Você está na página 1de 9

Chemical Papers 66 (11) 10101018 (2012)

DOI: 10.2478/s11696-012-0211-x

ORIGINAL PAPER

Design simulations for a biogas purification process using aqueous


amine solutions
Richard A. Gawel*
Department of Solid State Chemistry, Faculty of Materials Science and Ceramics, AGH University of Science and Technology,
Aleja Mickiewicza 30, 30-059, Krakow, Poland
Received 21 July 2011; Revised 11 April 2012; Accepted 18 April 2012

Using the simulation program CHEMCAD, performance characteristics, design optimization,


and costs of an absorption/stripping system used to purify 100 kg h1 of biogas in a biogas power plant were investigated. Potential absorbents used in the chemical absorption process
were the following aqueous solutions: pure diglycolamine, diglycolamine/piperazine, and diglycolamine/methyldiethanolamine/piperazine. Mixtures for agricultural biogas purification to below 1
vol. % of CO2 and 4 104 mass % of H2 S were determined via a simulation in the above mentioned program. The chosen mixtures were then entered into an absorption/desorption system and
simulations for each unit were provided by CHEMCAD. From the simulation results, the design
parameters were calculated and entered into each units cost estimation section in the aforementioned program to estimate the purchase costs of the apparatuses. Taking into account the
installation, maintenance, as well as other additional costs, the actual machine purchase costs were
multiplied by the Lang factor. Costs of additional streams were also calculated by multiplying the
ten-year utility losses by their respective cost factors. From these calculations, the absorbent mixture, allowing biogas production at the lowest estimated costs for ten years, was found.
c 2012 Institute of Chemistry, Slovak Academy of Sciences

Keywords: amine sweetening, biogas purification, chemical absorption, cost comparison

Introduction
Biogas is a promising alternative source of methane
(CH4 ) due to its availability and renewability. Unfortunately, high percentage of carbon dioxide (CO2 )
as well as traces of hydrogen sulde (H2 S) are also
found in this CH4 source. These compounds must be
removed to obtain biogas of acceptable quality. According to the Pacic Gas and Electric Company in
San Francisco, California, gas delivered for transportation must not contain more than 1 vol. % of CO2
and 4 104 mass % of H2 S. One of the most common methods of simultaneous CO2 and H2 S removal
is chemical absorption, i.e. penetration of gas into a
liquid phase, enhanced by the reaction of one or more
of the absorbed components with active component(s)
in the liquid phase. Currently, the most commonly
used active components for the above mentioned ab*Corresponding author, e-mail: ragaw@agh.edu.pl

sorption in industrial practice are amines such as:


monoethanolamine (MEA), diethanoloamine (DEA),
methyldiethanolamine (MDEA), and diglycolamine
(DGA), their behavior in the above mentioned chemical absorption process has been investigated (Savage
& Funk, 1981; Polasek & Bullin 1994; Salkuyeh & Mofarahi, 2011; Barreau et al., 2006). It has been determined that 5070 mass % of DGA aqueous solutions
are not only able to absorb CO2 , towards which it is
selective, but also capable of achieving the desired H2 S
specication (Polasek & Bullin 1994; Salkuyeh & Mofarahi, 2011). Unfortunately, high heat of the reaction
between DGA and CO2 , as well as that between DGA
and H2 S (Polasek & Bullin, 1994; Salkuyeh & Mofarahi, 2011) was also determined. On the other hand,
2050 mass % of MDEA aqueous solutions possess low
heat of the reaction compared to other amine solutions; also the selectivity towards the reaction with

R. A. Gawel/Chemical Papers 66 (11) 10101018 (2012)

H2 S (Savage & Funk, 1981; Polasek & Bullin 1994),


leaving large amounts of CO2 in the gaseous phase.
Using MDEA also leads to comparatively lower solution loss (Bullin et al., 1990). Therefore, despite other
amines being better suited for the purication of large
amounts of CO2 , the eect of MDEA addition in other
aqueous amine solutions was investigated (Pacheco
et al., 2000; Bullin et al., 1990; Sohbi et al., 2007;
Zare Aliabad & Mirzaei, 2009). Sohbi et al. (2007)
and Zare Aliabad and Mirzaei (2009) used simulation
programs (Aspen HYSYS, Aspen Plus) to compare
the results of several mixtures. The inuence of piperazine (PZ) addition on the rate of CO2 absorption
into aqueous MEA, MDEA, and MEA/MDEA solutions has also been the subject of investigation (Dang
& Rochelle, 2001; Bishnoi & Rochelle, 2002; Dubois &
Thomas, 2009). Also aqueous PZ/potassium carbonate (K2 CO3 ) mixtures have been studied (Cullinane &
Rochelle, 2003; Oyenekan & Rochelle, 2009). In this
manuscript, the ability of a 55 mass % DGA aqueous
solution to sweeten biogas was compared with that
of aqueous DGA/PZ and aqueous DGA/PZ/MDEA
mixtures in order to determine the eect of MDEA
and PZ additions to aqueous DGA on the CO2 and
H2 S removal, as well as on the overall system costs.

