Novel Shortcut Estimation Method For Regeneration Energy of Amine Solvents in An Absorption-Based Carbon Capture Process

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Article

pubs.acs.org/est

Novel Shortcut Estimation Method for Regeneration Energy of


Amine Solvents in an Absorption-Based Carbon Capture Process
Huiyong Kim, Sung June Hwang, and Kwang Soon Lee*
Department of Chemical and Biomolecular Engineering, Sogang University, 35 Baekbeom-ro, Mapo-gu, Seoul 121-742, Korea

ABSTRACT: Among various CO2 capture processes, the


aqueous amine-based absorption process is considered the
most promising for near-term deployment. However, the
performance evaluation of newly developed solvents still
requires complex and time-consuming procedures, such as
Downloaded by INST OF CHEMICAL TECHNOLOGY at 02:51:08:164 on May 24, 2019

pilot plant tests or the development of a rigorous simulator.


Absence of accurate and simple calculation methods for the
energy performance at an early stage of process development has
lengthened and increased expense of the development of
economically feasible CO2 capture processes. In this paper, a
novel but simple method to reliably calculate the regeneration
from https://pubs.acs.org/doi/10.1021/es504684x.

energy in a standard amine-based carbon capture process is


proposed. Careful examination of stripper behaviors and
exploitation of energy balance equations around the stripper
allowed for calculation of the regeneration energy using only vapor−liquid equilibrium and caloric data. Reliability of the
proposed method was confirmed by comparing to rigorous simulations for two well-known solvents, monoethanolamine (MEA)
and piperazine (PZ). The proposed method can predict the regeneration energy at various operating conditions with greater
simplicity, greater speed, and higher accuracy than those proposed in previous studies. This enables faster and more precise
screening of various solvents and faster optimization of process variables and can eventually accelerate the development of
economically deployable CO2 capture processes.

1. INTRODUCTION used in the industry.5,9−11 However, the immediate deployment


Preventing anthropogenic climate change by reducing of the current absorption-based technology has been delayed
emissions of CO2 without decreasing fossil fuel usage is a because of the large amount of energy required to run the
critical issue worldwide.1,2 Global energy usage is continuously process,5,10,12 while a lot of research is being carried out to
increasing, and the International Energy Agency (IEA) predicts develop advanced energy-efficient solvents and processes. The
that the reliance on fossil fuels will account for 80% of the adoption of any of the present absorption-based capture
energy supply in 2050.3 With the increase in the consumption processes in a power plant is expected to result in an energy
of fossil fuels, greenhouse gas emissions, which are thought to penalty of 20−30%,13−15 where the energy is consumed mostly
bring about severe climate change, are also expected to increase. for regeneration of absorbed CO2.
Efforts to solve this problem have led to numerous initiatives With various absorbent candidates being competitively
concerning the reduction of CO2 emissions. developed in many institutions worldwide, the performance
Carbon capture and sequestration (CCS) technology is (i.e., total required work including the regeneration energy)
considered to be a promising means of addressing the CO2 evaluation of developed solvents remains time-consuming and
emission problem. This technology includes the capture (and challenging. Typically, candidate solvents are pre-screened by
compression) of CO2 from fixed-point sources, its trans- comparing the absorption rate, absorption capacity, heat of
portation, and its storage. According to the IEA Blue Map absorption, degradation, viscosity, solvent cost, etc.16,17
Scenario, the amount of CO2 emitted in 2050 could be reduced However, the energy performance is not evaluated quantita-
by 48 billion tons from the projected total CO2 emissions tively at this stage, and only solvent characteristics are listed
through the use of all possible efforts, and CCS is predicted to without having a clear selection guideline for the best candidate
be responsible for 19% of this reduction.3 solvent. After pre-screening, pilot plant tests and rigorous
A large portion of carbon capture research is focused on modeling studies are usually conducted to observe solvent
absorption-based post-combustion CO2 capture technol- performance. Pilot plant testing is essential in the process
ogy.2,4−8 Other technologies, such as adsorption, solid
sorbent-based absorption, membrane separation, oxy-fuel Received: September 24, 2014
combustion, and pre-combustion capture, are not expected Revised: December 9, 2014
for near-term use because of their technical immaturity Accepted: December 11, 2014
compared to the absorption process, which has long been Published: January 20, 2015

