied from 1.1 to 7.2 bar; for 7 m MEA with 0.316 CO
2

loading P
t
is from 1.1 to 6.7 bar; for 7 m MEA with 0.479 CO
2
loading P
t
is from 3.3 to 15.9
bar.
2
at each experimental condition was estimated by subtracting
partial pressures of water and amines. The calculated results for 7 m MEA match well with
literature data. The regression based on data from 40 to 190 C gives empirical models for
CO
2
partial pressure over loaded aqueous PZ and MEA respectively and the models predict
the data well. Heat of absorption for CO
2
loaded aqueous PZ and MEA was calculated from
these empirical models.
For PZ:
2
2
1
ln 38.4 ( 102, 000 / ) 20.6 13, 200 3.23
CO
P J mol
RT T
= + + + (1)
ln
( 102, 000 13, 200 )( / )
1
( )
P
H R R J mol
T
= = +
(2)
For MEA:
2
2
1
ln 44.2 ( 116, 000 / ) 29.7 11, 600 17.3
CO
P J mol
RT T
= + + + (3)
ln
( 116, 000 11, 600 )( / )
1
( )
P
H R R J mol
T
= = +
(4)
165
2
Introduction
Figure 1 shows the conditions in a typical post-combustion CO
2
capture process with aqueous
MEA solution.

Figure 1: High Temperature Parts in CO
2
Capture Process
For concentrated PZ solution, thermal degradation is negligible up to 150 C (Freeman et al.,
2008). Many stripper configurations operate more efficiently at high temperatures. Thus
for concentrated PZ solutions, better energy performance may be obtained by increasing
stripper pressure without degradation of PZ (Rochelle et al., 2008). The stripper pressure of
the pilot plant campaign in 2008 approached 60 psia (Rochelle et al, 2009a). The non-ideal
behavior of concentrated PZ at high temperatures might make its volatility comparable to that
of MEA (Freeman et al., 2008).
For MEA solutions, due to the high thermal degradation rate, the stripper must operate below
120 C. However, the temperature in the reboiler is about 120 C and it can increase to
160 C in the thermal reclaimer. Many other aqueous amines also have relatively high
temperature and pressure processes in CO
2
capture. VLE research at high T and P can help
understand these processes.
In this period, the total pressure measurement only involved static operation. An autoclave
was used as the equilibrium cell. 2 runs for 7 m MEA, 4 runs for 8 m PZ, and 3 runs for
5 m PZ were conducted. All the experiments were performed by Martin Metzner, an
undergraduate research assistant, supervised by Qing Xu.
CO
2

Waste
Flue gas
ABS
HX
STRIP
Lean
Rich
Purified gas
Lean
Rich
Reclaimer
Reboiler
40-50 C
1 atm
100-110 C
1-2 atm
Up to 160 C
166
3

Figure 2: Total Pressure Measurement with an Autoclave
An autoclave (ZipperClave
, by Autoclave Engineers) was used. Its designed operating

range is up to 2000 psia and 232 C. The 500 mL pressure vessel is made of 316SS stainless
steel. Closure is effected by a resilient spring member (the Zipper) inserted through a
circumferential groove in the body and cover (Autoclave Engineers). A quick release/safety
lock ensures that the spring is fully inserted and makes it easy to open and close the
equilibrium cell. A magnetic agitator (MAG075, MagneDrive II Series, by Autoclave
Engineers) was used to get equilibrium without leaking to the atmosphere. It was driven by
a compressed air motor (2AM-NCC-16, by Gast
). The agitator is good for both liquid and

vapor phases. It has a hollow shaft, which draws the gas into the middle of the shaft when
the agitation starts. It is then dispersed through the impeller hub and mixes with the liquid.
A platinum resistance thermometer (model 5622-16, by Hart Scientific), installed inside the
thermal well of the vessel, was used for temperature measurement. It was connected to a
series data logger (PT-104, by Picotech), and the temperature was recorded by Picolog
software. Temperature was controlled by a Fuji Electric PXZ-4 temperature controller, with
connection to an Omega K-type thermocouple placed between the autoclave vessel wall
and the heating jacket. With 5 m PZ and with 8 m PZ at 0.465 CO
2
loading 8, the PRT
thermometer was replaced by an Omega K-type thermocouple, which was used for both
measurement and temperature control.
A pressure transducer (Validyne
DP15) was connected to a pressure indicator (Validyne

CD379). Because the indicator does not display pressure directly, calibration was
performed by heating water and correlating the readings with known vapor pressures.
About 330 to 350 mL water was added to the vessel, the cover was tightly sealed, and the
autoclave was heated up. At about 110 C, vapor was released from a valve on top of the
vessel directly to the back of hood until temperature dropped to 100 C. This was processed
3 times to purge all the air. Then the autoclave was heated to certain temperatures up to
200 C. Both temperature and the readings of the pressure indicator were recorded. The
vapor pressures of pure water at each temperature were found from DIPPR Chemical
Database. A calibration curve which relates the pressure indicator reading and the real
Data
Acquisition
system
Autoclave
Vent air
Compressed air
P transducer
Thermocouple
or PRT
Heating
jacket
Temperature
control
Liq.
Vap.
P indicator
Air motor for
the agitator
167
4
pressure in the vessel was regressed and used for further experiments. The calibration
method and results can be found in Appendix 1.
Before each run, about 330 to 350 mL of the CO
2
loaded aqueous amine solution was
prepared and added into the autoclave. To avoid the effects of O
2
, N
2
was used to purge air
and then the cell was sealed. Initial pressure and temperature readings were recorded for
later data correction. Then the cell was heated. Data recording of both temperatures and
pressures started at around 100 C and the intervals of the data points were about 10 C.
After it reached 160 C for MEA and up to 190 C for PZ, heating was stopped and the
system started to cool down, but the heating jacket was still used from time to time to
maintain certain temperatures. Data were also recorded when cooling down. Liquid
samples were collected before and after each experiment and analyzed by TIC and acid
titration. The agitation rate varied from 1500 RPM to 2500 RPM, depending on the
viscosity of the mixture.
Analytical Methods
Total Inorganic Carbon (TIC)
The concentration of CO
2
in solution was determined by TIC analysis. The liquid samples
collected before and after each run were diluted by a factor of 100. About 1015 L diluted
sample was injected into a CO
2
analyzer (Model 525, Horiba PIR 2000). Details can be
found in Appendix B.2 of Hilliards dissertation (2008).
Acid Titration
The total alkalinity of solution was determined by acid titration using a Metrohm-Peak 835
Titrando equipped with an automatic dispenser, Metrohm-Peak 801 stirrer, and 3M KCl pH
probe. Details are available in Appendix A.3 of Hilliard (2008) and Appendix F of Sexton
(2008).
Results
Total Pressure
Table 1 shows the measured total pressure for each run, including the data from last quarter.
Table 1: Measured Total Pressure in This Work
Amine Total Pressure(bar)
Type Concentration(m)
ldg
@100C @150C
6.97 0.427 1.8 14.9
6.86 0.479 3.3 15.9
6.82 0.389 1.1 9.1
MEA
6.86 0.316 1.1 6.7
7.78 0.314 1.9 13.0
7.43 0.312 1.4 11.8
7.93 0.33 N/A 11.9
7.92 0.306 0.9 9.0
PZ
7.94 0.424 3.9 20.6
168
5
Amine Total Pressure(bar)
Type Concentration(m)
ldg
@100C @150C
7.75 0.378 3.2 18.8
7.93 0.374 N/A 17.7
7.86 0.252 0.7 6.7
8.00 0.465 7.4 25.8*
4.93 0.293 1.1 7.2
4.97 0.339 1.1 8.7
4.96 0.374 1.8 12.6
* at 147 C.

Aqueous PZ
In this work only total pressure can be measured directly. The partial pressure of CO
2
was
obtained by subtracting P
water
and P
amine
from the corrected P
total
. The raw data and
calculation examples are in Appendices 2 and 4. Table 2 shows the partial pressure of CO
2

in aqueous PZ from this work, including data from the last period.
CO
2
loading is defined as:
2 2
moles of CO ( )
moles of equivalent amine ( ) 2 ( )
n CO
ldg
n MEA n PZ
= =
+
(5)

Table 2: Partial Pressure of CO
2
in Aqueous PZ in This Work
PZ T CO2 ldg PCO2 Pt PZ T CO2 ldg PCO2 Pt PZ T CO2 ldg PCO2 Pt
m C mol/molalk Pa Pa m C mol/molalk Pa Pa m C mol/molalk Pa Pa
7.78 110 0.314 122004 250946 7.94 130 0.424 890663 1134289 8.00 146.7 0.465 2190458 2583187
7.78 120 0.314 206448 385261 7.94 140 0.424 1232321 1558442 8.00 140.5 0.465 1890708 2221231
7.78 130 0.314 261222 504652 7.94 146 0.424 1439628 1825179 8.00 128.3 0.465 1415352 1646638
7.78 140 0.314 507117 832979 7.94 150 0.424 1631711 2061510 8.00 117.8 0.465 1061530 1228207
7.78 150 0.314 881085 1310544 7.94 157 0.424 1893836 2410800 8.00 112.2 0.465 886736 1025533
7.78 160 0.314 1230159 1788109 7.94 150 0.424 1624908 2054707 8.00 100.6 0.465 593660 686342
7.78 150 0.314 851237 1280696 7.94 139 0.424 1217419 1534395 4.93 110.6 0.293 37363 174033
7.78 140 0.314 536965 862826 7.94 130 0.424 936923 1180550 4.93 131.1 0.293 77414 236125
7.78 130 0.314 335841 579272 7.94 120 0.424 676762 855720 4.93 138.9 0.293 145351 338550
7.78 120 0.314 206448 385261 7.94 110 0.424 474636 603683 4.93 150.0 0.293 263278 473104
7.78 110 0.314 122004 250946 7.94 100 0.424 318940 410152 4.93 159.4 0.293 435371 708830
7.43 120 0.312 90102 270140 7.94 90 0.424 195049 258119 4.93 169.4 0.293 649478 1380592
7.43 130 0.312 277354 522437 7.94 82 0.424 125038 171188 4.93 180.0 0.293 970061 1909424
7.43 140 0.312 521298 849354 7.75 100 0.378 170068 262580 4.93 191.1 0.293 1409788 2614787
7.43 150 0.312 863332 1295661 7.75 110 0.378 276202 407013 4.93 180.6 0.293 1014296 1966727
7.43 160 0.312 1299707 1861361 7.75 120 0.378 436736 618119 4.93 170.0 0.293 694638 1436458
7.43 170 0.312 1930831 2650918 7.75 134 0.378 711388 989360 4.93 160.6 0.293 444675 1032286
7.43 180 0.312 2976199 3888194 7.75 140 0.378 898772 1229235 4.93 150.0 0.293 286258 731810
7.43 169 0.312 1351290 2054103 7.75 151 0.378 1396338 1843641 4.93 140.0 0.293 169141 507307
169
6
PZ T CO2 ldg PCO2 Pt PZ T CO2 ldg PCO2 Pt PZ T CO2 ldg PCO2 Pt
7.43 160 0.312 956457 1518110 7.75 160 0.378 1894297 2460012 4.93 130.0 0.293 100479 353178
7.43 150 0.312 639473 1071803 7.75 160 0.378 1867497 2433213 4.93 120.0 0.293 56453 242135
7.43 140 0.312 416830 744886 7.75 150 0.378 1486603 1922081 4.93 108.9 0.293 28143 157210
7.43 130 0.312 187810 432894 7.75 140 0.378 1070682 1401145 4.93 100.6 0.293 18161 114694
7.43 119 0.312 140659 315053 7.75 130 0.378 787574 1034471 4.97 130.0 0.339 155641 408218
7.43 110 0.312 111869 241702 7.75 120 0.378 570081 751464 4.97 140.0 0.339 261383 599378
7.93 104 0.330 122886 289325 7.75 110 0.378 414122 544933 4.97 150.6 0.339 431175 883729
7.93 111 0.330 177648 347203 7.93 121 0.374 298160 482894 4.97 160.0 0.339 664657 1243089
7.93 120 0.330 289389 462958 7.93 137 0.374 750749 1050095 4.97 170.0 0.339 957045 1698510
7.93 130 0.330 465124 643168 7.93 146 0.374 1093812 1479419 4.97 180.6 0.339 1362623 2314623
7.93 140 0.330 755280 937817 7.93 152 0.374 1353818 1807285 4.97 183.9 0.339 1465469 2491812
7.93 150 0.330 1192741 1379792 7.93 161 0.374 1796602 2369457 4.97 180.0 0.339 1314079 2253030
7.93 160 0.330 1777501 1969091 7.93 163 0.374 1913345 2515848 4.97 170.0 0.339 974279 1715745
7.93 167 0.330 2472784 2667568 7.93 150 0.374 1301407 1731268 4.97 160.0 0.339 679019 1257451
7.93 161 0.330 1736266 1928314 7.93 139 0.374 890953 1207975 4.97 148.9 0.339 433183 865522
7.93 149 0.330 1183986 1370584 7.93 129 0.374 601494 837945 4.97 140.0 0.339 278617 616613
7.93 139 0.330 747838 929925 7.93 119 0.374 384305 557678 4.97 130.0 0.339 167130 419707
7.93 125 0.330 326615 502420 7.93 110 0.374 219026 348092 4.97 118.9 0.339 86984 266187
7.92 130 0.306 127680 371356 7.86 129.9 0.252 51537 294670 4.96 100.0 0.374 74309 168981
7.92 140 0.306 245208 571395 7.86 140.9 0.252 113667 448507 4.96 109.4 0.374 137126 268343
7.92 150 0.306 446316 876201 7.86 151.5 0.252 222010 669834 4.96 120.0 0.374 239208 424820
7.92 160 0.306 704144 1262643 7.86 163.4 0.252 426793 1035847 4.96 130.0 0.374 383427 636038
7.92 170 0.306 1135744 1851814 7.86 171.9 0.252 638306 1388487 4.96 140.0 0.374 570972 909014
7.92 174 0.306 1394408 2182589 7.86 180.5 0.252 925639 1843775 4.96 150.0 0.374 825638 1271034
7.92 170 0.306 1226905 1942975 7.86 191.8 0.252 1440747 2623128 4.96 160.0 0.374 1136407 1714917
7.92 160 0.306 830679 1389179 7.86 182.9 0.252 1059682 2029640 4.96 170.0 0.374 1522077 2263643
7.92 150 0.306 495978 925863 7.86 173.2 0.252 723127 1497001 4.96 175.0 0.374 1756862 2592581
7.92 140 0.306 280584 606771 7.86 159.8 0.252 398841 954906 4.96 170.6 0.374 1564217 2316616
7.92 130 0.306 148769 392445 7.86 149.5 0.252 238340 662796 4.96 160.6 0.374 1181920 1769326
7.92 120 0.306 83481 262476 7.86 141.1 0.252 154799 491536 4.96 148.3 0.374 814471 1239909
7.94 81 0.424 92604 136936 7.86 131.1 0.252 91982 343991 4.96 140.0 0.374 599696 937738
7.94 89 0.424 149554 210261 8.00 100.0 0.465 709231 799937 4.96 130.6 0.374 410284 667468
7.94 94 0.424 198010 271311 8.00 110.0 0.465 876977 1005924 4.96 120.6 0.374 268515 457686
7.94 101 0.424 274363 368890 8.00 120.0 0.465 1169529 1348350 4.96 110.0 0.374 167312 301209
7.94 111 0.424 442568 576028 8.00 129.4 0.465 1499210 1738306 4.96 100.6 0.374 100818 197539
7.94 120 0.424 617576 796534 8.00 140.6 0.465 1917040 2248496

CO
2
partial pressure data over 0.912 m PZ have been previously reported by Hilliard (2008),
Dugas and Nguyen (Rochelle et al., 2008), and Ermatchkov (2006). The original data can
be found in Appendix 2. Based on those data and the high temperature data in this work, an
empirical model was developed:
170
7
2
2
1
ln 38.4 ( 102, 000 / ) 20.6 13, 200 3.23
CO
P J mol
RT T
= + + +
ln
( 102, 000 13, 200 )( / )
1
( )
P
H R R J mol
T
= = +

2
0 9902 R . =
Figures 3 and 4 show the prediction of P
CO2
by this empirical model. P
calc
was calculated
using the model while P
exp
was measured in experiments.
0.4
0.8
1.2
1.6
2.0
40 60 80 100 120 140 160 180 200
T (C)
P
c
a
l
c
/
P
e
x
p

Figure 3: Prediction of PCO2 over Aqueous PZ by Empirical Model, T Dependence
0.4
0.8
1.2
1.6
2.0
0.1 0.2 0.3 0.4 0.5
CO
2
loading (mole/equiv PZ)
P
c
a
l
c
/
P
e
x
p

Figure 4: Prediction of PCO2 over Aqueous PZ by Empirical Model, Ldg Dependence
Data at low T
New data at high T
Data at low T
New data at high T
171
8

Figure 5 shows CO
2
solubility in aqueous PZ at temperatures from 40 to 160 C and the heat
of absorption over 0.20.5 CO
2
loading range. The points are experimental data and curves
in different colors are the values predicted by the empirical model. According to this figure
the model fairly predicts the data at both high and low temperature, and the CO
2
partial
pressure strongly depends on loading and temperature. Points at 80, 100, and 120 C
indicate that results from this work match well with former data. The heat of absorption
decreases at increased CO
2
loading. H
abs
is 47 to 80 kJ/mol at 0.5 to 0.2 loading. Both
PCO2 and H are independent of PZ concentration in this empirical correlation.
1.E+01
1.E+02
1.E+03
1.E+04
1.E+05
1.E+06
1.E+07
0.20 0.25 0.30 0.35 0.40 0.45 0.50
CO
2
loading (moles/equiv PZ)
P
C
O
2

(
P
a
)
45
50
55
60
65
70
75
80
H
a
b
s

(
k
J
/
m
o
l
)
Figure 5: CO2 solubility over aqueous PZ
Aqueous MEA
In this work, P
CO2
was obtained using the same method as with aqueous PZ. The raw data
and calculation samples are in Appendices 3 and 4. Table 3 shows the partial pressure of
CO
2
in aqueous MEA from this work.
Table 3: Partial Pressure of CO
2
in MEA in This Work
MEA T CO2 ldg PCO2 Pt MEA T CO2 ldg PCO2 Pt MEA T CO2 ldg PCO2 Pt
6.97 100 0.427 100613 191250 6.86 160 0.479 1426541 1983628 6.82 120 0.389 108188 286547
6.97 110 0.427 137624 265870 6.86 166 0.479 1568194 2216066 6.82 110 0.389 60929 189520
6.97 120 0.427 237224 415109 6.86 161 0.479 1441713 2013184 6.82 100 0.389 34293 125176
6.97 130 0.427 351956 594196 6.86 149 0.479 997132 1414288 6.86 100.5 0.316 15881 108340
6.97 140 0.427 613050 937446 6.86 139 0.479 715741 1031654 6.86 111.3 0.316 22059 156294
6.97 150 0.427 1061899 1489631 6.82 111 0.389 22177 155167 6.86 121.8 0.316 45436 228767
Filled points: this work
Open points: former work
40 C
160 C
80 C
120 C
H
172
9
MEA T CO2 ldg PCO2 Pt MEA T CO2 ldg PCO2 Pt MEA T CO2 ldg PCO2 Pt
6.97 140 0.427 463811 788207 6.82 120 0.389 62433 240791 6.86 131.9 0.316 73217 330103
6.97 130 0.427 292260 534500 6.82 131 0.389 153209 403473 6.86 141.4 0.316 130830 468989
6.97 120 0.427 192452 370337 6.82 140 0.389 288705 613945 6.86 150.2 0.316 213155 644022
6.97 110 0.427 122700 250946 6.82 150 0.389 463202 892032 6.86 159.0 0.316 332091 875068
6.97 100 0.427 85689 176326 6.82 160 0.389 723822 1281241 6.86 152.0 0.316 235784 687920
6.86 101 0.479 94696 188823 6.82 170 0.389 1090007 1805101 6.86 142.3 0.316 144827 491656
6.86 111 0.479 170514 303420 6.82 170 0.389 1050788 1765882 6.86 129.9 0.316 77312 319308
6.86 121 0.479 283022 466998 6.82 160 0.389 731665 1289085 6.86 120.4 0.316 50175 230689
6.86 130 0.479 448092 690820 6.82 150 0.389 500460 929291 6.86 109.0 0.316 35729 159965
6.86 140 0.479 683104 1008146 6.82 140 0.389 307661 632901 6.86 101.1 0.316 12239 106695
6.86 150 0.479 1011870 1440443 6.82 129 0.389 176362 412044

CO
2
partial pressure data over 3.513 m MEA have been reported by Hilliard (2008), Jou
(1995), and Dugas (Rochelle et al., 2008). The original data can be found in Appendix 3.
Based on those data and the high temperature data in this work, an empirical model was
developed:
2
2
1
ln 44.2 ( 116, 000 / ) 29.7 11, 600 17.3
CO
P J mol
RT T
= + + +
ln
( 116, 000 11, 600 )( / )
1
( )
P
H R R J mol
T
= = +

2
0 9884 R . =
Figures 6 and 7 show the prediction of P
CO2
by this empirical model. P
calc
was calculated
using the model while P
exp
was measured in experiments.
0.2
0.7
1.2
1.7
2.2
40 60 80 100 120 140 160 180
T(C)
P
c
a
l
c
/
P
e
x
p

Figure 6: Prediction of PCO2 over Aqueous MEA by Empirical Model, T Dependence
Data at low T New data at high T
173
10

0.2
0.7
1.2
1.7
2.2
0.05 0.15 0.25 0.35 0.45 0.55
Loading (mole CO
2
/mole MEA)
P
c
a
l
c
/
P
e
x
p

Figure 7: Prediction of PCO2 over Aqueous MEA by Empirical Model, Ldg Dependence
Figure 8 shows CO
2
solubility in aqueous MEA solutions at temperature from 40 to 160 C
and the heat of absorption over 0.3-0.5 CO
2
loading range. The points are experimental data
and curves in different colors are the predicted values by the empirical model. The heat of
absorption is about 67 to 86 kJ/mol at 0.5 to 0.3 loading. Both PCO2 and H are
independent on MEA concentration.
1.E+01
1.E+02
1.E+03
1.E+04
1.E+05
1.E+06
1.E+07
0.30 0.35 0.40 0.45 0.50
Loading (moles CO
2
/mole MEA)
P
C
O
2

(
P
a
)
65
70
75
80
85
90
H
a
b
s

(
k
J
/
m
o
l
)

Figure 8: CO
2
solubility over aqueous MEA
Data at low T
New data at high T
40 C
80 C
120 C
160 C
H
Open points: former work
174
11

Comparison with literature data
Jou et al. (1995) has reported the solubility of CO
2
in 7 m MEA between 0 and 150 C at
partial pressure of CO
2
ranging from 0.001 to 20,000 kPa. Table 4 shows their data at 100,
120, 150 C.
Table 4: Solubility of CO
2
in 7 m MEA by Jou et al, 1995
100 C 120 C 150 C
Loading PCO2(kPa) Loading P CO2 (kPa) Loading P CO2 (kPa)
0.941 19812 0.863 17723 0.755 16441
0.914 14842 0.829 14741 0.727 11504
0.856 9871 0.7815 9770 0.6505 8525
0.81 5891 0.7205 5809 0.583 4544
0.705 2899 0.639 2804 0.484 570
0.6135 909 0.536 822 0.388 560
0.571 509 0.473 422 0.358 420
0.589 376 0.444 222 0.301 123
0.481 109 0.4115 122 0.153 34.2
0.477 69 0.349 46.8 0.0509 2.63
0.422 39 0.119 2.29 0.0134 0.184
0.381 19 0.0247 0.0984 0.00992 0.0843
0.188 1.43 0.0112 0.0221 0.00496 0.0239
0.0566 0.136 0.00333 0.00202 0.00199 0.00477
0.0117 0.00724

Figure 9 shows the comparison of data in this work and in Jous work. The CO
2
loading
range is restricted to 0.30.5 in this graph. The curves are from the empirical model. The
new data fairly match Jous work, and the model predicts CO
2
solubility well. The data at
100 C are not as good because the CO
2
partial pressure estimation method in this work gets
closer results to the true values when it is at high pressure.
175
12
1.E+04
1.E+05
1.E+06
1.E+07
0.3 0.35 0.4 0.45 0.5
Loading (mole CO
2
/mole MEA)
P
C
O
2
(
P
a
)

Figure 9: CO
2
solubility over aqueous MEA, comparison with literature data
Conclusions
The total pressure of 8 m PZ with 0.465 CO
2
loading varied from 7.4 to 25.8 bar at 100 to
147 C. At 100 to 150 C, for 5 m PZ with 0.293 CO
2
loading P
t
varied from 1.1 to 7.2 bar;
for 7 m MEA with 0.316 CO
2
loading P
t
is from 1.1 to 6.7 bar; for 7 m MEA with 0.479 CO
2

loading P
t
is from 3.3 to 15.9 bar.
The calculated CO
2
partial pressure matches well with previous data at lower temperatures.
The calculated results for 7m MEA also match well with the work by Jou et al (1995) at high
temperature.
For a specific amine solution P
CO2
is a function of temperature and CO
2
loading. A new
empirical model for aqueous PZ was developed based on data from 40 to 190 C:
2
2
1
ln 38.4 ( 102, 000 / ) 20.6 13, 200 3.23
CO
P J mol
RT T
= + + +
Another empirical model for aqueous MEA was developed based on data from 40 to 160 C:
2
2
1
ln 44.2 ( 116, 000 / ) 29.7 11, 600 17.3
CO
P J mol
RT T
= + + +
Both the models predict the data well. Heat of absorption for CO
2
loaded aqueous PZ and
MEA were calculated from these empirical models.
For PZ:
ln
( 102, 000 13, 200 )( / )
1
( )
P
H R R J mol
T
= = +

Open points: Jou et al, 1995
120 C
150 C
100 C
176
13
For MEA:
ln
( 116, 000 11, 600 )( / )
1
( )
P
H R R J mol
T
= = +

Future Work
The work planned for the next period includes modifying the total pressure measurement,
adding LabView into the data acquisition system, as well as using a deadweight tester to do
independent calibration for the pressure transducer. We will then conduct more experiments
at various CO
2
loadings for more concentrated PZ and other aqueous amines. The empirical
model for PCO2 over aqueous PZ will be modified.
Based on the total pressure data, the Aspen PZ model, developed by Hilliard (2008) and
modified by Van Wagener for PZ will be tested and further modified.
We will build an apparatus for (x, y, P, T) measurement at high temperature and pressure. A
framework has been developed, but detail problems need to be fixed. The same autoclave
will be used as the equilibrium cell; vapor sample will be depressurized, diluted with nitrogen,
and analyzed by FTIR; liquid sample will be drawn out, cooled, and analyzed by TIC and
titration.
Further goals include exploring the temperature dependence of heat of absorption of CO
2

loaded amine solutions.

