Chemical Engineering and Processing 46 (2007) 262–270

Acid gas absorption in trickle flow columns—Modelling of the

residence time distribution of a pilot plant
Tobias Pröll a,∗ , Teodor Todinca b , Marcel Şuta b , Anton Friedl a
a
Institute of Chemical Engineering, Vienna University of Technology, Getreidemarkt 9/166, 1060 Vienna, Austria
b Department of Chemical Engineering, University “Politehnica” of Timişoara, Piaţa Victoriei 2, 30006 Timişoara, Romania
Received 7 September 2005; received in revised form 25 April 2006; accepted 2 June 2006
Available online 27 June 2006

Abstract
A simple mathematical model has been developed and validated for the dynamic behaviour of a pilot plant absorber with random packing
(stainless steel construction, column diameter: 0.08 m, height of packing zone: 1.3 m, random packing: 15 mm ceramic spheres). The model is
based upon film theory, axial dispersion, and rate enhancement. Simplifications are made which are justified for the conditions of the experimental
investigations in the pilot plant of this work. Simulation of other conditions like for example acid gas removal in industrial applications will require
additional consideration of water evaporation and of the energy balance. The experimental part concentrates on measurements of the residence time
distribution of several sections of the pilot installation. Combination of the different transition functions allows the determination of the dynamic
behaviour of the packing zone. The model has been validated by experiments with absorption of CO2 into water and into carbonate solution at
moderate conditions (5 barabs , 290–333 K). Although the mass transfer parameters have been strictly prescribed according to theory, the simulation
results agree well with experimental data.
© 2006 Elsevier B.V. All rights reserved.

Keywords: Acid gas removal; Trickle flow; Packing column; Dynamic modelling; Residence time distribution

1. Introduction difficulty concerning absorption into carbonate solution is a

relatively slow reaction in the liquid phase causing low mass
The removal of acid gases like CO2 , H2 S, or SO2 from indus- transfer rates and therefore requiring a large contact surface. In
trial gas streams is important either to meet certain process order to increase efficiency of the potassium carbonate process
specifications or to keep emission limits. The major approach for CO2 removal, the absorption is carried out at high temper-
towards acid gas removal is absorption into solutions of alka- atures (90–105 ◦ C, 30 bar) and rate-promoting additives (e.g.,
nolamines or potassium carbonate. In practical application, such diethanolamine) are commonly used (Benfield/Carsol Process).
absorption plants consist of two gas–liquid reactors operated at Better comprehension about the mass transfer in the trickle flow
different temperatures and pressures: absorber and regenerator. column, which is determined by hydrodynamics and chemistry,
The solvent is recycled in a closed loop and heat is exchanged is important for the dimensioning of new plants and to increase
between the hot, lean solvent after the regenerator and the cooler, the efficiency of existing installations by the appropriate use of
rich solvent after the absorber. rate promoters. In the present study, the behaviour of a pilot plant
When gases with high loads of acid components are to be absorber is studied with emphasis on residence time distribution
treated (bulk removal), the counter-current trickle flow column (RTD) of each phase and the absorption parameters for both
with either structured or random packing is the preferred type water and carbonate solution (active surface area, mass transfer
of apparatus. The main advantages of carbonate solutions for coefficients). The results represent a basis for further investiga-
CO2 removal are the high chemical solubility of CO2 in the tions evaluating rate-enhancing chemicals for the hot potassium
carbonate/bicarbonate system and low solvent costs. The major carbonate process. If industrial conditions prevail (higher tem-
peratures and pressures, high absorption rates) the important
effects of water evaporation and reaction enthalpies (local energy
∗ Corresponding author. Tel.: +43 1 58801 15965; fax: +43 1 58801 15999. balances) have to be considered. A rigorous approach towards
E-mail address: tobias.proell@tuwien.ac.at (T. Pröll). the modelling of reactive absorption that does not only cover