© 2015 American Chemical Society 1478 DOI: 10.1021/es504684x


Environ. Sci. Technol. 2015, 49, 1478−1485
Environmental Science & Technology Article

development but is labor-intensive and costly. In addition, assumptions, a reliable estimation of Qregen was enabled using
further scale-up is required to obtain an accurate energy only VLE and caloric data. Calculated values of Qregen were
assessment because of the heat loss problem in small-scale compared to those from rigorous simulations for two well-
plants. Rigorous modeling is another labor-intensive job, studied solvents, MEA and piperazine (PZ), at two different
demanding a lot of experimental work for thermodynamic stripper operating pressures to verify the pertinence of the
and kinetic modeling. Thus, developing a method that can proposed method.
reliably estimate the regeneration energy at the pre-screening In the remainder of the paper, section 2 describes the
stage can reduce trial and error and act as an important agent standard CO2 absorption process and stripper behaviors,
that can accelerate carbon capture research. section 3 describes the derivation of the proposed shortcut
The total required work of a CO2 capture process is method, and section 4 demonstrates the performance of the
composed of the regeneration energy (i.e., reboiler duty of the shortcut method by applying it to four different cases.
stripper), compression energy to 150 bar, solvent pumping
energy, and flue gas blowing energy, as in eq 1.18 The 2. ABSORPTION PROCESS AND STRIPPER BEHAVIORS
regeneration energy, Qregen, can be further decomposed into Before introduction of the proposed shortcut method, stripper
energies required for solvent heating (sensible heat, Qsens), behavior is investigated in this section. For the base case study,
water and amine evaporation (latent heat, Qlatent), and CO2 an absorption process with a typical absorber−stripper
desorption (heat of reaction, Qrxn), as shown in eq 219 and also configuration is assumed for 30 wt % MEA solvent, and
in Figure 2a. In this study, we restricted ourselves to amines stripper behavior of the process is studied by rigorous
that have much lower volatility than water, and the amine simulation using Aspen Plus.23 Because Qregen is provided at
evaporation was neglected. the reboiler of the stripper, understanding the stripper behavior
is crucial, and only the stripper is studied. Thermal energy
Wtotal = Wcomp + Wpump + Wblower + 0.75Q regen balance of the stripper discussed in section 2.3 shows that
Treboiler + 10 − Tambient knowing the top-stage temperature of the stripper is the key to
(GJelec /tCO2 ) estimating Qregen, and stripper behavior investigated in section
Treboiler + 10 (1)
2.4 reveals that the top-stage temperature can be easily
Q regen = Q sens + Q latent + Q rxn (GJth /tCO2 ) obtained.
(2) 2.1. Standard Carbon Capture Process. Figure 1 shows a
Individual energy terms in eqs 1 and 2 can be easily calculated, standard configuration of the absorption-based carbon capture
except Qlatent, the latent heat for water vaporization. Required
works for compression, pumping, and blowing can be easily and
reliably estimated from the respective performance equations.
Among the three energy terms comprising Qregen, Qrxn can be
predicted quite accurately using the Gibbs−Helmholtz
equation19,20 and a small number of vapor−liquid equilibrium
(VLE) data. Consistency of evaluating Qrxn using the Gibbs−
Helmholtz equation with VLE data was confirmed by Mathias
and O’Connell.20 Qsens can also be estimated from a simple
enthalpy balance relation. The most troublesome part of the
energy assessment is the latent heat for water evaporation,
Qlatent.
There have been some studies dedicated to the estimation of
Qlatent. Rochelle et al. proposed a method to calculate Qlatent Figure 1. Typical configuration of the absorption-based carbon
when regeneration is conducted in a flash drum instead of a capture process.
stripper.21 Notz et al. proposed a shortcut method to estimate
Qregen in a stripper using a modified Kremser equation.22 This process, which is most widely employed in industrial processes.
method can estimate Qregen for various solvent flow rates when The main equipment used in Figure 1 includes the absorber,
VLE data at absorber and stripper temperatures and caloric data the stripper, and the heat exchanger. The cool CO2-rich solvent
are available. The method can yield reasonable results in limited stream (RICHOUT) exits from the bottom of the absorber,
situations only because of underlying assumptions, such as gaining energy by exchanging heat at the heat exchanger with
constant temperature and vapor flow within the stripper. As the hot CO2-lean solvent stream (LEANOUT) exiting from the
demonstrated in Figures 2b and 3a for a monoethanolamine bottom of the stripper. As a result of the heat exchange, CO2
(MEA) case, the maximum temperature difference within the and water in the RICHIN stream, which enter the stripper at
stripper is greater than 20 °C. The vapor flow rate also varies the top stage, are partially vaporized. The minimum temper-
significantly along the stripper column. Such discrepancies ature approach (ΔTMTA) is formed at the RICHOUT side of
between assumed idealities and practical situations have limited the heat exchanger because of a larger mass flow rate of the
the dependability of the Qregen prediction. RICHOUT stream than that of the LEANOUT stream. The
On the basis of the considerations mentioned above, the liquid portion in the RICHIN stream flows downward,
purpose of this research is to propose a novel shortcut method transferring CO2 to and receiving water from the upward
to estimate Qregen for an absorption-based CO2 capture process vapor flow inside the stripper. The CO2-stripped LEANOUT
using a stripper. An examination of stripper characteristics stream is withdrawn from the reboiler.
revealed that the stripper operation can be separated into two 2.2. Simulation Conditions for a MEA-Based Process.
regions. With the introduction of suitable energy balance Some basic assumptions and simulation conditions for the
equations for each region and some simplified but reasonable rigorous simulation of the MEA-based process are described in
1479 DOI: 10.1021/es504684x
Environ. Sci. Technol. 2015, 49, 1478−1485
Environmental Science & Technology Article