Theoretical
Composition of raw biogas is dependent on the
source of the biomass from which it is generated. In
this work, it was assumed that raw biogas used to
obtain methane originating from farms, i.e. from agricultural crops and residues and farm animal manure,
consists of 5565 vol. % and 6070 vol. % of CH4 , respectively. The remaining part consists of CO2 with
102 0.5 mass % of H2 S, and trace amounts of ammonia (NH3 ), water vapor (H2 O), nitrogen (N2 ), and
oxygen (O2 ). By careful monitoring of the anaerobic
digestion process, so that no air is introduced into the
fermentation tank, the introduction of N2 and O2 , and
the necessity of the aforementioned compounds removal via a costly membrane separation process can
be avoided. Upon exiting the digestion tank, before
chemical absorption, biogas can be cleaned of H2 O and
NH3 by one of the following methods: simple pressure
increase, raw biogas cooling via burying the gas line
equipped with a condensate trap in the soil, or components adsorption on activated carbon or molecular
sieves. Taking all the information above into account,
it can be assumed that the composition of raw biogas
entering the bottom of the absorption column (absorber) shown in Fig. 1 ranges from 5570 vol. % of
CH4 , 3045 vol. % of CO2 and 102 0.5 mass % of
H2 S. The average biogas composition, after purication, in this work was assumed to be 60 vol. % of CH4 ,
about 40 vol. % of CO2 , and 0.3 mass % of H2 S and
in the worst case scenario: 55 vol. % of CH4 , about 45
vol. % of CO2 , and 0.5 mass % of H2 S.

1011

Fig. 1. Scheme of simplied amine sweetening system.

Fig. 1 presents a schematic picture of the amine purication (sweetening) system used to separate CO2
and H2 S from raw biogas, and regenerate the absorbent, without taking into account the capture and
storage of CH4 and CO2 . Two streams exit the system: puried biogas (stream 3) consisting mainly of
methane, some water vapor, and trace amounts of
amine(s), and distillate vapor (stream 7) consisting
mainly of CO2 and H2 S, as well as of water vapor and
CH4 . Separation of CO2 , H2 S, and CH4 in stream 7
takes place via bioltration, i.e. a ltration process,
where microorganisms on the lters separate and oxidize H2 S into hydrogen sulde (H2 SO4 ) and CH4 into
CO2 . H2 SO4 can be sold as a component of drying
agents and acidic drain cleaners. The remaining CO2
can be used to produce more crops, i.e. more biomass
to be used for further biogas production. In this work,
only the costs of chemical absorption/desorption were
taken into account. Costs concerning biomass digestion, NH3 /H2 O removal and bioltration were not calculated. However, it can be assumed that they are
approximately constant for any variant of the absorption/desorption process.
Mathematical modeling
In order to simulate the chemical absorption and
desorption processes in CHEMCAD, electrolyte reactions taking place during the processes have to be
determined. Reactions taking place in the presence
of DGA and MDEA can be found in the CHEMCAD database. Reactions of PZ were determined
from the information in Table 1 in Cullinane and
Rochelle (2003). The simulator generates electrolyte
reactions concerning DGA, MDEA, H2 O, CO2 , and
H2 S, whereas electrolyte reactions concerning PZ were
manually inserted, after the components were selected.
The program simulates the absorption and desorption
of CO2 and H2 S into and from the liquid phase on
basis of the reaction equilibrium constants and the
Henry constants of CO2 and H2 S. The Henry con-

1012

R. A. Gawel/Chemical Papers 66 (11) 10101018 (2012)

stant is used by the simulator to determine the physical absorption and desorption, and the equilibrium
constants of the electrolyte reactions taking place in
the liquid phase determine the inuence of these reactions on the absorption and desorption processes.
Values of these constants are determined by temperature dependence equations, also included in CHEMCAD, within the electrolyte package. The temperature dependence constants of the equations were either
taken from the CHEMCAD database or they were inserted manually. The above mentioned constants for
electrolyte reactions of PZ were taken from Cullinane
and Rochelle (2003). Temperature dependence of the
piperazine electrolyte reaction enthalpies was determined using the equilibrium temperature dependence
and the Vant Ho equation. Standard enthalpies of
piperazine electrolytes formation were then calculated
by applying the Hess law to the above mentioned reaction enthalpies at 25 C and they were then introduced
into the CHEMCAD component database.
In order to determine the activity coecients of
DGAMDEAPZH2 OCO2 H2 S during the absorption and desorption processes, the electrolyte NRTL
model proposed by Chen and Evans (1986) was used.
In this model, the interaction between the above mentioned components is determined by binary parameters and the nonrandomness factor (Chen & Evans,
1986; Cullinane & Rochelle, 2003). The nonrandomness factor values were set as: 0.2 for waterion pairs,
0.1 for amineion interactions, and 0.0 in for molecule
molecule interactions (Cullinane & Rochelle, 2003).
Further information on this electrolyte model can be
found in Chen and Evans (1986) and Cullinane and
Rochelle (2003). Temperature dependence constants
of the binary parameters were taken either from the
CHEMCAD database or from Cullinane and Rochelle
(2003) and Bishnoi and Rochelle (2002).
After specifying the design parameters of each operating unit and the material of each unit, CHEMCAD
generates the purchase costs of each respective unit.
In order to determine the total costs, the sum of the
purchase costs of all units is multiplied by the Lang
factor (Lang, 1948), assumed in this case as 5. Parameters considered in the Lang factor were taken from
Table 9-6 in Baasel (1990). The costs of additional
streams (steam, cooling water, makeup feed compensating for absorbent losses), as well as the contingency
allowance (Baasel, 1990), assumed to be 25 %, are also
taken into account when determining the ten-year biogas production costs. Moreover, the energy penalty, of
around 28 % on average for an amine capture system,
leads to additional avoidance costs of $55 per 1000
kg of CO2 assuming that the replacement power costs
are $100 per MWh (Chung, 2009). The ten-year additional stream costs are calculated by multiplying the
additional stream ow rates (kg h1 ) by the appropriate cost factors ($ per kg) of the components in the
additional streams and then by 24, 365, and 10. The