References
Autoclave Engineers, Zipperclave 500&1000 mL stirred reactor,
http://www.autoclaveengineers.com/ae_pdfs/SR_500_1000_Zip.pdf
DIPPR, 1998-Provo, UT: BYU DIPPR, Thermophysical Properties Laboratory, 1998-Version
13.0.
Ermatchkov V et al. "Solubility of Carbon Dioxide in Aqueous Solutions of Piperazine in the
Low Gas Loading Region." J Chem Engr Data. 2006;51(5):17881796.
Freeman SA et al. "Carbon dioxide capture with concentrated, aqueous piperazine." GHGT-9,
Washington D.C. 2008.
Gas. The University of Texas at Austin. Ph.D. Dissertation. 2008;1083.
Jou F-Y, Mather AE et al. "The Solubility of CO
2
in a 30 Mass Percent Monoethanolamine
Solution." Can J Chem Eng. 1995;73(1):140147.
2
2008." Luminant Carbon Management Program. The University of Texas at Austin.
2008.
177
14
2
2009a.
2
Capture by Aqueous Absorption, First Quarterly Progress Report
2009b.
Sexton AJ. Amine Oxidation in CO
2
Capture Processes. The University of Texas at Austin.
Ph.D. Dissertation. 2008.

Appendices
Appendix 1: Calibration
Calibration for Autoclave - nitrogen
The following calibration was used for run MEA-3 and PZ-6. About 1 atm of nitrogen at
room temperature was purged into the autoclave before calibration and before each
experiment run. Thus corrections were made to the measured total pressure.
Table 5: Calibration for Autoclave - nitrogen
T(C) Indicator reading P water(kPa) P N
2
(Pa) P(Pa) T(C)Indicator readingP water(kPa) P N
2
(Pa) P(Pa)
105 0.2225 120.7 128513 249213 200 1.357 1551.6 160798 1712398
110 0.242 143.12 130212 273332 190 1.131 1252.5 157400 1409900
121 0.295 204.64 133950 338590 180 0.942 1000.5 154001 1154501
131 0.351 277.88 137349 415229 169 0.769 771.45 150263 921713
140 0.417 360.75 140407 501157 160 0.6505 616.82 147204 764024
150 0.502 475.09 143806 618896 150 0.541 475.09 143806 618896
160 0.613 616.82 147204 764024 140 0.4505 360.75 140407 501157
171 0.7655 809.66 150942 960602 130 0.3795 269.71 137009 406719
180 0.923 1000.5 154001 1154501 120 0.322 198.29 133610 331900
190 1.1215 1252.5 157400 1409900 110 0.2775 143.12 130212 273332
200 1.364 1551.6 160798 1712398 99 0.239 97.702 126474 224176

178
15
y=13.07x0.65
R
2
=1.00
2
4
6
8
10
12
14
16
18
0.2 0.4 0.6 0.8 1 1.2 1.4
Transducerreading
P
r
e
s
s
u
r
e
(
b
a
r
)

Figure 10: Calibration for Autoclave Nitrogen

Calibration Calculation Example:
At 105 C, the vapor pressure of water is 120700 kPa, the transducer reading is 0.2225
(average value).
At 25 C before calibration, the 1 atm nitrogen inside the vessel has a reading of 0.113.
Assume the nitrogen behaves as an ideal gas during calibration, according to PV=nRT, ignore
the volume change, P/Pi=T/Ti, thus:
2
105 273.15( )
101325( ) 128513( )
25 273.15( )
N
K
P Pa Pa
K
+
= =
+

The total pressure in the equilibrium cell at 105 C:
2
128513 120700 249213( )
total N water
P P P Pa = + = + =
Then correlate the P
total
with readings from the indicator and get the calibration curve in
Figure 10.

Calibration for Autoclave - vacuum
The following calibration was used for run MEA-5 and PZ 7-12. In the beginning of this
calibration at about 110 C water vapor was released from a valve on top of the vessel
directly to the back of hood until temperature dropped to 100 C. This was processed 3 times
to purge all the air. So no correction was conducted to the total pressure measured in these
experiments.
179
16
Table 6: Calibration for Autoclave - vacuum
Temperature (C) Transducer Reading Pressure (kPa)
102.0 0.155 108.70
120.0 0.219 198.29
139.3 0.319 353.66
159.4 0.487 607.46
173.0 0.744 849.35
200.1 1.153 1554.90
220.8 1.719 2351.00
226.0 1.899 2593.40
209.1 1.378 1869.80
189.3 0.958 1233.40
171.2 0.659 813.56
150.6 0.434 482.78
130.9 0.292 277.05
y = 1436.2x - 128.17
R
2
= 0.9985
0
500
1000
1500
2000
2500
3000
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Transducer Reading
P
r
e
s
s
u
r
e

(
k
P
a
)

Figure 11: Calibration for Autoclave Vacuum

180
17
Appendix 2: PZ Data Tables
Raw data for PZ 1 to 5 can be found in the section by Xu in the last quarterly progress report.
(Rochelle et al., 2009)
Table 7: P
CO2
from Hilliard Dissertation (Table D.5, 2008)
PZ Temp. CO
2
ldg P
CO2
PZ Temp. CO
2
ldg P
CO2
PZ Temp. CO
2
ldg P
CO2

m C mol/mol
alk
Pa m C mol/mol
alk
Pa m C mol/mol
alk
Pa
0.9 40 0.208 44 2 40 0.227 106 3.6 60 0.385 13600
0.9 40 0.217 70.5 2 40 0.257 184 3.6 60 0.4 19300
0.9 40 0.241 103 2 40 0.309 526 3.6 40 0.146 21.1
0.9 40 0.284 234 2 40 0.372 1950 3.6 40 0.217 62.8
0.9 40 0.344 987 2 40 0.431 10100 3.6 40 0.272 211
0.9 40 0.418 4850 2.5 40 0.166 31.7 3.6 40 0.318 687
0.9 60 0.111 29 2.5 40 0.228 88.4 3.6 40 0.384 4370
0.9 60 0.217 299 2.5 40 0.278 247 3.6 40 0.412 8420
0.9 60 0.242 841 2.5 40 0.328 662 5 40 0.172 28.7
0.9 60 0.325 1930 2.5 40 0.423 7510 5 40 0.22 60.5
0.9 60 0.37 8290 2.5 40 0.437 10600 5 40 0.274 211
0.9 60 0.383 14700 2.5 60 0.164 141 5 40 0.339 798
2 60 0.132 92.4 2.5 60 0.196 263 5 40 0.409 5710
2 60 0.193 296 2.5 60 0.251 725 5 40 0.413 6990
2 60 0.275 1400 2.5 60 0.341 3960 5 60 0.164 137
2 60 0.33 3950 2.5 60 0.4 16900 5 60 0.226 365
2 60 0.37 9910 2.5 60 0.443 27400 5 60 0.296 1290
2 60 0.412 24700 3.6 60 0.158 129 5 60 0.33 3310
2 60 0.169 142 3.6 60 0.217 431 5 60 0.386 18300
2 60 0.383 13700 3.6 60 0.277 1050 5 60 0.417 51400
2 40 0.146 21.5 3.6 60 0.338 3490

Table 8: P
CO2
from Ermatchkov (2006)
PZ Temp. CO
2
ldg P
CO2
PZ Temp. CO
2
ldg P
CO2
PZ Temp. CO
2
ldg P
CO2

m C mol/mol
alk
Pa m C mol/mol
alk
Pa m C mol/mol
alk
Pa
2.281 80 0.067 111 1.969 80 0.362 25620 4.168 80 0.264 4470
2.281 80 0.154 870 1.969 80 0.368 28420 4.168 80 0.288 6860
2.281 80 0.251 3580 1.969 80 0.382 37950 4.168 80 0.308 10150
2.156 80 0.271 5600 3.95 80 0.077 154 4.199 80 0.342 19850
2.156 80 0.312 10820 3.95 80 0.138 480 4.199 80 0.362 31400
2.156 80 0.340 18360 3.95 80 0.201 1530 4.199 80 0.404 77630

181
18
Table 9: P
CO2
from Dugas (Rochelle et al., 2008)
PZ Temp. CO
2
ldg P
CO2
PZ Temp. CO
2
ldg P
CO2
PZ Temp. CO
2
ldg P
CO2

m C mol/mol
alk
Pa m C mol/mol
alk
Pa m C mol/mol
alk
Pa
2 40 0.240 96 5 40 0.402 4563 8 60 0.360 7454
2 40 0.316 499 5 60 0.226 385 8 60 0.404 30783
2 40 0.352 1305 5 60 0.299 1814 8 80 0.253 3255
2 40 0.411 7127 5 60 0.354 5021 8 80 0.289 9406
2 60 0.240 559 5 60 0.402 17233 8 100 0.253 13605
2 60 0.316 2541 5 80 0.238 2192 8 100 0.289 32033
2 60 0.352 5593 5 80 0.321 9699 12 60 0.231 331
2 60 0.411 25378 5 100 0.238 8888 12 60 0.289 1865
2 80 0.239 2492 5 100 0.321 36960 12 60 0.354 6791
2 80 0.324 12260 8 40 0.231 68 12 80 0.222 2115
2 100 0.239 9569 8 40 0.305 530 12 80 0.290 9141
2 100 0.324 39286 8 40 0.360 1409 12 100 0.222 7871
5 40 0.226 65 8 40 0.404 8153 12 100 0.290 33652
5 40 0.299 346 8 60 0.231 430
5 40 0.354 1120 8 60 0.305 2407

Table 10: Raw Data for Run PZ-6
T(C) transducer reading Pt(Pa) T(C)
transducer
reading
Pt(Pa) T(C)
transducer
reading
Pt(Pa)
100 0.251 262580 151 1.4605 1843641 130 0.8415 1034471
110 0.3615 407013 160 1.932 2460012 120 0.625 751464
120 0.523 618119 160 1.9115 2433213 110 0.467 544933
134 0.807 989360 150 1.5205 1922081 100 0.3335 370432
140 0.9905 1229235 140 1.122 1401145

transducer
reading
Pt(Pa) T(C)
transducer
reading
Pt(Pa)
121 0.538 482894 161 1.863 2369457 129 0.7875 837945
137 0.9375 1050095 163 1.9655 2515848 119 0.5895 557678
146 1.239 1479419 150 1.4155 1731268 110 0.441 348092
152 1.469 1807285 139 1.048 1207975

182
19
transducer
reading
Pt(Pa) T(C)
transducer
reading
Pt(Pa)
100.3 0.22 74518.1 163.4 0.875 1035847 159.8 0.815 954906
110.1 0.251 139819 171.9 1.12 1388487 149.5 0.612 662796
118.5 0.288 193348 180.5 1.437 1843775 141.1 0.493 491536
129.9 0.359 294670 191.8 1.98 2623128 131.1 0.39 343991
140.9 0.466 448507 182.9 1.564 2029640
151.5 0.62 669834 173.2 1.193 1497001

transducer
reading
Pt(Pa) T(C)
transducer
reading
Pt(Pa)
100.0 0.705 799937 140.6 1.720 2248496 117.8 1.006 1228207
110.0 0.850 1005924 146.7 1.954 2583187 112.2 0.864 1025533
120.0 1.090 1348350 140.5 1.701 2221231 100.6 0.626 686342
129.4 1.363 1738306 128.3 1.299 1646638

transducer
reading
Pt(Pa) T(C)
transducer
reading
Pt(Pa)
101.1 0.222 100211 169.4 1.125 1380592 140.0 0.512 507307
110.6 0.275 174033 180.0 1.495 1909424 130.0 0.403 353178
121.1 0.320 236125 191.1 1.988 2614787 120.0 0.324 242135
131.1 0.393 338550 180.6 1.535 1966727 108.9 0.263 157210
138.9 0.488 473104 170.0 1.164 1436458 100.6 0.232 114694
150.0 0.654 708830 160.6 0.881 1032286
159.4 0.862 1005288 150.0 0.670 731810

transducer
reading
Pt(Pa) T(C)
transducer
reading
Pt(Pa)
100.0 0.232 103304 170.0 1.356 1698510 140.0 0.597 616613
109.4 0.280 169679 180.6 1.787 2314623 130.0 0.458 419707
122.2 0.364 286831 183.9 1.911 2491812 118.9 0.349 266187
130.0 0.450 408218 180.0 1.744 2253030 110.0 0.284 175261
140.0 0.585 599378 170.0 1.368 1715745 100.0 0.236 109049
150.6 0.785 883729 160.0 1.047 1257451
160.0 1.037 1243089 148.9 0.772 865522

183
20
transducer
reading
Pt(Pa) T(C)
transducer
reading
Pt(Pa)
100.0 0.279 168981 160.0 1.367 1714917 140.0 0.822 937738
109.4 0.350 268343 170.0 1.751 2263643 130.6 0.632 667468
120.0 0.461 424820 175.0 1.981 2592581 120.6 0.484 457686
130.0 0.610 636038 170.6 1.788 2316616 110.0 0.373 301209
140.0 0.802 909014 160.6 1.405 1769326 100.6 0.299 197539
150.0 1.056 1271034 148.3 1.034 1239909

Appendix 3: MEA Data Tables
Raw data for MEA-1 and MEA-2 can be found in the section by Xu in the last quarterly
progress report. (Rochelle et al., 2009)

Table 17: P
CO2
from Hilliard Dissertation (D4, 2008)
MEA T CO2 Loading PCO2 MEA T CO2 Loading PCO2 MEA T CO2 Loading PCO2
(m) (C) (mol/molalk) (Pa) (m) (C) (mol/molalk) (Pa) (m) (C) (mol/molalk) (Pa)
3.5 40 0.121 5.55 7 40 0.153 5.7 7 40 0.501 1870
3.5 40 0.212 14 7 40 0.17 7.21 7 40 0.491 1100
3.5 40 0.3 36.2 7 40 0.163 6.64 7 40 0.518 3030
3.5 40 0.369 116 7 40 0.194 9.85 7 40 0.326 48.5
3.5 40 0.467 879 7 40 0.191 9.95 7 40 0.348 66.2
3.5 40 0.552 8560 7 40 0.272 22.4 11 40 0.115 5.05
3.5 60 0.159 21.2 7 40 0.232 14.6 11 40 0.201 10.8
3.5 60 0.219 78 7 40 0.246 19.1 11 40 0.298 29.5
3.5 60 0.307 244 7 40 0.269 23.1 11 40 0.373 104
3.5 60 0.38 794 7 40 0.36 96.6 11 40 0.485 1620
3.5 60 0.477 4320 7 40 0.35 72.1 11 40 0.545 22300
3.5 60 0.504 14800 7 40 0.386 120 11 60 0.136 15.5
7 60 0.114 19.4 7 40 0.389 113 11 60 0.225 73.1
7 60 0.191 58.9 7 40 0.4 128 11 60 0.291 199
7 60 0.291 209 7 40 0.382 131 11 60 0.415 847
7 60 0.386 763 7 40 0.466 574 11 60 0.464 6980
7 60 0.485 4860 7 40 0.591 28300 11 60 0.502 26500
7 60 0.544 25800 7 40 0.481 883 11 40 0.115 5.05
7 60 0.565 50200 7 40 0.464 750 11 40 0.201 10.8

184
21
Table 18: P
CO2
from Jou et al., 1995
7 40 0.0888 1.47 7 60 0.438 2010 7 100 0.0566 136
7 40 0.203 8.96 7 60 0.504 11000 7 100 0.188 1430
7 40 0.365 67.7 7 60 0.565 34100 7 100 0.381 19000
7 40 0.461 604 7 80 0.118 99.2 7 100 0.422 39000
7 40 0.513 2570 7 80 0.187 278 7 100 0.477 69000
7 40 0.557 8090 7 80 0.348 2670 7 100 0.481 109000
7 60 0.119 19.3 7 80 0.46 16000 7 100 0.589 376000
7 60 0.206 57.9 7 80 0.517 56000
7 60 0.389 528 7 80 0.576 235000

Table 19: P
CO2
from Dugas, (Rochelle et al., 2008)
7 40 0.252 15.7 9 60 0.231 61 11 80 0.256 860
7 40 0.351 77 9 60 0.324 263 11 80 0.359 3923
7 40 0.432 465 9 60 0.382 892 11 100 0.256 4274
7 40 0.496 4216 9 60 0.441 2862 11 100 0.359 18657
7 60 0.252 109 9 60 0.496 21249 13 40 0.252 12.3
7 60 0.351 660 9 80 0.265 979 13 40 0.372 84
7 60 0.432 3434 9 80 0.356 4797 13 40 0.435 491
7 60 0.496 16157 9 100 0.265 4940 13 40 0.502 8792
7 80 0.271 1053 9 100 0.356 21534 13 60 0.252 100
7 80 0.366 4443 11 40 0.261 14.0 13 60 0.372 694
7 100 0.271 5297 11 40 0.353 67 13 60 0.435 3859
7 100 0.366 19008 11 40 0.428 434 13 60 0.502 29427
9 40 0.231 10.4 11 40 0.461 1509 13 80 0.254 873
9 40 0.324 34 11 60 0.261 96 13 80 0.355 3964
9 40 0.382 107 11 60 0.353 634 13 100 0.254 3876
9 40 0.441 417 11 60 0.428 3463 13 100 0.355 18406
9 40 0.496 5354 11 60 0.461 8171

Table 20: Raw Data for Run MEA-3
T(C) transducer reading Pt(Pa) T(C) transducer reading Pt(Pa) T(C) transducer reading Pt(Pa)
101 0.2235 91477 160 1.15 1281241 129 0.4765 412044
111 0.275 155167 170 1.5535 1805101 120 0.378 286547
120 0.343 240791 170 1.5235 1765882 110 0.301 189520
131 0.4705 403473 160 1.156 1289085 100 0.249 125176
140 0.634 613945 150 0.878 929291
150 0.8495 892032 140 0.6485 632901

185
22
Table 21: Raw Data for Run MEA-5
T(C) transducer reading Pt(Pa) T(C) transducer reading Pt(Pa) T(C) transducer reading Pt(Pa)
100.5 0.255 108340 150.2 0.64 644022 120.4 0.345 230689
111.3 0.291 156294 159.0 0.803 875068 109.0 0.293 159965
121.8 0.344 228767 152.0 0.671 687920 101.1 0.254 106695
131.9 0.417 330103 142.3 0.532 491656
141.4 0.516 468989 129.9 0.409 319308
Appendix 4: Total Pressure Correction and CO
2
Partial Pressure Calculation
In this period all the experiments used nitrogen to purge air. The total pressure of the
solution was corrected by subtracting P
N2
.
Calculation example:
In run PZ-6, at 100 C, the transducer reading is 0.251. Based on the equation from
calibration for calorimeter - air: P=1307301*(reading)-64677.3 (Pa). The initial pressure
reading of nitrogen is 0.05 at 20 C, which corresponds to 688 Pa. Assume nitrogen
behaves as an ideal gas during the run and ignore the vapor volume change,
2
100 273.15( )
688( ) 875( )
20 273.15( )
N
K
P Pa Pa
K
+
= =
+

The vapor pressure at 100 C for pure water is 101260 Pa, and for pure PZ it is 21296 Pa.
These can be found in DIPPR Chemical Database. Assume an ideal mixture in the solution,
and CO
2
is combined with PZ (or MEA, in MEA case), then the vapor pressure of water and
PZ in vapor can be calculated by Raoults Law:
*
101260 0.8775 88853( )
water water water
P P x Pa = = =
*
, , .
21296 0.1225 2609( )
PZ PZ
PZ PZCOO etc
P P x Pa
= = =
2
:
2 2
263455 875 88853 2609 171118( )
CO N water PZ
P P P P P Pa = = =
186

1
Amine Volatility

Progress Report for April 1

June 30, 2009
by Thu Nguyen
and the
Industrial Associates Program for CO2 Capture by Aqueous Absorption
July 1, 2009

Abstract
Amine volatility is a crucial screening criterion which affects fugitive emission and calls for
appropriate water wash design. At a lean CO
2
partial pressure of 500 Pa at 40 C, the ranking of
amine volatility is: 7 m MDEA/2 m PZ (7/6 ppm) < 12 m EDA (9 ppm) < 8 m PZ (14 ppm) <
7 m MEA (28 ppm) < 5 m AMP (112 ppm). The less volatile amines appear to have higher heats
of amine solution than the more volatile amines. There is no apparent correlation between
volatility and the amine heat of desorption.
Introduction
This report discusses the volatility of amine solvents at the entrance to the water wash section at
40 C and nominal lean loading. Amine volatility is analyzed in terms of apparent activity
coefficient. Both the heats of solution and vaporization of the amines are also estimated. The
amine systems that have been studied include: monoethanolamine (MEA), piperazine (PZ),
methyldiethanolamine/piperazine (MDEA/PZ), ethylene diamine (EDA), and 2-amino-2-methyl-
1-propanol (AMP).
Experimental Apparatus
Amine volatility is measured by using a setup which includes a stirred reactor coupled with a hot
gas FTIR analyzer (Fourier Transform Infrared Spectroscopy technique) manufactured by
Gasmet Inc. Figure 1 shows this VLE experimental setup.