this subsection. Numbers are assigned to process streams If the heat of reaction, Δhrxn (kJ/mol of CO2), is a function of
around the stripper in Figure 1, in addition to names used to CO2 loading and the reference states associated with the four
simplify the designation of the associated heat and mass terms. streams in eq 4 are unloaded liquid solvent, gaseous CO2, and
Some basic assumptions of the process for simulation include liquid water at TR and 1 bar, the enthalpy terms would be
the following: (i) The flue gas contains 15% CO2, saturated reasonably expressed as follows:
water vapor at 40 °C, and N2 for balance. It enters the process
at a flow rate of 510 kg/h or equivalently 17.1 kmol/h. (ii) The H1 = (cam + rwc w + αrichcCO2)(T1 − TR )
solvent is an aqueous solution of 30 wt % MEA. (iii) Rate-based rich
simulation is used for the absorber, and VLE-based simulation + ∫0 Δhrxn(α) dα
with 20 equilibrium stages is applied to the stripper. The
operating pressures for the absorber and stripper are chosen as H2 = (cam + rwc w + αleancCO2)(T2 − TR )
1 and 2 bar, respectively. (iv) The temperature of the flue gas, lean
the LEANIN stream, and the condensate recycle stream of the + ∫0 Δhrxn(α) dα
stripper is 40 °C. (v) The CO2 recovery rate (percentage of
CO2 removed from the flue gas stream) is 90%, which H3 = (McCO2(Ttop − TR ) + m w c w(Ttop − TR ) + m w λ)
corresponds to a CO2 capture rate of 100 kg/h. (vi) ΔTMTA for
/(M + m w )
the main heat exchanger is 10 °C.
The assumptions above are typical, and the resulting stripper H4 = c w(T4 − TR )
behaviors would be generally applied to amine-based carbon (5)
capture processes. Rate and equilibrium models and data
provided in Aspen Plus were used without modifications. where λ (kJ/mol) denotes the latent heat of water and Ttop =
The amount of CO2 absorbed in the RICHOUT stream T3. Note that the heat capacity, c, is molar-based, having the
(rich loading) and the LEANIN stream (lean loading) can have unit of kJ mol−1 K−1. It is assumed that c for a species has the
various values depending upon equipment design, operating same value for both gas and liquid states and c for a mixture has
parameters, and solvent properties. In this study, rich loading a molar-weighted average value of pure species, regardless of
was fixed at 0.54 (mol of CO2/mol of amine), and lean loading the existence of reaction products inside the mixture.
was varied over a range. The chosen value of rich loading Substitution of eq 5 into eq 4 yields the following:
corresponds to CO2 vapor pressure of 5 kPa at 40 °C. Because Q regen = Q sen + Q rxn + Q latent
the CO2 partial pressure in the flue gas is 15 kPa (15% CO2 in
the 101 kPa flue gas), setting the CO2 vapor pressure of the
Q sen = mam(cam + rwc w + αleancCO2)(T2 − T1)
solvent at 5 kPa can offer a reasonable driving force for mass
transfer from the flue gas to the MEA solvent.18,19 The rich + McCO2(Ttop − T1) + m w c w(Ttop − T4)
loading value of 0.54 was maintained during simulation by rich
adjusting the absorber height and the amount of interstage
cooling. The CO2 recovery was maintained at 90% by adjusting
Q rxn = −mam ∫lean Δhrxn(α) dα
the solvent circulation rate. Q latent = m w λ (6)
2.3. Thermal Energy Balance around the Stripper.
Equations for Qregen computation can be derived from the It is interesting to note that Qlatent is determined by the amount
energy balance for the control surface around the stripper of water content at the top of the stripper, mw, instead of the
shown in Figure 1. It is convenient to write the energy terms amount of water boil-up at the reboiler.
based on molar amount of unloaded solvent wherever In eq 6, T2 − T1 is set to have ΔTMTA because it minimizes
applicable. For this, let us define mam (mol/tCO2) as the Qsen and all the other terms, except Ttop and mw, are known or
molar amount of amines in the circulating solvent streams per can be pre-specified. The value of Δhrxn can be obtained
tonne of CO2 product. In addition, let us define rw as the molar experimentally using an instrument, such as the reaction
ratio of water to amines in the unloaded solvent. Also, let us calorimeter, or using VLE data and the following Gibbs−
define M and mw as the molar amount of CO2 and the molar Helmholtz equation:
amount of water vapor in stream 3 per tonne of CO2 product,
respectively. Obviously, M = 2.2727 × 104 mol of CO2 (1 tonne Δhrxn ⎡ ∂ ln P* ⎤
=⎢ ⎥
CO2
of CO2). Finally, let α be the CO2 loading defined as moles of −
CO2 per mole of amines. Then, the following holds:
R ⎣⎢ ∂(1/T ) ⎥⎦α (7)

