Energy Conversion and Management 254 (2022) 115216

Contents lists available at ScienceDirect

Energy Conversion and Management

journal homepage: [Link]/locate/enconman

A direct contact cooler design for simultaneously recovering latent heat and
capturing SOx and NOx from pressurized flue gas
Piyush Verma , Zhiwei Yang *, Richard L. Axelbaum
Dept. of Energy, Environmental & Chemical Engineering, Washington University in St. Louis, St. Louis, MO 63130, USA

A R T I C L E I N F O A B S T R A C T

Keywords: Integrated SOx and NOx removal technology using a direct contact cooler is a promising alternative for cleaning
SOx coal combustion flue gas, especially for pressurized combustion systems. Past investigations of integrated SOx
NOx and NOx removal have been limited to low-temperature processes, where the flue gas has been cooled upstream
Latent heat
of the removal unit, and the moisture in the flue gas has been condensed out. This work presents a novel design
Pressurized
Combustion
that recovers the flue gas heat while simultaneously removing SOx and NOx in a single reactive-absorption
Oxyfuel column – direct contact cooler. A validated model was used to evaluate different direct contact cooler de­
signs. Modeling results suggested that a counter-current column with a single water inlet can scrub 83% of the
NOx and 96.5% of the SOx in 115 s, while an optimized design with multiple water inlets enhances the scrubbing
of NOx by 9 percentage points and SOx by 3.5 percentage points, due to a 34% increase in residence time. It is
also observed that the liquid-phase reaction between absorbed SOx and NOx plays a significant role in the
removal of SO2 from the flue gas for high-temperature scrubbing compared with low-temperature scrubbing. The
design reinforces the potential of high-temperature SOx-NOx removal and heat recovery and suggests a means of
reducing capital costs for such columns.

upstream combustion process.

The second significant benefit of high pressure is the potential to
1. Introduction remove SOx and NOx simultaneously and economically in a reactive-
absorption column. The SOx and NOx in the flue gas are major pollut­
The goal of rapidly decarbonizing the electricity grid while main­ ants formed during coal and biomass combustion that also have the
taining its reliability has led to significant interest in carbon capture, potential to corrode plant equipment [6]. In conventional atmospheric
utilization, and storage (CCUS) technologies for power plants [1,2]. pressure combustion processes, the removal of SOx and NOx is achieved
Pressurized oxy-combustion (POC), where fuel is burned at elevated in two steps - a selective catalytic reduction (SCR) unit to remove NOx
pressure in the presence of oxygen and recycled flue gas, has emerged as and a flue gas desulfurization (FGD) unit to remove SOx [7,8]. In a POC
a promising lower-cost technology for the implementation of CCUS [3]. process, the partial pressure of O2 is higher in the flue gas because of the
The combustion products from solid-fuel-based POC processes consist higher total pressure. This helps to more rapidly oxidize NO into water-
primarily of CO2 and H2O, with small amounts of sulfur and nitrogen soluble NO2, making it easily removable in an absorption column [9].
oxides. The principal motivation for high-pressure combustion is the The solution-based reactions between nitrogen and sulfur species further
benefit of condensing the flue gas moisture at a higher temperature enhance removal of NOx and SOx, such that POC has brought the op­
compared to atmospheric pressure combustion [3,4]. The latent heat portunity for the integrated pollutant removal (IPR), where NOx and SOx
recovered from the flue gas at higher temperature can be integrated into are simultaneously scrubbed in a pressurized water-wash column, i.e.,
the steam cycle to increase plant efficiency. In a POC process operating direct contact cooler (DCC). A cost reduction of about 70% has been
at 15 bar, the integration of moisture latent heat can lead to a potential reported for this technology, compared with conventional cleaning
increase of 1.5–2.5 percentage points in net plant efficiency [5]. Since technologies using SCR and FGD [10].
the CO2 product needs to be compressed downstream for transportation Previous work on simultaneous removal of SOx and NOx focused on
and storage, and the cost of compressing oxygen is comparable to that of flue gas from atmospheric pressure oxy-combustion of coal. White et al.
compressing CO2, there is no additional cost for pressurizing the

* Corresponding author.
E-mail address: [Link]@[Link] (Z. Yang).

[Link]
Received 6 October 2021; Received in revised form 22 December 2021; Accepted 4 January 2022
Available online 22 January 2022
0196-8904/© 2022 Published by Elsevier Ltd.
P. Verma et al. Energy Conversion and Management 254 (2022) 115216