cost factors of cooling water and process water are


$0.013 per m3 and $0.13 per m3 , respectively, according to Table 17.1 in Seider et al. (2003), p. 566. From
the cost factors of steam saturated at dierent pressures in the same table, steam saturated at 500 kPa
can be estimated as $6.3 per 1000 kg. Amine loss is
extremely low (usually below 104 kg h1 ) compared
to the amount of water and steam used in the absorption/desorption system. Therefore, even when assuming very high amine cost factor, the impact of amines
on the overall ten-year costs is insignicant. The price
for relatively small amounts of amine during the ten
years of the system operation can be estimated from
the amine price given by various chemical suppliers
on the internet. Finally, the total predicted ten-year
estimated costs can be calculated by adding together
the sum of all purchase costs multiplied by the Lang
factor and the sum of all additional stream costs, multiplied by 1.25, due to the 25 % contingency allowance
mentioned above; the obtained value was added to the
additional avoidance costs.

Method
After entering the missing piperazine electrolyte
information into the program database, all components present in the amine system were selected. Then,
the electrolyte package was used and the electrolyteNRTL 1986 model was chosen; the simulator automatically generated electrolyte reactions and their equilibrium coecients. The temperature dependence constants for CO2 and the Henry constants for H2 S were
generated providing the binary parameters for the
above mentioned model. The missing PZ data concerning the equilibrium constants and binary parameters were entered manually. Upon completing the electrolyte model information, the units and the input and
output feeds were chosen, which resulted in the nal
template design shown in Fig. 1. Then, the components in the absorber input feeds, as well as their ow
rates (kg h1 ), temperature and pressure, were entered into CHEMCAD. The temperature was specied
as 40 C. Reaching lower temperatures requires the use
of costly chilled water, the cost factor of which is given
in Table 17.1 in Seider et al. (2003), p. 566. Furthermore, the theoretical number of stages was chosen as
3 for the absorber and 9 for the desorption unit, where
stages 1 and 9 pertain to the separation of vapor and
liquid in the partial condenser and partial reboiler,
respectively. Theoretical stages at which the feeds enter the respective columns were also specied. In the
absorber, the absorbent was entered to stage 1 and
the raw biogas to stage 3, whereas in the desorption
unit, the feed was entered to stage 3, i.e. to the theoretical stage below the top of the stripper, where the
process water was entered. The pressure at the top
of each column took into account the assumed pressure drop of around 17 kPa, specied manually. The

1013

R. A. Gawel/Chemical Papers 66 (11) 10101018 (2012)

Table 1. Absorption results as a function of ow rate and column pressure


Absorbent ow rate/(kg h1 )

Column pressure

Content in sweetened biogas/mass %

kPa

DGA

H2 O

CO2

H2 S

250
250
300
300

350
400
400
390

280
320
320
300

1.50
< 0.01
< 0.01
< 0.01

0.610
2.558 103
1.610 104
2.222 102

Table 2. Desorption unit costs, total system costs, and methane loss with selected absorbents
Absorbent ow rate/(kg h1 )

Ten-year costs in thousands of $

Methane loss

DGA

PZ

MDEA

H2 O

Desorption unit

Whole system

mass %

400
240
130

0
60
60

0
0
380

320
300
300

657.11
373.20
295.30

1350.02
1078.80
951.33

1.20
0.81
0.40

pressure drop calculated was lower in each case; however, it is best to assume the worst case scenario because the pressure drop slightly decreases the CO2 and
H2 S absorption eciencies. Pressure at the bottom of
the desorption unit was set to 210 kPa, which is in
the operation pressure range recommended for amine
strippers by Arnold and Stewart (1989) and Kohl and
Nielsen (1997). The desired temperature of the distillate was specied as 90 C and the desired bottom
mass ow rate of CO2 was chosen as 0.2 kg h1 . The
vaporliquid equilibrium (VLE) model, described in
Green and Perry (2008), was chosen as the model by
which CHEMCAD determined the ow rates of the
components in each theoretical stage of the absorption or desorption process. In heat exchangers, the
outlet temperature of one stream was selected. The
outlet stream of impure absorbent after exiting the
two-sided heat exchanger was entered at a temperature between 104110 C, so that the regenerated absorbent could reach the lowest possible temperature
before entering the one-sided heat exchanger, where
its outlet stream temperature was determined as 39 C.
Temperature of the regenerated absorbent after exiting the pump and entering the mixer was approximately 41 C. The pump outlet pressure was determined as the pressure at which the absorbent enters
the absorber, which is similar to the outlet specication in the mixer. After entering the specications
into the program, simulation of each individual unit
was run, resulting in output streams and other data
necessary for the calculations shown in the Mathematical modeling section of the manuscript. After
entering the calculated data into the cost estimation
sections of the units, selecting the most cost-ecient
materials (carbon steel for the columns and heat exchangers, cast iron for the pump) and specifying sieve
trays, chosen due to their relatively low costs, as the
type of column tray, CHEMCAD generated the es-