m1 = mam(1 + rw + αrich), m2 = mam(1 + rw + αlean) where P*CO2(T, α) is the CO2 vapor pressure.
m3 = M + m w , m4 ≈ m w The value of mw can be estimated using the pressure
relationship. The column pressure at the top is the sum of the
(3) * 2(Ttop, αtop), and the partial pressure of
CO2 vapor pressure, PCO
Let us denote Hi (kJ/mol) as the enthalpy of the ith stream per water.
mole of amines for i = 1 and 2 and per mole of the stream itself
for i = 3 and 4. According to the energy balance around the * (Ttop , αtop) + Pw(Ttop)
Ptotal = PCO (8)
2
control surface, Qregen can be represented as follows:
When P*CO2(Ttop, αtop) and Pw(Ttop) are known, mw can be
Q regen = mamH2 + (M + m w )H3 − mamH1 − m w H4 computed as follows:

(GJ/tCO2 ) (4) * (Ttop , αtop)


m w = MPw(Ttop)/PCO (9)
2

1480 DOI: 10.1021/es504684x


Environ. Sci. Technol. 2015, 49, 1478−1485
Environmental Science & Technology Article

Pw(Ttop) can be approximated by Raoult’s law, xw,topPvap w (Ttop), (region A), while they begin to deviate significantly as αlean
where xw is the mole fraction of water, because xw is high (>0.8) becomes smaller than 0.2 (region B). In region A, the flow rate
for most aqueous amine solvents. It is observed from simulation and enthalpy of the RICHIN stream dominate those of the
studies that xw inside the stripper remains nearly constant other incoming streams to the top stage (condensed water and
around the mole fraction of water in the unloaded fresh solvent, upward vapor from the lower stages) and Ttop inevitably follows
xw,fresh, irrespective of the stage number, especially when the TRICHIN. In contrast, in region B, a larger amount of thermal
bicarbonate formation is not excessive. On the basis of this energy is provided to the reboiler to discharge CO2 by
observation, eq 8 can be rewritten as follows: increasing the reboiler temperature and the upward vapor
* (Ttop , αtop) + x w,freshPwvap(Ttop) stream has a stronger effect on the top stage than in region A.
Ptotal = PCO2 (10) Consequently, Ttop of region B is determined to be higher than
which has two unknowns, Ttop and αtop. Because Ptotal is given, TRICHIN, and the difference becomes more severe as αlean is
αtop can be obtained from eq 10 if Ttop is known and VLE data reduced further. The stripper behavior demonstrates that Ttop,
for P*CO2(T, α) is available. This enables the computation of mw the important variable for estimation of Qlatent, can be simply
approximated by TRICHIN for region A.
in eq 9 and then Qregen from eq 6. Figure 3 illustrates the temperature, CO2 loading, and vapor
Additionally, when Δhrxn is given as a function of α only, mole fraction profiles inside the stripper to better comprehend
* 2(Ttop, αtop) can be expressed, as in eq 11, by integrating the
PCO
Gibbs−Helmholtz equation using VLE data at only 313 K.
* (Ttop , αtop) = PCO
PCO * (313, αtop)
2 2

⎛ ⎞
Δh ⎛ 1 1 ⎞⎟⎟
exp⎜⎜ − rxn ⎜⎜ − ⎟
⎝ R ⎝ Ttop 313 ⎠⎟⎠ (11)
In short, knowing one unknown, Ttop, is sufficient for
calculating Qlatent and then Qregen. Examining stripper behaviors
can provide a clue to estimate the unknown variable Ttop.
2.4. Stripper Behaviors. Stripper behaviors were inves-
tigated for different values of αlean, from 0.15 to 0.375, while T1
and αrich from the absorber were fixed at 40 °C and 0.54,
respectively. Three energy terms and Qregen are shown in Figure Figure 3. Profiles of (a) temperature, (b) CO2 loading in the liquid
2a as αlean varies. Qrxn remains nearly constant over the phase, and (c) mole fractions of CO2 and water in the vapor phase
inside the stripper for selected values of αlean.