0255-2701/$ – see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.cep.2006.06.006
 T. Pröll et al. / Chemical Engineering and Processing 46 (2007) 262–270 263

the phenomena mentioned above but further avoids the need for treated as non-volatile carrier of the liquid stream, i.e. there is
empirical rate enhancement parameters by discrete description no evaporation of water into the gas phase modelled. In real-
of the film region has been presented by Górak et al. [1,2]. ity, the gas phase can be expected to be rapidly saturated with
steam inside the gas/liquid contact area. The errors resulting
2. Mathematical model from this simplification (with respect to the partial pressure of
CO2 ) increase with temperature and decrease with pressure and
Within the present work, a simple mathematical model is are still relatively low in the operating range of the pilot plant
used to describe the phase flow in the entire pilot plant installa- (maximum 4% at 5 bar (abs), 60 ◦ C). The mass balance for the
tion as well as the mass transfer taking place in the packing of absorbing component (e.g. CO2 ) is illustrated in Fig. 1.
the absorber. The model parameters are determined in specific In a dimensionless form, the two mass balance equations for
experiments on the pilot plant by comparison with computer the two phases can be written as the following:
simulation.
The fluid phases pass additional volumes apart from the • Gas side mass balance for CO2 [molCO2 /molinert ]:
packed bed (in bottom and top of the column, pipes, etc.), which inert H
βG ∂Y ∂Y 1 ∂2 Y KG aH
are described by a network of simple flow models like ideally =− + − (pCO2 −p∗CO2 )
mixed volumes, plug flow with axial dispersion, and stagnant (1 + Y )G0 ∂t ∂z PeG ∂z 2 G0
(1)
zones. The model equations for these additional zones are not
mentioned here in detail since the RTD models for these reactor • Liquid side mass balance for CO2 [molCO2 /kgH2 O ]:
types are well known [3]. The model structure with respect to
the sequence of additional flow models, which was developed βL H ∂X ∂X 1 ∂2 X KG aH
based on experimental results, is presented in a later section of =+ + + (pCO2 − p∗CO2 )
L0 ∂t ∂z PeL ∂z2 ρ H2 O L 0
this text. (2)
The main modelling task is the description of phase flow
dynamics and mass transfer in the packing zone of the absorber.
Therefore, a one-dimensional, continuous model will be applied In Eqs. (1) and (2), z is the dimensionless height coordinate
scaled from 0 (bottom) to 1 (top). βL and βG inert , respectively
to describe the concentration profiles along the height of the
trickle flow column. The axial mixing effects are described in represent the hold up of liquid and gas in the packing while X
analogy to molecular diffusion (plug flow with axial dispersion). and Y are the molar loads of CO2 on the respective bulk phase
The presented model is based on differential mass balances flow. The extent of axial dispersion is described for each phase
only, while the energy balances in the packing zone have been by the dimensionless Peclet number:
removed again from the original version. The reason is that the wH
thermal effects occurring together with absorption show only Pe = (3)
DAX
little impact at pilot plant size, where mainly the heat trans-
fer through the column walls determines the local temperature. For high values of Pe, the reactor behaves close to ideal plug
Instead of solving the energy balance, the temperature profile flow, while for low Peclet numbers the mixing effects dominate.
is prescribed along the height of the column. Further, water is Alternatively, the axial dispersion specific for a specific packing

Fig. 1. Partial mass balance for the transition component CO2 in the differential element of volume S dh.
264 T. Pröll et al. / Chemical Engineering and Processing 46 (2007) 262–270

is characterised by the Bodenstein number where m is the molality of the carbonate solution and He0 is the
wdP dP Henry’s constant in pure water, which is strongly dependent on
Bo = = Pe (4) temperature. Plummer and Busenberg [5] give a correlation for
DAX H
He0 valid between 273 and 523 K (0–250 ◦ C). The coefficients
where dP is the nominal packing size. of the general equation Eq. (13) are summarized in Table 1.
The boundary conditions by Danckwerts are formulated for
A3 A4
each phase: log10 (He0 , Ki ) = A0 + A1 T + A2 log10 (T ) + + 2 (13)
T T
• Entry conditions: The physical mass transfer coefficients kL0 and kG as well
 as the actual interfacial area a are determined by hydrodynam-
1 ∂Y 
Yin = Y (0+, t) − (5) ics. The present study uses the correlations by Onda et al. [7],
Pe ∂z z=0+ which are recommended for random packing with continuous

1 ∂X  flow configuration (i.e., uninterrupted rings, spheres, saddles).
Xin = X(1, t) + (6)
Pe ∂z z=1
The influence of liquid phase chemical kinetics on mass trans-
fer, expressed by the rate enhancement factor E in Eq. (11), is
• Exit conditions: treated extensively in the literature [8–10]. The value of E can