187

2

Figure 1. Amine Volatility Experimental Setup

The 1L glass reactor is well-stirred and kept isothermal by use of dimethylsilicone oil circulating
from the oil bath. The reactor is insulated from the surrounding with aluminum foil. As the
experiment proceeds, vapor from the headspace of the reactor is continually being drawn off into
a heated line kept at an elevated temperature of 180 C which is also the FTIR operating
temperature. The gas sampling rate is between 5 and 10 L/min. Both the line and analyzer are
maintained at 180 C to prevent possible condensation or adsorption of vapor amine to any of the
inner surfaces. The FTIR is capable of multi-component analysis as it is able to measure both
CO
2
solubility and volatility of the rest of the gaseous species present, including the amines of
interest. After the gas passes through the FTIR, it is taken back to the reactor via a different line
kept at approximately 55 C higher than the equilibrium reactor temperature. It was determined
that the 55 C difference is sufficient for two reasons: (1) to ensure that the return gas does not
upset the solution that is in equilibrium with the gas inside the reactor; (2) to prevent potential
heat loss at the bottom of the reactor.
Loading is initially determined gravimetrically by weighing the amount of CO
2
that is sparged
into the amine solution. At the end of the VLE experiment, the loading is again verified by the
Total Inorganic Carbon method which measures the amount of CO
2
evolution into 30 wt %
H
3
PO
4
.
Theory
Amine volatility is quantified by the apparent activity coefficient (
amine
) as defined by the
modified Raoults law.

amine
= P
amine
/ [x
amine
* P
amine
sat
]

The reference value for
amine
is 1 which is the case of a solution having ideal species interaction.
A coefficient that is less than 1 generally indicates a low volatile system while a value greater
than 1 indicates the opposite. Note that the activity coefficients presented in this report are
188

3
apparent values, instead of being actual values, as they are computed using x
amine
that are not the
true liquid phase mole fractions of free amine present in solution, but only estimates.
The heat of amine solution, and similarly, the heat of amine vaporization (desorption from
solution), are calculated by the Gibbs-Helmholtz relations.
d (ln P
amine
) / d (1/T) = -H
vaporization
/ R
d (ln
amine
) / d (1/T) = -H
solution
/ R
Data
Table 1: 3.5 m, 7.0 m, 11.0 m MEA Volatility
MEA (m) T ( C ) Loading P
CO2
(Pa) P
MEA
(Pa)
MEA

3.50 60.0 0.00 0.0 13.20 0.34
3.57 59.9 0.16 21.2 11.00 0.28
3.63 60.1 0.22 78.0 9.26 0.23
3.53 60.0 0.31 244.0 7.20 0.19
3.57 60.0 0.38 794.0 5.08 0.13
3.55 59.9 0.48 4320.0 3.23 0.08
3.54 60.0 0.50 14800.0 2.19 0.06
3.50 40.0 0.00 0.0 4.19 0.44
3.53 40.0 0.12 5.6 3.91 0.41
3.46 40.0 0.21 14.0 3.41 0.36
3.51 39.9 0.30 36.2 2.81 0.30
3.54 40.1 0.37 116.0 2.24 0.24
3.57 40.0 0.47 879.0 1.68 0.18
3.49 40.0 0.55 8560.0 0.98 0.11
7.00 40.0 0.00 0.0 10.00 0.55
6.88 40.0 0.15 5.7 6.58 0.37
6.98 40.0 0.17 7.2 6.36 0.36
6.95 40.1 0.16 6.6 6.36 0.36
6.85 40.0 0.19 9.9 6.45 0.37
6.97 40.1 0.19 10.0 6.23 0.35
6.93 40.4 0.27 22.4 5.11 0.29
7.06 40.0 0.23 14.6 5.63 0.32
7.08 40.1 0.25 19.1 5.53 0.31
7.10 40.0 0.27 23.1 5.16 0.29
7.12 39.9 0.36 96.6 3.55 0.20
7.05 40.0 0.35 72.1 4.23 0.24
7.06 39.9 0.39 120.0 3.62 0.21
7.05 39.9 0.39 113.0 3.38 0.19
7.05 40.0 0.40 128.0 3.50 0.20
7.58 40.1 0.38 131.0 3.32 0.18
7.00 39.9 0.47 574.0 2.70 0.16
7.11 40.0 0.59 28300.0 1.46 0.08
7.06 40.0 0.48 883.0 2.47 0.14
7.17 40.0 0.46 750.0 2.66 0.15
7.06 40.0 0.50 1870.0 1.99 0.12
7.11 39.9 0.49 1100.0 1.93 0.11
189

4
7.06 40.0 0.52 3030.0 1.72 0.10
7.06 39.9 0.33 48.5 4.58 0.26
7.04 39.9 0.35 66.2 4.23 0.24
7.00 60.0 0.00 0.0 27.10 0.37
7.00 59.9 0.11 19.4 21.50 0.29
7.08 60.0 0.19 58.9 18.60 0.25
7.07 60.0 0.29 209.0 14.10 0.20
7.03 59.9 0.39 763.0 10.00 0.14
7.14 59.8 0.49 4860.0 4.94 0.07
7.17 60.1 0.54 25800.0 3.16 0.04
7.38 59.9 0.57 50200.0 2.88 0.04
11.00 40.0 0.00 0.0 12.00 0.45
11.00 40.0 0.12 5.1 10.40 0.40
10.75 40.0 0.20 10.8 8.42 0.33
10.90 39.9 0.30 29.5 6.03 0.24
11.28 40.1 0.37 104.0 4.39 0.17
11.06 40.0 0.49 1620.0 1.98 0.08
11.12 40.0 0.55 22300.0 0.95 0.04
11.00 60.0 0.00 0.0 40.20 0.37
11.21 60.0 0.14 15.5 36.09 0.33
11.17 60.0 0.23 73.1 28.38 0.27
11.12 60.0 0.29 199.0 22.52 0.21
11.36 60.0 0.42 847.0 14.30 0.14
11.32 59.9 0.46 6980.0 6.55 0.06
10.98 60.0 0.50 26500.0 4.16 0.04

Table 2: 2 m, 5 m, 8 m PZ Volatility
8 m PZ
T ( C ) Loading P
CO2
(Pa) P
PZ
(Pa)
PZ

40 0.00 0.0 14.21 0.10
40 0.28 512.6 2.14 0.02
40 0.38 6500.8 1.07 0.01
60 0.00 0.0 109.31 0.25
60 0.28 4930.0 16.31 0.04
60 0.38 30620.8 7.39 0.02

5 m PZ
T ( C ) Loading P
CO2
(Pa) P
PZ
(Pa)
PZ

40 0.00 0.0 5.12 0.05
40 0.17 28.7 3.12 0.03
40 0.22 60.5 2.88 0.03
40 0.27 211.0 2.20 0.02
40 0.34 798.0 1.03 0.01
40 0.41 5710.0 0.82 0.01
190

5
40 0.41 6990.0 0.86 0.01
60 0.00 0.0 17.20 0.06
60 0.16 137.0 10.20 0.04
60 0.23 365.0 7.45 0.03
60 0.30 1290.0 5.59 0.02
60 0.33 3310.0 4.86 0.02
60 0.39 18300.0 2.86 0.01
60 0.42 51400.0 2.23 0.01

2 m PZ
T ( C ) Loading P
CO2
(Pa) P
PZ
(Pa)
PZ

40 0.00 0.0 2.17 0.06
40 0.15 21.5 2.12 0.05
40 0.23 106.0 1.80 0.04
40 0.26 184.0 1.68 0.04
40 0.31 526.0 1.49 0.04
40 0.37 1950.0 1.38 0.04
40 0.43 10100.0 1.09 0.03
60 0.00 0.0 6.78 0.06
60 0.13 92.4 5.55 0.05
60 0.19 296.0 4.80 0.04
60 0.28 1400.0 2.93 0.02
60 0.33 3950.0 2.24 0.02
60 0.37 9910.0 1.77 0.02
60 0.41 24700.0 1.28 0.01
60 0.17 142.0 5.13 0.05
60 0.38 13700.0 1.87 0.02

Table 3: 7 m MDEA/2 m PZ Blend Volatility
T (C ) Loading P
CO2
(Pa)
P
MDEA

(Pa) P
PZ
(Pa)
MDEA

PZ

40 0.00 0.0 0.81 2.04 1.78 0.06
40 0.10 398.9 0.69 0.59 1.55 0.02
40 0.19 2580.9 0.60 0.09 1.37 0.00
60 0.00 0.0 7.37 12.37 2.90 0.12
60 0.10 2967.1 6.53 4.80 2.61 0.05
60 0.19 18639.7 6.48 2.47 2.63 0.02

Table 4: 8 m and 12 m EDA Volatility
EDA (m) T (C ) Loading P
CO2
(Pa) P
EDA
(Pa)
EDA

8 40 0.00 0.0 24.14 4.7E-02
8 40 0.41 248.1 0.93 1.8E-03
8 40 0.49 4090.2 0.46 9.2E-04
8 60 0.00 0.0 165.69 1.2E-01
191

6
8 60 0.41 2695.6 6.19 4.4E-03
8 60 0.49 25054.1 2.90 2.1E-03
12 40 0.00 0.0 57.00 7.9E-02
12 40 0.42 203.5 1.41 2.0E-03
12 40 0.50 10747.8 0.16 2.3E-04
12 60 0.00 0.0 430.31 2.1E-01
12 60 0.42 2420.2 8.64 4.4E-03
12 60 0.50 32959.6 1.20 6.2E-04

Table 5: 5 m AMP Volatility
T (C ) Loading P
CO2
(Pa) P
AMP
(Pa)
AMP

40 0.30 1090.7 9.84 0.68
40 0.55 7768.1 4.57 0.33
60 0.30 6816.2 55.53 0.86
60 0.55 30818.2 26.40 0.42

Results
Figure 2 shows MEA partial pressure for 7.0 m and 11.0 m MEA systems at 40 C and 60 C,
respectively.

0
5
10
15
20
25
30
0 0.1 0.2 0.3 0.4 0.5 0.6
Loading (mol CO
2
/mol total alkalinity)
P
M
E
A

(
P
a
)
7m MEA 40C
11m MEA_40C
7m MEA_60C

Figure 2: MEA Partial Pressure with respect to T and Loading

192

7
MEA partial pressure is greater for higher amine concentration. Evidently, 11 m MEA solution
has greater partial pressure than 7 m MEA at the same temperature. The presence of greater
amine in solution gives rise to higher partial pressure. Additionally, partial pressure increases
with temperature. At 60 C MEA partial pressure is higher than at 40 C for the 7 m solution.
With higher temperature there is more heat energy present to volatilize more amine into the
vapor phase; therefore, the partial pressure is greater.
Figure 3 displays MEA volatility for 3.5 m, 7 m, and 11 m solutions in terms of MEA apparent
activity coefficient at 40 C and 60 C.

0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.0 0.1 0.2 0.3 0.4 0.5 0.6
Loading (mol CO
2
A
p
p
a
r
e
n
t

M
E
A

A
c
t
i
v
i
t
y

C
o
e
f
f
i
c
i
e
n
t
28 ppm
40C
60C
Figure 3: 3.5 m, 7 m, 11 m MEA Volatility

The effect of MEA concentration has been normalized as volatility is now expressed as an
apparent activity coefficient instead of partial pressure. Therefore, all the data points
representing different MEA concentrations (3.5 m, 7 m, and 11 m) collapse onto individual lines
representing only the effect of temperature. Furthermore, the effect of temperature has also been
greatly reduced by presenting the results as apparent activity coefficient.

193

8
0.001
0.01
0.1
1
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40
Loading (mol CO
2
A
p
p
a
r
e
n
t

P
Z

A
c
t
i
v
i
t
y

C
o
e
f
f
i
c
i
e
n
t
60C
40C
21.4 ppm

Figure 4: 8 m PZ Volatility

The apparent PZ activity coefficient in 8 m PZ also decreases with loading (Figure 4), as free PZ
is converted by reaction with CO
2
to other nonvolatile species. The apparent PZ activity
coefficient is somewhat greater at 60 C than at 40 C, exhibiting endothermic behavior. In
subsequent plots, it can be seen that the temperature behavior for lower concentrations of PZ
(5 m and 2 m) is different from 8 m due to unique and complex speciation occurring at each
concentration. At the absorber operating condition of interest (40 C and nominal lean loading of
~0.30), 8 m PZ volatility is approximately 21.4 ppm.
With 5 m PZ (Figure 5) there is little effect of temperature on the apparent activity coefficient.

194

9
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45
Loading (mol CO
2
A
p
p
a
r
e
n
t

P
Z

A
c
t
i
v
i
t
y

C
o
e
f
f
i
c
i
e
n
t
40C
60C


With 2 m PZ (Figure 6) there is a reversed effect of temperature on the apparent activity
coefficient.

195

10
0
0.01
0.02
0.03
0.04
0.05
0.06
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45
Loading (mol CO
2
A
p
p
a
r
e
n
t

P
Z

A
c
t
i
v
i
t
y

C
o
e
f
f
i
c
i
e
n
t
40C
60C


In this case, the apparent PZ activity coefficient is greater at 40C than at 60C which indicates
an exothermic behavior. In summary, as PZ concentration is lowered from 8 m to 5 m and then
to 2 m, the solution transitions from being endothermic to exothermic as is indicated by the
temperature behavior of the apparent PZ activity coefficient in each case. This is possibly due to
different speciation effects occurring in each solution. At zero loading, however, the apparent
PZ activity coefficient is always higher for 60C than for 40C, due to the lack of speciation from
the CO
2
reactions.

The volatility of PZ in 7 m MDEA (methyldiethanolamine) / 2 m PZ is presented in Figure 7.

196

11
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0 0.05 0.1 0.15
CO
2
Loading (mol CO
2
/mol tot alk)
A
p
p
a
r
e
n
t

P
Z

A
c
t
i
v
i
t
y

C
o
e
f
f
i
c
i
e
n
t
40 C
60 C
5.9 ppm

Figure 7: PZ Volatility in 7 m MDEA/2 m PZ Blend

7 m MDEA/2 m PZ shows temperature and loading effects similar to those of MEA. The
apparent PZ activity coefficient of the blend decreases with loading as there is less free amine
present in solution in the presence of greater CO
2
at higher loading. PZ in this blend shows an
endothermic behavior with the apparent PZ activity coefficient being greater at 60 C than at
40 C. At the operating condition of interest, PZ volatility for this blend is roughly 6 ppm which
is much less than that of 7 m MEA and 8 m PZ.
Figure 8 exhibits the volatility of MDEA in the 7 m MDEA/2m PZ.

197

12
1
10
0 0.05 0.1 0.15
CO
2
Loading (mol CO
2
/mol tot alk)
A
p
p
a
r
e
n
t

M
D
E
A

A
c
t
i
v
i
t
y

C
o
e
f
f
i
c
i
e
n
t
40 C
60 C
6.9 ppm

Figure 8: MDEA Volatility in 7 m MDEA/2 m PZ Blend

Unlike PZ, MDEA in the 7 m MDEA/2 m PZ blend does not show a very strong dependence on
loading as the apparent MDEA activity coefficient remains rather constant throughout the
loading range shown. This behavior is due to CO
2
reacting more preferentially with PZ than
with MDEA in the blend, and thereby MDEA volatility stays roughly constant. Again the
temperature behavior is endothermic. The volatility of MDEA for this blend is about 6.9 ppm at
the relevant operating condition.
Figure 9 presents ethylenediamine (EDA) volatility for the 8 m EDA and 12 m EDA at 40 C and
60 C.

198

13

0.0001
0.001
0.01
0.1
1
0 0.1 0.2 0.3 0.4 0.5
CO
2
Loading (mol CO
2
/equiv EDA)
A
p
p
a
r
e
n
t

E
D
A

A
c
t
i
v
i
t
y

C
o
e
f
f
i
c
i
e
n
t
8m EDA_40C
12m EDA_40C
8m EDA 60C
14.1 ppm
12m EDA_60C

Figure 9: EDA Volatility

The apparent EDA activity coefficient is greater at higher temperature than at lower temperature
again an indication of endothermic behavior. Moreover, at the same temperature, EDA
volatility is greater at 12 m EDA than at 8 m EDA as there is more free amine present in the
former to volatilize. As loading is increased, EDA volatility decreases as has been seen with all
the other amines. At the operating condition of interest, 12 m EDA volatility is approximately
14.1 ppm.
Figure 10 displays 5 m AMP (2-amino-2-methyl-1-propanol) volatility.

199

14
0.1
1
0.3 0.35 0.4 0.45 0.5 0.55
CO
2
Loading (mol CO
2
/mol AMP)
A
p
p
a
r
e
n
t

A
M
P

A
c
t
i
v
i
t
y

C
o
e
f
f
i
c
i
e
n
t
40 C
60 C
97 ppm

Figure 10: 5 m AMP Volatility

5 m AMP displays an endothermic temperature behavior as far as the apparent amine activity
coefficient is concerned. Volatility for this solution is seen to decrease with CO
2
loading as has
always been expected. At 40 C and a nominal lean loading of 0.3, 5 m AMP volatility is as
much as 97 ppm which is remarkably the highest of all the solutions studied so far.
The following graph summarizes the volatility of all the systems in terms of amine partial
pressure as a function of CO
2
partial pressure at 40 C.

200

15
0.1
1
10
100
0 2000 4000 6000 8000 10000 12000
P
CO2
(Pa) at 40C
P
a
m
i
n
e

(
P
a
)
7m MDEA/2m PZ
8m PZ
7m MEA
5m AMP
12m EDA
Figure 11: Amine Volatility Comparison

The least volatile system observed thus far is 7 m MDEA/2m PZ. The system with the next
higher volatility is 12 m EDA; however, surprisingly enough, unloaded 12 m EDA has a greater
volatility than all of the systems. 8 m PZ is seen to have comparable volatility to the baseline
7 m MEA solvent. 5 m AMP is noted to be the most volatile system studied to date.
Conclusions
Table 6 ranks the amine systems in order of increasing volatility starting from the top. It also
tabulates estimated figures of the amine heats of solution (the amount of heat given off as the
amine comes into solution with water and CO
2
) and the heats of vaporization (the amount of heat
it takes to desorb the amine from the solution). Calculations for these quantities are done per the
Gibbs-Helmholtz relations presented in the Theory section.

201

16
Table 6. Summary of Amine Volatility

Note that these amine systems are all ranked on the same basis this being the nominal lean
loading that corresponds to a CO
2
partial pressure of approximately 500 Pa at 40C in each case.
In order of increasing volatility, 7 m MDEA / 2 m PZ is the least volatile system, followed by 12
m EDA, 8 m PZ, baseline 7 m MEA, and 5 m AMP being the most volatile. In regard to the
amine heat of solution, it appears that the less volatile systems have higher heats of solution than
those that are more volatile. Meanwhile, there is no apparent correlation between amine
volatility and the amine heat of vaporization.
Future Work
Screening of additional amine systems is to be continued. Future systems that will be considered
for screening include: 2-PE, MAPA, DGA, and HEP. Furthermore, there is plan to resume
collecting additional experimental data (unloaded) for the MDEA/PZ blend to support Aspen
Plus
modeling activity. Heat capacity and NMR speciation for the blend at different loadings
will also be studied. In the long run, an attempt will be made to construct a generalized amine
volatility model.

202
1
Degradation of Amine Solvents for Carbon Capture

by Alexander Voice
and the
2
July 17, 2009
Abstract
Oxidative degradation experiments in the high gas flow (HGF) apparatus continued this quarter.
An experiment was conducted to determine the sensitivity of the oxidative degradation rate of 7
m MEA under kinetically limited conditions in the high gas flow apparatus on various operating
conditions. Temperature was by far the most important parameter, showing a 100500% change
in ammonia production for a 10 C temperature change.
Amine screening for oxidative degradation rate was also conducted in the HGF apparatus.
Ethylenediamine (EDA, 8 m) degraded at a rate of 1 mM/hr for 0.11 mM Fe as evidenced by
NH
3
production. Addition of 1 mM Fe produced a burst of 0.15 mM of NH
3
, although the steady
state rate remained unchanged. 4.8 m 2-amino-2-methyl-1-propanol (AMP) and 17.7 m diglycol
amine (DGA
) did not degrade to produce NH

3
in significant quantities. AMP did not produce
significant quantities of heat stable salts. The presence of other degradation products, including
NO
2
and formaldehyde, had a low signal to noise ratio in the FTIR spectrum, and will require
verification. 8 m PZ did not produce NH
3
or heat stable salts in the HGF. 11 m MEA produced
ammonia at a rate of 4.1 mM/hr, compared with 1.8 mM/hr for previous 7 m experiments
conducted by Sexton.
Formic acid and 7 m MEA at 0.4 loading reacted rapidly to produce hydroxyl-ethyl-formamide
at 135 C. Formate and formamide concentrations were stable after 24 hours (the first sample
taken). The formate to formamide ratio was 1:1 in a solution which initially contained only
formic acid. In the solution loaded with formic acid and formaldehyde, the ratio was 2:1.
Formate was also detected in the solution containing only formaldehyde at a concentration of
22 mM, well above the ~1 mM formate detected in the neat solution.
The metal dissolution rate (metal concentration divided by time) in thermal degradation
experiments was found to be a clean Arrhenius function of T.
Introduction
Degradation is a serious problem for most carbon capture systems employing amine absorption
stripping technology. Solvent degradation can be divided into two main categories: oxidative
203
2
degradation, which occurs primarily in the absorber (due to the presence of dissolved oxygen),
and thermal degradation, which occurs primarily in the stripper (due to elevated temperature). In
the case of monoethanolamine (MEA), oxidative degradation produces ammonia and formate
(Sexton, 2008; Goff, 2005). Thermal degradation is not known to produce any heat stable salts.
Instead, MEA undergoes a polymerization reaction to produce larger molecules which have not
been observed in oxidative degradation experiments (Davis, 2009).
Degradation reduces the amount of solvent available for absorbing carbon dioxide, thus reducing
the capacity of the solution and the performance of the system. In addition, degradation products
must eventually be removed and disposed of. Solvent reclaiming systems can be used to remove
heat stable salts from solution; however it is preferable to prevent degradation from occurring in
the first place.
Oxidative degradation of MEA has been shown to be mass transfer controlled in situations where
high concentrations of catalysts are present (Goff, 2005). Iron, Cr, and Ni are all catalysts for
oxidative degradation, and have been observed in high concentrations in thermal degradation
experiments using stainless steel reactors (Rochelle et al., 2008). High temperature, low loading,
high agitation rate, high MEA concentration, and high oxygen concentration also increase the
degradation rate (Goff, 2005). Degradation can be limited by selecting a solvent with a low
affinity for dissolved oxygen, which limits the driving force for oxygen mass transfer. Inhibitors
which scavenge oxygen, chelate metals, or absorb free radicals can reduce the degradation rate.
Proprietary inhibitors A and B have been shown to decrease the rate of ammonia production
from MEA and EDA solutions in the presence of dissolved iron and copper. Oxidative
degradation can be different for different solvents and inhibitors; thus it is an important factor to
consider in selecting a solvent.
Oxidative Degradation Experiments
A procedure for characterizing oxidative degradation of various amine solvents in the high gas
flow apparatus was developed this quarter. The procedure and apparatus are similar to those
employed by Goff and Sexton. 350mL of loaded amine solution were placed in a 1L glass
reactor with a heated dimethylsilicone oil jacket. The jacket is set at 63 C to maintain a reactor
temperature of 55 C. Air and carbon dioxide (98%v./2%v.) are fed into the bottom of the
reactor by two Brooks mass flow controllers at a rate of approximately 0.23 mol/min (dry). The
total gas flow rate is calculated from the water content of the gas (measured by an FTIR
analyzer), the temperature of the reactor, and by assuming ideal gas behavior and total pressure
to be 1 atmosphere (1).
Gas rate = n*R*T/P/(1-f) = 6.19L/min (1)
where
n = molar dry gas rate (.23 mol/min)
R = gas constant (.08206 L*atm/mol/K)
T = Reactor temperature (328.15K)
P = ambient pressure (1 atm)
f = water fraction (.1 - .2)

204
3
The rate of species produced from the solution (including volatile losses and degradation
products) can be calculated using the gas rate and concentration reported by the FTIR.
Production rate [=] mM/hr = gas rate (gmol/min) *
concentration (mol/mol) * (1 mol/1e6 mol) / solution volume (liters)
* 1000 (mmol/mol) * 60min/hour (2)
The water balance was maintained on the reactor in one of two ways: by passing the dry gas
through a saturator bath before entering the reactor or by feeding dry gas and makeup water
directly to the reactor. Liquid samples were taken every 2448 hours. An FTIR analyzer
provided online analysis of the gas phase, allowing degradation products in the gas phase and
volatile losses to be well accounted for. Liquid phase samples were analyzed using anion and
cation chromatography using the procedure described by Sexton. Operating conditions for the
high gas flow apparatus are summarized in Table 1
Table 1: Operating Conditions for the HGF System
Operating Condition Operating Range
Agitation rate (rpm) 1000 5
Temperature (
o
C) 551
Pressure (mm Hg) 76015
Dry gas composition (%) 2 CO
2
, 98 air
Dry gas flow rate (gmol/min) 0.23 .04
Saturator temperature*(
o
C) 53-63C
Water makeup rate* (ml/min) 0.1-0.5
*Continuous water makeup is only used as a replacement for saturating the reactor gas
Thermal Degradation Experiments
A new apparatus was designed for thermally degrading large quantities of amine solutions. The
device is a stainless steel container with a sampling port, valve, and quick-connect sampling
tube. When the container is heated pressure builds inside the device. When the valve on the
sampling port is opened, pressure forces liquid into the sampling tube. The sampling tube is
cooled and then disconnected. The device was designed to facilitate degradation of relatively
large quantities (up to 400 mL) of amine at a time (Figure 1).