timated purchase costs of each unit. The total estimated ten-year system costs were then determined by
the method presented above in Mathematical modeling.
In order to nd an aqueous DGA solution capable of cleaning the worst case scenario raw biogas,
several DGA/H2 O mixtures were simulated at various pressures until the minimum amount of DGA
necessary to clean the worst case scenario raw biogas to the desired level of purity was determined
at the minimum required pressure. The chosen mixture at the chosen pressure was then entered into
the absorption/desorption system shown in Fig. 1.
This procedure was repeated with DGA/PZ and
DGA/PZ/MDEA aqueous mixtures. Every time the
ow rate of PZ or MDEA was increased, the DGA
ow rate, in turn, was decreased.

Results and discussion


Table 1 presents some of the results for the CO2
and H2 S absorption eciency analysis for the worst
case scenario raw biogas at dierent ow rates and different column pressures for approximately 55 mass %
DGA solutions.
From Table 1 it can be determined that the minimum amount of DGA necessary for the purication
of the worst case scenario raw biogas is 400 kg h1
of DGA dissolved in 320 kg h1 of H2 O entering the
absorber at the unit operation pressure of 300 kPa.
The results conrm the selectivity of DGA towards
the CO2 absorption. Using the same procedure,
DGA/PZ/H2 O and DGA/PZ/MDEA/H2 O mixtures
capable of purifying the worst case scenario biogas
were determined and the minimum absorber pressure
for the aqueous DGA/PZ and DGA/PZ/MDEA solutions were determined to be 250 kPa and 350 kPa, respectively. Aqueous DGA/PZ and DGA/PZ/MDEA

1014

R. A. Gawel/Chemical Papers 66 (11) 10101018 (2012)

Table 3. Column design parameters for the chosen aqueous amine mixtures
DGA/PZ

Design parameter

Column cross-sectional diameter/m


Column height/m
Number of trays
Downcomer area to column area ratio
Hole area to active area ratio
Weir length/m
Column thickness/m

DGA/PZ/MDEA

Absorber

Stripper

Absorber

Stripper

0.120
18.500
20
0.155
0.080
0.099
0.005

0.230
16.400
20
0.102
0.100
0.168
0.006

0.140
22.200
22
0.158
0.060
0.115
0.006

0.190
19.200
20
0.104
0.060
0.139
0.006

Table 4. Comparison of design parameters for each heat exchanger in both chosen DGA/PZ and DGA/PZ/MDEA aqueous amine
mixtures
Heat
exchanger

Aqueous
mixture

Heat exchanger Heat exchanger Outer tube B.W.G. Nr. of Nr. of Distance between
inner diameter/m diameter/m gage tubes baes
baes/m
area/m2

Partial
condenser

DGA/PZ
DGA/PZ/MDEA

1.400
0.500

0.203
0.203

0.013
0.010

16
18

121
96

4
4

0.300
0.100

Partial
reboiler

DGA/PZ
DGA/PZ/MDEA

11.000
13.200

0.305
0.203

0.013
0.010

16
18

130
126

18
22

0.500
0.700

Two-sided heat DGA/PZ


exchanger
DGA/PZ/MDEA

3.400
10.600

0.203
0.203

0.010
0.010

18
18

163
151

20
46

0.200
0.300

One-sided heat DGA/PZ


exchanger
DGA/PZ/MDEA

5.600
8.300

0.203
0.203

0.010
0.010

18
18

178
178

20
32

0.300
0.300

Table 5. Centrifugal pump and reux drum design parameters


Centrifugal pump

Reux drum

Aqueous mixture

DGA/PZ
DGA/PZ/MDEA

Pump head/m

Overall eciency

Inner diameter/m

Length/m

Thickness/m

4.140
14.400

0.017
0.021

0.220
0.060

0.880
0.024

0.006
0.006

solutions ensuring the best ten-year total system costs,


in their respective mixtures, were compared with the
selected aqueous DGA solution in Table 2 after their
simulation by CHEMCAD in the system shown in
Fig. 1.
From Table 2 it follows that the DGA/PZ/MDEA/
H2 O mixture not only ensures the lowest estimated
system costs, but, despite operating at a higher pressure than the other mixtures, also decreases methane
loss which, in turn, increases the CH4 selling prots.
From this Table it can also be concluded that an addition of MDEA decreases not only the desorption costs
but also other aspects of the total ten-year system
costs.
Design parameters for DGA/PZ and DGA/PZ/
MDEA aqueous mixtures for the absorption and desorption columns, heat exchangers, and the centrifugal
pump and reux drum are given in Tables 35.

From Table 3 it follows that while high MDEA


content decreases the stripper diameter and therefore
the stripper costs, it also increases the absorber diameter and height leading to higher absorber costs.
However, the presence of MDEA also ensures that the
absorbent is regenerated with the lowest additional
stream costs in the desorption system of the mixtures
used in the simulations provided by CHEMCAD, as
conrmed by the desorption costs in Table 2. Unfortunately, from Table 4 it can be concluded that
an addition of MDEA into the absorbent mixture increases the area of the heat exchangers, resulting in
an increase in their purchase costs. Table 5 conrms
that the addition of MDEA decreases the reux drum
costs. Additionally, it can be concluded that using the
mixture with MDEA, at a higher ow rate, results in
better overall centrifugal pump eciency and a larger
pump head.