∂Y  be approached by the introduction of the dimensionless Hatta
=0 (7)
∂z z=1 number, which relates the chemical reaction rate to the trans-
 port rate by molecular diffusion. For the operation conditions
∂X  achieved on the pilot plant in the present work (absorption of
=0 (8)
∂z z=0+ CO2 into carbonate solution at temperatures of about 55 ◦ C) a
pseudo-first-order reaction mechanism can be assumed, if the
The two balance equations are coupled by the interfacial mass following condition is fulfilled [8]:
transfer of CO2 . According to Eqs. (1) and (2), the mass transfer  
is determined by the three parameters KG , a, and p∗CO2 The [CO2 ]aq.,i
1
+
2
interfacial area a depends on hydrodynamics. The CO2 partial [CO2−
3 ] [HCO− 3]
pressure that would be in chemical equilibrium with the bulk of  
the liquid (p∗CO2 ) depends on absorption chemistry in the liquid × 1+
DL,CO2 kI,CO2
−1 1 (14)
phase. The gas side overall mass transfer coefficient KG includes kL2 0
the liquid side mass transfer resistance and therefore depends on
both hydrodynamics and liquid phase chemistry. Cents et al. [11] found that the absorption reaction can be
While the calculation of the differential mass balance equa- treated as first order if the left hand side of Eq. (14) is smaller
tions represents the core of the simulation model, the relatively than 0.1. As the results will show, this criterion is fulfilled in the
complex, but algebraic, calculation of local equilibrium and operating range of the pilot plant within the present study. In this
mass transfer kinetics is done in an inner shell of the model case, the relatively simple correlation reported by Danckwerts
structure. and Sharma [8]:
According to film theory, the overall interfacial mass transfer
Ha
can be described by the two parameters KG and p∗CO2 , defined E= (15)
by tanh(Ha)
where
JA = KG (pCO2 − p∗CO2 ) (9) 
kI,CO2 DL,CO2
The gas phase partial pressure of CO2 is directly linked to the Ha = (16)
CO2 load on the inert gas: kL2 0
pCO2 yields results equal to those of other, more complex correlations
Y= (10)
p − pCO2 tested. The binary diffusion coefficient of CO2 in water DL,CO2 is
calculated using the method of Wilke and Chang [12] proposed
The overall gas side mass transfer coefficient KG can be calcu-
by Danckwerts [9]. In order to calculate the pseudo first order
lated from the actual mass transfer coefficients on both sides of
reaction rate constant kI,CO2 , liquid phase chemistry must be
the interface as
taken into account.
1 He 1 Savage et al. [10] present the chemistry of the system in a
= + (11)
KG EkL0 ρH2 O kG detailed way. The chemical state is determined by the equilib-
Astarita et al. [4] report a correlation for the dependency of rium of the three formal reactions
Henry’s law constant He of CO2 on ionic strength in potassium CO2 + H2 O ⇔ H+ + HCO3 − (17)
carbonate solution:
He HCO3 − ⇔ H+ + CO3 2− (18)
log10 = 0.125m, (12)
He0 H2 O ⇔ H+ + OH− (19)
 T. Pröll et al. / Chemical Engineering and Processing 46 (2007) 262–270 265

Table 1
Effect of temperature on Henry’s law constant and dissociation constants Ki according to Eq. (13)
Constant Unit A0 A1 (K−1 ) A2 A3 (K) A4 (K2 ) Valid (◦ C) Ref.

He0,CO2 = (pyCO2 )/(mCO2 (aq.) ) Pa kg/mol −103.3808 −0.01985076 40.45154 6919.53 −669365.0 0–250 Plummer and Busenberg [5]
Kw = mH+ mOH− mol2 /kg2 61.2062 – −22.4773 −5839.5 – 0–225 Tsonopoulos et al. [6]
KCO2 = (mH+ mHCO− )/(mCO2 (aq.) ) mol/kg −356.3094 −0.06091964 126.8339 21834.37 −1684915.0 0–250 Plummer and Busenberg [5]
3
KHCO− = (mH+ mCO2− )/(mHCO− ) mol/kg −107.8871 −0.03252849 38.92561 5151.79 −563713.9 0–250 Plummer and Busenberg [5]
3 3 3