studied by several researchers, which makes the study of removal of SOx

Nomenclature and NOx even more relevant [16,17]. Co-removal of SOx and NOx was
studied using the addition of ClO2 in the gas phase, which enhances the
List of Symbols and Abbreviations oxidation rate of NO, but the reaction is limited to a gas temperature of
SOx Sulfur Oxides 155 ◦ C [18]. Another IPR process that uses seawater/urea peroxide to
NOx Nitrogen Oxides remove SOx and NOx while simultaneously reducing the concentration
CCUS Carbon Capture, Utilization, and Storage of nitrate in the liquid phase, has also been proposed by Zhou et al. [19].
POC Pressurized Oxy-Combustion Although, these additives promise good removal efficiency they are
IPR Integrated Pollutant Removal limited by temperature, as well as add to the operational cost of the
ri Rate of reaction ‘i’ system. In this work, we propose a novel design for a DCC operated at 15
Ci Molar Concentration of Species ‘i’ bar for a 550 MWe pressurized oxy-combustion power plant, and we use
DCC Direct Contact Cooler a validated model for performance evaluation. The conventional DCC
SCR Selective Catalytic Reduction design typically contains a counter-current absorber with one water
FGD Flue Gas Desulfurization inlet at the top and one gas inlet at the bottom of the column. The design
L/G Liquid to Gas Ratio proposed in this work allows for multiple water inlets. A model is used to
kconventional Rate constant at each stage for NO oxidation in optimize the number of inlets and the water flow rate for each inlet to
conventional design maximize the NOx and SOx removal. The reaction kinetics used in the
koptimized Rate constant at each stage for NO oxidation in model are validated against results from a 100 kWth pilot-scale DCC
optimized design experiment. We compare the temperature and volumetric flow profiles
RNO_conv Rate of NO conversion to NO2 in conventional design for gas and liquid in the conventional and new designs and discuss the
RNO_opt Rate of NO conversion to NO2 in optimized design effect of moisture condensation on the residence time and gas and liquid
reaction kinetics. Finally, comparison and analysis of the differences in
SOx and NOx removal efficiency are presented. This work is the first
effort to model and optimize a single DCC column that integrates heat
[11] used a slipstream from an atmospheric-pressure, 160 kWth oxyfuel recovery and SOx and NOx removal. Moreover, the work explores the
combustor to evaluate SOx and NOx removal. The flue gas was cooled to potential to simultaneously remove SOx and NOx in a POC power plant at
room temperature such that the majority of the moisture was removed high temperature, without the addition of reactive additives or catalysts.
before compressing to 5 – 15 bar to prevent acid condensation in the
compressors. They concluded that it was possible to remove SOx and
2. Methodology
NOx components from oxy-fuel-derived CO2 without employing FGD or
SCR technologies. Murciano et al. [12] performed lab-scale experiments
2.1. Modeling methodology
on dry synthetic flue gas to understand the effect of pressure and resi­
dence time on NOx and SOx scrubbing. They reported that NO is readily
The design of the DCC is based on flue gas from a 550 MWe pres­
oxidized to NO2 at high pressure in the presence of oxygen. However,
surized oxy-combustion plant using Montana PRB coal [20]. The flue gas
SO2 was not oxidized to SO3. They also noted an interaction between
conditions and composition are presented in Table 1. The 1.2% oxygen
SO2 and NO2 in the presence of water vapor, however, the analysis was
concentration in the flue gas is based on the work of Zhiwei et al. [21],
complicated by the formation of acid droplets. A similar study by White
which showed that complete combustion of coal particles could be
et al. [13] on an atmospheric oxy-combustion slipstream analyzed SOx
achieved in oxy-combustion with an excess oxygen concentration of just
and NOx removal in a two-stage system of 15 and 30 bar. They reported
1% when operating at 15 bar. The inlet concentration of NOx and SOx to
that NO reduction was a function of residence time and pressure, while
the DCC depends on the amount of sulfur and nitrogen in the coal. Inlet
SO2 removal was affected by both residence time and make-up water
concentrations of 900 ppm SO2 and 900 ppm NO are assumed for this
flow rate. These studies confirmed that SOx and NOx removal at high
analysis, which is in the same range as the values assumed by Ajdari
pressure, low temperature, and high liquid-to-gas ratio (L/G) was
et al. [15,22] and IIoeje et al. [15,22] for similar analyses. The trace
possible, without the objective to recover the latent heat of moisture.
elements, such as chlorine and mercury, are not considered in the model
IIoeje et al. and Tumsa et al. [14,15] modeled and parametrically
since they are in small quantities and do not affect the chemistry of the
analyzed similar DCC processes, both of which established the possi­
process.
bility of pressurizing the flue gas and then removing SOx and NOx using a
The DCC was modeled as a rate-based reactive absorption column
water contact column. They suggested pressurizing the gas to 30 bar for
using a RadFrac unit in Aspen plus V9, as a rate-based model is required
greater NOx removal. Similar to the experimental system, these models
to analyze residence time constraints. The ionic interaction in the liquid
did not consider recovering the latent heat from the flue gas. The
phase requires the use of an electrolytic equation of state for the
moisture in the flue gas was removed in a cooler before compression,
and cold flue gas was fed into the column at high L/G ratios.
Table 1
Although these studies confirm the possibility of removing SOx and
Inlet modeling parameters.
NOx at high pressure, a detailed analysis is required for POC systems
where the flue gas is already at high pressure and temperature. As Parameter Value

proposed by Gopan et al. [3], the recovery of latent and sensible heat, DCC pressure (bar) 15
along with the removal of SOx and NOx in one single column, will further Inlet gas flowrate (kg/s) 205.3
Inlet gas temperature (◦ C) 200
reduce the complexity and cost of the POC power plant. In this case, the
Inlet water flowrate (kg/s) 212.5
DCC can act both as a heat exchanger to recover latent and sensible heat Inlet water temperature (◦ C) 24
from the flue gas and an absorber for SOx and NOx. Moreover, to Inlet gas composition (% v/v)
maximize plant efficiency, the water temperature exiting the DCC must CO2 55.4
be maintained at a relatively high temperature, which is accomplished H2O 40.2
O2 1.2
by constraining the L/G ratio, making scrubbing a high-temperature
NO 0.09
process [4,5]. The performance of SOx and NOx removal in a DCC at NO2 0.01
high water temperature and low L/G ratio has received little attention. SO2 0.09
Pressurized oxy-combustion systems for carbon capture are being SO3 0.01