Figure 1: Schematic of Thermal Degradation Device
205
4
A thermal degradation experiment was designed to determine the effect of formate and
formaldehyde on the thermal degradation rate. 10 mL SS316L tubes with Swagelok endcaps
(thermal bombs) were filled with one of four different solutions. A batch of neat 7 m MEA
with 0.4mol CO
2
/mol MEA was used to prepare all four solutions (Table 2).
Table 2: Thermal Bomb Solutions Summary
Solution Contents
1 7 m MEA
0.4mol/CO
2
per mol MEA
0.1mol formic acid/mol MEA

2 7 m MEA
0.4mol/CO
2
per mol MEA
0.1mol formaldehyde acid/mol MEA

3 7 m MEA
0.4mol/CO
2
per mol MEA
0.05mol formic acid/mol MEA
0.05mol formaldehyde/mol MEA

4 7 m MEA
0.4mol/CO
2
per mol MEA

Samples were taken in roughly geometric intervals (1, 2, 4, 8days) and analyzed for the
presence of formate, formamide (by sodium hydroxide treatment), total inorganic carbon content
(by acid evolution), and dissolved iron (by flame atomic absorption). Samples will eventually be
analyzed by cation chromatography to determine amine loss and the presence of thermal
degradation products.
Analytical Methods
Analytical methods have been described extensively in previous quarterly reports (Rochelle et
al., 2008; Rochelle et al., 2009). Methods used this quarter are summarized in Table 3.
Table 3: Analytical Methods Overview
Method Analyte (s)
Total inorganic carbon (by phosphoric acid
release)
Dissolved carbon dioxide
Cation chromatography Amines (ethanolamine, ethylene diamine,
piperazine, etc.)
Thermal degradation products (MEA-
oxazolidone, MEA-urea)
Anion chromatography Organic carboxylic acids (formate, oxalate,
glycolate, acetate)
Amides (formamide, oxalamide, glycolamide,
acetamide)
Inorganic salts (nitrates, nitrites, sulfates,
206
5
chloride)
Cation chromatography mass spectroscopy Unknown cation degradation products
Titration Total alkalinity amine concentration, exact
for unloaded neat solns
High pressure liquid chromatography Hydroxyethyl imidazole, N-formyl-
ethanolamine (MEA-formamide)
Fourier transform infrared analyzer Gas phase degradation products (ammonia,
N
2
O, NO
x
)
Gas phase inerts (water, carbon dioxide)
Other gas phase components (including, not
limited to, methane, ethylene, formaldehyde,
acetaldehyde, methanol, methylamine)

Results
Degradation of 7 m MEA Kinetic Control
MEA was degraded under kinetically controlled conditions: 0.1-1 mM Fe with 50100 mM
inhibitor A. Although NH
3
production rates are not a good estimate of the absolute degradation
rate under these conditions, the data demonstrate that temperature is a major factor in oxidative
degradation of MEA.
Table 4: Ammonia Production from MEA
Agit.
(RPM)
Fe (mM) A (mM)
Dry Gas O
2

(%)
Temp.
(C)
NH
3
Rate
(mM/hr)
Change (%)
1500 0 50 17.2% 54 0.99 --
1500 .1 50 17.2% 54 1.16 17
1000 .1 50 17.2% 54 1.01 -12
1000 .1 50 20.5% 54 1.20 18
1000 .1 50 15.0% 54 1.00 -16
1000 .1 50 17.2% 64 3.04 204

500 .1 50 17.2% 64 1.66 -45
1500 1 100 17.2% 64 0.63 -62
500 1 100 17.2% 64 0.5 -20
1500 1 100 17.2% 46 -0.03 -106
207
6
1700 1 100 17.2% 54 0.04 -233
1700 1 100 17.2% 64 0.23 475

Corrosion Data
Metals concentrations found in Davis SS-316L thermal degradation bombs were evaluated
using flame AA last quarter. The results were re-interpreted this quarter to explore the metal
dissolution rate (metal concentration in g/g divided by the time in days) as a function of
temperature. For this particular data set, the dissolution rate was a clean Arrhenius function for
100150 C (Figure 2). For the 135 C data, the rate was calculated using the slope retrieved
using the LINEST function in MS Excel. The slope of ln(g/g*day) vs. 1/T was -7698, -10408,
and -7708 for Fe, Ni, and Cr, respectively.

Degradation Bombs

Figure 2: Metal Dissolution Rate as a Function of Temperature
208
7
Table 5: Present in Thermally Degraded MEA Metals Present in Thermally Degraded
MEA
Time
(Days)
Temperature (C) Fe
(ppm)
Ni
(ppm)
Cr
(ppm)
28 100 116 26 38
14 120 167 49 69
4 135 196 57 78
8 135 302 111 142
14 135 409 142 184
28 135 765 241 310
14 150 673 386 217
Oxidative Degradation Screening Experiments
The following is a summary of the results of screening studies conducted in the HGF system.
8 m EDA, 4.8 m AMP, 8 m PZ, 11 m MEA, and 17.7 m DGA were degraded oxidatively.
Liquid samples were taken every 2448 hours and analyzed using anion chromatography.
Cation chromatography will be conducted on all samples to gauge amine loss.
EDA was the only amine besides MEA to produce significant quantities of ammonia (Figure 4).

Figure 4: Oxidative Degradation of EDA in the HGF

209
8
Table 6: 8 m EDA Degradation Product Concentrations (mmols/kg solution)
Time (min) 0 207 1227* 1762** 3202 4547
7131
8687
Formate 0.36 0.67 0.77 0.72 0.76 2.03 7.40 10.49
Formamide 2.26 1.52 1.97 2.02 3.56 7.15 25.33 31.38
Oxalate 0.04 0.06 0.07 0.04 0.03 0.04 0.08 0.13
Oxamide 0.00 0.00 0.02 0.03 0.04 0.12 0.69 0.97
Nitrate 0.04 0.06 0.13 0.17 0.47 1.03 1.56 1.60
Nitrite 0.09 0.20 0.28 0.33 0.51 0.99 1.55 1.59
Volatility 0.00 0.47 2.61 3.84 6.05 13.32 22.11 27.89
EDA (%wt.) 32.05 23.61
*1 mM Fe added, **5 mM Cu added,
50 mM A added,
50 mM A added
EDA produced ammonia at a rate of 1 mM/hr for concentrations of 0.1 and 1.1 mM Fe. 5 mM
copper caused a roughly 4-fold increase in the steady state ammonia production rate. Addition
of 50 mM Inhibitor A reduced the ammonia rate to 2.2 mM/hr; 100 mM Inhibitor A reduced the
ammonia rate to 1.2 mM/hr in the presence of 1.1 mM Fe and 5 mM Cu.
MEA produced significant concentrations of ammonia (quantified in the gas phase with the
FTIR) and formate (quantified using anion chromatography), in addition to lower concentrations
of other products (Table 6). Of the heat stable salts, formate, formamide, oxalate, oxalamide,
nitrate, and nitrite all exhibited a linear trend over the course of the experiment. The ratio of
formamide to formate fluctuated between 7 and 12, with no clear trend. Glycolate and acetate
were observed in low concentrations, and appeared to remain constant over the course of the
experiment.
Table 7: 11 m MEA Degradation Product Concentrations (mmols/kg solution)
Time (hours) 0 22.82 40.03 47.30 64.97
Glycolate 0.00 3.35 3.05 3.81 2.68
Glycolamide 0.00 0.66 1.87 2.89 3.50
Acetate 0.00 2.41 2.00 1.83 2.28
Acetamide 0.00 0.00 0.00 0.00 0.00
Formate 0.28 2.34 3.60 5.41 8.06
Formamide 0.00 19.85 42.46 49.19 61.49
Oxalate 0.00 0.04 0.08 0.12 0.22
Oxalamide 0.00 1.45 4.47 6.09 10.26
Nitrate 0.00 2.24 3.92 5.25 6.59
Nitrite 0.00 0.49 0.78 1.00 1.44
210
9
NH
3
0.00 96.86 169.71 199.59 267.75
MEA 0.00 30.08 50.00 58.08 76.63
N
2
O 0.00 0.31 0.50 0.58 0.73
NO
2
0.00 3.21 5.81 6.83 9.14

Table 8: 4.8 m AMP Degradation Product Concentrations (mmols/kg solution)
Time (hrs) 0 3.1 10.5 44.5 50.8 67.9 72.4
Formate 0.12 1.11 1.00 0.56 0.89 0.75 1.13
Volatility 0.00 46.05 133.77 570.81 636.44 799.91 844.12

Finally samples from oxidatively degraded amine solutions were evaluated using HPLC. The
program uses acetonitrile-H
2
O from 2% to 20% acetonitrile. No significant peaks (besides the
amine peak) were observed in any of the solutions except MEA, which had well defined peaks at
t = 3.069, 3.315, 4.159, and 4.292 min. None of these peaks correspond with those qualified by
Sexton (n-formyl-ethanolamine t = 2.66min, hydroxyethyl-imidazole t = 5.91min). Therefore,
these peaks will be qualified by LC/MS.
PZ and DGA had some peaks in the 67min and 611min noise range. These peaks were not
well defined and will be difficult to qualify using LC/MS. Methanol was investigated as an
alternative solvent for the HPLC. The same gradient was employed using methanol instead of
water as the polar solvent in the mobile phase. This program did not achieve better separation of
peaks in the MEA sample.
Thermal Degradation Experiments
Formate and total formamide concentrations were determined in solutions of 7 m MEA at 0.4
loading after thermal degradation for t = 1, 2, 3, and 6 days at 135 C. Solutions 1 and 3 were
spiked with 0.1 and 0.05 mols of formic acid per mol of MEA, respectively. Solution 2
contained 0.1mol of formaldehyde per mol MEA, and solution 4 contained only 7 m MEA.
Solutions 1 and 3 quickly equilibrated converting some formate to formamide in less than 24
hours. Solution 2, which initially contained no formate, had converted some formaldehyde to
formate and formamide in less than 24 hours. Formate and formamide were observed in trace
concentrations in the neat solution (Figures 5 and 6)

211
10

Figure 5: Formate concentrations in formic acid spiked MEA solutions (1 and 3), 7m MEA
at 0.4 loading at 135 C

Figure 6: Formate concentrations in MEA solutions (2 and 4), 7 m MEA at 0.4 loading at
135 C

Table 9: Formate (mmol/kg solution) in MEA Solutions
Time
(Days)
Solution 1 Solution 2 Solution 3 Solution 4
0 559 3 330 1
1 362 22 189 1
212
11
2 377 22 182 1
3 331 22 173 2
6 359 24 200 --

Table 12: Formamide (mmol/kg solution) in MEA Solutions
Time
(Days)
Solution 1 Solution 2 Solution 3 Solution 4
0 41 11 0 5
1 268 15 98 0
2 265 13 117 1
3 222 14 100 4
6 314 13 76
Total inorganic carbon content was determined for the first four samples. For the neat solution,
4.40mol base/kg solution corresponds to 7.07 m MEA.
Table 11: Total Inorganic Carbon in Thermal Bombs (mol CO
2
/kg solution)
Soln t=1 days t=2 days t=3 days t=6 days
1 1.657 1.539 1.681 1.630
2 1.683 1.821 1.765 1.690
3 2.022 1.729 1.716 1.683
4 1.785 1.829 1.811 --
Table 12: Total Alkalinity of 7 m MEA with Additives
Solution Total Alkalinity (mol
base / kg solution)
Additive
1 3.78 Formic acid
2 4.26 Formaldehyde
3 3.93 Formic acid and formaldehyde
4 4.40 None

Discussion
The high gas flow apparatus provides a valuable tool for evaluating the volatility and oxidative
degradation products of amines. The FTIR analyzer provides information about the gas phase
exiting the reactor that cannot otherwise be easily obtained. The device can be calibrated to
detect any gas phase compound. However, if unknown components are in the gas phase, they
can interfere with the analysis. Once the existence of a particular compound has been verified,
213
12
the FTIR is an excellent tool for quantification. Volatile losses from the HGF can be reliably
accounted for because the instrument can be calibrated to detect the amine, and because the gas
phase contains significant quantities of amine. Furthermore, analysis of FTIR spectra using
various settings (i.e. having the machine simultaneously analyze for different components) has
shown the amine concentration to be relatively insensitive to the analysis settings. In summary,
the FTIR can quantify amine volatility and gaseous degradation products, but the former
measurement is more reliable because the component is known and exists in significant
quantities.
Cation chromatography is the most reliable way to interpret oxidative degradation, because it
relies on amine loss rather than the identification and quantification of unknown degradation
products. However, significant degradation must occur for this method to yield useful results for
several reasons. The error in the cation analysis may be 100 mM, which corresponds to roughly
2% for 7 m MEA at 0.4 loading, therefore amine losses should be at least 5% to be significant.
In addition, volatile losses, which may exceed degradative losses, must be accurately quantified.
Volatile losses are calculated from the concentrations of amine given by the FTIR analyzer, the
gas rate, and the solution volume. The amine concentration and gas rate can be precisely
determined, but the solution volume may fluctuate if water is evaporated or condensed during the
experiment. Changes in the water balance affect both the total solution volume and the amine
concentration. Cation chromatography can therefore only be used to quantify oxidative
degradation losses in experiments where significant amounts of degradation occur and where the
water balance has been well maintained.
Conclusions
Temperature is the most important factor in oxidative degradation experiments where the
reaction is kinetically limited. It outweighs catalyst, reagent, and inhibitor concentration, as well
as agitation rate and loading. In 7 m MEA with 0.1 Fe
++
/50 mM A, the NH
3
production rate
increased from 1 to 3 mM/hr as T increased from 54 to 64
o
C.

EDA degrades to produce ammonia with 0.1mM Fe. High Fe concentrations do not increase the
degradation rate. 5 mM Cu produces roughly 4x the degradation rate of 0.1-1mM Fe. 11 m
MEA produces ammonia at a 2x higher rate than 7 m MEA. DGA, AMP, and PZ do not produce
significant quantities of ammonia or heat stable salts.
Formate reacts with MEA to form n-formyl-ethanolamine in less than 24 hours. Formaldehyde
also reacts with MEA to form formate and formaldehyde in less than 24 hours.
The metal dissolution rate in thermal experiments can be fitted with an Arrhenius function. The
slope of the Arrhenius function was similar for Fe and Cr.
Future Work
LC/MS is a top priority to qualify unknowns in oxidatively degraded MEA solutions. LC/MS
will offer an opportunity to identify additional unknowns in MEA degradation and potentially
close the gap in the mass balance.
In the thermal degradation experiments, formate was observed to react with MEA to form
formamide faster than anticipated. A second experiment will be conducted to evaluate
formamide formation over a shorter time period (<24hrs). In addition, the formation of formate
in the MEA/formaldehyde solution was unexpected. This result will be explored further.
214
13
Screening of additional amine solvents will continue. Cation chromatography will be the main
tool employed in this study. Amines studied will include MDEA/PZ blend, 2-piperadine ethanol
(2-PE), 2-methyl piperazine, 2,5-methyl piperazine, hydroxy-ethyl-piperazine (HEP), glycine,
and methyl-amino-propyl amine (MAPA).

References
Davis J. Thermal Degradation of Amines Used in Carbon Dioxide Removal Applications. The
Goff G. Oxidative Degradation of Aqueous Monoethanolamine in CO
2
Capture Processes: Iron
and Copper Catalysis, Inhibition, and O
2
Mass Transfer. Ph.D. Dissertation. 2005.
2
2
Capture by Aqueous Absorption, First Quarterly Progress Report 2009.
2
Dissertation. 2008.

215
1
Model Validation and Dynamic Simulation of Absorption
System in Aspen Custom Modeler

by Sepideh Ziaii Fashami
and the
2
July 9, 2009

Abstract
The dynamic model of the absorber created in ACM has been run at the design and
operating conditions of a pilot plant run with 9 m MEA. The steady state results give
58.8 % CO
2
removal, which is 1.7% less than a reconciled value (59.9%). The
temperature profile is not completely matched with experimental data because of
inaccurate calculation of heat of absorption for 9 m MEA.
In addition, the dynamic model of the absorber has been run for two stripper ratio control
strategies in flexible CO
2
capture. The CO
2
removal, rich loading and temperature
profiles were calculated for each strategy and overall dynamic and steady state behavior
were compared.
Pilot Plant Model Validation
The dynamic model of the absorber in ACM
was run at the design basis of a pilot plant

run with 9 m MEA at the University of Texas at Austin. In this simulation, the inlet
conditions of the lean solvent and flue gas are set at the reconciled values by Plaza et al.
(2009). (For details of the absorber model, see Rochelle et al., 2009)
The absorber packing (Flexipac AQ Style 20) was modeled using Flexipac 1Y with 29
mixed flow model. The segments were spaced with a Gaussian distribution pattern with
L
max
/L
min
=10, which was shown by Rochelle et al. (2009). The interfacial area was
calculated using a new correlation developed by Tsai et al. (2009).
Reconciled inlet conditions by Plaza et al. (2009) for simulation of pilot plant run along
with the resulting values from this model are presented in Table 1.

216
2
Table1: Pilot Plant Model Validation, 9 m MEA, 6.1 m packing,
0.43m diameter
Variable
Reconciled value
by Plaza (2009)
Inlet and resulting
values
Inlet gas (mol/hr)

33346 33346
Y
CO2
-In 0.1192 0.1192
Y
CO2
-Out 0.0501 0.044
Y
H2O
-In 0.022 0.022
T
G
-In(C) 25.1 25.1
T
G
-Out(C) 46.1 55.18
T
L
-In(C) 38.2 38.2
T
L
-Out(C) 46.7 43.03
Water-Lean(mol/hr) 143700 143700
CO
2
-Lean(mol/hr) 8307 8307
CO
2
removal (%) 59.9 58.9
Rich loading (mol CO
2
/mol
MEA)
0.469 0.457
As indicated in Table 1, resulting values give 58.9% CO
2
removal, which is 1.7% less
than the reconciled value. The outlet temperature values and profiles (Figure 1) do not
match the data, which may reflect an inaccurate heat of absorption for 9 m MEA.

Figure1: Temperature profiles, () reconciled values, (-) and
(..) resulting values from the model

25
30
35
40
45
50
55
60
65
70
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Z/Ztotal
T
(
C
) T
vapor
T
liquid
217
3
In this model, the heat of absorption is calculated from Clausius-Clapeyron equation
using a P
eq
CO2
regression from Hilliards Aspen Plus
model points (2008). The equation

is a linear function of MEA concentration and is regressed around 7 m MEA,
consequently it gives an independent equation for H
abs
from MEA concentration.
Therefore, the model could not predict an accurate H
abs
for a broad range of solvent. In
order to get better results, the calculated H
abs
is multiplied by a correction factor, which
is [MEA] (m) divided by 7 m.
Dynamic Analysis
As discussed in previous quarterly reports, flexible capture is a system that is
dynamically operated at variable load based on electricity market conditions. An
important part of this discussion is to find a control strategy that minimizes the energy
consumption and maximizes the profits. Some dynamic strategies have been proposed
and analyzed on the stripper in the absence of an absorber model in our previous reports.
In this report, two strategies were simulated with the current dynamic model of the
absorber with the same design and inlet conditions presented in the pilot plant model
validation section. This report assumes that stripper is operated with ratio control: when
the reboiler duty decreases, only a portion of rich solvent (proportional to the stripper
vapor rate) is regenerated and the remaining portion is by-passed from the stripper or
stored. Using this stripper strategy, there are three options for running the absorber.
1. Run the absorber with constant lean solvent rate and variable lean loading by
returning to the stripper by-passed rich solution and regenerated solvent to the
absorber (full flue gas flow rate).
2. Run the absorber with variable lean solvent rate and constant lean loading by
returning the regenerated portion of solvent to the absorber (full flue flow gas
rate).
3. Run the absorber with constant lean loading and reduced solvent and flue gas rate,
which is option 2 with partially by-passed flue gas from the absorber.
In options 2 and 3 we kept the lean loading constant because we have shown that it
remains almost unchanged if we use ratio-control strategy for the stripper when the
reboiler heat duty is reduced (Ziaii et al., 2009). In this report, we looked at steady state
results and dynamic behavior of the absorber for options 2 and 3.
Option 2: 50% step decrease in solvent flow rate
To simulate option 2 with 50% reduction in reboiler heat duty, a negative 50% step
change was made to the lean solution rate. As shown in Figure 2, CO
2
removal goes
down, rich loading goes up as liquid rate decreases, and both reach steady state in 56
minutes. A delay is seen for the rich loading at the initial point, which represents the
time needed for the liquid to get to the bottom and changes to be sensed.
In Figure 3, the liquid temperature profile is shown at 0, 1.8 min, and 8 min. As also
shown in this figure, the temperature at the top of the column initially increases and then
decreases while from the middle to the bottom, temperature goes down monotonically.
The initial temperature rise can be interpreted with the faster response of the liquid hold-
up relative to the absorption of CO
2
, especially at the top of the column. Furthermore,
218
4
the upward movement of the temperature bulge with the decreasing liquid rate is clear in
this figure.

Figure 2: CO
2
removal and rich loading responses to a -50% step change in lean
solution rate

Figure 3: Liquid temperature profile at 0, 1.8, and 8 min in response to -50% step
change in lean solution

Option 3: 50% ramp decrease in both solvent rate and flue gas rate
To simulate option 3 with 50% reduction in reboiler heat duty, a -50% ramp change in 3
min was made to both lean solution rate and inlet gas rate with fixed inlet compositions.
A 50% step change is not a reasonable change and causes the absorber to show very
sharp changes versus time. In order to get smooth dynamic results, we have chosen the
ramp change.
Similar to option 2, for option 3 rich loading and CO
2
removal responses are plotted in
Figure 4. CO
2
removal takes less than 30 seconds to reach the steady state after 3 mins
ramping. However, rich loading takes 56 min to get to the steady state. This means that
the CO
2
hold-up time should be much smaller in the vapor phase than the liquid phase.
0.45
0.46
0.47
0.48
0.49
0.5
0.51
0.52
0.53
0 0.6 1.2 1.8 2.4 3 3.6 4.2 4.8 5.4 6 6.6 7.2 7.8 8.4 9
Time (min)
R
i
c
h

l
o
a
d
i
n
g
0.46
0.48
0.5
0.52
0.54
0.56
0.58
0.6
C
O
2

R
e
m
o
v
a
l
CO
2
Rich
loading
30
35
40
45
50
55
60
65
70
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Z/Ztot
L
i
q
u
i
d

T
e
m
p
e
r
a
t
u
r
e

(
C
)
t= 0 (Initial state)
t= 1.8
t= 8 min (Final state)
219
5
Temperature increases at first along the column and then decreases as shown in Figure 5.
When gas and liquid rates are changed at the same time, no temperature bulge movement
is observed.

Figure 4: CO
2
removal and rich loading responses to -50% ramp change in lean
solution rate and flue gas rate

40
45
50
55
60
65
70
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Z/Ztotal
L
i
q
u
i
d

T
e
m
p
e
r
a
t
u
r
e

(
C
)

Figure 5: Liquid temperature profile at 0, 1.8, and 8 min in response to -50% ramp
change in lean solution rate and flue gas rate
Steady State comparison of options 2 and 3
Table 2 summarizes initial and final steady state values for options 1 and 2. Comparing
the steady state values for the options, bypassing the gas (option 3) results in much lower
final overall CO
2
removal and lower rich loading compared to option 2. The lower the
lean loading, the more specific energy required for solvent regeneration. Therefore
option 2 may result in a lower regeneration energy consumption and higher compression
energy due to more CO
2
being compressed, provided stripper operating conditions do not
change significantly.