Temperature dependent equilibrium constants of reactions ∂2 x xi+1,j − 2xi,j + xi−1,j
(17)–(19) can as well be calculated by Eq. (13) and the coeffi- = (28)
∂z2 i,j (z)2
cients are summarised in Table 1. Introduction of the carbonate
conversion rate or partial saturation where the index “i” is referring to location and the index “j” to
time.
[HCO−
3]
α= (20) At each of the two boundaries, a hypothetical out-of-area cell
2m is added in order to fulfil the boundary conditions in the discrete
yields the partial pressure of the gas phase in equilibrium with system.
the bulk of the liquid: The solving algorithm starts at the time t = 0 with known ini-
KHCO− 4mα2 tial distributions and proceeds step by step along the time axis.
p∗CO2 = He 3
(21) The explicit algorithm uses only the known values at time t to
KCO2 1 − α calculate the gradients leading to the values at time t + t. The
Combination of reactions (17) and (19) yields the practically system remains relatively easy to handle despite the non-linear
occurring and rate limiting reaction step terms in the model equations. A disadvantage of the applied
explicit method is the high risk of instability, which requires
CO2 + OH− ⇒ HCO3 − (22) small time steps. Finally, the solution of the model equations
provides the dynamic behaviour of the concentration distribu-
The following correlation for the second order constant kOH for
tions of the two phases in the column.
reaction (22) based on data between 273 and 313 K is reported
[8]:
3. Experimental set-up and procedure
2895
log10 kOH− = 13.635 − (23)
T The general arrangement of the installation in the laboratory
The influence of other reactions on the global rate is negligi- is shown in Fig. 2. Some geometric data and packing characteris-
ble low [10]. For the conditions occurring in the pilot plant,
the kinetics can be well approximated by a pseudo-first-order
reaction
rCO2 = kI,CO2 [CO2 ] (24)
with the pseudo-first-order reaction constant kI,CO2 depending
on temperature and acidity in the bulk of the liquid phase:
kI,CO2 ≈ kOH− [OH− ] (25)
Eqs. (9)–(25) represent the algebraic inner part of the model.
All parameters are determined by the local concentrations in the
bulk of the phases at a certain cross-section of the column.The
system of differential equations is solved by a finite differences
algorithm with explicit integration. The differential equations
are transformed into equations of differences by writing the par-
tial derivatives as quotients of differences. If the location axis
step size is z and time axis step size t, the derivatives in any
node (i, j) of the discrete area are transformed according to

∂x xi,j+1 − xi,j
= (26)
∂t i,j t

∂x xi+1,j − xi−1,j
= (27)
∂z i,j 2z Fig. 2. Experimental set-up.
266 T. Pröll et al. / Chemical Engineering and Processing 46 (2007) 262–270

Table 2 be obtained for the gas phase RTD by partially filling up the
Geometric data and packing properties column with water (e.g. until the top of the packing during deter-
Column geometry mination of the RTD in the gas phase exit zone) and for the liquid
Total height of column 2290 mm phase by introduction of the tracer solution at different points
Total height of bed (both sections), H 1300 mm of the installation (e.g. right below the packing zone for the
Height of bottom zone 540 mm
Height of top zone 340 mm
determination of the RTD in the liquid phase exit zone).
Interior diameter in packing zone 80 mm For the liquid phase, an aqueous solution of 0.5 mol/kgH2 O
Interior diameter in bottom zone 160 mm KCl, which can be quantitatively detected by a conductivity cell,
Interior diameter in top zone 80 mm has been selected as tracer. In the gas phase, the reactive compo-
Sectional area of packing zone S 5.027 × 10−3 m2 nent CO2 has been used and, because of the physical solubility
Total volume of packing zone, Vtotal 6.535 × 10−3 m3
of the tracer in water, only the gas phase flow behaviour without
Packing properties (ceramic spheres) liquid counter-current has been determined in the experiments.
Nominal diameter, dP 15 mm
The irrigation impact of the liquid counter flow on the gas phase
Number of spheres per m3 325,000
Porosity of the random packing, ε 0.45 m3 /m3 axial mixing in the packing zone has been estimated according
Specific surface of the packing, aS 230 m2 /m3 to data reported by Charpentier and Midoux [13].
Ratio actual/critical surface tension, σ C /σ 1 (estimated) In order to evaluate the mass transfer model, absorption
experiments are carried out using pure water and an aqueous
solution of K2 CO3 as solvents. The CO2 /water experiments are
tics are presented in Table 2. It must be emphasized that the pilot carried out at ambient temperature. The absorption experiment
plant represents a test facility for experiments to determine the into carbonate solution is run at an increased temperature of
effectiveness of different rate promoters on the hot potassium about 328 K (55 ◦ C). The absolute pressure in the column is
carbonate process. Therefore, the geometric properties (ratio 5 bar for all absorption experiments. The gas concentration is
between packing size and column diameter, liquid redistribution, measured in the middle of the column between the two packing
etc.) are not perfectly accorded to industrial design and the pack- sections and at the gas exit at the top of the column (Fig. 2).
ing material has been chosen in order to allow simple calculation
of the maximum interfacial area. Wall effects and channelling 4. Results and discussion
can therefore be expected in the pilot plant to a larger extent
than this is acceptable for industrial installations. However, the In the development of the dynamic model for the pilot plant,
measured liquid RTD indicates that no significant bypassing of the natural shape of the installation as well as the results of
liquid due to maldistribution is observed. The hydrodynamic preliminary RTD measurements have been taken into account.
parameters determined in the present work will allow proper Fig. 3 shows the dynamic model structure for the pilot plant.
interpretation of future experiments on the same installation. The additional zones represent free volumes in the column
The presented method for determination of the hydrodynamic and the pipes between signal generation and column as well as
parameters can directly be adapted to other pilot or even indus- between column and signal registration. The idealized types of
trial plants. phase flow behaviour used in the modelling are ideally mixed
In order to describe the dynamic behaviour of the pilot plant, zones and plug flow with axial dispersion. In the bottom of the
measurements of the residence time distribution (RTD) have column appears a dead volume (stagnant zone) for the gas phase.
been performed for different sections of the installation. The Because of the linear character of the dynamic transition func-
division of the geometry in different characteristic zones could tions of the modelled volumes, the final signal is not influenced