2
P. Verma et al. Energy Conversion and Management 254 (2022) 115216

thermodynamic properties of the components. Hence, ELECNRTL was stage, were introduced in the column. The motivation behind this
used as the property method because of its compatibility with the system optimization was based on the fact that, since cooling and condensation
in both the liquid and gas phases. The column was modeled as a ten- of the gas seems to have a strong impact on the reaction kinetics of the
stage column, with gas introduced below the bottom-most stage. It column, a different heat transfer profile may result in a better scrubbing
should be noted the column design is a packed bed design, and the performance for SOx and NOx. The objective of the optimization routine
number of stages used in this work is only for mathematical calculation. was the minimization of the two target variables ‘NO outlet mole flow
A sensitivity analysis on the number of stages was performed (10–30 rate’ and ‘SO2 outlet mole flow rate,’ which were the exit molar flow rate
stages); however, the analysis found no relevant increase in accuracy of NO and SO2, respectively, in the gas phase from the top of the column.
with more than ten stages. For the conventional DCC design, water is Two constraints were imposed on the optimization routine 1) The total
introduced above the top stage. For the modified DCC design with water flow rate of the water should not change from the conventional
multiple water inlets, water can be introduced above each stage. The design, i.e., an L/G ratio of 1.03 kg/kg should be maintained, and 2) the
results are presented with the bottom-most stage numbered as Stage 1. water inlet from Stage 10, (the topmost stage) should be sufficient
The columns were assumed to be adiabatic with no heat loss to the enough to support a reasonable pressure drop and optimum mass
surroundings. transfer in the upper section. The water inlet flow rate in each of the
stages was allowed to vary to optimize the target variable. The details of
the optimization parameters are provided in Table 2.
2.2. Column design
The optimization results suggested that splitting the overall water
inlet into four inlet streams provided the best minimization of NO and
2.2.1. Conventional column
SO2 in the outlet flue gas. However, we found that the water distribution
A conventional design, presented in Fig. 1(a), is a counter-current
was majorly distributed into two streams, one stream at the top with
packed column with water coming from the top and gas from the bot­
23.5% of the flow, and another one at the 4th stage with 76% of the flow,
tom. For this study, the L/G ratio was determined by fixing the moisture
the rest of the 0.5% was distributed between stage 3 and stage 5. For
content of the outlet flue gas to be 1.5 % v/v and maximizing water
operational simplification and the fact that reducing water inlets to just
outlet temperature. The resulting L/G ratio was close to 1.03 kg/kg, and
two major streams made no significant difference to the optimized re­
the corresponding outlet water temperature was around 160 ◦ C. Based
sults, a simple design with only two water inlets is proposed and dis­
on this L/G ratio, the column diameter was sized for a 75% approach to
cussed in this study. The outlet SOx concentration and NOx
the flooding velocity for optimum mass transfer. The flooding velocity in
concentration in the optimized design were achieved below 50 ppm and
Aspen was calculated using the Eckert generalized pressure drop cor­
100 ppm, respectively. The required outlet concentrations are based on
relation (GPDC) [23]. The empirical chart developed by Eckert for
the CO2 transport pipeline quality guidelines [24,25]. These split ratios
packed towers provides a correlation between the flow parameter and
and injection locations could vary with the total moisture content in the
(G2 Fp φμ0.2 /ρg ρl gc ), where G is the gas mass flow rate per unit area, Fp is
flue gas, inlet gas and liquid temperature, and the inlet SOx and NOx
the packing factor, φ is the density of water/density of the liquid, μ is the
concentration. Hence, for a different coal and combustion design,
viscosity of the liquid, ρg andρl are the densities of gas and liquid,
further optimization would be needed.
respectively, and g is the gravitational constant. Using the correlation,
the diameter for the conventional column was calculated to be 5.6 m for
a 75% approach to the flooding velocity. The height of the conventional 2.3. Chemical kinetics and model validation
column was determined based on the targeted outlet concentration of
SOx (<50 ppm). The packing material selected was 80 mm Raschig rings In addition to the thermophysical water condensation process,
since this provides good mass transfer with a reasonable pressure drop in several gas- and liquid-phase reactions occur inside the DCC. The most
the column. important reaction for NOx removal is the oxidation of NO to NO2.
Tsukahara et al. [26] reviewed several studies on the gas-phase oxida­
2.2.2. Optimized column tion of NO to NO2 and reported that NO oxidation is a third-order re­
The proposed column, shown in Fig. 1(b), uses the same dimension action, depending on the partial pressure of NO and O2. They reported
and packing material as those of the conventional column in Fig. 1(a). that the kinetic constant for the reaction decreases slightly up to a
The optimization was performed using a multivariate optimization tool temperature of 500 ◦ C and then increases at higher temperatures. Since
in Aspen Plus. To begin with, ten auxiliary water inlets, one above each the DCC is designed to operate between gas temperatures of 20 to
200 ◦ C, it is important to consider the dependency of the kinetic constant
on temperature. Ting et al. [9] also reported similar reaction kinetics
with rate constants relevant for high-pressure operations.
In the DCC column, NO2 produced via,
2NO(g) + O2 (g)→2NO2 (g) (R1)

is readily absorbed in water to form HNO2 and HNO3 in a 1:1 ratio,

according to

Table 2
Optimization Parameters.
Optimization Target Variables
Parameters

Objective 1. Minimize molar concentration of NO in the outlet gas

2. Minimize molar concentration of SO2 in the outlet gas
Constraints 1. L1 + L2 +… L10 = Constant*
2. L10 should satisfy the minimum mass transfer value
(calculated separately)
Variables 1. L1, L2, L3 … L9

Fig. 1. (a) Conventional and (b) Proposed multi-water-inlet column design. *L = Mass flow rate of water.