0.45
0.455
0.46
0.465
0.47
0.475
0 1.2 2.4 3.6 4.8 6 7.2 8.4 9.6 10.8
Time (min)
R
i
c
h

l
o
a
d
i
n
g
0.58
0.6
0.62
0.64
0.66
0.68
C
O
2

R
e
m
o
v
a
l
Rich loading
CO
2
Removal
t= 0 (Initial state)
t= 3 min
t= 8 min (Final state)
220
6
Table 2: Steady state results of dynamic options 2 and 3
Variable Initial value
Final value
Option 2
Final value
Option 3
Rich loading 0.457 0.495 0.471
CO
2
removal 0.589 0.513 0.335
T
G
-Out (C) 55.18 56.25 57.22
T
L
-Out (C) 43.03 31.88 43.55

Dynamic responses indicate that both rich loading and CO
2
removal reach steady state in
about 5 mins with 98% approach when the lean solvent rate is decreased (option 2).
Soon after both gas and liquid rate are ramped in option 3, it takes about 3.8 mins for the
removal and 4.8 mins for rich loading to reach steady state with 98% approach.
Conclusions and Future Work
The ACM
time-variant model for the absorber was validated with steady state
reconciled pilot plant data with 9 m MEA. A 1.7% difference between reconciled
measured, and calculated values was achieved for CO
2
removal. Temperature profiles do
not match the reconciled data well because of the lack of an accurate model to calculate
H
abs
for 9 m MEA.
Two dynamic strategies were run with the current absorber model to simulate varying
steam consumption. The first option, where the absorber works at the full load of flue
gas and partial solvent rate, may give lower specific energy consumption compared to the
second option where the absorber operates at partial flue gas load. However, option 2
needs a bigger and more expensive heat exchanger and shows slower response time when
the system changes from full load to the partial load and vice versa. Comparing the
response time of the above dynamic options, option 2 indicates about 1 min faster
response for the removal (3.8 mins). Rich loading gets to the steady state in about 5 mins
in both strategies.
Next quarter, I plan to provide realistic multi-stage compressor and steam pressure
characteristic curves and incorporate them into the while integrated system. I plan to use
optimization tools in ACM
to do multivariable optimization to optimize the lean

loading, stripper pressure, and other variables.
References
Hilliard MD. 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.
Plaza J, Van Wagener D, Rochelle GT. "Modeling CO
2
Capture with Aqueous
Monoethanolamine". In 9th International Conference on Greenhouse Gas Control
Technologies. Washington D.C., Elsevier. 2008.
2
Capture by Aqueous Absorption, First Quarterly
Progress Report 2009". Luminant Carbon Management Program. The University
of Texas at Austin. 2009.
221
7
Tsai R, Seibert F, Eldridge B, Rochelle GT. "Influence of Viscosity and Surface Tension
on the Effective Mass Transfer Area of Structured Packing". 9th International
Conference on Greenhouse Gas Control Technologies. Elsevier: Washington D.C.,
2008.
Ziaii S, Cohen S, Rochelle GT, Edgar TF, Webber ME. "Dynamic operation of amine
scrubbing in response to electricity demand and pricing". 9th International
Conference on Greenhouse Gas Control Technologies. Washington D.C., Elsevier.
2008.

222
1
Electric Grid-Level Implications
of Flexible CO
2
Capture Operation

by Stuart Cohen
Supported by EPA STAR Fellowship
July 13, 2009
Abstract
A flexible carbon dioxide (CO
2
) capture system with large-scale solvent storage allows
continuous high CO
2
removal from flue gas while power output is increased by turning stripping
and CO
2
compression systems to partial- or zero-load. In such a system, the basic tradeoff is that
longer solvent storage times provide more opportunity to increase plant output when electricity
prices are high, but at the expense of additional energy cost to regenerate stored solvent when
electricity prices are low as well as additional capital costs of solvent inventory, storage tanks,
and larger stripping, compression, and auxiliary equipment.
For storage times of 112 hours using monoethanolamine (MEA) solvent, 10s million kg MEA,
10s100s m
3
storage capacity, and tens to hundreds of millions of dollars in incremental capital
costs are required. Storage tanks are a relatively insignificant capital cost relative to that of
solvent inventory and larger stripping/compression equipment. A preliminary optimization study
finds that solvent storage of a few hours or more is attractive when the electricity market
experiences large differences between high and low electricity prices, even if only in the form of
infrequent price spikes. Lower capital costs can improve the economics of a solvent storage
system, but capital cost reductions are insufficient to justify installation of a solvent storage
system without a favorable electricity price distribution. However, 2008 annual average
electricity prices in the Electric Reliability Council of Texas (ERCOT) along with base case
parameters and the preliminary modeling methodology do not provide the conditions required for
solvent storage to be desirable.
Background and Introduction to Flexible CO
2
Capture
Flexible operation of a post-combustion amine absorption and stripping system could allow plant
operators to recover some or all of the energy required for carbon dioxide (CO
2
) capture to
increase power output under appropriate electricity market conditions. Flexible operation would
consist of redirecting some or all of the steam being used for solvent regeneration back to the
power cycle, thereby also reducing the need for energy to compress CO
2
.
In this manner, operating the energy intensive components of the amine absorption/stripping
process at partial- or zero-load could allow the plant operator to choose the CO
2
capture
223
2
operating point that provides the most economical combination of cost and output for the current
electricity market conditions. For instance, it may be profitable to operate CO
2
capture at partial-
or zero-load when electricity prices are high if additional electricity sales are greater than any
increase in CO
2
emissions costs under a CO
2
regulatory framework (Ziaii, Cohen et al., 2008).
In addition, operating CO
2
capture at zero-load during annual peak electricity demand can
eliminate the need to spend billions of dollars to replace generation capacity lost when CO
2

capture operates at full-load (Cohen, Rochelle et al., 2008). If a flexible CO
2
capture system was
able to respond fast enough that the energy typically used for CO
2
capture is considered available
to the electric grid as reserve capacity, CO
2
capture systems may not even have to turn to zero-
load during annual peak demand time periods unless the grid experienced an unplanned power
plant outage or other reliability event.
There are two basic concepts for flexible CO
2
capture with an amine absorption/stripping system.
My previous work has almost exclusively analyzed the concept shown in Figure 1, where the
steam and rich solvent flow rates to the stripper are reduced equally and simultaneously during
partial- or zero-load. At partial-load, rich solvent that is not sent to the stripper is recycled to the
absorber, so CO
2
removal in the absorber will decrease as solvent becomes saturated with CO
2
.
Zero-load could involve recirculating all solvent through the absorber, or the CO
2
capture system
could be bypassed completely. The incremental capital cost of flexibility is relatively low in this
configuration and would primarily consist of costs for additional piping and possibly valves and
control devices. This designs primary disadvantage is increased CO
2
emissions at partial- or
zero-load, which would incur additional CO
2
costs under a CO
2
regulatory regime.
S
t
r
i
p
p
e
r
Rich
Solvent
Steam Flow of
100% to 0%
A
b
s
o
r
b
e
r
Steam
Control Valve
Lean
Solvent
100% to 0%
CO
2
Flow
Flue Gas With
90% to 0%
CO
2
Removal
CO
2
for
Transport
& Storage
Bypass
Stream
0% to
100%
Flow
Flue
Gas In
CO
2
Compressor
Heat
Ex.

Figure 1: Simultaneously reducing steam and rich solvent flow to the stripper allows
increased output but at the expense of additional CO
2
emissions.
Figure 2 describes another concept for flexible CO
2
capture. At high electricity prices, stripping
and compression systems turn to partial- or zero-load while the absorber remains at full-load
(Chalmers, Chen et al., 2006). The absorber is fed lean solvent from a lean solvent storage
system and sends rich solvent to a rich solvent storage system in what may be termed storage
mode. When electricity prices are low, stored rich solvent is then regenerated and the product
CO
2
is compressed in regeneration mode. In regeneration mode, stripping and compression
systems, including appropriate pumps and heat exchange devices, must treat both the current
process stream as well as stored solvent, so these systems would be larger and total energy
requirements would be greater than in a baseline system built for inflexible CO
2
capture or
224
3
flexible CO
2
capture with CO
2
venting. This configuration is environmentally attractive because
it maintains low CO
2
emissions; however, total energy requirements for CO
2
capture may be
greater in regeneration mode, and there are significant additional capital costs for solvent
inventory, solvent storage tanks, and larger stripping, compression, and auxiliary equipment.
A
b
s
o
r
b
e
r
L
a
r
g
e
r
S
t
r
i
p
p
e
r
Flue Gas
With 90%
CO
2
Removal
Flue
Gas In
Rich
Solvent
Lean
Solvent
CO
2
for
Transport
& Storage
Variable CO
2
0%-100%
Steam Flow
Optional
Lean
Solvent
Storage
Optional
Rich
Solvent
Storage
Steam
Control Valve
Larger CO
2
Compressor
Larger
Heat
Ex.

Figure 2: Including large-scale solvent storage incurs significant capital costs but allows
continued high CO
2
removal at partial- and zero-load.
Optimizing Solvent Storage Time
Problem Description
This report focuses on recent work involving the solvent storage configuration in Figure 2 where
the analysis seeks to find the optimal amount of time that a plant should operate in storage
mode (storage time) by finding the most profitable combination of time spent in storage mode
and regeneration mode. Previous calculations, some of which are presented in the Luminant
Program 2
nd
Quarterly Report in 2008, find that solvent inventory costs for monoethanolamine
(MEA) are on the order of tens of millions to hundreds of millions of dollars (tens of millions of
kgMEA at $2/kgMEA), and tank volumes are on the order of tens to hundreds of cubic meters

(5-50 million gallons) for storage times of 1-12 hours (Rochelle, 2008). A baseline amine
absorption/stripping CO
2
capture system has capital costs on the order of several hundreds of
millions of dollars or more depending on system size, so solvent storage is assumed to be cost
prohibitive for time periods longer than one day (USNETL, 2007).
Figure 3 describes the general optimization problem by plotting the average diurnal variation in
electricity prices in the Electric Reliability Council of Texas (ERCOT) in 2008 along with
several hypothetical output curves for a 500MW plant with 0, 2, 6, or 10 hours of solvent
storage. Zero storage hours represents an inflexible CO
2
capture facility. As solvent storage
time increases, more time can be spent at maximum output during high electricity price periods,
but because of additional stripping and compression requirements in regeneration mode, the
output in regeneration mode decreases with solvent storage time. Thus, there is a basic tradeoff:
as solvent storage time increases, more profits are earned during high electricity prices, but
capital costs increase along with the energy costs of CO
2
capture and compression in
regeneration mode.
225
4
200
300
400
500
600
700
800
-20
0
20
40
60
80
100
120
140
0:00 4:00 8:00 12:00 16:00 20:00 0:00
P
l
a
n
t

O
u
t
p
u
t

(
M
W
)
E
l
e
c
t
r
i
c
i
t
y

P
r
i
c
e

(
$
/
M
W
h
)
Time (24 hour clock)
Annual Avg.
Prices
Output Curves (hrs Storage Shown)
2 10 6
0

Figure 3: The annual average diurnal electricity price variation in ERCOT is shown with
hypothetical output curves for different amounts of solvent storage to illustrate the tradeoff
between storage mode and regeneration mode (ERCOT, 2008).
Modeling Methodology and Assumptions
A first-order MATLAB model is created to find the number of storage mode hours in a 24-hour
period that maximize plant profits earned by selling electricity in the energy market. Ancillary
service markets are not considered in this analysis but could be a subject for future work because
a system that could transition quickly between storage and regeneration modes could be in a
good position to profit from short-lived high prices in ancillary service markets (Chalmers, Chen
et al., 2006). In contrast to previous work that used a first-order dispatch model of the ERCOT
electric grid, this work is a single plant analysis that uses average diurnal electricity price
variations to represent electric grid dynamics.
Since this study does not include a detailed electricity market model, output at the power plant
has no influence on electricity prices. The balancing energy market in ERCOT clears every 15
minutes, meaning that electricity prices make discrete jumps at 15-minute intervals, so plant
profits are calculated at 15-minute intervals. Only a single price curve is used, so seasonal
variations in electricity price are not accounted for.
In the initial formulation of this analysis, storage mode is assumed to turn all necessary stripping,
compression, and auxiliary systems to zero-load, and the only available operating points are
zero-load storage mode and full-load regeneration mode. Intermediate operating points, as well
as partial-load of the power cycle itself, may be incorporated later. Though some auxiliary
components of the CO
2
capture system remain operational to continue CO
2
absorption when
stripping and compression is at zero-load, initial analysis will ignore any residual energy penalty,
and plant output and efficiency in storage mode is assumed equal to that of the base plant
without a CO
2
capture system.
In actual operation, there will be a transition time between storage and regeneration modes that
depends on the response time of various system components. However, this initial analysis
ignores transition performance and allows the plant to choose operating mode at each time
interval without regard to system response time. Since profits are always greater in storage
226
5
mode, the model will find the highest price intervals to operate in storage mode and operate in
regeneration mode at all other time intervals, and all available solvent storage capacity is used in
the day. Future problem development may consider transition performance.
Profit at each time interval accounts for electricity sales, marginal costs, and capital costs scaled
by the capital charge factor. Marginal costs account for fuel costs as well as any CO
2
costs that
would be incurred under a CO
2
regulatory policy. In storage mode, output and efficiency are the
same as a base plant without CO
2
capture, and the CO
2
emissions rate is reduced by an amount
specified by the CO
2
removal rate. In regeneration mode, the energy penalty for CO
2
capture
scales with storage time by the factor (24/(24-x)), where x is the number of hours in storage
mode. For instance, if storage time is 12 hours, there is twice as much CO
2
to be stripped and
compressed throughout all hours in regeneration mode. In regeneration mode, overall plant
efficiency and output decrease, and the CO
2
emissions rate increases slightly as a result. CO
2

transport and storage costs and other operating and maintenance costs are not included in the
current analysis since they are relatively small and would vary little if at all on a cost per unit
electricity basis.
Capital charges at each time interval are the same for both modes. The capital cost of solvent
inventory increases linearly with total daily solvent storage time and is calculated from the
design difference between rich and lean solvent loading, solvent price, plant output and
emissions rate, and the desired CO
2
removal rate. From the quantity of solvent, the volume and
cost of solvent storage tanks is determined from a plant design handbook that describes the cost
of large field-erected storage tanks, keeping in mind that there must be storage facilities on both
the lean and rich side of the absorber. For all other components whose capital costs increase
with storage time, the base capital cost without any solvent storage is assumed to increase with
storage time by the factor (24/(24-x)) adjusted for economies of scale. Initially, the base capital
cost of all of these components is taken together and scaled as one value, but future development
of this problem may include a more detailed representation of the capital costs of each relevant
system component. All capital costs are continuous functions of solvent storage time, so the
current analytical model does not incorporate discrete changes in component quantity as solvent
storage time increases.
An upper bound on the number of hours spent in storage mode is set by the constraint that the
energy requirement for CO
2
capture in regeneration mode must not exceed the total power plant
capacity. In practice, the upper bound on storage time will likely be set by a specific power plant
component specification such as the minimum load on the low pressure turbine, but this level of
detail has not yet been included.
Formulas have not been included in this report in order to keep document length to a minimum,
but I would be happy to furnish them upon request.
Base Case Model Input Parameters
The base case electricity prices used in this study were the 2008 annual average ERCOT
electricity prices in the balancing energy market
1
, shown previously in Figure 3.
The base plant output is chosen arbitrarily at 500 MW. Its heat rate of 10.8 MMBTU/MWh
2
and
CO
2
emissions rate of 1.03 tCO
2
/MWh
3
are weighted average values for all coal-fired plants in

1
These prices can be found on the ERCOT website labeled MCPEL for Market Clearing Prices for Energy for Load.
2
Units are million British thermal unts (MMBTU) per megawatt-hour (MWh).
227
6
the ERCOT grid from the Environmental Protection Agencys (EPA) most recent eGRID
database (USEPA, 2007).
Table 1 contains a summary of major parameters used to calculate capital charges. The capital
charge factor is chosen from the National Energy Technology Lab (NETL) CCS Systems
Analysis Guidelines. A CO
2
capture plant analysis performed by Trimeric Corp. is used to
determine the base capital costs for stripping and compression equipment that scale with solvent
storage time, including requisite pumps and heat exchange equipment. The economy of scale
factor is chosen as a relatively high 0.85 because increasing the system size may require a larger
number of components rather than simply increasing the size of components, so the average
economies of scale may not be very substantial. Storage tank capital costs are calculated using
an equation determined from the plot of purchased cost of large field-erected tanks in Plant
Design and Economics for Chemical Engineers by Peters, Timmerhaus, and West and are scaled
to 2008 dollars using the Chemical Plant Index.
Table 1: These parameters are used to determine capital charges (Peters, Timmerhaus et
al., 2002; NETL, 2005; Fisher, 2007; CES, 2009).
Description (units) Value
capital charge factor (%/yr) 17.5
base capital cost of stripping/compression equipment (millions)
4
$192
economy of scale factor for stripping/compression equipment 0.85

The coal price used in this study is the average coal price in Texass electricity power sector in
2008 as reported by the Energy Information Administration, MEA price is taken from Chemical
Industry News and Intelligence, and the CO
2
price is somewhat arbitrarily chosen at a price
commonly thought to be near that which is required for economic viability of CO
2
capture and
sequestration.
Table 2: These prices are assumed in the base case (Rao and Rubni, 2002; ICISpricing,
2008; USEIA, 2009).
fuel price ($/MMBTU) 1.53
2 2
CO Price ($/tCO ) 50
MEA price ($/kgMEA) 1.98

CO
2
removal is set to 90%, a commonly cited value since much greater CO
2
removal can
become prohibitively expensive (Rubin, 2007). MEA mass fraction and capacity to absorb CO
2

(the difference between rich and lean loading) is found in Oyenekan Ph.D. dissertation in the
University of Texas at Austin Chemical Engineering Department, and the energy requirement for
CO
2
capture is taken from a recent dynamic analysis of MEA absorption and stripping done by
Ziaii.

3
All CO
2
quantities are shown in metric tons.
4
All monetary values are in 2008 U.S. dollars.
228
7
Table 3: These parameters are used to define CO
2
capture performance (Oyenekan, 2006;
Ziaii, Cohen et al., 2008).
2
fraction of CO removed from flue gas 0.9
mass fraction of solvent in solution 0.3
2
solvent capacity - difference between rich and lean loading (molCO /molMEA) 0.12
2 2
energy requirement for CO capture (MWh/tCO ) 0.269
Solution and Results
Base Case Results: The maximum storage time based on using the plants entire output for
CO
2
stripping and compression in regeneration mode is 18.0 hours, so data are not displayed for
longer storage times.
Figure 4 plots the marginal costs of electricity production in storage mode and regeneration
mode as a function of solvent storage time. Also plotted is the daily average cost of electricity
production for each solvent storage time when the output and amount of time spent in each mode
is taken into account. Since storage mode assumes output and costs to return to base plant levels
without CO
2
capture, marginal cost in storage mode is not a function of solvent storage time. In
regeneration mode, increasing solvent storage time requires the plant to process more stored
solvent in a shorter amount of time, so marginal costs can increase to very high levels. However,
when the output and amount of time spent in each mode is taken into account and costs of
electricity production are averaged over a 24-hour period, the daily average cost is not a function
of storage time and is equivalent to the marginal cost of an inflexible CO
2
capture system.

0
40
80
120
160
200
0 4 8 12 16
C
o
s
t
s

o
f

E
l
e
c
t
r
i
c
i
t
y

P
r
o
d
u
c
t
i
o
n

(
$
/
M
W
h
)
Storage Time (hours)
MC in
Regeneration
Mode
MC in
Storage
Mode
Daily Average =
MC of Inflexible
System

Figure 4: Marginal costs in regeneration mode increase with solvent storage time, but the
daily average cost of electricity production remains constant at the marginal costs of an
inflexible CO
2
capture facility.
The incremental capital costs of solvent storage are plotted in Figure 5 with regions to represent
each of the three major capital cost components. Capital costs increase with storage time largely
due to costs of stripping and compression equipment and solvent inventory. The cost of storage
229
8
tanks is relatively small. Costs of stripping and compression equipment do not increase linearly
because an economy of scale factor is used in calculations. As a point of reference, a full CO
2

capture system without solvent storage for a 500 MW plant is expected to cost around $470
million (USNETL, 2007).
It is also important to note that the MEA inventories required for several hours of solvent storage
are on the order of tens of millions of kilograms, so a serious practical concern with large-scale
solvent storage is the chemical manufacturing capacity available to supply solvent inventory.
The stoichiometry of MEA production requires production of diethanolamine (DEA) and
triethanolamine (TEA) in appreciable quantities, so unless there is sufficient demand for all three
chemicals, MEA manufacturers may have to charge extra for MEA to account for production of
chemicals with lower market demand. In general, physical and economic concerns in solvent
manufacturing must also be addressed if flexible CO
2
capture with solvent storage is to become a
practical system.
0
100
200
300
400
500
600
700
0 4 8 12 16
I
n
c
r
e
m
e
n
t
a
l

C
a
p
i
t
a
l

C
o
s
t

o
f

S
o
l
v
e
n
t

S
t
o
r
a
g
e

(
m
i
l
l
i
o
n
s
)
Additional Equipment for
Stripping/Compression
Storage
Tanks
Solvent Inventory

Figure 5: Solvent inventory and stripping and compression equipment constitute the
primary incremental capital costs of flexible CO
2
capture with solvent storage.
When all cost components are accounted for along with electricity prices and output, the daily
profit vs. solvent storage time can be calculated (Figure 6). Since maximum profits are desired,
the base case optimal storage time is actually 0 hours with daily profits of about $237,000; a
solvent storage system does not improve profits with base case parameters and this initial
optimization formulation.
230
9
$0
$50
$100
$150
$200
$250
0 4 8 12 16
D
a
i
l
y

P
r
o
f
i
t
s

(
t
h
o
u
s
a
n
d
s
)
Base Case Parameters

Figure 6: With base case parameters, profits are greatest without a solvent storage system.
Sensitivity to Capital Cost Reductions: Since base case results do not support the installation
of solvent storage systems, decreasing the capital costs of solvent inventory or stripping and
compression equipment may be one way to make solvent storage desirable. Keeping all other
parameters constant, a 42% decrease in solvent price is required for a nonzero optimal storage
time, and a 50% decrease allows a 2 hour optimal solvent storage time. These MEA price
decreases may be unrealistic, but an increase in the design difference between rich and lean
loading would accomplish the same effect. In general, a solvent storage system would require a
high capacity, low cost solvent.
If MEA price is returned to its base value and a similar sensitivity analysis is performed on the
base capital cost of stripping and compression equipment, a 62% cost reduction is required for a
nonzero solvent storage time. This level of price reduction would largely have to come in the
prices of concrete, steel, and construction materials, and such a steep drop in these commodity
prices seems unlikely.
Sensitivity to Changes in Electricity Price: Since the tradeoff in a large solvent storage
system is between capital costs and operating profits, another important determinant of optimal
solvent storage time is likely to be electricity prices. In order to investigate the effects of
changes in electricity prices on solvent storage optimization, the base case prices in each interval
are multiplied by a constant factor greater than one. Rather than uniformly shift electricity
prices, the price multiplier increases the disparity between high and low electricity prices. For
example, if electricity is $30/MWh at 2:00am and $50/MWh at 8:00pm, a price multiplier of 2
would double each price, thereby doubling the difference in the two prices from $20/MWh to
$40/MWh.
Figure 7 displays the increase in optimal storage time with electricity price multiplier along with
an approximate lifetime profit improvement offered by the solvent storage system. The
minimum electricity price multiplier for a nonzero optimal storage time is 1.29, where optimal
storage time is 1 hour and daily profits are $408,000. Optimal storage time increases to about 6
hours with a price multiplier of 2.4, remains at about 6 hours up to a 3.8 multiplier, and then
jumps considerably to the 1214-hour range at price multipliers from four to five. Approximate
231
10
lifetime profit improvement is determined by finding the difference in daily profits earned with
optimal solvent storage time and without solvent storage, then multiplying this quantity by 365
days of a year and 20 years to approximate additional profits over an economic lifetime. Taxes,
capital depreciation, and other detailed economic concerns are not included in this preliminary
calculation.
$0
$200
$400
$600
$800
$1,000
$1,200
0
2
4
6
8
10
12
14
16
1 2 3 4 5
A
p
p
r
o
x
.