that region A is governed by the RICHIN stream while region


B is governed by the vapor flow from the reboiler. Centering
around αlean = 0.2, the temperature, CO2 loading in the liquid
phase, and vapor mole fractions of CO2 and water at the
bottom stage (reboiler) continue to a near-top stage when αlean
< 0.2, and the reverse holds when αlean > 0.2. The column
profiles are best balanced, and minimum Qregen is achieved,
when αlean = 0.2.

3. SHORTCUT ESTIMATION METHOD FOR


REGENERATION ENERGY
In the previous section, it was revealed that the stripper behaves
differently in high and low lean loading regions. On the basis of
this observation, the shortcut estimation method for Qregen is
developed separately for each region.
3.1. Shortcut Estimation Method for Region A. For
region A, Ttop ≅ TRICHIN = T5 and accurate prediction of T5 is
the key to the estimation of Qregen. T5 can be easily obtained
Figure 2. (a) Individual energy terms contributing to Qregen and (b) from the energy balance around the main heat exchanger
comparison of Ttop with TRICHIN from rigorous simulation for different shown in Figure 4a. The RICHIN stream is in a partially
values of αlean when T1 = 40 °C and αrich = 0.5383. vaporized state and can be expressed into vapor and liquid
streams.
considered αlean range. Qsens increases with αlean as the solvent The energy balance equation can be written as
circulation rate is increased to achieve 90% CO2 recovery. Qlatent mam(H1 − H2 − H5,liq − H5,vap + H6) = 0 (12)
increases sharply as αlean is lowered below 0.2 because of the
increased boiling in the reboiler. Qregen is strongly affected by where H1 and H2 are as given in eq 5. Assuming that the vapor
Qlatent and shown to have a minimum at αlean = 0.2. portion of the RICHIN stream is composed of only CO2 and
In Figure 2b, Ttop is compared to TRICHIN or equivalently T5. water vapor, the remaining three enthalpy terms are expressed
Ttop and TRICHIN agree quite closely in the region of αlean > 0.2 as follows:
1481 DOI: 10.1021/es504684x
Environ. Sci. Technol. 2015, 49, 1478−1485
Environmental Science & Technology Article

Figure 5. (a) Temperature estimates and (b) estimates of the energy


terms by the rigorous simulation and the shortcut method for region A
(RIG, rigorous simulator; SC, shortcut method).
Figure 4. (a) Streams around the main heat exchanger and (b) energy
balance boundary for region B.

H5,liq = (cam + rw,5lc w + α5lcCO2)(T5 − TR ) from the proposed shortcut method for region A. Three energy
α5l terms that comprise Qregen calculated by the shortcut method
+ ∫0 Δhrxn(α) dα are also displayed in Figure 5b together with Qregen from the
rigorous simulation. Quite satisfactory agreement of the
H5,vap = ((rw − rw,5l)c w + (αrich − α5l)cCO2)(T5 − TR ) reboiler and RICHIN temperatures and the regeneration
+ (rw − rw,5l)λ energy from the rigorous simulation and the shortcut method
manifests for region A.
H6 = (cam + rwc w + αleancCO2)(T6 − TR ) 3.2. Shortcut Estimation Method for Region B.
lean Accurate estimation of Qregen for region B is not important
+ ∫0 Δhrxn(α) dα because it is not a recommended operating region as a result of
high regeneration energy. Nevertheless, Qregen for region B,
(13) together with Qregen for region A, enables us to figure out the
Rearranging eq 12 using T2 − T1 = ΔTMTA and m5,vap = mam((rw overall variation of Qregen with respect to αlean and, especially, to
− rw,5) + (αrich − α5)) yields the following relationship: locate the region boundary, where Qregen is likely to have the
minimum value.
0 = (cam + rwc w )(ΔTMTA + T6 − T5) To estimate Qregen for region B, energy balance around the
reboiler as shown in Figure 4b is used with an additional
+ cCO2(αlean(T6 − T2) − αrich(T5 − T1)) assumption that the temperature difference between the
rich reboiler (stage 20) and the upper next stage (stage 19) is
− (rw − rw,5l)λ + ∫α Δhrxn(α) dα
(14) small, which can be observed from Figure 3a for αlean = 0.15.
This assumption is valid, because upward vapor flow with a
5l