3
P. Verma et al. Energy Conversion and Management 254 (2022) 115216

2NO2 (g) + H2 O(aq)→HNO2 (aq) + HNO3 (aq) (R2) Table 3. Two main findings were reported in this work: 1) the gaseous
species such as N2O4 have a negligible impact on the kinetics, and 2) the
The SO2 in the gas is absorbed in water to form HSO 3, according to
−
dissociation of HNO2 to HNO3 and NO is not significant in DCC condi­
SO2 (g) + H2 O(aq) ↔ HSO−3 (aq) + H + (aq) (R3) tions. The SO3 absorption reaction is not considered in the kinetic
models discussed above because SO3 is highly soluble in water and is
Lee et al. [27,28] studied the dissolution of NO2 in water at low removed at the entrance to the column; hence it does require a kinetic
partial pressure and reported that the dissolution was fast. The rate of analysis. However, it was included in the Aspen model presented in this
Reaction (2) used in the model is derived from their work. Reaction 3 is work. The reaction rates of the kinetically controlled reactions were
instantaneous and is treated as equilibrium reactions in the kinetic determined in Aspen Plus based on the reduced power-law expression,
model [27,28]. Similarly, SO3 dissolves instantaneously in water to form ( )
H2SO4 according to Reaction 4 and is also treated as an equilibrium E
ri = ki T n exp − Ca Cb Cd (1)
reaction. RT i j l

SO3 (g) + H2 O(aq) ↔ HSO−4 (aq) + H + (aq) (R4) where ki is the rate constant, T is the temperature of the phase the re­
Although the liquid-phase interactions have received less attention, action is occurring, and C is the concentration of the species determining
Normann et al. [29] found that complex interactions between HNO2 and the rate of the reaction. The product formation from the liquid phase
HSO− 3, which are capable of enhancing the oxidation rate to produce reaction was calculated using a sub-routine based on the pH at each
HNO3 and H2SO4, play an important role in the scrubbing chemistry. stage, which determines the stoichiometry.
They presented the Raschig mechanism to explain the interaction of This proposed kinetic model was validated in a small pilot-scale DCC
HNO2 and HSO3− in the liquid phase: by Stokie et al. [33]. The experiments were performed in a counter-
current packed bed with an 0.2 m diameter and 2.3 m of packing.
HNO2 (aq) + HSO−3 (aq)→NOSO−3 (aq) + H2 O(aq) (R5) Several parametric studies were performed at 15 bar pressure and inlet
gas temperatures of 24 and 202 ◦ C using synthetic flue gas containing
NOSO−3 (aq) + H + (+H2 O)→HNO(aq) + H2 SO4 (aq) (R6) SO2, NO, O2 and CO2. Experimental and model results for 15 bar pres­
sure and 24 ◦ C and 202 ◦ C inlet gas temperature are presented in Fig. 2.
NOSO−3 (aq) + HSO−3 (aq)→HNO(SO3 )2−2 (aq) (R7) In this experiment, the NO concentration was varied from 0 to 1000
ppm, with SO2 and O2 concentrations fixed at 500 ppm and 3% v/v,
HNO(aq) + HNO(aq)→N2 O(g) + H2 O(aq) (R8) respectively. The residence time of the gas in the column was main­
tained at 120 s for each condition. For these conditions, the NO removal
HNO(SO3 )2−2 (aq) + H + (aq)→HNOHSO−3 (aq) + H + (aq) + HSO−4 (aq) (R9) increased with increasing inlet NO concentration, and SO2 removal also
They concluded that the liquid phase interaction between HNO3– and increased with increasing NO for each case. The kinetic model presented
HNO2 should be further investigated in the range of pH relevant to DCC. in Table 3 was capable of accurately predicting the outlet concentration
Based on a sensitivity analysis of the primary reactions, Ajdari et al. of NO and SO2. Experimental system details and additional results can
[22,30] proposed a reduced mechanism of between 7 and 12 reactions be found in the study by Stokie et al. [33].
for the relevant pH range. They concluded that the gas-phase reactions
are limited by NO oxidation, whereas the liquid-phase interactions are 3. Result and discussion
predominantly pH controlled. They reduced the liquid-phase mecha­
nism to two major reactions, Reaction 10 and 11: For the base case analysis, a conventional column with one water
inlet at an L/G ratio of 1.03 kg/kg was simulated for the targeted outlet
1 1
HNO2 + HSO−3 →HSO−4 + N2 O + H2 O (R10) moisture and SOx concentration. An analysis of the time scales for heat
2 2 and mass transfer and the rate of chemical reaction revealed that the
removal of SOx and NOx in the column is limited by the oxidation of NO
HNO2 + 2HSO−3 →HON(SO3 )2−2 (HADS) + H2 O (R11)
to NO2. This result was also observed in the experiments by Stokie et al.
In our previous work [31], we further reduced the entire mechanism [33] and Stanger et al. [34]. Since the L/G ratio and the flue gas oxygen
to a total of 5 reactions (R1, R2, R3, R10, and R11) through an experi­ content are both fixed, the removal of NO and SO2 depends on the
mental and mathematical optimization of the model, as shown in residence time and temperature profile in the column. The residence
time enhances the overall reaction time for NO oxidation, while the
temperature profile impacts the local kinetic constant of both gas- and
Table 3 liquid-phase reactions. The proposed new column design, with the same
Reduced reaction mechanism used in the modeling. column height, diameter, and packing, was modeled and operated to
Reaction Type Rate (mol m− 3 − 1
s ) achieve a similar moisture outlet concentration and NOx (<100 ppm)
1 2NO + O2 → 2NO2 Kinetic [9]
(
530.4
) and SOx (<50 ppm) concentrations less than the target.
3
r = 2.4 × 10− × exp ×
T
C2NO × CO2
3.1. Gas temperature profile comparison
2 2NO2 + H2O → HNO2 + Kinetic [27]
r = 1.6 × 106 ×
HNO3
(
− 13109.67
) The Aspen results indicate that in a conventional column, it is
exp × C2NO2
T possible to remove SOx and NOx to the required standards. These results
3 SO2 + H2O ↔ HSO3− + H+ Equilibrium Keq from Aspen
also reveal that only a small fraction of this residence time is required for
4 SO3 + H2O ↔ HSO− 4 + H+ Equilibrium Keq from Aspen
5 HNO2 + HSO3− →HSO4− Kinetic*
(
− 6981
) cooling and condensing the moisture in the gas, owing to a faster time
r = 1.1 × 10− 6 exp × constant for heat transfer compared to NO oxidation. A majority of the
+ ½ N2O + 0.5 H2O [32] T
CHNO2 CHSO3 CH
− + cooling and condensation is achieved in about 40% of the column height
( )
6 HNO2 + 2HSO3− →HNO Kinetic*
r = 1.1 × 10− 6 exp
− 6981
×
with the specified L/G ratio. In the conventional column, most of the
(SO3)22– + H2O [32] T cooling and condensation occurs in the top section of the column due to
CHNO2 CHSO−3 CH+
the counter-current, one-water-inlet design.
*The rate of reactions 5 and 6 are equal, but stoichiometry of consumption of In Fig. 3, a comparison between the gas temperature profile in the
HSO3− varies with pH [32]. conventional column and the optimized column is presented. In the