L
i
f
e
t
i
m
e

P
r
o
f
i
t

I
m
p
r
o
v
e
m
e
n
t

U
s
i
n
g

O
p
t
.

S
t
o
r
a
g
e

(
m
i
l
l
i
o
n
s
)
O
p
t
i
m
a
l

S
t
o
r
a
g
e

T
i
m
e

(
h
o
u
r
)
Electricity Price Multiplier
Optimal
Storage Time
Lifetime Profit
Improvement

Figure 7: Optimal storage time and the overall economic benefit of solvent storage
increases with high/low electricity price disparity induced by a price multiplier.
To better understand the shape of the optimal storage time curve in Figure 7, the base case
electricity prices from Figure 3 are plotted from low to high in Figure 8. As the electricity price
multiplier increases from 1 to 2.4, solvent storage becomes economical during about 6 hours of
the highest price intervals. Then, because there is a significant drop in price to the next highest
price intervals, the price multiplier must increase to above 4 for any additional time intervals to
utilize solvent storage. However, since there is a large group of price intervals beyond the 6
hours with the highest prices that have similar electricity prices, all of these price intervals
become economical for solvent storage at roughly the same electricity price multiplier, hence the
sudden jump from 6 to 14 hours of solvent storage.
232
11
0
20
40
60
80
100
120
140
0 12 24 36 48 60 72 84 96
A
v
e
r
a
g
e

A
n
n
u
a
l

E
l
e
c
t
r
i
c
i
t
y

P
r
i
c
e

(
$
/
M
W
h
)
Price Interval
6 Hours:
Opt storage
time with
2.4-3.8 price
mult.
14 Hours:
Opt storage
time with 4+
price mult.

Figure 8: A plot of the base case annual average electricity prices in order from low to high
provides insight into the calculated variation in optimal storage time with electricity price
multiplier.
Additional Analysis of Electricity Price Effects: In order to gain an even better
understanding of the effects of electricity price distribution on solvent storage time, several
hypothetical electricity price curves based on a sinusoidal distribution were created (Figure 9).
The base sine curve is means to capture the maximum and minimum actual electricity prices
along with their general diurnal variation. From this base sine curve, a shifted curve uniformly
raises electricity prices $25/MWh throughout the day, a high amplitude curve systematically
increases electricity price disparity throughout the day, and a price spike curve follows the base
sine curve except for three 15-minute intervals where electricity prices are set to $500/MWh.
0
40
80
120
160
200
0:00 4:00 8:00 12:00 16:00 20:00 0:00
E
l
e
c
t
r
i
c
i
t
y

P
r
i
c
e

(
$
/
M
W
h
)
Time (24 hour clock)
Actual Annual
Avg. Prices
Base Sine
Curve
Shifted
High
Amplitude
Three
$500/MWh
Price Spikes

Figure 9: Several hypothetical sine curves are used to investigate the effects of electricity
price distribution on optimal solvent storage time.
233
12
Figure 10 shows the daily profits for each of the four hypothetical price curves as a function of
solvent storage time. The base sine curve results are similar to those shown in Figure 6; profits
are greatest without solvent storage. A uniform upward shift in electricity prices raises profits
across all solvent storage times, but the shape of the profit curve does not change, and solvent
storage is still uneconomical. Adding three price spikes, however, makes it optimal to store
solvent during these three time intervals, and the profits earned by doing this allow a large range
of solvent storage times to be more profitable than a plant without solvent storage. With the
systematic increase in price disparity induced by the high amplitude sine curve, the optimal
solvent storage time is 9.25 hours.
150
250
350
450
550
0 4 8 12 16
D
a
i
l
y

P
r
o
f
i
t
s

(
$

t
h
o
u
s
a
n
d
s
)
Base Sine Curve
Shifted
High Amplitude
Three $500/MWh
Price Spikes

Figure 10: Uniform price shifts do not affect solvent storage optimization, but disparity
between high and low electricity prices makes solvent storage valuable.
Conclusions
Flexible CO
2
capture using large-scale MEA storage requires tens of millions kgMEA and
storage capacities of tens to hundreds m
3
(550 million gallons) for storage times of 112 hours.
In this range of storage times, the incremental capital costs of solvent storage are in the tens to
hundreds of millions of dollars; storage tank costs are relatively small, as capital costs are
dominated by solvent inventory and the need for larger stripping, compression, and auxiliary
equipment to process stored solvent in regeneration mode. Solvent manufacturing capacity is a
significant limitation to large-scale solvent storage. Low cost and high capacity are desirable
solvent characteristics for a solvent storage system.
The need for additional energy in regeneration mode to strip and compress CO
2
from stored
solvent can result in high marginal costs in regeneration mode, but daily average electricity
production costs do not exceed that of a plant using inflexible CO
2
capture. In contrast,
depending primarily on current fuel and CO
2
prices, a flexible capture system that vents
additional CO
2
may have average production costs greater than, equal to, or less than the cost of
operating an inflexible system.
Solvent storage of a few hours or more is attractive when there are large differences between
high and low electricity prices over the course of a day, especially if capital or other cost
reductions are possible. Even if there are only a few time periods with significantly high
234
13
electricity prices each day, a power plant with flexible CO
2
capture using solvent storage can be
more profitable than one with an inflexible CO
2
capture facility, assuming system response time
on the order of 15 minutes. However, base case input parameters and the preliminary modeling
methodology do not provide the conditions required for solvent storage to be desirable, and a
uniform increase in electricity prices does not change the optimal solvent storage time.
In the absence of a more favorable electricity price distribution, capital cost reductions alone are
insufficient to promote solvent storage. However, if base plant stripping and compression
equipment was overdesigned and operates at a high rich-lean delta loading, the incremental
capital costs of solvent storage would only include storage tanks and solvent inventory. If the
comparatively inexpensive storage tanks were built upfront and solvent is slowly stockpiled
when prices are lower, the capital cost economics of solvent storage could be much more
favorable than is indicated in this analysis.
Other Activities and Future Work
M.S. Thesis
My M.S. thesis was completed in May 2009 and can be found on the Rochelle Group website at
http://www.che.utexas.edu/rochelle_group/Pubs/Cohen_MS_Thesis_2009.pdf.
Long-Term Analysis
In order to investigate the performance, environmental impacts, and economics of flexible CO
2

capture systems over an investment lifetime, several 20-year fuel and CO
2
prices paths have been
chosen as input into the first-order ERCOT dispatch model explained in previous work (Cohen,
2009). Data collection and analysis on several of these price paths has been completed, and a
full analysis of results is expected to be completed in summer 2009.
Imperial College Collaboration to Compare/Contrast ERCOT and UK Grids
Collaboration with Hannah Chalmers of the Imperial College of London is underway to compare
and contrast the implications of flexible CO
2
capture in the ERCOT and United Kingdom electric
grids. Chalmers has completed some preliminary qualitative comparison of the two grids, and I
have begun creating model inputs necessary to study flexible CO
2
capture in the UK using my
electric grid modeling code.
Analyzing Several Replacement Capacity Scenarios and Partial/Zero-Load Points
As another extension on previous work with electric grid dispatch modeling, I plan to investigate
inflexible CO
2
capture scenarios with various types of capacity used to replace output lost to CO
2

capture energy and compare these replacement capacity scenarios with flexible CO
2
capture
cases. Where previous calculations have been limited to a 20% partial-load point for the CO
2

capture system (Ziaii, Cohen et al., 2008), I hope to study several other partial-load points such
as 0% load with and without a residual energy penalty.
Continuation of Solvent Storage Optimization
After completing the other tasks mentioned above, I intend to extend the solvent storage
optimization analysis to incorporate electricity price seasonality as well as a finite system
response time required to transition from storage mode to regeneration mode. I may also include
a more detailed representation of power plant economics as well as consider the possibility of
partial-load boiler operation in conjunction with flexible CO
2
capture.
235
14
References
CES. "CHEMICAL ENGINEERING PLANT COST INDEX (CEPCI)." Chem Eng Sci.
2009:116(3):1/3.
Chalmers H, Chen C et al. Initial Evaluation of Carbon Capture Plant Flexibility. 8th
International Conference on Greenhouse Gas Control Technologies. Trondheim, Norway,
Elsevier. 2006.
Cohen SM. The Implications of Flexible CO
2
Capture on the ERCOT Electric Grid. The
University of Texas at Austin. M.S. Thesis. 2009:154.
Cohen SM, Rochelle GT et al. Turning CO
2
Capture On & Off in Response to Electric Grid
Demand: A Baseline Analysis of Emissions and Economics. ASME 2nd International
Conference on Energy Sustainability. Jacksonville. 2008.
ERCOT. Balancing Energy Services Market Clearing Prices for Energy Annual Report.
MCPER_MCPEL_2008.xls. 2008.
Fisher KS, et al. Advanced Amine Solvent Formulations and Process Integration for Near-Term
CO
2
Capture Success, Trimeric Corp. 2007.
ICISpricing. Retrieved May 28, 2008, 2008, from http://www.icispricing.com/. 2008.
NETL. Carbon Capture and Sequestration Systems Analysis Guidelines. USDOE. 2005.
Oyenekan B. Modeling of Strippers for CO
2
capture by Aqueous Amines. The University of
Texas at Austin. Dissertation. 2006:291.
Peters MS, Timmerhaus KD et al. Plant Design and Economics for Chemical Engineers.
McGraw-Hill Professional. 2002.
Rao AB, Rubin ES. "A Technical, Economic, and Environmental Assessment of Amine-Based
CO
2
Capture Technology for Power Plant Greenhouse Gas Control." Environ. Sci. Technol.
2002;36(20):44674475.
2
2008. Luminant Carbon Management Program, University of Texas at Austin. 2008.
Rubin ES, Chen C, Rao AB. Cost and performance of fossil fuel power plants with CO
2
capture
and storage. Energy Policy. 2007;35:44444454.
Ziaii S, Cohen SM et al. Dynamic operation of amine scrubbing in response to electricity
demand and pricing. 9th International Conference on Greenhouse Gas Technologies.
Washington, DC, Elsevier. 2008.
USEIA (2009). Average Cost of Coal Delivered for Electricity Generation by State, Year-to-
Date through March 2009 and 2008 (Dollars per Million Btu). epmxlfile4_10_b.xls,
USDOE.
USEPA. Emissions & Generation Resource Integrated Database (eGRID).
eGRID2006_Version_2_1. 2007.
USNETL. Cost and Performance Baseline for Fossil Energy Plants. Bituminous Coal and
Natural Gas to Electricity. J. M. Klara. 2007;1.
236
1
Modeling Absorber/Stripper Performance with MDEA/PZ

by Peter Frailie
and the
2
July 1, 2009
Abstract
The goal of this study is to evaluate the performance of an absorber/stripper operation
that utilizes the MDEA/PZ blended amine. Due to the complexity of this system the
model will be developed in several smaller, more manageable parts that can later be
combined to form the final model. The first section that will be developed is an
MDEA/PZ model based on thermodynamic data, which must initially be developed as
separate MDEA and PZ models. Once the MDEA/PZ model has been completed it must
be incorporated into separate absorber and stripper models similar to those developed by
Van Wagener and Plaza. Those models can then be combined to form the final
MDEA/PZ absorber/stripper model. This study is currently in the process of developing
the MDEA/PZ model based on thermodynamic data. Separate MDEA and PZ models
have already been completed, and the process of combining them has begun. The
MDEA/PZ thermodynamic model should be completed early in the next quarter, which
will allow work to commence on modeling an absorber/stripper operation.
Introduction
The removal of CO
2
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
CO
2
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. The report also makes the recommendation that more
kinetic and thermodynamic data must be acquired concerning the MDEA/PZ blend
before the model can be significantly improved. Two researchers in the Rochelle lab are
currently acquiring this data, but it has not yet been incorporated into an absorber/stripper
237
2
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 of
this study will be to combine absorber and stripper models to evaluate the overall system
performance.
Methods and Discussion
This quarters work dealt primarily with the MDEA model. As mentioned in the
previous quarterly report, Aspen Plus
currently has a rate-based MDEA model that uses

equilibrium constants to predict speciation. Initially, the approach was to develop a
thermodynamic MDEA model that matched the Aspen Plus
models predictions, but

that proved to be problematic when combining the MDEA and PZ models. Rather than
develop the MDEA and PZ models in separate files, it was more effective to develop
them in the same file. Most of the data used to regress the thermodynamic parameters
was cited by Aspen Plus
in its model.
The first set of parameters regressed concerned the viscosity of the MDEA/H
2
O/CO
2

mixture. The experimental data (Teng, 1994) provided the viscosity of MDEA/H
2
O
mixtures with MDEA mole fraction ranging from 0 to 1 between 25
o
C and 80
o
C. The
parameters selected for this regression were the first terms of the MUKIJ and MULIJ
binary parameters for a H
2
O/MDEA mixture and the first five DIPPR liquid viscosity
terms (MULDIP) for pure MDEA. Figure 1 compares the mixture viscosity as a function
of CO
2
loading at 25
o
C for the MDEA model to that of experimental data at 30, 40, 50,
and 60 wt % MDEA (Weiland, 1998), and Tables 1 and 2 report the regressed values for
the above mentioned parameters.
0
5
10
15
20
25
30
0 0.1 0.2 0.3 0.4 0.5 0.6
Loading (mol CO
2
/mol Alkalinity)
V
i
s
c
o
s
i
t
y

(
m
P
a
.
s
)
60 wt%
50 wt%
40 wt%
30 wt%

Figure 1: Liquid viscosity as a function of CO
2
loading for the MDEA model (lines)
and experimental data (points) at 25
o
C
238
3

Table 1: Regressed binary parameters for MDEA liquid viscosity
Parameter Value Reference Temperature (K)
MUKIJ/1 -1.71 298.15
MULIJ/1 -12.74 298.15

Table 2: Regressed pure component parameters for MDEA liquid viscosity (Note:
temperature units are
o
C and property units are Pa.s)
Parameter Value
MULDIP/1 -267.1
MULDIP/2 16910
MULDIP/3 36.5
MULDIP/4 -0.038
MULDIP/5 -1.0E5

The next set of parameters concerned the density of liquid MDEA. The parameters
regressed in the Aspen Plus
MDEA model were the cation/anion pair parameters for the

Clarke liquid density model (VLCLK) for the MDEAH
+
/HCO
3
-
and MDEAH
+
/CO
3
2-

pairs, as well as the interaction parameter VLQKIJ for the MDEA/H
2
O mixture. Because
the Aspen Plus
model was developed in the most recent version of Aspen Plus
and the
University of Texas still used Aspen Plus
v2006.5, some of these parameters (i.e.

VLQKIJ) have not yet been regressed. This primarily affects the density of the
MDEA/H
2
O/CO
2
mixture at low loading. Before the fall semester begins we will be
upgrading to the newest version of Aspen Plus
, at which point VLQKIJ will be

regressed. Because liquid density does not significantly affect vapor-liquid equilibrium
(VLE) predictions, VLQKIJ can be added after the rest of the thermodynamic model has
been developed. The experimental data used in the regression (Weiland, 1998) provides
the density of MDEA/H
2
O/CO
2
mixtures containing 30, 40, 50, and 60 wt % MDEA
between 0 and 0.5 loading at 25
o
C. Figure 2 compares the mixture density as a function
of CO
2
loading at 25
o
C for the MDEA model to that of experimental data at 30, 40, 50,
and 60 wt % MDEA, and Table 3 reports the regressed values for the above mentioned
parameters.
239
4
1
1.02
1.04
1.06
1.08
1.1
1.12
1.14
1.16
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Loading (mol CO
2
/mol Alkalinity)
D
e
n
s
i
t
y

(
g
/
m
L
)
30 wt%
60 wt%
50 wt%
40 wt%

Figure 2: Liquid density as a function of CO
2
loading for the MDEA model (lines)
and experimental data (points) at 25
o
C

Table 3: Regressed parameters for MDEA liquid density
Property Cation Anion Value Property Units
VLCLK/1 MDEAH
+
HCO
3
-
0.17 m
3
/kmol
VLCLK/1 MDEAH
+
CO
3
2-
0.18 m
3
/kmol

The final set of parameters that needed to be determined was the core thermodynamic
parameters (Gibbs energies of formation, enthalpies of formation, heat capacities, and
ionic interaction parameters). The data used to determine these values concerned
MDEA/H
2
O/CO
2
mixture VLE (Jou, 1982) and heat capacity (Weiland, 1997). After
initial attempts to regress all of the data simultaneously were unable to converge
satisfactorily, a handful of parameters was selected to be set manually to fit the VLE data.
The parameters selected were the Gibbs free energy of formation for MDEAH
+
, the
enthalpy of formation for MDEAH
+
, the first two heat capacity parameters for MDEAH
+

(CPAQ0), and the ionic interaction parameter for MDEAH
+
and HCO
3
-
in water
(GMELCC). All of the other ionic interaction parameters were set at their regressed
values which gave the best fit for the VLE data. Figure 3 compares the partial pressure of
CO
2
as a function of loading for the MDEA model to that of experimental data at 25, 40,
70, 100, and 120
o
C, Figure 4 compares the heat capacity of a MDEA/H
2
O/CO
2
mixture
as a function of loading for the MDEA model to that of experimental data at 30, 40, 50,
240
5
and 60 wt % MDEA, and Table 4 reports the regressed values for the above mentioned
parameters.
1.E-03
1.E-02
1.E-01
1.E+00
1.E+01
1.E+02
1.E+03
1.E+04
1.E-03 1.E-02 1.E-01 1.E+00
Loading (mol CO
2
/mol alkalinity)
P
a
r
t
i
a
l

P
r
e
s
s
u
r
e

C
O
2

(
k
P
a
)
25
o
C
120
o
C
100
o
C
70
o
C
40
o
C

Figure 3: CO
2
partial pressure as a function of CO
2
loading for the MDEA model
(lines) and experimental data (points)
2
2.5
3
3.5
4
4.5
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Loading (mol CO
2
/mol Alkalinity)
H
e
a
t

C
a
p
a
c
i
t
y

(
J
/
g
.
K
)
30 wt%
40 wt%
50 wt%
60 wt%

Figure 4: MDEA/H
2
O/CO
2
mixture heat capacity as a function of CO
2
loading for
the MDEA model (lines) and experimental data (points)
241
6

Table 4: Core thermodynamic parameters for MDEA model
Parameter Species Value Units
DGAQFM MDEAH
+
-2.528E8 J/kmol
DHAQFM MDEAH
+
-4.92E8 J/kmol
CPAQ0/1 MDEAH
+
920000 J/kmol.K
CPAQ0/2 MDEAH
+
-6000 J/kmol.K
GMELCC H
2
O (MDEAH
+
/HCO
3
-
) 9.637 N/A
GMELCC (MDEAH
+
/HCO
3
-
) H
2
O -4.5 N/A
GMELCC MDEA (MDEAH
+
/HCO
3
-
) 37.3 N/A
GMELCC (MDEAH
+
/HCO
3
-
) MDEA -2.47 N/A
GMELCD H
2
O (MDEAH
+
/HCO
3
-
) 485.4
o
C
GMELCD (MDEAH
+
/HCO
3
-
) H
2
O -139.1
o
C
GMELCD MDEA (MDEAH
+
/HCO
3
-
) 2069
o
C
GMELCD (MDEAH
+
/HCO
3
-
) MDEA 360.9
o
C

It should be noted that for the heat capacities of the 30, 40, and 50 wt % MDEA mixtures,
the percent difference between the literature values and the model values was never
greater than 15% over the range of tested values.
Conclusions
Overall the conversion of the Aspen Plus
equilibrium constant model to a

thermodynamic model went well. The thermodynamic model can adequately predict
viscosity and VLE at useful loadings (00.25 mol CO
2
/mol alkalinity), and after the
incorporation of the VLQKIJ parameter it should adequately predict molar volume. The
biggest concern is the heat capacity predicted by the model, which is particularly
important when evaluating absorber and stripper performance.
Future Work
The main goal over the next three months will be to finish developing the MDEA/PZ
rate-based model using experimentally obtained thermodynamic data. Once the rate-
based model has been finalized, it will be applied to an absorber/stripper operation. As in
the development of the rate-based model, this model will first be split into two sections
(absorber and stripper) for early development, which will be combined later to form the
final model. This is similar to the approach used by Jorge Plaza and David Van
Wagener, though they did not combine their models into a single operation. Their work
and methodology will serve as the basis for the development of the separate absorber and
stripper models.
242
7

References
Bishnoi S. Carbon Dioxide Absorption and Solution Equilibrium in Piperazine Activated
Methyldiethanolamine. The University of Texas at Austin. Ph.D. Dissertation. 2000.
Jou FY, Mather AE, Otto FD. Solubility of hydrogen sulfide and carbon dioxide in
aqueous methyldiethanolamine solutions. Ind Eng Chem Process Des Dev.
1982;21(4):539544.
Teng TT, Maham Y, Hepler LG, Mather AE. Viscosity of Aqueous Solutions of N-
Methyldiethanolamine and of Diethanolamine. J Chem Eng Data. 1994;39:290
293.
Weiland RH, Dingman JC, Cronin DB. Heat Capacity of Aqueous Monoethanolamine,
Diethanolamine, N-Methyldiethanolamine, and N-Methyldiethanolamine-Based
Blends with Carbon Dioxide. J Chem Eng Data. 1997;42:10041006.
Weiland RH, Dingman JC, Cronin DB, Browning GJ. Density and Viscosity of Some
Partially Carbonated Aqueous Alkanolamine Solutions and Their Blends. J Chem
Eng Data. 1998;43:378382.