Because the column pressure varies only slightly from top to high enthalpy content from the reboiler renders the temper-
bottom, the same pressure relationship as in eq 10 holds at the atures at stages above the reboiler similar.
reboiler with respect to αlean and T6. The control surface and streams around the reboiler are
defined as shown in Figure 4b. Then, Qregen can be represented
* (T6 , αlean) + x w,freshPwvap(T6)
Ptotal = PCO (15)
2 as follows:
T6 can be determined from eq 15. Consequently, eq 14 has Q regen = −mamH6 − m8H8 + mamH7 (18)
three unknowns, T5, rw,5l, and α5l. They can be uniquely
determined using the following two more relationships: where H6 is in eq 13 and
* (T5 , α5l) + x w,freshPwvap(T5)
Ptotal = PCO (16)
H7 = (cam + rwc w + α7cCO2)(T7 − TR )
2 α7

* (T5 , α5l)
PCO
+ ∫0 Δhrxn(α) dα
αrich − α5l 2
=
rw − rw,5l x w,freshPwvap(T5) (17) H8 = ((m w,R c w + mCO2,R cCO2)(T6 − TR ) + m w,R λ)/m8

Figure 5a compares temperatures at the reboiler, the top stage, m8 = m w,R + mCO2,R
and the RICHIN stream from the rigorous simulation to those (19)

1482 DOI: 10.1021/es504684x


Environ. Sci. Technol. 2015, 49, 1478−1485
Environmental Science & Technology Article

where mw,R and mCO2,R represent molar flow rates of water Figure 7 compares the estimates of regeneration energy from
vapor and CO2 gas from the reboiler. They can be estimated by the shortcut method and rigorous simulation under various
solving the following:
mCO2,R = mam(α7 − αlean)

mCO2,R * (T6 , αlean)


PCO2
=
m w,R x w,freshPwvap(T6) (20)
Finally, α7 can be computed from the VLE relationship if T7 is
known.
* (T7 , α7) + x w,freshPwvap(T7)
Ptotal = PCO (21)
2

In Figure 3a, we can see that T7 (at stage 19) is slightly lower
than T6 (at stage 20) for region B. The difference is
approximately a few degrees Celsius at best and becomes Figure 7. Estimates of regeneration energy of MEA and PZ systems
smaller as αlean becomes smaller. If T7 can be reasonably from the shortcut method and rigorous simulation under various
assumed, Qregen for region B can be estimated using eqs 18−21. operating conditions.
Figure 6 compares Qregen estimates for both regions. In the
shortcut estimation for region B, four different T6 − T7 values operating conditions. Estimates of the regeneration energy by
the shortcut method agree well with those from rigorous
simulation, especially in region A. Qregen in region B was
calculated for various values of T6 − T7, between 0 and 5 °C,
and the highest value at each αlean was selected.
For PZ, rigorous simulation was prone to fail in region B, and
the results could not be provided in Figure 7. VLE data for PZ
available in the literature are limited for αlean < 0.2, and the
associated VLE model can be unreliable in region B. This might
be the cause for the failure of simulation.
Three energy terms comprising the regeneration energy
calculated by the shortcut method and rigorous simulation are
compared in Figure 8. Because caloric data such as the heat of
Figure 6. Comparison of the estimate of regeneration energy from the
shortcut method with that from the rigorous simulation.

were assumed. The results for 0.3, 1.0, and 3.0 °C show
consistency in that αlean at the region boundary becomes
smaller, while the Qregen curve becomes steeper. The result for
5.0 °C of T6 − T7 is out of the consistency mentioned above.
This may indicate that a 5.0 °C temperature difference is
unrealistic for this system. On the basis of this observation, we
suggest a method to estimate Qregen such that Qregen is
computed for various values for T6 − T7 (between 0 and 5
°C, for example) simultaneously and select the highest value at
each αlean.

4. APPLICATIONS OF THE SHORTCUT METHOD TO


OTHER CASES
Reliability of the proposed shortcut method was confirmed by
applying the method to additional cases and comparing the Figure 8. Estimates of three energy terms comprising regeneration
results to those from rigorous simulations. As additional cases, energy from the shortcut method and rigorous simulation for (a) MEA
7 mol of MEA at 1 bar of stripper pressure and also 8 mol of at 2 bar, (b) MEA at 1 bar, (c) PZ at 5 bar, and (d) PZ at 2 bar.
piperazine (PZ) at 2 and 5 bar of stripper pressure were
considered. The proposed method is applicable to other
aqueous amine solvent systems as long as VLE and caloric data reaction and heat capacity are available or can be predicted
are available and volatility of amine is much lower than water. easily from VLE data, precise estimates of the sensible heat and
Rate and equilibrium models for PZ necessary for rigorous the heat of reaction can be performed without much difficulty.
simulation are provided in Aspen Plus and were used without Evaluation of the latent heat, the most troublesome part, could
modifications.24 The VLE information from Aspen Plus was also be reasonably performed using the proposed method, as
used for shortcut simulation. The rich loading corresponding to shown in Figure 8. Some deviation exists in the estimation of
CO2 vapor pressure of 5 kPa was assumed for all cases, while the latent heat between the two methods. Underlying
lean loading was varied. assumptions, such as m4 ≈ mw in eq 3, xw ≈ xw,fresh in eq 10,
1483 DOI: 10.1021/es504684x
Environ. Sci. Technol. 2015, 49, 1478−1485
Environmental Science & Technology Article