4
P. Verma et al. Energy Conversion and Management 254 (2022) 115216

Fig. 2. Kinetic model validation.

Fig. 3. Gas temperature profile in the two designs.

Fig. 4. Comparison of the rate constants for NO oxidation for each stage of the
two designs.
optimized column, most of the water (76%) is fed above the 4th stage,
which is a sufficient height for the majority of heat transfer. This
NOx in the optimized design.
modification leads to a faster temperature drop in the bottom part
(below the lower water inlet) of the optimized column. The gas outlet
temperature is slightly higher than that of the conventional design 3.2. Change in volumetric flow rate and residence time
because a smaller L/G ratio in the upper section leads to a reduced
surface area of contact. A different packing in the upper column could As seen in Fig. 5a, the total volumetric gas flow rate in the optimized
yield better heat transfer. design decreases significantly faster in the lower section. This is due to
The kinetics for NO oxidation is significantly affected by the gas earlier condensation of moisture and faster cooling of the gas, both of
temperature profile. In Fig. 4, the rate constant for the NO oxidation which reduce volumetric flow rate. In the conventional design, most of
reaction for the optimized column is compared to that of the conven­ the cooling and condensation occurs in the upper section of the column.
tional DCC for each stage. Except for the top stage (Stage 10), the rate In contrast, in the optimized design, they are largely completed by Stage
constant ratio (koptimized/kconventional) is higher than unity throughout 5, leading to a much lower flow rate in the remaining part of the column.
the length of the column because the flue gas temperature is lower in the It should be noted that the inlet flue gas for this study contains close to
optimized design than in the conventional design. The highest ratio is in 40% moisture, which leads to a sharp drop in volumetric flow rate when
Stages 5 through 7, where the rate constant is 35% higher in the opti­ condensation occurs. Depending on the amount of moisture in the flue
mized column, as this is where the gas temperature difference between gas, the design can be modified to optimize the water inlet for fast
the two columns is greatest. On average, over the entire column, there is condensation. On the conventional column, the volumetric flow rate
a 13.2 % increase in the rate constant for NO oxidation. This increased drops close to 60% in the entire length of the column, from 15 m3/s to
NO oxidation rate is one factor that contributes to the higher removal of close to 6 m3/s. However, in the optimized column, most of this drop is
accomplished by Stage 5.

5
P. Verma et al. Energy Conversion and Management 254 (2022) 115216

Fig. 5. Comparison of a) Volumetric flow profile of gas and b) Volumetric flow profile of water, and c) Mole Fraction of H2O in the gas phase, in the conventional and
optimized design.