Fig. 3. Model of the laboratory setup including the parameters describing the dynamic behaviour.
 T. Pröll et al. / Chemical Engineering and Processing 46 (2007) 262–270 267

Table 3
Parameters describing the dynamics in the additional zones
Section in the model Symbol Value or correlation,
respectively

VIM,G,bottom 0.50 dm3

Gas phase entry in the
VSZ,G 2.15 dm3
bottom of the column
kSZ 1.55 × 10−2 s−1
VIM,G ,Top 0.98 dm3
Gas phase exit at the top of the
VAD,G 1.70 dm3
column and pipes to gas analyzer
PeAD,G 49
VIM,L 1.68·hB − 0.224 (dm3 ),
Liquid phase pipes and exit volume
hB in dm
in the bottom of the column
VAD,L 0.31 dm3
PeAD,L 9.07L0 + 196, L0 in
cm3 s−1

by the sequence of the additional RTD units. This allows chang-

ing of the order of the models from Fig. 3 in the simulation. In
more than 100 tracer experiments, which are carried out on sep-
arate parts of the installation and on the whole installation, the
13 dynamic parameters from Fig. 3 can be successively quan-
tified starting with the additional zones. The parameters of the
Fig. 4. Typical gas phase tracer signals after parameter optimisation for the
additional zones are summarized in Table 3. The liquid volume
packing zone (p = 5 bar, V̇N,G = 1.0 m3N /h).
in the bottom of the column depends on the liquid level in the
bottom zone hB and the liquid exit Peclet number depends on
the liquid flow rate L. The rest of the values from Table 3 are obtained are within a ±20% range around the average value of
constant within the tested operating range. Bo0G = 1.17.
In order to determine the dynamics of the packing zone, the This value agrees very well with the gas phase Bodenstein
RTD of the entire installation is measured and the known RTD of number found by Charpentier and Midoux [13] for 10 mm
the additional zones is taken into account for model evaluation. spheres and Raschig rings of the dimensions 10.3 and 6.4 mm.
Practically, the simulation creates a signal that reflects the RTD These authors also report values for the gas phase axial mix-
of the additional zones according to Table 3 (both before and ing when liquid countercurrent occurs [13]. If the irrigation is
after the packing zone) and uses this signal as input for the quantified in terms of the liquid phase Reynolds number, the
packing zone RTD model. The parameters of the packing zone influence of the liquid flow rate on the gas phase Bodenstein
can then be fit in order to obtain agreement between measured number is given by
and simulated final response. The fitting procedure is based on
the minimisation of square errors between the signals (Fig. 4). BoG = Bo0G ezReL , (29)
The results for the packing zone are presented in terms of the
liquid hold up and the Bodenstein numbers (Eq. (4)). The flow
rates are expressed in terms of the Reynolds number. This way
of description decouples the results from the shape of the pilot
plant and comparison with reported results is directly possible.
Sater and Levenspiel [14] found, that both, liquid hold up and
liquid phase axial dispersion, are not significantly influenced by
the gas flow rate, and that the form of equation proposed by
Otake and Okada [15] to describe the liquid hold up can as well
be used for the dispersion group, i.e. for the Bodenstein number.
Liquid hold up and Bodenstein number for the 15 mm spheres
are correlated to ReL and compared to the dynamic hold up
results obtained by Sater and Levenspiel [14] for 12.7 mm Berl
saddles and 12.7 mm Raschig rings in Figs. 5 and 6.
The main part of the free volume is occupied by the gas phase.
The gas phase hold up is, therefore, calculated as the difference
between free volume and liquid hold up. As far as the axial
mixing of the gas flow in the dry packing is concerned, there
is, according to Fig. 7, no significant dependency of Bo0G on the
gas flow rate or on pressure to be observed. Most of the values Fig. 5. Liquid hold up vs. liquid phase Reynolds number.
268 T. Pröll et al. / Chemical Engineering and Processing 46 (2007) 262–270

Fig. 6. Liquid phase Bodenstein number vs. liquid phase Reynolds number for Fig. 7. Gas phase Bodenstein number vs. gas phase Reynolds number in the dry
different gas countercurrents. packing for different pressures.