and non-volatility of amines, may have led to the uncertainty of


estimation.
■ AUTHOR INFORMATION
Corresponding Author
Reliable estimation of the latent heat by the shortcut method
in region A can be achieved when top and bottom temperatures *Telephone: +82-2-705-8477. Fax: +82-2-3272-0319. E-mail:
of the stripper are correctly predicted and top and RICHIN kslee@sogang.ac.kr.
temperatures agree well with each other, which is one of the Notes
key prerequisites of the shortcut method. Figure 9 exhibits the The authors declare no competing financial interest.


estimates of top, bottom, and RICHIN temperatures from the
two calculation methods and shows that the required ACKNOWLEDGMENTS
performances are satisfactorily fulfilled.
This work was supported by the Korea Carbon Cpature and
Sequestration Research and Development Center (KCRC)
Grant funded by the Korean Government (Ministry of
Education, Science and Technology, 2012-0008886) and the
Sogang University Research Grant (201214007).

■ REFERENCES
(1) Wang, M.; Lawal, A.; Stephenson, P.; Sidders, J.; Ramshaw, C.;
Yeung, H. Post-combustion CO2 capture with chemical absorption: A
state-of-the-art review. Chem. Eng. Res. Des. 2011, 89 (9), 1609−1624.
(2) Richadson, K.; Steffen, W.; Schellnhuber, H. J.; Alcamo, J.;
Barker, T.; Kammen, D. M.; Leemans, R.; Liverman, D.; Munasinghe,
M.; Osman-Elasha, B.; Stern, L. N.; Wæver, O. Synthesis Report:
Climate Change, Global Risks, Challenges & Decisions; IARU Group:
Copenhagen, Denmark, 2009.
(3) International Energy Agency (IEA). Energy Technology
PerspectivesScenarios & Strategies to 2050; IEA: Paris, France, 2010.
(4) Tobiesen, F. A.; Svendsen, H. F. Study of a modified amine-based
Figure 9. Prediction of top and bottom temperatures of the stripper regeneration unit. Ind. Eng. Chem. Res. 2006, 45 (8), 2489−2496.
and RICHIN temperature from the shortcut method and rigorous (5) Rochelle, G. T. Amine scrubbing for CO2 capture. Science 2009,
simulation for (a) MEA at 2 bar, (b) MEA at 1 bar, (c) PZ at 5 bar, 325 (5948), 1652−1654.
and (d) PZ at 2 bar. (6) Feron, P. H. M. The potential for improvement of the energy
performance of pulverized coal fired power stations with post-
combustion capture of carbon dioxide. Energy Procedia 2009, 1 (1),
1067−1074.
(7) Markewitz, P.; Kuckshinrichs, W.; Leitner, W.; Linssen, J.; Zapp,
Performance evaluation of a solvent is perhaps the most P.; Bongartz, R.; Schreiber, A.; Muller, T. E. Worldwide innovations in
important but a quite complex and time-consuming step in the the development of carbon capture technologies and the utilization of
development of an economically feasible CO2 capture process. CO2. Energy Environ. Sci. 2012, 5 (6), 7281−7305.
Through this study, a novel but simple method to reliably (8) Neveux, T.; Moullec, Y. L.; Corriou, J. P.; Farve, E. Modeling of
estimate the regeneration energy, Qregen, for an aqueous amine CO2 capture in amine solvents: Prediction of performance and insights
solvent in a typical absorption-based CO2 capture process with on limiting phenomena. Ind. Eng. Chem. Res. 2013, 52 (11), 4266−
a stripper is proposed. The method requires only VLE and 4279.
(9) Yang, H.; Xu, Z.; Fan, M.; Gupta, R.; Slimane, R. B.; Bland, A. E.;
caloric data at various CO2 loadings and temperatures and can
Wright, I. Progress in carbon dioxide separation and capture: A review.
provide estimates of Qregen at various conditions of lean loading, J. Environ. Sci. 2008, 20 (1), 14−27.
rich loading, and stripper pressure. The method was derived (10) MacDowell, N.; Florin, N.; Buchard, A.; Hallett, J.; Galindo, A.;
from energy balance equations around the stripper under some Jackson, G.; Adjiman, C. S.; Williams, C. K.; Shah, N.; Fennell, P. An
ideal assumptions plus some crucial approximations drawn overview of CO2 capture technologies. Energy Environ. Sci. 2010, 3
from careful examination of stripper characteristics. The (11), 1645−1669.
assumptions and approximations include non-volatility of (11) Cousins, A.; Wardhaugh, L. T.; Feron, P. H. M. A survey of
amines, two distinctive stripper behaviors depending upon process flow sheet modifications for energy efficient CO2 capture from
flue gases using chemical absorption. Int. J. Greenhouse Gas Control
the value of lean loading, coincidence of feed and top-stage
2011, 5 (4), 605−619.
temperatures, invariance of water content in the liquid phase, (12) Puxty, G.; Rowland, R.; Allport, A.; Yang, Q.; Bown, M.; Burns,
Raoult’s law for water, heat capacity of a mixture as a molar R.; Maeder, M.; Attalla, M. Carbon dioxide postcombustion capture: A
average heat capacity of pure components, etc. Reliability of the novel screening study of the carbon dioxide absorption performances
proposed shortcut method was verified by comparing it to the of 76 amines. Environ. Sci. Technol. 2009, 43 (16), 6427−6433.
results for MEA and PZ systems from rigorous simulation using (13) Freguia, S.; Rochelle, G. T. Modeling of CO2 capture by
Aspen Plus. aqueous monoethanolamine. AIChE J. 2003, 49 (7), 1676−1686.
It is believed that the proposed method enables more reliable (14) Rubin, E. S.; Chen, C.; Rao, A. B. Cost and performance of fossil
fuel power plants with CO2 capture and storage. Energy Policy 2007, 35
solvent screening with minimum experimental data, much faster
(9), 4444−4454.
“shortcut” calculation compared to rigorous process simu- (15) Moullec, Y. L.; Kanniche, M. Screening of flowsheet
lations, easier optimization of operating conditions, and modifications for an efficient monoethanolamine (MEA) based post-
eventually acceleration of the development of economically combustion CO2 capture. Int. J. Greenhouse Gas Control 2011, 5 (4),
feasible absorption-based CO2 capture processes. 727−740.