Fig. 5(b) presents the total water flow rate at each stage. It should be validates the fact the majority of the decrease in volumetric gas flow
noted that the water flow is from Stage 10 to Stage 1. The total water at comes from water condensing from the gas phase to the liquid phase.
the outlet (Stage 10) increases from 0.22 m3/s to 0.28 m3/s due to water The mole fraction of H2O in the gas phase drops sharply in the optimized
condensation from the flue gas in both columns. For the conventional design from 0.42 to <0.02. Although the extent of the recovery of latent
column, the increase in the flow occurs between Stages 6 and 9, while in heat remains the same in both designs, the time at which it occurs
the optimized design, the addition of water due to moisture condensa­ changes the dynamics in the system. Doing this increases both the
tion is completed before the second water inlet (between Stages 1 to 4). residence time of the gas (cooler gas has much less volume) and the
The sudden rise of water flow between stages 4 and 5 in the optimized kinetic rate constant for NO oxidation. It should be noted that, due to
column is because of the 2nd water inlet. The height required for the decrease in the volumetric flow of the both gas and liquid in the upper
cooling and condensing of the flue gas is relatively small, as can be seen section of the optimized column, the pressure drop decreases. Although,
in the temperature plot of the original design, where most of the cooling the model suggests that the contact between liquid and gas remains close
and condensation is completed in the top 3 – 4 stages. A similar rate of to the optimum value, engineering modifications, especially a more
gas temperature decrease can be seen in the optimized design as well, suitable packing size, may be considered.
only, at a different location. The optimized design, therefore, exploits The different gas flow rates over the ten stages lead to different cu­
the fact that heat transfer is a fast process compared to NO oxidation rate mulative gas residence times in the two columns, as shown in Fig. 6. The
which determines the height of the column. A column with the sole total residence time in the optimized column is 39 s longer (or 34%
objective of recovering the latent heat would require about 40% of the larger) than that in the conventional column. The relative increase in the
total length of the design presented in this work. The change in the residence time is apparent above Stage 4. Since the removal of NOx, and
concentration of H2O in both the designs is presented in Fig. 5(c), which consequently SOx, is limited by the NO oxidation rate, a longer residence

6
P. Verma et al. Energy Conversion and Management 254 (2022) 115216

conversion over the ten stages are presented in Fig. 7(a) and (b),
respectively. As seen in Fig. 7(b), between Stages 4 and 6, the total NO
converted to NO2 in the optimized column is 2–4 times higher than the
conversion in the conventional column. As discussed in relation to Fig. 4,
the ratio of the rate constants also peaks between Stages 4 to 7. The
overall impact of the increase in the rate constant, residence time, and
NO partial pressure in these stages makes the NO removal high. Even
though the residence time is higher in the upper stages, the NO con­
centration in the optimized column has been reduced to a smaller value,
and hence the overall rate becomes slightly smaller than the conven­
tional column for Stage 7 and beyond. It should be noted that the benefit
of the optimized design is not limited to the pressurized oxy-combustion
processes and should be applicable to other removal technologies having
similar kinetic and heat removal constraints.
SO2 removal depends upon the interaction between absorbed NO2
(through HNO2) and absorbed SO2 (through HSO3− ). The liquid phase
reaction is first order in both HNO2 and HSO3− , hence an increase in the
concentration of either reactant increases the reaction rate. Fig. 8
demonstrates the comparison of SO2 flow in the two designs. For the
conventional column, SO2 is removed slowly and almost linearly in the
bottom stages but rapidly in Stages 7–10. However, in the optimized
column, SO2 reduction is rapidly in the first five stages. This is due to
two reasons. First, in the conventional design, the NO oxidation rate is
Fig. 6. Comparison of gas residence time in the conventional and opti­ high in the top section, which results in higher HNO2 concentration in
mized design. the liquid phase in Stages 7–10, and a higher liquid phase reaction rate.
In the optimized design, the NO oxidation is high between Stages 3–6,
time increases the total amount of NO converted to NO2. This results in resulting in high HNO2 concentration and a higher reaction rate in the
higher NOx removal and, consequently, higher SOx removal because of bottom stages. Also, since most of the SO2 is already consumed in the
the liquid phase interactions. bottom stages, the unreacted HNO2 from the top stages flows down and
reacts with HSO3− leading to higher removal rates. The second reason
for rapid SO2 reduction in the first five stages is that SO2 absorption is a
3.3. Impact on NOx and SOx reaction function of SO2 partial pressure and liquid temperature. Due to the faster
moisture condensation in the bottom stages of the optimized design, the
There are three factors that affect the greater NO to NO2 oxidation in partial pressure of SO2 increases sharply compared to conventional
the optimized column and thus the larger NO removal. These factors are
design, resulting in higher absorption of SO2 in the bottom stages and,
1) an increased rate constant due to the sharper drop in gas temperature consequently, a faster liquid-phase reaction. The water temperature also
in the optimized design, 2) an increase in gas residence time due to the
drops rapidly in the optimized design (Fig. 3), which results in higher
faster cooling and moisture condensation in the flue gas, and 3) an in­ absorption of SO2 in lower stages of the optimized design compared to
crease in partial pressure of the reactants due to the reduced gas volume.
conventional design. The overall impact of these factors results in
The molar flow rate of NO and the comparison of total NO to NO2

Fig. 7. (a) Molar flow rate of NO and (b) the ratio of NO consumption between optimized and conventional column (Stage ‘0’ in Fig. 7(a) represents the
inlet conditions.)