243
1
Effective Area and
Mass Transfer Coefficients of Packing

Chao Wang
Supported by the Luminant Carbon Management Program,
2
and the Process Science and Technology Center
July 3, 2009
Abstract
Packings are widely used in distillation, stripping, and scrubbing processes because of their
relatively low pressure drop, good mass transfer efficiency, and ease of installation. Packings
are also being investigated for the post-combustion carbon capture process for these reasons.
Research continues to focus on development of high performance packing, especially on
minimizing pressure drop, maximizing mass transfer efficiency, and minimizing costs. The
design of packed absorbers for carbon dioxide capture will require the reliable measurement and
accurate prediction of the effective area, gas and liquid film mass transfer coefficient. A variety
of experimental methods for measuring effective area, gas and liquid film mass transfer
coefficient k
L
a has been reported. Consistent measurements of these important design
parameters will begin this summer.
Absorption of CO
2
with NaOH is applied to measure the effective area of packings.
Atmospheric CO
2
in air is used as gas phase and 0.1 M NaOH is used as liquid phase. This is a
liquid phase controlled mass transfer system so the liquid phase mass transfer coefficient k
l
or
often referred to as k
g
can be assumed as the overall mass transfer coefficient K

G
. In the
proposed summer work, the gas flow rate set points are 180, 300, 450 ACFM and liquid flow
rate set points are 1, 2.5, 5, 7.5, 10, 15, and 25 gpm/ft
2
(same operating conditions explored by
Robert Tsai). The effective area can then be calculated by the equation:
RT Zk
y
y
u
RT ZK
y
y
u
a
g
out CO
in CO
G
G
out CO
in CO
G
e
'
2
2
2
2
) ln( ) ln(
= (1)
Absorption of sulfur dioxide (SO
2)
with NaOH is applied to measure the gas phase mass transfer
coefficient. SO
2
, blended with ambient air at a composition of approximately 80 ppm, will be
absorbed by 1 M NaOH solution. The reaction is instantaneous and the mass transfer process is
controlled by the gas phase. Thus the overall mass transfer coefficient K
G
can be replaced by gas
phase mass transfer coefficient k
G
. This experiment can be combined with the effective area
244
2
experiment as long as the gas and liquid flow rates are set at the same level. The gas phase mass
transfer coefficient can be calculated by the equation:
e
out SO
in SO
G
G
ZRTa
y
y
u
k
) ln(
2
2
= (2)
Desorption of toluene in water with air is applied to measure the liquid phase mass transfer
coefficient. Ambient air is used to strip toluene from water. As a result of the high Henrys
constant, the mass transfer resistance is controlled by the liquid phase. The gas flow rates and
liquid flow rates are set at the same value with the effective area measurement to make the 3
experiments consistent. The liquid phase mass transfer coefficient can be calculated by the
equation:
) / ln(
2 1 LA LA
e
L
L
c c
Za
u
k = (3)
Introduction
This quarterly report mainly focused on the experimental methods of measurement of effective
area, gas-side mass transfer coefficient, liquid-side mass transfer coefficient which will be used
in the summer project. Three structured packings will be tested this summer. They are: Raschig
RSP 250 wSE, Flexipac 1.6 Y HC, and Mellapak 2X. Limited experiments will be carried out
using the Mellapak 252Y.
Raschig RSP 250 wSE is a structured packing developed by RASCHIG-JAEGER. The surface
area of this packing is approximately 250 m
2
/m
3
. The rows of sinusoidal waves within vertical
packing sheets are surface enhanced to encourage greater turbulent radial spread of thin liquid
film flows on the front and back of the waves on each sheet within an element. The effect of
surface enhancement on the mass transfer efficiency of RSP-250 wSE compared to Raschig
Super-Pak 300 without surface enhancement (RSP-300 woSE) is significant with consistently
lower HETP values (Schultes, 2006). The pressure drop of RSP-250 wSE is consistently lower
by an order of magnitude than the 250 m
2
/m
3
surface area standard structured packings and
significantly less than the high capacity types at high flow rates (Olujic, 2001).
Flexipac 1.6 Y HC is a structured packing developed by Koch-Glitsch. The surface area of this
packing is 295 m
2
/m
3
. An improvement has been made at the base region of the sheets to reduce
the liquid holdup at the interface between two sets. The corrugation angle is 90 degrees at the
base region of the sheets which is different from the corrugation angle at the bulk region. The
effect of this design is to reduce the gas velocity and permit improved draining of liquid at the
interface between two stacked elements and increase ultimate usable hydraulic capacity. The
capacity increases 1040% and efficiency increases 1020% compared to conventional packing
of similar geometry (Hausch, 2005).
Mellapak 2X is a structured packing developed by Sulzer Chemtech. The surface area of this
packing is 205 m
2
/m
3
. The corrugation angle of this packing is 60 degrees which is larger than
the conventional packings. This design is to reduce the resistance of the gas flow inside the
packing and hence reduce the pressure drop.
The packing properties are as follows:
245
3
Table 1: Packing properties
Packing Surface area (m
2
/m
3
) Void fraction (%) Channel side (mm) HETP (m)
RSP 250 wSE 250 0.92 5.08
Flexipac 1.6Y HC 295 0.91 9.8 0.2~0.3
Mellapak 2X 205 0.99 21.5 0.63~0.9

Hydraulic data for these packings are as follows:
Hydraulic Performance of RSP 250 wSE
0.0100
0.1000
1.0000
10.0000
0.100 1.000 10.000
F-factor (ft/s) (lb/ft
3
)
0.5
P
r
e
s
s
u
r
e

D
r
o
p
,

i
n
H
2
O
/
f
t

p
a
c
k
i
n
g
5 gpm/ft2
10 gpm/ft2
15 gpm/ft2
20 gpm/ft2
30 gpm/ft2
DRY DP
300 Dry
300 5
300 10
300 15
Figure 1: Hydraulic performance of RSP 250 wSE (from SRP database)

246
4

Figure 2: Hydraulic performance of Flexipac 1.6 Y HC (Hausch, 2005)
Measurement of effective area
Absorption of CO
2
in air with 0.1 M NaOH solution is used for the measurement of effective
area. It is a common method used to measure effective area for packings. Revising my previous
literature review (SepDec), a large variety of 2-fluid systems has been used for this purpose.
Among these, absorption of CO
2
into dilute caustic solution can be considered one of the more
attractive options. The Separations Research Program (UT-SRP) used this method to measure
effective area for structured packings in air/water column for years. Therefore, we decided to
choose this method for our experiment. The reaction
CO
2
+ 2OH
-
CO
3
2-

can be considered irreversible. When CO
2
partial pressure is low and hydroxide ion is present in
relative excess, the reaction can be treated as pseudo-first order. The reaction rate is:
r=k
1
[CO
2
]
where
k
1
=k
OH
-
[OH
-
].
According to 2-film theory, the overall mass transfer resistance is:
247
5
'
1 1 1
g g G
k k K
+ = (4)
The gas side mass transfer coefficient could be expressed as:
) ( ) ( 075 . 1
, 2 85 . 0
, 2
2
RTd
D
hD
d u
k
G CO
G CO
G
g
= (5)
where u
G
is the gas velocity, D
CO2,G
is the diffusivity of CO
2
in the gas phase, h is the exposed
length of wetted wall column (Bishnoi, 2000).
The gas side resistance was calculated to account for less than 10% of the overall resistance in
this situation in the air/water column (Rocha, 1996).
The liquid side mass transfer coefficient could be expressed as:
2
, 2
2
'
) (
] [
1
o
L
L CO
OH
CO
o
L
g
k
D OH k
H
k
k
+ = (6)
where k
L
o
is the physical mass transfer coefficient in the liquid phase, H
CO2
is the Henrys
constant for CO
2
in the liquid, k
OH
-
is the second order reaction rate constant, D
CO2,L
is the
diffusivity of CO2 in the liquid phase. In the WWC experiments and in our previous air/water
column experiments,
2
, 2
) (
] [
o
L
L CO
OH
k
D OH k

>>1 (Tsai, 2008), so the 1+ can be ignored. Then the

liquid side mass transfer coefficient could be:
2
, 2
'
] [
CO
L CO
OH
g
H
D OH k
k
= (7)
Diffusivity could be calculated by the following equations (Barrett, 1966):

2
5
, 2 10
10 591 . 2 5 . 712
1764 . 8 log
T T
D
W CO
+ = (8)
. ) * ( ) * (
, 2 , 2
const D D
T W W CO T L CO
= = (9)
Henrys constant could be calculated by the following equations (Danckwerts, 1970):

2 5 2
, 2 10
10 8857 . 7 10 9044 . 5 1229 . 9 ) ( log T T H
W CO

+ = (10)
=
i i
W CO
CO
h I
H
H
, 2
2
10
log (11)
g i
h h h h + + =
+
(12)
k
OH
-
is calculated by the following equations (Pohorecki, 1988):
T
k
OH
2382
895 . 11 log
10
=
(13)
2
10
016 . 0 211 . 0 log I I
k
k
OH
OH
=
(14)
With these equations, the overall mass transfer coefficient could be calculated. Then the
effective area could be calculated from this equation:
248
6
RT Zk
y
y
u
RT ZK
y
y
u
a
g
out CO
in CO
G
G
out CO
in CO
G
e '
2
2
2
2
) ln( ) ln(
= (15)
Measurement of gas side mass transfer coefficient
Absorption of SO
2
with NaOH solution is used for the measurement of gas side mass transfer
coefficient. The inlet SO
2
concentration is around 80 ppm which is realized by mixing 2% SO
2

in a cylinder with air. The initial NaOH concentration is 0.1 M. Large varieties of experimental
systems for the measurement of gas side mass transfer coefficient have been reviewed (Rochelle
et al., 2009). Among these systems, the SO
2
/NaOH system has advantages. The property of SO
2

is similar to CO
2
and they could use the same liquid solution which means these two experiments
can be performed at the same time. It is an instantaneous reaction and the equilibrium is easy to
reach. The SO
2
concentration could be read from an installed SO
2
analyzer.
The reaction is:
SO
2
+ 2OH
-
SO
3
2-
It is an irreversible instantaneous reaction. The mass transfer resistance at the liquid phase is
much smaller compared with the resistance at the gas phase. Thus, the liquid side mass transfer
resistance could be ignored. The overall mass transfer coefficient could be assumed equal to the
gas side mass transfer coefficient. The volumetric gas side mass transfer coefficient could be
calculated by the following equation:
ZRT
y
y
u
a k
out SO
in SO
G
G
) ln(
2
2
= (16)
where u
G
is the superficial gas velocity, y
SO2in
and y
SO2out
is the concentration of SO
2
in the gas
phase at the inlet and outlet of the column (Sharma, 1966). The SO
2
concentration is measured
by Thermo 43i SO
2
analyzer which has a range of 0100ppm and can reach as low as 0.5 ppb.
The inlet SO
2
concentration is adjusted to around 80 ppm by mixing 2% SO
2
from the cylinder
with air. The outlet SO
2
concentration could be around 10 ppb with 10 feet packings. Since
from our previous measurement of effective area, the k
G
value could be calculated by:

e
G
G
a
a k
k = (17)
Therefore, we can get the k
G
value from direct measurement.
These experiments will be combined with the effective area measurements to refine the
determination of the effective area and will also be used to determine k
g
. The inlet air should be
spiked with SO
2
with air and both the SO
2
concentration and the CO
2
concentration should be
measured. The gas flow rates and liquid flow rates for both experiments are the same. Both
experiments use the same NaOH solution, so the concentration of OH
-
is calculated by
] [ 2 ] [ 2 ] [ ] [
2
3
2
3

= SO CO OH OH
initial remaining
(18)
The concentration of carbonate is calculated from the FTIC analysis:
[ ]
( )( ) 011 . 12 1000 1
1000
1
1
011 . 12
1
ln 10
6
2
3
ppmC x
L
mL
mL
g
gC
molC
so grams
gramsC x
CO = =

249
7
x ppm C could be read from the standard curve.
The concentration of sulfite is calculated from the material balance of S:
t y y G
RT
P
SO
out SO in SO
* ) ( * ] [
2 2
2
3
=
(19)
where G is the gas flow rate, and t is the time for each run.

Measurement of liquid side mass transfer coefficient
Desorption of toluene from water using air is adopted for the measurement of liquid side mass
transfer coefficient. According to the 2-film theory:

a mk a k a K
y x x
1 1 1
+ = (20)
where K
x
is the overall mass transfer coefficient, k
x
is the liquid side mass transfer coefficient; k
y

is the gas side mass transfer coefficient; m is the slope of equilibrium curve (Basmadjian, 2004).
If we choose a system with a large m, the liquid side mass transfer resistance will be much
greater than the gas side mass transfer resistance. Therefore we can assume the overall liquid
phase mass transfer coefficient equals the liquid film mass transfer coefficient. Thus, we can get
the value of liquid side mass transfer coefficient from the overall mass transfer coefficient.
Toluene/water system is one such system. The Henrys constant of toluene in water is 353.1
atm/mol fraction (Chapoy, 2008) and the solubility is 542 ppm (Carl, 1990) at room temperature
and atmosphere. m equals Henrys constant times activity coefficient divided by total pressure.
So in this system m is large enough to ignore the gas side resistance. The system used is water
saturated with toluene; samples are taken from the inlet and the outlet of the column and
analyzed in a GC. The gas flow rate is 180,300 450 CFM and liquid flow rate is 1-25 gpm/ft
2
.
The liquid side mass transfer coefficient could be calculated by the following equation:
) / ln(
2 1 LA LA
L
L
c c
Z
u
a k = (21)
where u
L
is the superficial liquid velocity, Z is the packing height and c
LA1
and c
LA2
is the liquid
phase toluene concentration at the inlet and outlet of the column (Linek, 1984). Since we have
the value of effective area from our previous experiment, we can calculate the k
L
by:
e
L
L
a
a k
k = (22)
Future Work
We will perform these experiments starting around mid-July. Measurements of effective area, k
G

and k
L
as a function of liquid and gas load will be performed.
An alternative liquid phase film control system will be investigated which requires the
construction of pH control system. This method uses NaOH/CO
2
system. According to equation
23:
250
8
2
, 2
2
'
) (
] [
1
o
L
L CO
OH
CO
o
L
g
k
D OH k
H
k
k
+ = (23)
If the concentration of hydroxyl is controlled at a low level so the chemical term k
OH
-
[OH
-
]D
CO2,L

is much smaller than the physical term k
L
o
, the liquid side mass transfer coefficient measured
should be the physical mass transfer coefficient. Since this system is liquid controlled we can
get the liquid side mass transfer coefficient from the overall mass transfer coefficient. The
concentration of hydroxyl should not exceed 10
-4
mol/L to satisfy this condition.
The key requirement for this system is to control the pH at 10. My proposal is to continuously
add 0.1 mol/L NaOH solution to the bulk liquid phase. pH control valve is used to control the
flow rate of the buffer solution. The k
L
o
is estimated to be around 0.8-2*10
-4
m/s using the
following equation:

2 / 1
) ( 2
S
U C D
k
LE E L o
L
= (24)
where D
L
is the diffusivity of CO
2
in liquid phase, C
E
is correction factor for surface renewal,
usually equal to 0.9, U
LE
is the effective liquid velocity, m/s and S is the side dimension of
corrugation (Rocha, 1996).
The rate of CO
2
absorbed:
Z ac k r
CO
o
L A
*
2
= (25)
where k
L
o
is the physical mass transfer coefficient, a is the effective area, c
CO2
*
is the equilibrium
concentration of CO
2
in liquid phase and Z is the packing height.
Make the material balance of CO
2
in the total system:
Rate of CO
2
absorbed = rate of CO
2
accumulated in the liquid phase.
) (
0
*
2
2
3
2
3
=

=
t CO t CO
L
CO
o
L
c c
t
V
Z ac k (26)
Where V
L
is the total volume of liquid in the system, t is the time period,
t CO
c
2
3
and
0
2
3
=
t CO
c is the
concentration of carbonate at t time and t=0 time. So if we take samples every few minutes, we
can get the carbon concentration curve with time and the slope is
L CO
o
L
V Z ac k /
*
2
. Then we can
get the value of k
L
o
a. Since we have measured the effective area, the liquid side mass transfer
coefficient is:

e
o
L o
L
a
a k
k = (27)
Conclusions
For the measurement of effective area, absorption of CO
2
in air with 0.1 M NaOH proves to be a
reasonable method. The effective area is calculated by the equation:
RT Zk
y
y
u
RT ZK
y
y
u
a
g
out CO
in CO
G
G
out CO
in CO
G
e
'
2
2
2
2
) ln( ) ln(
= (28)
251
9
For the measurement of gas side mass transfer coefficient k
G
, absorption of SO
2
with NaOH
solution is adopted. k
G
is calculated by the equation
e
out SO
in SO
G
G
ZRTa
y
y
u
k
) ln(
2
2
= (29)
For the measurement of liquid side mass transfer coefficient k
L
, desorption of toluene in water
with air is adopted. k
L
is calculated by the equation
) / ln(
2 1 LA LA
e
L
L
c c
Za
u
k = (30)
References
Barrett PVL. Gas absorption on a sieve plate, University of Cambridge, Cambridge, England,
1966.
Basmadjian D. Mass transfer: principles and applications. CRC Press: Boca Raton, FL, 2004.
Bishnoi S. Carbon dioxide absorption and solution equilibrium in piperazine activated
methyldiethanolamine, University of Texas at Austin. Ph.D. Dissertation. 2000.
Carl LY, Haur-Chung Y. Water solubility data for organic compounds. Poll Engr.
1990;22(10):7975
Chapoy A, Haghighi H, Tohidi B. Development of a Henrys constant correlation and solubility
measurements of n-pentane, i-pentane, cyclopentane, n-hexane, and toluene in water. J
Chem Thermod. 2008;40:10301037
Danckwerts PV. Gas-Liquid Reactions. McGraw-Hill: New York, 1970.
Hausch G, Nieuwoudt I, Sommerfeldt RA. Advances in styrene fractionation with InTALOX
packed tower systems: Part 2-FLEXIPAC HC Structured Packing. AIChE Spring Natl Meet
Conf Proc. 2005. 1813.
Linek V, Petericek P. Effective interfacial area and liquid side mass transfer coefficients in
aborption columns packed with hydrophilised and untreated plastic packings. Chem Eng
Res Des. 1984;62:1321.
Olujic Z, Seibert AF, Kaibel B, Jansen H, Rietfort T, Zich E. Performance of a New High
Capacity Structured Packing. AIChE Spring National Meeting, Houston, TX. 2001.
Pohorecki R, Moniuk W. Kinetics of Reaction between Carbon Dioxide and Hydroxyl Ions in
Aqueous Electrolyte Solutions. Chem Eng Sci. 1988;43(7):1677.
Rocha JA, Bravo JL, Fair JR. Distillation columns containing structured packings: A
comprehensive model for their performance. 2. Mass-Transfer model. Ind Engr Chem Res.
1996;35:16601667
2
252
10
Schultes M, Chambers S. How To Surpass Conventional and High Capacity Structured
Packings with Raschig Super-Pak. AIChE Spring National Meeting, Orlando, FL, 2006.
Tsai R, Schultheiss P et al. Influence of Surface Tension on Effective Packing Area. Ind Engr
Chem Res. 2008;47:12531260

253
1
Pilot Plant Testing of Advanced Process Concepts using
Concentrated Piperazine

by Eric Chen
and the
2
July 12, 2009
Abstract
Pilot plant testing of 8 m piperazine in a two-stage heated flash is planned for the Fall of 2009.
Substantial modifications to the existing pilot plant at SRP will be needed. Process flow
diagrams (PFD) and piping and instrument diagrams (P&ID) have been developed for the new
process. Preliminary specifications for the high pressure pump, cross-exchanger, and steam
heaters have been developed and are in the process of being formally designed and quoted by
various vendors.
Introduction
The concept of concentrated (8 m) piperazine with high temperature stripping in a two-stage
heated flash is ready for pilot plant testing. The concept is described by Freeman et al. (2009).
A successful three-week campaign with a simple stripper was completed in the SRP pilot plant in
the Fall of 2008. A campaign is proposed for the SRP pilot plant in the Fall of 2009. This
campaign will require substantial equipment modifications for the high temperature stripping,
including three heat exchangers, two gas-liquid separators, a high pressure pump. Absorber
intercooling will also need to be added and will consist of a pump and a heat exchanger. This
system will be operated for three weeks with 8 m PZ to determine mass transfer, heat transfer,
and energy performance.
Pilot Plant Modifications
Process Flow Diagram
The existing pilot plant at SRP will be modified to evaluate high temperature stripping and
absorber intercooling using 8 m piperazine. A process flow diagram has been developed and is
shown in Figure 1. The high temperature stripping process will replace the existing simple
stripper with a two-stage heated flash. The new process will consist of a high pressure pump, a
cross-exchanger, two steam heaters, and two gas-liquid separator vessels and will be installed
downstream of the existing cross-exchanger.
254
2

Figure 1: Process flow diagram for the two-stage flash process
255
3
Rich solvent from the existing cross-exchanger will be pressurized by the high pressure pump
and fed into the high pressure cross-exchanger. The rich solvent will be further heated to 150 C
in the high pressure steam heater. The first stage of the flash process is designed to operate at
150 C and 13.5 atm. Flashing will occur in both the cross-exchanger and steam heater. The gas
and liquid will be separated in the first separator vessel. Liquid from the bottom of the separator
will pass through a throttling valve, dropping the pressure from 13.5 to 8 atm. The solvent will
be reheated to 150 C in the second steam heater, resulting in additional flashing of CO
2
and
H
2
O from the liquid.
The CO
2
and H
2
O vapor will exit off the top of both gas-liquid separator vessels and will be
combined into a single gas stream. The overhead condenser will condense out the water vapor
and pumped to the absorber feed tank. The CO
2
vapor will be recycled back to the gas
accumulator.
Absorber intercooling will be necessary to obtain the optimal rich loading for the high
temperature stripping process. A pump and heat-exchanger will be added. Solvent will be
pumped out from the bottom of the chimney tray at the middle of the absorber into a heat
exchanger and back into a nozzle above the chimney tray.
Two-Stage Flash Equipment Design
The two-stage heated flash process will be designed to be mounted on a portable skid that will
enable it to be transported for field testing. Equipment size was specified based on a nominal
liquid flow rate of 15 gpm. The skid will consist of a high pressure pump, high pressure cross-
exchanger, two steam heaters, and two gas-liquid separators. To accommodate field testing, the
skid will be designed to fit inside a standard 20 ft ocean shipping container. The dimensions for
the inside length, door width, and door height are 194, 78, and 76, respectively.
Specifications for the two-stage flash process were determined using a spreadsheet model
developed by Rochelle and Van Wagener. Both stages of the flash process are designed to
operate at a temperature of 150 C. The pressure of the first and second stage will be 13.5 and 8
atm, respectively. The pressures in the two-stage flash were optimized to vaporize
approximately the same amount of CO
2
in each stage. The amount of water vaporized in the
second flash is about 2.4 times that of the first flash.
The current heat exchanger in the SRP pilot plant has a temperature and pressure rating of
135 C and 10 atm, respectively. Since the two-stage flash requires higher operating pressures
and temperatures, a new high pressure cross-exchanger will be purchased. A high pressure pump
will also be purchased and installed downstream on the outlet of the rich stream from the existing
cross-exchanger.
Piping and Instrumentation Diagram
The piping and instrumentation diagram (P&ID) for the skid is shown in Figure 2. Rich solvent
on the cold side of the existing cross-exchanger will first be pumped through a bag filter before
being heated in the high pressure cross-exchanger. The high pressure steam heater will further
heat the rich solvent to 150 C. Steam flows to the two steam heaters will be controlled to
maintain a temperature of 150 C in both gas-liquid separator vessels.
256
4
FV-257
TT
221B
TT
220D
H-114-DI
H-115-DI
P-109
TT
221A
TT
220C
TT
220B
VSC
104
V-109
FV-260
PDT
260
LT
222
TT
222
LC
222
FE-217
FT
217
PC
260
FV-226
Steam
Check Valve Relief Valve
Vent to Absorber
Feed Tank
TT
207
F-108
FT
207
DT
207
From LP Cross Exchanger
Cold Side (Rich)
To LP Cross-Exchanger
Hot-Side (Lean)
Condensate
Flash Sep
T = 150 C
P = 13.5 atm
Rich Amine Stream
Q = 15 gpm
T = 120 C
P = 15 atm
Rich Amine Stream
Q = 15 gpm
T = 142 C
P = 14 atm
Rich Amine Stream
Q = 15 gpm
T = 150 C
P = 13.5 atm
Lean Amine Stream
Q = 15 gpm
T = 150 C
P = 8 atm
Lean Amine Stream
Q = 15 gpm
T = 123 C
P = 7 atm
HP Flash Gas Stream
Q = 0.08 m
3
/min
T = 150 C
P = 13.5 atm
76% CO2, 24% H2O
TT
223B H-116-DI
TT
223A
FV-227
Steam
Condensate
FV-258
V-110
PDT
261
LT
224
TT
224
LC
224
TT
208
FT
208
DT
208
Flash Sep
T = 150 C
P = 8 atm
FV-261
FE-218
FT
218
PC
261
Condenser
Check Valve
Bypass to Condenser
Filter
Vent Relief to
Absorber Feedtank
TT
220A
V-175 V-176 V-177 V-178
FE-219
FT
219
Steam
Trap
Steam
Trap
FE-226
FT
226
TT
227A
FE-227
FT
227
PT
227
TT
227B
TT
226A
PT
226
TT
226B
Pressure
Relief
Pressure
Relief
FV-259
Downstream
Pressure Control
PC
228
PDT
228
F-107
Filter