R. A. Gawel/Chemical Papers 66 (11) 10101018 (2012)

Conclusions
The addition of PZ to the DGA/H2 O absorbents
not only signicantly improves the rate of the absorption process but also leads to much lower absorbent regeneration costs caused by decreased heat duties of the
partial condenser and reboiler, which, in turn, leads
to lower additional stream ow rates and costs. However, an MDEA addition results in larger amounts of
weeping at the top of the absorber compared to the
other absorbents, which leads to additional absorber
construction costs. However, MDEA also successfully
reduces the heat of the electrolyte reactions and therefore the temperature of the absorbent exiting the absorber, which decreases the cooling water ow rate
in the one-sided heat exchanger. Furthermore, MDEA
minimizes the component losses during the absorption process. The eect of MDEA on the condenser
and reboiler heat duties in the desorption system is
also a positive one, though not as much as that of the
addition of PZ. Nevertheless, the presence of MDEA
in an aqueous mixture increases the viscosity of the
absorbent leading to laminar ows inside the tubes,
which require higher heat exchange area. Moreover,
signicant pressure losses in the tube-side ows are
observed when large amounts of MDEA are present
in the given aqueous mixture.
Acknowledgements. I would like to thank the AGH University of Science and Technology for granting me access to
the CHEMCAD simulation program. I would also like to thank
Prof. Volkmar Jordan for overseeing my Master Thesis, which
this work follows from.

References
Ahmad, Z. (2006). Principles of corrosion engineering and corrosion control (pp. 125). Oxford, UK: Elsevier.
AISC (1989). AISC Manual of steel construction: Allowable
stress design (9th ed.). Chicago, IL, USA: American Institute of Steel Construction.
Arnold, K., & Stewart, M. (1989). Surface production operations: Design of gas handling systems and facilities (2nd
ed.). Houston, TX, USA: Gulf.
Baasel, W. D. (1990). Preliminary chemical engineering plant
design (2nd ed., pp. 272). Melbourne, Australia: Elsevier.
Barreau, A., Blanchon le Bouhelec, E., Habchi Tounsi, K. N.,
Mougin, P., & Lecomte, F. (2006). Absorption of H2 S and
CO2 in alkanolamine aqueous solution: Experimental data
and modeling with the electrolyte-NRTL model. Oil & Gas
Science and Technology Revue de lIFP, 61, 345361. DOI:
10.2516/ogst:2006038a.
Bell, K. J., & Mueller, A. C. (1984). Wolverine engineering data
book II. Shawnee, OK, USA: Wolverine Tube Inc.
Bishnoi, S., & Rochelle, G. T. (2002). Thermodynamics of
piperazine/methyldiethanolamine/water/carbon dioxide. Industrial & Engineering Chemistry Research, 41, 604612.
DOI: 10.1021/ie0103106.
Brownell, L. E., & Young, E. H. (1959). Process equipment design: Vessel design (pp. 172). New York, NY, USA: Wiley.
Bullin, J. A., Polasek, J. C., & Donnelly, S. T. (1990). The use
of MDEA and mixtures of amines for bulk CO2 removal. In
Proceedings of the 69th Gas Processors Association Annual

1015

Convention, March 1213, 1990 (pp. 135139). Tulsa, OK,


USA: Gas Processors Association.
Chen, C. C., & Evans, L. B. (1986). A local composition model
for the excess Gibbs energy of aqueous electrolyte systems.
AICHE Journal, 32, 444454. DOI: 10.1002/aic.690320311.
Chung, T. S. (2009). Expert assessments of retrofitting coalfired power plants with carbon dioxide capture technologies
(pp. 2629). Masters project, Nicolas School of the Environment of Duke University, Durham, NC, USA.
Cullinane, J. T., & Rochelle, G. T. (2003). Properties of concentrated aqueous potassium carbonate/piperazine for CO2
capture. In The 2nd Annual Conference on Carbon Sequestration, May 58, 2003. Alexandria, VA, USA.
Dang, H., & Rochelle, G. T. (2001). CO2 absorption rate and
solubility in monoethanolamine/piperazine/water. In The
1st National Conference on Carbon Sequestration, May 14
17, 2001. Washington, DC, USA.
Dubois, L., & Thomas, D. (2009). CO2 absorption into aqueous solutions of monoethanolamine, methyldiethanolamine,
piperazine and their blends. Chemical Engineering & Technology, 32, 710718. DOI: 10.1002/ceat.200800545.
Green, D. W., & Perry, R. H. (2008). Perrys Chemical engineers handbook (8th ed.). Portland, ME: Tata McGraw-Hill.
Kohl, A. L., & Nielsen, R. B. (1997). Gas purification (5th ed.).
Houston, TX, USA: Gulf.
Lang, H. J. (1948). Simplied approach to preliminary cost estimates. Chemical Engineering, NY, 55(6), 112113.
Lindeburg, M. R. (2006). Mechanical engineering reference
manual for the PE exam (12th ed., pp. 188). Belmont, CA,
USA: Professional Publications.
Luyben, W. L. (2011). Principles and case studies of simultaneous design. Hoboken, NJ, USA: Wiley.
Manning, F. S., & Thompson, R. E. (1991). Oilfield processing of petroleum: Natural gas (pp. 116). Tulsa, OK, USA:
PennWell.
Oyenekan, B. A., & Rochelle, G. T. (2009). Rate modeling of
CO2 stripping from potassium carbonate promoted by piperazine. International Journal of Greenhouse Gas Control, 3,
121132. DOI: 10.1016/j.ijggc.2008.06.010.
Pacheco, M. A., Kaganoi, S., & Rochelle, G. T. (2000).
CO2 absorption into aqueous mixtures of diglycolamine and
methyldiethanolamine. Chemical Engineering Science, 55,
51255140. DOI: 10.1016/s0009-2509(00)00104-4.
Pitts, D. R., & Sissom, L. E. (1998). Schaums outline of theory
and problems of heat transfer (2nd ed., pp. 54). New York,
NY, USA: McGraw-Hill.
Polasek, J. C., & Bullin, J. A. (1994). Selecting amines for
sweetening units. In Proceedings of the GPA Regional Meeting Process Considerations in Selecting Amine, September, 1994. Tulsa, OK, USA: Gas Processors Association.
Rousseau, R. W. (1987). Handbook of separation process technology (pp. 293). New York, NY, USA: Wiley.
Salkuyeh, Y. K., & Mofarahi, M. (2011). Comparison of MEA
and DGA performance for CO2 capture under dierent operational conditions. International Journal of Energy Research, 36, 259268. DOI: 10.1002/er.1812.
Savage, D. W., & Funk, E. W. (1981). Selective absorption of
H2 S and CO2 into aqueous solutions of methyldiethanolamine. In 90th AIChE National Meeting, April 59, 1981.
Houston, TX, USA.
Seader, J. D., & Henley, E. J. (2006). Separation process principles (2nd ed.). Danvers, MA, USA: Wiley.
Seider W. D., Seader, J. D., & Lewin, R. L. (2003). Product and
process design principles: Synthesis, analysis and evaluation
(2nd ed.). New York, NY, USA: Wiley.
Sinnott, R. K., & Towler, G. (2009). Chemical engineering design (5th ed.). Oxford, UK: Elsevier.