7
P. Verma et al. Energy Conversion and Management 254 (2022) 115216

Fig. 9. Comparison of NOx and SOx removal in the conventional and opti­
mized designs.

from the top. The gas temperature remains high until the upper section,
where most of the moisture condensation and gas temperature reduction
occur. To optimize the design, the water supply to the column is split
into multiple streams, one for each of the column stages, and then the
Fig. 8. SO2 mole flow over ten stages in the conventional and optimized design.
flow rates for each water stream are optimized to maximize the SOx and
NOx removal efficiency. The optimal design for the system studied has
enhanced removal of SO2 in the optimized column.
only two water streams, with 23.5% of the water fed from the top and
It should be noted that almost all the SO2 is removed in the process
the remaining fed from Stage 4. The optimized design proved to be
for the conditions specified, and it would be possible to remove an even
significantly better in terms of SOx and NOx scrubbing efficiency. In this
greater amount of SO2 if more were supplied at the inlet. The removal is
design, the heat transfer in the lower sections is enhanced, leading to
only limited by the amount of NO in the system. However, the SO2
faster temperature drop and earlier moisture condensation. The lower
removal is also very sensitive to the amount of NOx in the inlet; there­
temperature and moisture content both contribute to a lower volumetric
fore, with a reduced N/S ratio, the SO2 removal may decline. An
gas flow rate and hence a longer residence time (34% longer). Since the
enhanced pressure or a modified L/G ratio can be employed to achieve
NO oxidation reaction occurs faster at lower temperatures, the opti­
similar objectives in such cases.
mized column has a higher NO oxidation rate, in addition to a longer
residence time for NO to oxidize, leading to significantly higher NOx and
3.4. Comparison of NOx and SOx removal efficiencies SOx removal efficiencies.

Fig. 9 summarizes the scrubbing efficiency of SOx and NOx for both CRediT authorship contribution statement
designs. The scrubbing efficiencies of SOx and NOx are calculated using
Eq. (2), based on the change in the molar flow between inlet and outlet. Piyush Verma: Conceptualization, Methodology, Investigation,
( ) Writing – original draft. Zhiwei Yang: Supervision, Investigation,
Scrubbing Efficiency(%) =
Inlet Molar flow − Outlet Molar flow
*100 Writing – review & editing. Richard L. Axelbaum: Supervision, Writing
Inlet Mole flow – review & editing, Funding acquisition.
(2)
The increase in gas residence time and rate constant of the NO Declaration of Competing Interest
oxidation reaction in the optimized design enhances the scrubbing of NO
by almost nine percentage points compared to the conventional design. The authors declare that they have no known competing financial
Consequently, the increased formation and subsequent absorption of interests or personal relationships that could have appeared to influence
NO2 and the reduced temperature and increased partial pressure (due to the work reported in this paper.
faster moisture condensation) enhance the liquid phase reaction with
HSO3− , resulting in faster and greater absorption of SO2 in water. The Acknowledgement
scrubbing efficiency of SOx increases by almost three percentage points
to 99.98%. The optimized two-water-inlet design is much more efficient This work has been supported by the U.S. Department of Energy
and, consequently, more economical than the conventional counter- (DOE) (FE0025193), the U.S.-China Clean Energy Research Center
current design. (CERC)(DE-PI0000017), and the Consortium for Clean Coal Utilization
(CCCU) at Washington University in St. Louis.
4. Conclusion
References
A direct contact cooler that simultaneously recovers flue gas heat and [1] Guo J-X, Huang C. Feasible roadmap for CCS retrofit of coal-based power plants to
removes SOx and NOx can significantly reduce the cost of pollutant reduce Chinese carbon emissions by 2050. Appl Energy 2020;259:114112. https://
removal in pressurized oxy-combustion systems. This study modeled [Link]/10.1016/[Link].2019.114112.
[2] Rissman J, Bataille C, Masanet E, Aden N, Morrow WR, Zhou N, et al. Technologies
and optimized this type of direct contact cooler using a reaction mech­
and policies to decarbonize global industry: Review and assessment of mitigation
anism and kinetic data validated by pilot-scale experiment results. A drivers through 2070. Appl Energy 2020;266:114848. [Link]
time scale analysis of the heat and mass transfer and chemical reactions apenergy.2020.114848.
indicates that the NO removal is limited by the NO oxidation reaction. [3] Gopan A, Kumfer BM, Phillips J, Thimsen D, Smith R, Axelbaum RL. Process design
and performance analysis of a Staged, Pressurized Oxy-Combustion (SPOC) power
A conventional column design was compared with a novel optimized plant for carbon capture. Appl Energy 2014;125:179–88. [Link]
design. In the conventional column, gas is fed from the bottom and water [Link].2014.03.032.