Figure 2: Piping and instrument diagram for two-stage flash system mounted on skid
257
5
The liquid level in the separators will be controlled separately by a valve downstream of the flow
measurement. Control valves in the vapor line downstream of each separator will be used to
maintain the pressure at 13.5 and 8 atm in high and low pressure gas-liquid separators. The
vapor flow rate will also be measured at the outlet of the each separator and also in the combined
vapor stream. The lean solvent exiting the bottom of the low pressure separator flows through a
bag filter before being used to preheat the rich solvent in the high pressure cross-exchanger.
The skid will incorporate the latest wireless process instrumentation technology by Emerson
Process Management. The benefits of wireless technology will be two-fold. First, it will
streamline the installation of the process instruments into the skid. Second, once installed,
wireless technology will facilitate incorporation of the skid instrumentation into existing process
control systems at a field test site. The wireless technology will eliminate the need to run
conduit and wires for all of the process instrumentation.
Wireless level, pressure and temperature transmitters will be used on the two-stage flash skid.
Micromotion flowmeters will be used to measure liquid flow rate and density. Density will be
used to indirectly measure CO
2
loading and control pilot plant operations. Orifice meters will be
used to measure the vapor flow rate leaving the top of each gas-liquid separator. An additional
orifice meter will measure the flow rate of the combined vapor stream. Orifice meters will also
be used to measure the flow rate of steam for each of the steam heaters.
High Pressure Pump
A high pressure pump will be needed to provide the 13.5 atm pressure required for the first stage
of the flash. A Grundfos multi-stage centrifugal pump has been identified that will meet the
desired specifications. The pump specifications are listed in the table below.
Table 1: High pressure pump specification
Model Grundfos CRNE5-24
Flow 20 GPM @ 250 psig
Matl. of Const. 316 Stainless Steel
Power 7.5 HP / 480V
Motor TEFC
Controller Variable Speed
Mechanical Seal Kalrez
High Pressure Cross-Exchanger
The majority of CO
2
capture pilot plants are currently designed using MEA as the solvent, which
inherently operates at lower temperatures (~120 C) and pressures to minimize thermal
degradation. In contrast, piperazine can be operated at higher temperatures (up to 150 C) and
higher pressures before thermal degradation becomes significant. In order to evaluate the high
temperature two-stage flash, a new cross-exchanger will need to be purchased. To reduce capital
costs and remain within the practical design limits of cross-exchangers, the new high pressure
exchanger will be installed in series with the existing cross-exchanger.
In order to prevent flashing from occurring in the existing low pressure cross-exchanger, the high
pressure cross-exchanger will be designed to have an inlet rich solvent (cold-side) temperature of
105 C. A bypass on the hot-side of the existing cross-exchanger will be used to control and
prevent high temperature excursions. The rich solvent is expected to begin flashing at
258
6
approximately 133 C. The outlet temperature of the rich solvent is expected to be 143 C if we
design the lean solvent (hot-side) to have inlet and outlet temperatures of 150 and 110 C,
respectively. The cross-exchanger will be designed to have a pressure drop of 10 psi. A heat
release curve was developed for the high pressure cross-exchanger (Figure 3).
0
200,000
400,000
600,000
800,000
1,000,000
1,200,000
1,400,000
40 60 80 100 120 140 160
Temperature (
C)
C
u
m
u
l
a
t
i
v
e

H
e
a
t

R
a
t
e

(
B
t
u
/
h
r
)
Lean
Rich

Figure 3: Heat release curve for high pressure cross-exchanger
A preliminary design for the cross-exchanger was developed by Alfa Laval. The original design
called for using a single cross-exchanger to heat the rich solvent from 45 to 145 C. However,
the specified temperature range was beyond the practical design limits. Only an outlet
temperature of 136 C was achievable, which was designed without accounting for flashing.
Since the high pressure cross-exchanger will be used in series with the existing exchanger, the
design should be more than adequate. Table 2 lists the specifications of the high pressure cross-
exchanger.
259
7
Table 2: High pressure cross-exchanger specification
Vendor. Alfa Laval
Model M6-MFD
Flow 15 GPM
Heat Exchanged 1045 kBTU/hr
LMTD 22.1 F
Heat Transfer Area 120.6 ft
2

Matl. of Const. 316 Stainless Steel
Gasket Matl. EPDMP
Design Pressure 250 psi
Design Temperature 320 F
Steam Heaters
Two heat exchangers using 120 psia steam will be used to heat the rich solvent to 150 C, one
for each stage in the high temperature stripping process. The first, high pressure steam heater
will have a two-phase feed because flashing will have occurred in the high pressure heat
exchanger. CO
2
and H
2
O will continue to vaporize as the solvent is heated from 143 to 150 C.
In the second, low pressure steam heater, the feed will again be two-phase because a throttling
valve will be used to drop the pressure from 13.5 to 8 atm. Preliminary calculations indicate that
the solvent temperature will drop to about 144 C across the valve. Heat release curves were
developed for both steam heaters (Figure 4).
0
20,000
40,000
60,000
80,000
100,000
120,000
140,000
160,000
142 143 144 145 146 147 148 149 150 151
Temperature (
o
C)
C
u
m
u
l
a
t
i
v
e

H
e
a
t

R
a
t
e

(
B
t
u
/
h
r
)
HP Steam Heater
LP Steam Heater

Figure 4: Heat release curve for high pressure and low pressure steam heaters
260
8
Absorber Intercooling Design
Absorber intercooling will be needed to obtain the optimal rich loading for the two-stage high
temperature stripping process. The absorber will need to be operated such that the temperature
bulge will be located at the middle of the column. The intercooling system will consist of a
pump and heat-exchanger. Liquid will be pumped from the bottom of the chimney tray in the
middle of the absorber, cooled in the exchanger, and pumped back above the chimney tray. This
design will eliminate the need for level control. An existing pump, P-105, will need to be rebuilt
and the Alfa Laval M6-FG heat exchanger used in a previous pilot plant campaign will be used.
Cooling water will be used to cool the amine stream.
Pilot Plant Schedule
Pilot plant tests using 8 m piperazine and the two-stage high temperature stripping process are
planned for Fall 2009. Activities will include the design, layout, and construction of the skid-
mounted two-stage flash system and installation of the absorber intercooling system. The pilot
plant will be operated for three weeks. The proposed schedule for the planned pilot plant
campaign is shown in the table below.
Table 3: Pilot plant schedule for 2009
June Order equipment
July Equipment and instrument procurement
August-September Begin equipment installation and skid construction
October Complete equipment installation
November Begin pilot plant campaign
December Complete campaign

After the completion of testing at SRP, long term testing of concentrated piperazine with real
flue gas is planned. The system will be operated for 3 to 6 months at a coal-fired facility. Some
possible locations include the Wilsonville pilot plant constructed by Southern Company and the
Tarong pilot unit being constructed by CSIRO in Australia.
Future Work
Additional bids will need to be obtained for the all of the major equipment and then ordered.
Designs for the gas-liquid separator will also need to be finalized and then go out for bids. Once
the dimensions for the high pressure pump, cross-exchanger, steam heaters and separators have
been finalized, layout and construction of the skid can begin.
References
Freeman SA, Dugas R, Van Wagener D, Nguyen T, Rochelle GT. "Carbon dioxide capture with
concentrated, aqueous piperazine". IJGGC; in press.

261
Degradation of Concentrated,
Aqueous Piperazine (PZ) in CO
2
Capture
June 17
t h
, 2009
5
t h
Trondhei m Conf erence on CCS
262
2
Degradation of Concentrated PZ for CO
2
Capture
Outline
Introduction to Conc. PZ
Why Concentrated PZ?
Degradation Methods and Materials
Results of Degradation Studies
Thermal Degradation
Oxidation
Conclusions
263
3
2
Capture
Why Concentrated PZ?
Concentrated PZ = 8 m PZ = 40 wt% PZ
Oxidatively stable with Fe
2+
, V
5+
, Ni
2+
, & Cr
3+
Thermally stable up to 150C
Stripper: up to 15 atm
1.7X the capacity of 7 m MEA
Similar volatility to 7 m MEA
2X the kinetic rate of 7 m MEA
5-15% less energy intensive than 7 m MEA
264
Degradation Methods
and Materials
265
5
2
Capture
Thermal Deg. Thermal Cylinders
1
/
2
inch ID 316 SS
tubes with
Swagelock
endcaps
10 mL volume
1
/
2
inch
5 inch
Area : Volume = 315 m
2
/m
3
Placed in forced convection ovens
Maintains seal under pressure when heated
Easily conduct multiple experiments
simultaneously high throughput
266
6
2
Capture
Oxidation Low Gas Flow Apparatus
0.5 L jacketed reactor
Agitation at 1400 rpm
Operates continuously
for 2 to 5 weeks
Accelerates expected
oxidation
Used to quickly get
degraded samples for
analysis
100 mL/min
98%O
2
-2%CO
2
1400 rpm
55C
Water
267
7
2
Capture
Analytical Methods
PZ conc. acid titration (total alkalinity)
CO
2
conc. total inorganic carbon (TIC)
Anion Ion Chromatography (IC)
Formate, acetate, oxalate, glycolate, sulfate, etc.
Cation IC
PZ, ethylenediamine (EDA), other amines
Mass Spectrometer (MS) (w/ Cation IC)
Atomic Absorption Spectrometry (AA) - metals
268
Results of Degradation
Studies
269
9
2
Capture
Thermal Deg: 8 m PZ, =0.3, 150C
PZ
Tot al Format e
Format e
EDA
0. 03 m
4. 9 mM
9
270
10
2
Capture
PZ and MEA Thermal Degradation
23 70X more
degradation
23 70X more
degradation
10
271
11
2
Capture
Thermal Screening (135C, =0.4)
[ Ami ne]
(m)
Ami ne Syst em
Ami ne Loss
(%/ week)
10 PZ Piperazine 0.3
7 AMP 2-amino-2-methyl-1-propanol 2.3
7 DGA Diglycolamine
2.3
7 HEP Hydroxyethyl PZ 3.3
7/2 MDEA/PZ Methyldiethanolamine/PZ 3.7
8 EDA Ethylenediamine 4.0
7 MEA Monoethanolamine 9.3
7 DETA Diethylenetriamine 23.5
(Davis, 2009)
272
12
2
Capture
Metal Catalyzed PZ Oxidation
Quantify metal catalyzed degradation and
inhibition in conc. PZ
Comparison of metals (all mM):
1.0 Fe
5.0 Cu
Stainless Steel Metals: 0.7 Cr + 0.3 Fe + 0.3 Ni
Inhibitor A (all mM)
1 Fe + 100 Inhibitor A
5 Cu + 0.1 Fe + 100 Inhibitor A
273
13
2
Capture
PZ Resists Oxidation over MEA (1)
Fe
MEA + Fe
(Sexton, 2008)
Reduction in oxidation
for concentrated PZ
for concentrated PZ
13
274
14
2
Capture
Fe
Cu
MEA + Fe +
Cu
MEA + Fe
(Sexton, 2008)
14
275
15
2
Capture
Fe
Cu
MEA + Cr +
Ni MEA + Fe +
Cu
MEA + Fe
Fe + Cr + Ni
8 m
PZ
7 m
MEA
(Sexton, 2008)
15
276
16
2
Capture
Inhibitor A Reduces PZ Oxidation
(Sexton, 2008)
due to Inhibitor A
due to Inhibitor A
16
277
17
2
Capture
Inhibitor A Reduces PZ Oxidation
(Sexton, 2008)
17
278
18
2
Capture
Conclusion
PZ demonstrates enhanced thermal
resistance
Stable up to 150C
Up to 70X less degradation compared to 7 m MEA
PZ oxidizes 3 to 5X slower than 7 m MEA
Inhibitor A is successful in reducing Fe and
Cu catalyzed degradation in conc. PZ
279
280
Thu Nguyen, Marcus Hilliard, Gary T. Rochelle
Luminant Carbon Management Program
The University of Texas at Austin, USA
June 17
th
, 2009
5
th
Trondheim Conference on CO
2
Capture, Transport, & Storage
Thermodynamics of CO
2
Capture:
Amine Volatility
281
Outline
Scope Amine Volatility
Background Why is Amine Volatility Important
Apparatus Hot Gas FTIR
Theory Apparent Activity Coefficient
Results 6 ppm 112 ppm
Conclusion MDEA Solvent is Least Volatile
282
Scope
Screen various amines for volatility
at 40C with nominal lean and rich loadings
7m MEA (monoethanolamine - industry baseline)
8m PZ (piperazine)
7m MDEA / 2m PZ (methyldiethanolamine / PZ)
12m EDA (ethylene diamine)
5m AMP (2-amino-2-methyl-1-propanol)
283
Why is Amine Volatility Important in CO
2
Capture?
(1) solvent loss into vapor phase --- capacity reduction, makeup amine cost
(2) cost to recapture lost amine --- capital & operating costs of water wash
(3) venting of residual amine into atmosphere --- environmental risks
Amine Volatility is an Important Screening
Criterion for New Amines
Volatility is Most Important at Lean Loading
Condition and 40C
Background
284
Amine Volatility: Stirred Reactor Coupled with FTIR Analyzer
5-10 L/min.
285
Modified Raoults Law for VLE
amine
= P
amine
/ [x
amine
* P
amine
sat
]
amine
is the apparent activity coefficient (key parameter)
Gibbs Helmholtz Relations
d (ln P
amine
) / d (1/T) = - H
vaporization
/ R
d (ln
i
) / d (1/T) = -H
solution
/ R
Theory
286
287
288
289
290
291
292
293
294
295
296
Conclusion
Loadings (mol CO
2
/mol tot alk) shown correspond to P
CO2
~500 Pa at 40C
297
0
50
100
150
200
250
300
350
80
90
100
110
120
130
0 0.5 1 1.5 2
t

(
s
)
F

(
1
0
-
3
m
2
s
)
[Fe
2+
] (mM)
Foaming Behavior of Piperazine
Aqueous Solutions for CO
2
Capture
Xi Chen, Stephanie Freeman and Gary Rochelle
Department of Chemical Engineering, The University of Texas at Austin, USA
Background
Solvent foaming is the #1 operational
problem in natural gas processing and
refinery sweetening processes.
Capacity reduction
Premature flooding
Solvent losses
Downstream process damage
Advantages of Piperazine over MEA
as a solvent for CO
2
capture
Experimental Apparatus
Less Packing
Richer Solutions
Lower Regeneration
Energy
Requirements
Lower Flow Rates
Lower sensible heat
requirements
Smaller Exchangers,
Pumps
Results
Conclusions
Oxidized PZ foamed more than
undegraded PZ.
Inhibitor A effectively mitigates foaming.
Formaldehyde significantly increases
foaming.
Silicone antifoam at 1ppm greatly
reduces foaming.
Fe
2+
increases foaming of PZ by 40%.
Fe
3+
, corrosion & oxidation inhibitors &
formic acid have little effects on PZ.
foaming.
Acknowledgements
The authors would like to thank the Luminant
Carbon Management Program for supporting
this project.
NH HN
Piperazine
Flowmeter
To
atmosphere
Nitrogen (335
ml/min)
Adapted from Standard Test Method for Foaming
Characteristics of Lubricating Oils (ASTM D892)
40C
400ml test
solution
Foaming tendency:
Foaminess coefficient
Foam stability:
Foam break time: t (sec)
G
V
G
V V
F
g
t
0
8m PZ, =0.3, 40 C
F*=F/F
0
t (s)
FeSO
4
HCHO
Antifoam
(Dow corning, Q2-
3183A)
0 0 0 1.00 34
1.5mM 0 0 1.32 >300
1.5mM 0 1ppm 0.09 <2
0 270mM 0 3.33 N/A
0 270mM 1ppm 0.18 20
0 270mM 2ppm 0.12 <8
Greater CO
2
Capacity
Faster Rates
Causes of foaming
Contaminants w/feed gas & makeup H
2
O.
Hydrocarbon
Suspended fine solid particles
Water-soluble surfactants & additives
Degradation products of alkanolamine.
Carboxylic acids
Heat stable salts
Product of formaldehyde & Piperazine
0
5
10
15
20
25
30
35
0
20
40
60
80
100
0 2 4 6 8 10
t

(
s
)
F

(
1
0
-
3
m
2
s
)
Amine (molal)
MEA
= 0.4
PZ
= 0.3
Formaldehyde
0
10
20
30
40
50
60
0
50
100
150
200
250
300
350
0 100 200 300
t

(
s
)
F

(
1
0
-
3
m
2
s
)
[Formaldehyde] (mM)
8m PZ
= 0.3
40 C
Antifoam
8m PZ, =0.3, 40 C F*=F/F
0
t (s)
+CuSO
4
(5mM) 0.98 30
+CuSO
4
(5mM) + A (100mM) 0.85 33
+CuSO
4
(5mM) + A (100mM)+
FeSO
4
(0.1mM)
0.90 35
+NaVO
3
(10mM)+
FeSO
4
(0.1mM)
0.77 28
+Formic acid (500mM) 1.06 30
Exp.
Solution Additives
Degradation
Time (hrs) F (m
2
/s)
8m PZ None 0 86
8m PZ 1 mMFe
2+
70 85
8m PZ 1 mMFe
2+
162 >>300
8m PZ
1 mMFe
2+
+
100 mMA 70 92
8m PZ
1 mMFe
2+
+
100 mMA 162 68
8m PZ
= 0.3
40 C
Corrosion & Oxidation Inhibitors
Amine Concentration
Degradation Inhibitor A
Ferrous Ion
NH N H O
+
N N
n
Degradation products Analysis
HO
NH
2
Monoethanolamine
25
27
29
31
33
35
37
39
41
72
74
76
78
80
82
84
86
88
0 0.2 0.4 0.6 0.8 1 1.2
t

(
s
)
F

(
1
0
-
3
m
2
s
)
[Fe
3+
] (mM)
8m PZ
= 0.3
40 C
Ferric Ion
6
6.5
7
7.5
8
8.5
9
0 50 100 150 200
P
Z

C
o
n
c
.

(
m
)

Exp. Time (hr)
1mM Fe
2+
+A
1mM Fe
2+
298
Jorge M. Plaza, Eric Chen, Gary T. Rochelle
Department of Chemical Engineering, The University of Texas at Austin
MODEL DESCRIPTION
SINGLE INTERCOOLING K
+
/PZ
BACKGROUND
Solvent VLE Kinetics
9m MEA Hilliard
Regression Aboudheir using
Laminar jet model in Aspen
4.5m K
+
4.5 m PZ
Hilliard
Modified Cullinane kinetics to
convert into activity based
Rigorous models in Aspen Plus
RateSep
TM
2 MEA + CO
2
MEAH
+
+ MEACOO
-
MEA + CO
2
+ H
2
O HCO
3
-
+ MEAH
+
4.5 m/4.5 m K
+
/PZ
9.8 m diameter
20 m CMR#2
5% hold up
Flue gas
5.49 kmol/s
40
o
C
12% CO
2
Lean flue gas
90% Removal
Lean Solvent
Variable loading
Rich Solvent
0.44
0.48
0.52
0.15 0.25 0.35 0.45
R
i
c
h

l
o
a
d
i
n
g

Lean loading
No Intercooling
Single Intercooling
Critical
L/G
PZ + CO
2
+ bPZCOO
-
+bH
+
PZCOO
-
+ CO
2
+ b PZ(COO
-
)
2
+ bH
+
CO
2
+ OH
-
HCO
3
-
CO
2
+b+H
2
O bH
+
+HCO
3
-
0
0.04
0.08
0.12
40
50
60
70
0.0 0.2 0.4 0.6 0.8 1.0
C
O
2
A
b
s
o
r
p
t
i
o
n

r
a
t
e

(
k
m
o
l
/
s
)
T

(
o
C
)
Z/Z
Total
CO
2
rate
Liquid T
Vapor T
Intercooling
Top Bottom
0
0.04
0.08
0.12
40
50
60
70
0.0 0.2 0.4 0.6 0.8 1.0
C
O
2
A
b
s
o
r
p
t
i
o
n

r
a
t
e

(
k
m
o
l
/
s
)
T
e
m
p
e
r
a
t
u
r
e

(
o
C
)
Z/Z
Total
CO
2
rate
Liquid T
Vapor T
Top Bottom
At the critical L/G the T bulge matches the
mass transfer pinch
Intercooling reduces absorber T & breaks
the pinch improving performance
CONCLUSIONS
The (L/G)
c
is predicted using energy
balances around the absorber and
between the T
bulge
assuming:
1. The energy due to CO
2
absorption
and vaporization of water is included
in the outlet gas enthalpy
2. T
bulge
is approximated using the case
in which all heat leaves with gas.
3. Water content of exiting gas is in
equilibrium with lean solvent
4. Change in liquid flow rate across the
column is neglected
5. CO
2
content at T
bulge
based on
results at various removals
Intercooling is commonly used in absorbers
Useful with high the heats of absorption
resulting in a T bulge affecting the P
vap
of
the dissolved species.
(L/G
i
)
c
is the critical ratio of liquid to inert gas
species.
Y is the fraction of H
2
O and CO
2
to inert species in
the gas stream (n
CO2
/G
i
, n
H2O
/G
i
)
T
b
is the bulge temperature
habs
Tb
is the CO2 heat of absorption at T
b
.
hvap
Tb
is the heat of vaporization of water at T
b
Cp
G
out
is the gas heat capacity at outlet conditions.
Cp
L
in is the liquid heat capacity at inlet conditions
R is the desired removal.
Removal at the bulge (R
b
) calculated
from the following empirical correlation:
R
b
= 1.47 R 0.70
Pack
H
(m)
R
(%)
(L/G)
c
T
b
(
o
C) y
out
H2O
T
out
G
(
o
C)
Aspen Appx Aspen Appx Aspen Appx Aspen Appx
5 90 3.9 4.0 69 68 0.09 0.06 46 40
10 90 4.1 4.0 72 68 0.10 0.06 46 40
20 90 4.1 4.0 72 68 0.09 0.06 46 40
10 80 3.8 4.1 69 66 0.10 0.06 48 40
10 60 3.5 4.5 61 62 0.09 0.06 47 40
Optimized loadings and flow
defined by stripper
11.4 m diameter
Flexipac 1Y
80% flooding
Variable height
Flue gas
6.1 kmol/s
40
o
C
13.3% CO
2
Lean flue gas
90% Removal
Lean Solvent
40
o
C
(57.6 kmol/s)
0.4 loading
Rich Solvent
0.495 loading
0
0.04
0.08
0.12
0.16
40
45
50
55
0 0.2 0.4 0.6 0.8 1
C
O
2
A
b
s
o
r
p
t
i
o
n

r
a
t
e

(
k
m
o
l
/
s
)
T

(
o
C
)
Z/Z
Total
Vapor T
Liquid T
CO
2
rate
Top
Bottom
15 m packing
84.7% CO
2
removal
Rich ldg = 0.489
0
0.04
0.08
0.12
0.16
40
45
50
55
0 0.2 0.4 0.6 0.8 1
C
O
2
A
b
s
o
r
p
t
i
o
n

r
a
t
e

(
k
m
o
l
/
s
)
T

(
o
C
)
Z/Z
Total
Liquid T
Vapor T
CO
2
rate
Bottom Top
Intercooling 5.16 m packing
90% CO
2
removal
Rich ldg = 0.495
Removal Intercooling
Packing
Height (m)
90%
None Infeasible
Mid column 6.07
Optimized 5.16
At (L/G)
c
intercooling improves
solvent capacity up to 45%
The model (L/G)
c
is within 10%
for removal >80% and for T
b
within 4
o
C for K
+
/PZ
For 9m MEA Intercooling
increased performance allowing
90% removal.
Optimum placement of the
intercooled stage can reduce
packing height by 13%.
C.W.
C.W.
SINGLE INTERCOOLING
MEA
b = PZ, PZCOO
-
,CO
3
-
,H
2
O, OH
-
299

Rochelle Q2 Report 2009

Cargado por

Información del documento

Descripción original:

Título original

Derechos de autor

Formatos disponibles

Compartir este documento

Compartir o incrustar documentos

Opciones para compartir

¿Le pareció útil este documento?

¿Este contenido es inapropiado?

Copyright:

Formatos disponibles

Rochelle Q2 Report 2009

Cargado por

Copyright:

Formatos disponibles

1

data by Dugas (Rochelle et al., 2008a) and previous pilot

can be assumed as the overall mass transfer coefficient

), N-Methyl-1,3-Propanediamine, and piperazine derivatives will be tested

predictions from the PZ model corrected for

to accompany the accurate VLE and heat capacity. For

for heat capacity corrected model (lines).

were not straightforward. They often

predictions of solvent properties to closely follow

was regressed to better fit recent VLE, heat

will be further evaluated to determine if it is ready for use in

data by Dugas (Rochelle et al., 2008a) and previous pilot

. It was implemented in Aspen Plus

using a FORTRAN subroutine. The obtained

for variable amine

data for 5 m PZ. Lines are k

data for 8 m PZ. Lines are k

will be used for this purpose. Additional absorber configurations

, by Autoclave Engineers) was used. Its designed operating

). The agitator is good for both liquid and

DP15) was connected to a pressure indicator (Validyne

) did not degrade to produce NH

was run at the design basis of a pilot plant

model points (2008). The equation

to do multivariable optimization to optimize the lean

currently has a rate-based MDEA model that uses

models predictions, but

MDEA model were the cation/anion pair parameters for the

model was developed in the most recent version of Aspen Plus

v2006.5, some of these parameters (i.e.

, at which point VLQKIJ will be

equilibrium constant model to a

can be assumed as the overall mass transfer coefficient K

>>1 (Tsai, 2008), so the 1+ can be ignored. Then the

También podría gustarte