1016

R. A. Gawel/Chemical Papers 66 (11) 10101018 (2012)

Sohbi, B., Meaka, M., Emtir, M., & Elgarni, M. (2007). The
using of mixing amines in an industrial gas sweetening plant.
World Academy of Science, Engineering and Technology,
31(1), 301305.
Souders, M., & Brown, G. G. (1934). Design of fractionating
columns I. Entrainment and capacity. Industrial & Engineering Chemistry, 26, 98103. DOI: 10.1021/ie50289a025.

Thakore, S. B., & Bhatt, B. I. (2007). Introduction to process


engineering and design. New Delhi, India: Tata McGraw-Hill.
Zare Aliabad, H., & Mirzaei, S. (2009). Removal of CO2 and
H2 S using aqueous alkanolamine solutions. World Academy
of Science, Engineering and Technology, 49(1), 194203.

Appendix

the distance between the trays. Its value can be obtained from Fig. 6.24 in Seader and Henley (2006),
p. 217. FLV was determined using Eq. (3), where
mL /(kg h1 ) and mV /(kg h1 ) are the mass ows of
liquid and vapor, respectively.

In order to design both the absorber and the stripper, the type of trays used in both columns had to
be determined. Sieve trays were selected due to their
relatively low installation and maintenance costs compared to other types of trays (valve trays, bubble cap
trays). According to Manning and Thompson (1991),
amine absorbers and strippers usually consist of 20
trays. Spacing between the amine absorber trays is either 0.46 m or 0.61 m, whereas the spacing between
the amine stripper trays is usually 0.61 m (Manning &
Thompson, 1991). The distance between the top of an
amine column and its top tray is between 0.91 m and
1.22 m (Manning & Thompson, 1991). Furthermore,
in amine absorption/desorption, tray absorption eciency is usually assumed to be below 20 % (Dang &
Rochelle, 2001) and tray desorption eciency usually
lies between 40 % and 50 % (Manning & Thompson,
1991). Taking all of the above into consideration, both
columns were designed under the following assumptions: tray spacing was 0.61 m, and space above the
top tray was assumed to be 0.91 m in the absorber and
1.22 m in the stripper. The space below the bottom
tray, i.e. the column base, should provide 5 min of liquid holdup at 50 % level (Luyben, 2011). It was also
assumed that the absorber possesses three theoretical stages, whereas the stripper has seven theoretical
stages.
The column was designed under the assumptions
that the F-factor (F), calculated from Eq. (1), is lower
than 2 and the fractional approach to ooding (f),
calculated from Eq. (2), is lower than 1.