8
P. Verma et al. Energy Conversion and Management 254 (2022) 115216

[4] Gopan A, Kumfer BM, Axelbaum RL. Effect of operating pressure and fuel moisture [20] Hume S, Yang Z, Kumfer B, Axelbaum R, Dhungel B, Slater JD, et al. Improving
on net plant efficiency of a staged, pressurized oxy-combustion power plant. Int J Flexibility and Performance of Staged Pressurized Oxy-Combustion. 14th Greenh
Greenh Gas Control 2015;39:390–6. [Link] Gas Control Technol Conf Melb 2018:21–6.
[5] Gopan A, Verma P, Yang Z, Axelbaum RL. Quantitative analysis of the impact of [21] Yang Z, Khatri D, Verma P, Li T, Adeosun A, Kumfer BM, et al. Experimental study
flue gas recirculation on the efficiency of oxy-coal power plants. Int J Greenh Gas. and demonstration of pilot-scale, dry feed, oxy-coal combustion under pressure.
Control 2020;95:102936. [Link] Appl Energy 2021;285:116367. [Link]
[6] Vannerberg N-G, Sydberger T. Reaction between SO2 and wet metal surfaces. [22] Ajdari S, Normann F, Andersson K, Johnsson F. Reduced mechanism for nitrogen
Corros Sci 1970;10(1):43–9. [Link] and sulfur chemistry in pressurized flue gas systems. Ind Eng Chem Res 2016;55
[7] Radojevic M. Reduction of nitrogen oxides in flue gases. Elsevier Ltd 1998. https:// (19):5514–25. [Link]
[Link]/10.1016/b978-0-08-043201-4.50097-6. [23] Green D, Southard M. Perry’s Chemical Engineers’ Handbook. 8th ed. McGraw
[8] Srivastava RK, Jozewicz W. Flue gas desulfurization: The state of the art. J Air Hill; 2008.
Waste Manag Assoc 2001;51(12):1676–88. [Link] [24] de Visser E, Hendriks C, Barrio M, Mølnvik MJ, de Koeijer G, Liljemark S, et al.
10473289.2001.10464387. Dynamis CO2 quality recommendations. Int J Greenh Gas Control 2008;2(4):
[9] Ting T, Stanger R, Wall T. Laboratory investigation of high pressure NO oxidation 478–84. [Link]
to NO2and capture with liquid and gaseous water under oxy-fuel CO2compression [25] Wetenhall B, Race JM, Downie MJ. The Effect of CO2 Purity on the Development of
conditions. Int J Greenh Gas Control 2013;18:15–22. [Link] Pipeline Networks for Carbon Capture and Storage Schemes. Int J Greenh Gas
ijggc.2013.06.016. Control 2014;30:197–211. [Link]
[10] Hume S. Enabling staged pressurized oxy-combustion. Improving Flexibility and [26] Tsukahara H, Ishida T, Mayumi M. Gas-phase oxidation of nitric oxide: Chemical
Performance at Reduced Cost 2019. kinetics and rate constant. Nitric Oxide - Biol Chem 1999;3(3):191–8. [Link]
[11] White V, Torrente-Murciano L, Sturgeon D, Chadwick D. Purification of oxyfuel- org/10.1006/niox.1999.0232.
derived CO2. Energy Procedia 2009;1(1):399–406. [Link] [27] Lee YN, Schwartz SE. Reaction kinetics of nitrogen dioxide with liquid water at low
egypro.2009.01.054. partial pressure. J Phys Chem 1981;85(7):840–8. [Link]
[12] Murciano LT, White V, Petrocelli F, Chadwick D. Sour compression process for the j150607a022.
removal of SOx and NOx from oxyfuel-derived CO2. Energy Procedia 2011;4: [28] Wang J, Himmelblau DM. Kinetic Study of Sulfur Dioxide in aqueous solution with
908–16. [Link] radioactive tracers. AICHE J 1964:574–80.
[13] White V, Wright A, Tappe S, Yan J. The Air Products Vattenfall oxyfuel [29] Normann F, Jansson E, Petersson T, Andersson K. Nitrogen and sulphur chemistry
CO2compression and purification pilot plant at schwarze pumpe. Energy Procedia in pressurised flue gas systems: A comparison of modelling and experiments. Int J
2013;37:1490–9. [Link] Greenh Gas Control 2013;12:26–34. [Link]
[14] Tumsa TZ, Lee SH, Normann F, Andersson K, Ajdari S, Yang W. Concomitant [30] Ajdari S, Normann F, Andersson K, Johnsson F. Gas phase oxidation of SO2 by NO2
removal of NOx and SOx from a pressurized oxy-fuel combustion process using a in pressurized flue gas systems - an experimental investigation. Energy Procedia
direct contact column. Chem Eng Res Des 2018;131:626–34. 2013;46:2011–3.
[15] Iloeje C, Field R, Ghoniem AF. Modeling and parametric analysis of nitrogen and [31] Verma P, Gromotka Z, Yablonsky G, Stokie D, Constales D, Kumfer B, et al.
sulfur oxide removal from oxy-combustion flue gas using a single column absorber. Optimizing complexity in the kinetic modelling of integrated flue gas purification
Fuel 2015;160:178–88. [Link] for pressurized oxy-combustion. Chem Eng J 2020;383:122875. [Link]
[16] Liu Q, Zhong W, Yu A, Wang C. Co-firing of coal and biomass under pressurized 10.1016/[Link].2019.122875.
oxy-fuel combustion mode: Experimental test in a 10 kW th fluidized bed. Chem [32] Axelbaum R, Stokie D, Verma P, Kumfer B, Min Y, Zhu Y, et al. Integrated Flue Gas
Eng J 2021;133457. [Link] Purification and Latent Heat Recovery for Pressurized Oxy-Combustion. 2019.
[17] Gardner SH, Gardner SH. Completion and Initial Testing of a Pressurized Oxy-Coal [33] Stokie D, Verma P, Kumfer BM, Yablonsky G, Suresh AK, Axelbaum RL. Pilot-scale
Reactor 2021. testing of direct contact cooler for the removal of SOx and NOx from the flue gas of
[18] Normann F, Andersson K, Johansson J, Heijnesson A. Simultaneous Removal of pressurized oxy-coal combustion. Chem Eng J 2021;414:128757. [Link]
NOx and SOx from Flue Gases Using ClO2. Process Scaling and Modeling 10.1016/[Link].2021.128757.
Simulations 2021. [Link] [34] Stanger R, Ting T, Spero C, Wall T. Oxyfuel derived CO2 compression experiments
[19] Zhou J, Ma J, Wang Z. Experimental study on removal performance of SO 2 and with NOx, SOx and mercury removal-Experiments involving compression of slip-
NOx in marine exhaust gas using seawater / urea peroxide solution and analysis of streams from the Callide Oxyfuel Project (COP). Int J Greenh Gas Control 2015;41:
ions concentration change. Fuel Process Technol 2022;227:107133. [Link] 50–9. [Link]
org/10.1016/[Link].2021.107133.