1484 DOI: 10.1021/es504684x


Environ. Sci. Technol. 2015, 49, 1478−1485
Environmental Science & Technology Article

(16) Davidson, R. M. Post-combustion Carbon Capture from Coal Fired


PlantsSolvent Scrubbing; International Energy Agency Clean Coal
Centre: London, U.K., 2007.
(17) Ma’mum, S.; Svendsen, H. F.; Hoff, K. A.; Juliussen, O.
Selection of new absorbents for carbon dioxide capture. Energy
Convers. Manage. 2007, 48 (1), 251−258.
(18) Oyenekan, B. A.; Rochelle, G. T. Energy performance of stripper
configurations for CO2 capture by aqueous amines. Ind. Eng. Chem.
Res. 2006, 45 (8), 2457−2464.
(19) Oexmann, J.; Kather, A. Minimising the regeneration heat duty
of post-combustion CO2 capture by wet chemical absorption: The
misguided focus on low heat of absorption solvents. Int. J. Greenhouse
Gas Control 2010, 4 (1), 36−43.
(20) Mathias, P. M.; O’Connell, J. P. The Gibbs−Helmholtz
equation and the thermodynamic consistency of chemical absorption
data. Ind. Eng. Chem. Res. 2012, 51 (13), 5090−5097.
(21) Rochelle, G. T.; Chen, E.; Freeman, S.; Wagener, D. V.; Xu, Q.;
Voice, A. Aqueous piperazine as the new standard for CO2 capture
technology. Chem. Eng. J. 2011, 171 (3), 725−733.
(22) Notz, R.; Tonnies, I.; Mangalapally, H. P.; Hoch, S.; Hasse, H. A
short-cut method for assessing absorbents for post-combustion carbon
dioxide capture. Int. J. Greenhouse Gas Control 2011, 5 (3), 413−421.
(23) Aspen Technology. Rate-Based Model of the CO2 Capture Process
by MEA Using Aspen Plus; Aspen Technology: Burlington, MA, 2008.
(24) Aspen Technology. Rate-Based Model of the CO2 Capture Process
by PZ Using Aspen Plus; Aspen Technology: Burlington, MA, 2008.

1485 DOI: 10.1021/es504684x


Environ. Sci. Technol. 2015, 49, 1478−1485

You might also like