F = u s V
(1)
f = us /uf

(2)

In the above equations, us /(m s1 ) is the supercial vapor velocity, V /(kg m3 ) the vapor density,
and uf /(m s1 ) the ooding velocity calculated by
Eq. (6-40) in Seader and Henley (2006), p. 216, where
qL /(kg m3 ) is the liquid density and C is the capacity parameter taken from Souders and Brown (1934).
The value of C was obtained by Eq. (6-42) in Seader
and Henley (2006), p. 217. In order to solve the aforementioned equation, the following parameters had to
be determined: surface tension factor (FST ), foaming
factor (FF ), hole area factor (FHA ), and CF . The last
parameter is a function of the abscissa ratio (FLV ) and

FLV = (mL /mV ) (V /L )

0.5

(3)

In turn, the surface tension factor was determined


from Eq. (4), where /(N m1 ) is the liquid surface
tension, a property estimated by CHEMCAD. According to Rousseau (1987), the foaming factor is 0.73 for
amine absorbers and 0.85 for amine strippers. On the
other hand, FHA was determined as 1, when the hole
area (Ah /m2 ) to the active tray area (Aa /m2 ) ratio
was at least 0.1, and it was calculated using Eq. (5) in
all other cases. The Ah /Aa ratio was assumed so that
the amount of weeping, an undesirable phenomenon
described in Seader and Henley (2006), p. 222, is minimal.
0.2
FST = (50)
(4)
FHA = 5 (Ah /Aa ) + 0.5

(5)

From the value of FLV , the downcomer area


(Ad /m2 ) to the total column cross-section area (A/m2 )
ratio can be determined. When FLV < 0.1, Ad /A =
0.1, whereas when FLV > 0.2, Ad /A = 0.2. In all
other cases, Ad /A was determined by Eq. (6). Knowing the value of Ad /A, the total cross-section diameter
(DT /m) as well as A can be calculated from Eqs. (7)
and (8), where V/(m3 h1 ) is the volumetric vapor
ow rate.
FLV 0.1
(6)
Ad /A = 0.1 +
9
4V
us (1 Ad A)
 2
A = 0.25 DT

DT =

(7)
(8)

The total vapor pressure drop during the absorption or desorption process can be obtained by multiplying the total pressure drop per tray (pt /kPa) by
the number of trays in the given column (n). In turn,
pt is the sum of the following pressure drops: dry
tray pressure drop (pd /kPa), pressure drop due to
liquid holdup on a tray (pl /kPa), and the pressure
drop due to the surface tension (p /kPa). Weeping
occurs when pd + p < pl . These pressure drops
can be calculated using Eqs. (6-50)(6-55) in Seader

1017

R. A. Gawel/Chemical Papers 66 (11) 10101018 (2012)

and Henley (2006), p. 219, and multiplying the obtained values, converted from inches into meters, by
the average density of the liquid owing through the
given column (L /(kg m3 )) and the gravitational acceleration (g/(m2 s1 )), and dividing the values by
1000. The orice coecient present in Eq. (6-50) in
Seader and Henley (2006), p. 219, can be determined
from Fig. 11.42. in Sinnott and Towler (2009) after
estimating the tray thickness (th/m) to the hole diameter ratio (DH /m). In this work, it was assumed
that th/DH = 0.6. DH was determined to be 0.003 m
in the absorbers case and 0.005 in the strippers case,
because the absorber column is more prone to weeping. In amine absorbers and strippers, the weir height
(hw /m) should be between 0.040.09 m (Thakore &
Bhatt, 2007). In this work, hw was assumed as 0.04 m.
Knowing hw and Ad /A, the weir length (Lw /m) was
determined using Table 8.34 in Thakore and Bhatt
(2007). All of the above mentioned calculations were
carried out for the liquid and vapor ow rates at the
rst tray at the top of the column and at the last tray
at the bottom of the column, as well as for the average
ow rates.
Minimum column thickness was determined using
Eq. (13.1) in Brownell and Young (1959). The general
practice is to allow a 0.0032 m corrosion allowance
(c/m) (Ahmad, 2006). Allowable stress (S/kPa) for
carbon steel was assumed as the value of allowable
shear stress which is 99284.5 kPa according to AISC
(1989). The maximal allowable joint eciency (E) of
0.85 was taken from Table 13.3 in Sinnott and Towler
(2009) under the assumption that the type of joint is
a double-welded butt. After obtaining the minimum
column thickness, the real column thickness was assumed.
Design calculations were carried out for heat exchangers with one-phase ows in the absorption/desorption system (Fig. 1) after the desired output temperature of the given heated or cooled stream was
specied. In one-phase heat exchangers, the cooled
liquid ows in the space between the tubes and the
heated liquid or cooling water ows inside the tubes.
The ow of liquids through each heat exchanger in
Fig. 1 is countercurrent. According to Seider et al.
(2003), p. 407, temperature of the cooling water entering the one-sided heat exchanger is 32 C. In this
work, the temperature of the cooling water was assumed to be 38 C after exiting the one-sided heat
exchanger. After running simulations for these heat
exchangers, CHEMCAD generated the streams at
the specied output temperature as well as the heat
duty (Q/(kJ h1 )) of the given heat exchanger. The
heat exchange area of the respective heat exchanger
(Ai ex /m2 ) as well as its length (Lex /m) were calculated from Eqs. (9) and (10). Taking into account the
shell thickness (ths /m) assumed in this work as 0.027
m, the value of which is dependent on the designed
pressure, the real heat exchanger area (Aex /m2 ) was

calculated by Eq. (11), where Di /m is the inner diameter of the heat exchanger.
Ai

ex

= Q (U TLM )1

(9)

= Di Lex

(10)

Ai

ex

Aex = (Di + 2ths ) Lex

(11)