Table 4 the estimation of the influence of the liquid counter current on

Coefficient Z in Eq. (29), based on data reported by Charpentier and Midoux
[13]
the gas phase flow characteristics in the present work, Z was set
to −0.020.
Source Packing type Coefficient Z In a next step, the prediction method for the actual interfacial
Charpentier and Midoux [13] Spheres (10 mm) −0.0318 area and for the physical mass transfer parameters is verified in
Charpentier and Midoux [13] Raschig rings (10.3 mm) −0.0168 experiments with absorption of CO2 into water. Another exper-
Sater and Levenspiel [14] Raschig rings (12.7 mm)/Berl −0.0067 iment is carried out with absorption into carbonate solution. In
saddles (12.7 mm)
DeMaria and White[16] Raschig rings (9.53 mm) −0.0110
this case, the chemical absorption model can be validated. The
parameters of the steady state experiments are summarised in
Table 5.
where Bo0G is the Bodenstein number without liquid countercur- The steady state gas phase concentration profiles are shown
rent. in Fig. 8. The agreement between experiments and simulation is
Different values of the coefficient Z in the exponent of Eq. good for the tested conditions in all three cases. The results for
(29) can be extracted from literature data. The results for the the absorption into water (experiments 1 and 2 in Table 5) indi-
measurements of different authors, all reported by Charpentier cate that the estimation methods for the physical mass transfer
and Midoux [13], are presented in Table 4. parameters kG , kL0 , and a are able to describe the behaviour of
The values for Z are widely dispersed despite the similar the pilot plant. The experiment with absorption into carbonate
shape and dimension of the investigated types of packing. For solution shows that the applied method of predicting the mass

Table 5
Operating parameters of the pilot plant during steady state absorption of CO2 into water and into potassium carbonate solution
Process parameter Unit Experiment 1 Experiment 2 Experiment 3

Absolute pressure in the column bar (abs) 5.0 5.0 5.0

Flow rate of gas at exit m3N /h 0.6 0.6 0.6
Specific flow rate of inert gas in absorber kmolinert /(m2 h) 5.1 5.1 5.1
Flow rate of liquid at entry dm3 h−1 80 100 60
Specific flow rate of liquid in absorber m3 /(m2 h) 16 20 12
Gas composition at entry (CO2 and N2 ) vol.% CO2 10.2 10.2 10.2
Gas composition at entry (Yin ) molCO2 /molN2 0.114 0.114 0.114
Molality of carbonate in aqueous solution (m) molinitial
K2 CO3 /kgH2 O 0.00 0.00 1.51
Liquid concentration at entry (Xin ) molCO2 /kgH2 O 0.000 0.000 0.335
Temperature of the liquid at entry K (◦ C) 290 (17) 290 (17) 333 (60)
Temperature in the bottom of the column K (◦ C) 291 (18) 291 (18) 323 (50)
Gas phase Reynolds number – 33 33 30
Gas phase Peclet number – 28 20 23
Liquid phase Reynolds number – 65 81 74
Liquid phase hold up m3 /m3total 0.071 0.085 0.078
Liquid phase Peclet number – 25 31 28
 T. Pröll et al. / Chemical Engineering and Processing 46 (2007) 262–270 269

model parameters have been determined by experiments. The

description of acid gas absorption is based on film theory, axial
dispersion, and rate enhancement. The model is kept simple
according to the limited operating parameters of the pilot plant.
This means on the one hand that the energy balance has been
omitted in favour of a measured temperature profile governed
by heat exchange to the environment. On the other hand, evap-
oration of water has not been taken into account because of the
low impact on the results in the investigated conditions. How-
ever, the good agreement between measurements and simulation
results shows that the model has got a suitable structure and that
it is of sufficient complexity for describing the hydrodynamic
behaviour of the pilot plant. Comparison of the obtained hold up
and axial dispersion data to values reported in the literature gives
a plausible picture within the range of investigated flow rates. In
experiments with absorption into water, the prediction method
for the physical mass transfer parameters and for the actual inter-
facial area has been verified. The results do not indicate a need for
Fig. 8. Steady state gas phase concentration profiles along the packing zone— changing the applied correlations. In the case of absorption into
comparison between simulation and experiment according to Table 4. potassium carbonate solution, the rate controlling effect of liquid
phase reactions is modelled. Comparison between simulation
and experiment shows good agreement at a temperature of 55 ◦ C.
For reliable evaluation of the enhancement model, further exper-
iments at varied conditions are required. For simulation of indus-
trial conditions, i.e. higher absorption rates and higher operating
temperatures, the energy balance and the impact of water evapo-
ration will have to be taken into account. The present work rep-
resents the basis for further studies on the absorption pilot plant.
The obtained knowledge about the axial dispersion behaviour
of the phases in the packing is important for evaluation of more
complex mass transfer models describing the rate promoting
effect of certain chemical additives to the carbonate solution.

Appendix A. Nomenclature

a actual interfacial area (m2 /m3 )

aS specific surface of the random packing (m2 /m3 )
Fig. 9. Rate enhancement factor and the criterion for pseudo first order reaction
by Danckwerts [8] along the packing zone. Parameters according to experiment A coefficient in Eq. (13) (div.)
3 in Table 5. Bo Bodenstein number Bo = wdP /DAX
Bo0G gas phase Bodenstein number without irrigation
transfer with chemical reaction works well at least for the chosen dP nominal diameter of packing bodies (m)
parameter set. Fig. 9 shows the local rate enhancement factor and DAX axial dispersion coefficient (m2 /s)
the left hand side of Eq. (14) (Danckwerts pseudo first order reac- DL,CO2 binary diffusion coefficient of CO2 in the solvent (m2 /s)
tion criterion) for the conditions of experiment 3. The increase of E rate enhancement factor (1)
the rate enhancement factor with height can be attributed to the G0 specific flow rate of inert gas in the column
chosen temperature profile. The Danckwerts criterion is well (molinert /(m2 s))
below 0.1 throughout the absorbtion column, what proves the h local height coordinate in the packing zone (m)
applicability of the pseudo first order reaction mechanism [11] hB liquid level in the bottom of the column (m)
followed in the present work. For a profound evaluation of the H height of packed bed (m)
chemically enhanced mass transfer model, however, more exper- Ha Hatta number defined by Eq. (16)
imental data will be required from varied operating parameters. He Henry’s law constant for CO2 in the solvent (Pa kg/mol)
He0 Henry’s law constant for CO2 in pure water (Pa kg/mol)
5. Conclusions kG gas side physical mass transfer coefficient
(mol/(Pa m2 s))
A dynamic model of a pilot plant absorber including the kI,CO2 pseudo first order reaction constant for reaction (22)
dynamics of entry and exit zones has been developed and the (1/s)
270 T. Pröll et al. / Chemical Engineering and Processing 46 (2007) 262–270