The overall heat transfer coecient (U/


(kW m2 K1 )) and the logarithmic mean temperature dierence (TLM /K) were calculated by the following equations:
TLM = (T1

ex T2

 
T1
)
ln
ex
T2

ex

1
(12)

ex


1
Do t
1

Di t
Di t

+
U =
+
+
Rf
i t

CS (Do t Di t )
Do t ss

tht Di t ln

(13)
In the above equations, T1 ex and T2 ex are the
temperature dierence at both sides of the heat exchanger. Thermal conductivity of carbon steel (CS )
is 0.043 kW m1 K1 according to Pitts and Sissom
(1998). The fouling factor (Rf/(m2 K kW1 )) was
taken from Table 11-3 in Green and Perry (2008).
Tube design parameters such as: inner tube diameter (Di t /m) and tube thickness (tht /m) can be obtained from Table 11-12 in Green and Perry (2008),
after estimating the outer diameter (Do t /m) and selecting a B.W.G. gage from the aforementioned table.
In this work, a triangular layout was chosen for each
heat exchanger and the tube pitch was assumed to be
1.25 times the tube outer diameter. The thermal conductivity (/(kW m1 K1 )), specic heat capacity
(cp /(kJ kg1 K1 )), and dynamic viscosity (/(Pa s))
of each stream was estimated by CHEMCAD. Thus,
the Prandtl number (Pr) of each stream could be calculated. The Nusselt number of the stream owing inside the tubes (Nui t ) was calculated by Eqs. (14) and
(15), in which fF is the Fanning friction factor, and
by Eq. (2.19) from Bell and Mueller (1984) assuming
the ow inside the tubes to be a transitional (2000
< Rei t < 105 ) one-phase ow. The heat transfer coecient (i t /(kW m2 K1 )) of the stream owing
inside the tubes can be calculated using the denition
of the Nusselt number
N ui t =

0.5fF ((Rei t 1000) P ri t )


1 + 12.7 (P ri0.67
t 1) 0.5fF
2

fF = (1.58 ln (Rei t ) 3.28)

(14)
(15)

When neither condensation nor boiling takes place,


the heat transfer coecient of the shell-side stream
owing through the area between the tubes

1018

R. A. Gawel/Chemical Papers 66 (11) 10101018 (2012)

(ss /(kW m2 K1 )) is calculated using Eqs. (2.15)


(2.17) and Fig. 2.13 in Chapter 2.3 in Bell and Mueller
(1984). The correction factors for bae leakage and
bae conguration eects as well as the inuence of
the factors on the heat transfer coecient value can be
determined using the gures and equations in Chapter 2.6 in Bell and Mueller (1984). The number and
distance between the baes should be assumed so
that a reasonably high heat transfer coecient can
be obtained, while ensuring that no signicant pressure drop of the shell-side stream occurs. It was also
assumed that the bae cut is 25 %, i.e. the height of
the baes is equal to 75 % of Di . This is the most
common bae cut according to Seider et al. (2003),
p. 418. The pressure drop of the shell-side stream can
also be calculated using Eqs. (2.57)(2.60) in Chapter
2.6 for a one-phase ow (Bell & Mueller, 1984).
In a partial condenser, where condensation takes
place at the shell-side, Eq. (16) was used to calculate
ss , where W/(kg s1 ) is the mass of the vapor condensed per unit time and subscripts L, V, and ss refer to liquid phase, vapor phase, and shell-side of the
heat exchanger, respectively. Furthermore, the pressure drop of the shell-side stream was determined using Eqs. (5.58)(5.61) and Fig. 5.32 in Chapter 5.6
in Bell and Mueller (1984). All other calculations remained the same as for the one-phase heat exchangers.
Also, the inlet and outlet temperatures of the cooling
water were the same as for the one-phase one-sided
heat exchanger.

ss = 0.951

3L

ss L ss

(L ss V
L ss Wss

ss ) 9.81Lex

0.33

(16)
In the partial reboiler, ss was calculated using Eq.
(5.7) in Bell and Mueller (1984). The pressure drop of

the partially boiled liquid owing through the shellside was determined using the same equations as in
the partial condenser. The heat transfer coecient of
the heat source (i t /(kW m2 K1 )), i.e. saturated
steam inside the heat exchanger tubes, was calculated
from Nui t . Eq. (2.27) in Bell and Mueller (1984) was
used to determine Nui t , assuming that the steam, entered into the tubes, condenses entirely before exiting.
All other calculations remained the same as in the two
other heat exchangers.
The reux drum vessel volume (v/m3 ) was determined on basis of the liquid residence time (t), which
should be at least 5 min (Seader and Henley (2006),
p. 279). In this work, the liquid residence time was
assumed as 6 min. The reux drum was designed so
that the inner volume of the drum is twice the amount
of liquid accumulated in the assumed residence time
(Seader and Henley (2006), p. 279). Furthermore, the
length of the drum (LRD /m) is usually assumed to be
four times the inner diameter (DRD /m) (Seader and
Henley (2006), p. 279). The thickness of the reux
drum (thRD /m) was determined in the same way as
the column thickness.
After running the pump simulation in CHEMCAD,
the program generated the necessary design parameters, such as the pump head (HP /m) and pump power
(NP /kW). The overall eciency () of the pump installation was obtained by multiplying the pump efciency ( P ) and the motor eciency ( M ) (Lindeburg, 2006). Pump eciency was determined from Table 15.5 in Seider et al. (2003) as a function of the
stream ow rate through the pump (q/(m3 h1 )) and
the dynamic viscosity of the stream. The motor eciency in this work was assumed to be 0.85, because,
according to Lindeburg (2006); up to 15 % of the motor horsepower can be lost.

Você também pode gostar