kL0 liquid side physical mass transfer coefficient (m/s) aq.,i aqueous CO2 (non-dissociated) at gas–liquid interface
kOH− second order rate constant of reaction (22) AD plug flow with axial dispersion
(kgH2 O /(mol s)) bottom referring to the bottom of the column
kSZ exchange coefficient stagnant/ideal mixed volume (1/s) G gas phase
Kw equilibrium constant for reaction (19): i location node; alias for water, CO2 , or HCO3 − in Eq.
Kw = [H+ ][OH− ] (mol2 /kgH2 O 2 ) (13)
KCO2 equilibrium constant for reaction (17): in at entry in the packing zone
KCO2 = [H+ ][HCO3 − ]/[CO2,aq. ] (mol/kgH2 O ) inert inert part of the gas phase, i.e. gas without CO2
KG overall gas side mass transfer coefficient initial initial K2 CO3 solution without any CO2 absorbed
(mol/(Pa m2 s)) IM ideally mixed
KHCO3 − equilibrium constant for reaction (18): j time node
KHCO3 − = [H+ ][CO3 2− ]/[HCO3 − ] (mol/kgH2 O ) L liquid phase
L0 specific liquid flow rate in the column (m3 /(m2 s)) top referring to the top of the column
m molality; initial K2 CO3 molality in the solvent total referring to total packing zone volume, total gas (i.e.
m = 2[K+ ] (mol/kgH2 O ) CO2 + inert)
p absolute pressure (Pa) SZ stagnant zone
pCO2 partial pressure of CO2 (Pa)
Pe Peclet number Pe = wH/DAX
R universal gas constant R = 8.314 (J/(mol K)) References
Re Reynolds number (–)
[1] E.Y. Kenig, R. Schneider, A. Górak, Rigorous dynamic modelling of
S sectional area of the column (m2 )
complex reactive absorption processes, Chem. Eng. Sci. 54 (1999)
t time (s) 5195–5203.
T absolute temperature (K) [2] L. Kucka, I. Müller, E.Y. Kenig, A. Górak, On the modelling and simulation
t time step size for integration (s) of sour gas absorption by aqueous amine solutions, Chem. Eng. Sci. 58
V volume (m3 ) (2003) 3571–3578.
[3] F. Kastanek, J. Zahradnik, J. Kratochvil, J. Cermak, Chemical reactors for
V̇NG normal volume flow of gas (normal: 1 bar, 273.15 K)
gas–liquid systems, in: Ellis Horwood Series in Chemical Engineering,
(m3N /h) 1993.
w phase flow velocity without packing (superficial veloc- [4] G. Astarita, D.W. Savage, A. Bisio, Gas Treating With Chemical Solvents,
ity) (m/s) John Wiley & Sons, New York, 1983.
x location and time dependent variable in Eqs. (26)–(28) [5] N. Plummer, E. Busenberg, The solubilities of calcite, aragonite, and
vaterite in CO2 –H2 O solutions between 0 and 90 ◦ C, and an evaluation
X liquid composition (molality of total CO2 combined)
of the aqueous model for the system CaCO3 –CO2 –H2 O, Geochim. Cos-
(molCO2 /kgH2 O ) mochim. Acta 46 (1982) 1011–1040.
yCO2 gas phase mole fraction of CO2 (mol/moltotal ) [6] C. Tsonopoulos, D.M. Coulson, L.B. Inman, Ionization constants of water
Y composition in the bulk of gas phase (CO2 load on inert pollutants, J. Chem. Eng. Data 21 (1976) 190–193.
gas) (molCO2 /molinert ) [7] K. Onda, H. Takeuchi, Y. Okumoto, Mass transfer coefficients between
gas and liquid phases in packed columns, J. Chem. Eng. Jpn. 1 (1968) 56–
z dimensionless height of column h/H
62.
z dimensionless height step size for integration [8] P.V. Danckwerts, M.M. Sharma, The absorption of carbon dioxide into
Z coefficient in Eq. (29) solutions of alkalis and amines (with some notes on hydrogen sulphide and
carbonyl sulphide), Chem. Eng. (1966) 244–280.
Greek symbols [9] P.V. Danckwerts, Gas–Liquid Reactions, McGraw-Hill Book Company,
UK, 1970.
α partial saturation of the carbonate system
[10] D.W. Savage, G. Astarita, S. Joshi, Chemical absorption and desorption of
(mol HCO3 − /molinitial
K2 CO3 ) carbon dioxide from hot carbonate solutions, Chem. Eng. Sci. 35 (1979)
βL liquid phase hold up (m3L /m3total ) 1513–1522.
[11] A.H.G. Cents, D.W.F. Brilman, G.F. Versteeg, CO2 absorption in carbon-
βGinert inert gas hold up (molinert /m3total )
ate/bicarbonate solutions: the Danckwerts-criterion revisited, Chem. Eng.
ε porosity of random packing (m3 /m3 ) Sci. 60 (2005) 5830–5835.
μ dynamic viscosity (Pa s) [12] C.R. Wilke, P. Chang, Correlation of diffusion coefficients in dilute solu-
ρ density (kg/m3 ) tions, AIChE J. 1 (1955) 264–270.
ρ H2 O density of pure water (kg/m3 ) [13] J.C. Charpentier, N. Midoux, Distribution des temps de séjour dans la phase
gazeuse d’un contacteur gaz–liquide, Chimie & Industrie—Genie Chim-
σ surface tension of the liquid (N/m)
ique 104 (1971) 1563–1574 (in French).
σC critical surface tension for the packing material (N/m) [14] V.E. Sater, O. Levenspiel, Two-phase flow in packed columns, I & EC
Fundam. 5 (1966) 86–92.
Indices [15] T. Otake, K. Okada, Liquid hold-up in packed towers; operating hold-up
* in equilibrium without gas flow, Kagaka Kogaku 17 (1953) 176.
[16] F. De Maria, R.R. White, Transient response study of gas flowing through
0 physical, dry packing, specific to cross section of
irrigated packing, AIChE J. 6 (1960) 473–481.
column