Intensification of CO2 capture using aqueous diethylenetriamine (DETA) solution from simulated flue gas in a rotating packed bed

Intensification of CO2 capture using aqueous diethylenetriamine (DETA) solution from simulated flue gas in a rotating packed bed

Fuel 234 (2018) 1518–1527 Contents lists available at ScienceDirect Fuel journal homepage: www.elsevier.com/locate/fuel Full Length Article Intens...

3MB Sizes 1 Downloads 5 Views

Fuel 234 (2018) 1518–1527

Contents lists available at ScienceDirect

Fuel journal homepage: www.elsevier.com/locate/fuel

Full Length Article

Intensification of CO2 capture using aqueous diethylenetriamine (DETA) solution from simulated flue gas in a rotating packed bed

T



Miaopeng Shenga, Chenxia Xiea,b, Xiaofei Zenga,b, Baochang Suna,b, , Liangliang Zhanga,b, ⁎ Guangwen Chua,b, Yong Luoa,b, Jian-Feng Chena,b, Haikui Zoua,b, a b

Research Center of the Ministry of Education for High Gravity Engineering and Technology, Beijing University of Chemical Technology, Beijing 100029, PR China State Key Laboratory of Organic-Inorganic Composites, Beijing University of Chemical Technology, Beijing 100029, PR China

A R T I C LE I N FO

A B S T R A C T

Keywords: CO2 capture Process intensification Diethylenetriamine Rotating packed bed Back-Propagation Neural Network model

Coal still has a vital role in power generation, and coal-fired power plants are considered to be a main source of CO2 emission. This work proposed a process intensification (PI) technology, combining the highly efficient diethylenetriamine (DETA) solvent for CO2 absorption and the PI device of rotating packed bed (RPB) for enhancing gas-liquid mass transfer, to improve CO2 capture performance. Experimental study was conducted in a lab-scale RPB, and the dependences of CO2 removal performance on various operation conditions were systematically investigated. It was founded that CO2 loading has a vital effect on removal efficiency, and increasing rotation speed and solvent flow rate is beneficial to CO2 removal. The comparison of mass transfer performance between RPB and packed column (PC) demonstrated that the gas retention time in RPB with a value of 1.5 s is far shorter than that in PC under the similar operation conditions, which means RPB possesses a great advantage of shrinking mass-transfer device’s size for the CO2 capture process. Additionally, a Back-Propagation Neural Network (BPNN) model was developed for predicting the value of overall volumetric mass-transfer coefficient (KGav), and the predicted values agree well with experimental data with a satisfactory average absolute relative derivation (AARD) of 7.85%. These results demonstrated that this PI technology is expected to be a competitive candidate for improving CO2 capture performance from flue gas.

1. Introduction Carbon dioxide (CO2) is a major contributor to global warming, and coal-fired power plants are considered to be a main source of CO2 emission. According to BP energy outlook 2017 [1], coal consumption in 2016 was 3732 Gt of oil equivalent, accounting for 28.1% of world’s global primary energy consumption, while Asia-Pacific region consumed 73.78% of the total as shown in Fig. 1. Despite the dramatic drop in the mount of coal power capacity in 2016, the number of the coalfired plants is still very large, and the amount of global coal power capacity in pre-construction planning and operating were 446.6GW and 1995.8GW, respectively, as of January 2017 [2]. Also, the continuous falling in coal price since 2011 enables coal power sector to obtain unprecedented excess profits, which may stimulate generation companies to invest in new coal power projects [3]. In view of growing increase in world population and demand of economic development, coal-fired power plant will still be a vital source of power generation in the near future, especially in developing countries like China and India, despite the growing utilization of natural gas and renewables. ⁎

Therefore, continually large CO2 emission as well as its impacts on climate change call for the efficient control of CO2 emission, and first of all, developing more advanced and cost-effective CO2 capture technologies becomes urgently needed. Post-combustion CO2 capture (PCC) using amine-based solvent is considered as the most mature and promising end-of-pipe treatment process for CO2 capture from coal-fired power plants, which has so far been applied in chemical industries [4]. However, high energy demand of this process is the main drawback, which leads to a reduction in power generation efficiency up to 13% [4] and high capital and operation cost, and thus hinders its commercialization. Therefore, the improvements, including process optimization, solvent development and design improvement in equipments, are vital to minimizing the energy penalty as well as the capital and operation cost [4–6]. Monoethanolamine (MEA), a typical chemical solvent, has been served as standard solvent for the CO2 capture process, but its drawbacks significantly expand the investment and operation costs, e.g. low capacity, high regeneration energy, corrosion and degradation [7]. In the past decades, many efforts have been made to develop new amine-

Corresponding authors at: P.O. Box 35, No. 15 Bei San Huan Dong Road, Beijing 100029, PR China. E-mail addresses: [email protected] (B. Sun), [email protected] (H. Zou).

https://doi.org/10.1016/j.fuel.2018.07.136 Received 10 June 2018; Received in revised form 25 July 2018; Accepted 30 July 2018 0016-2361/ © 2018 Elsevier Ltd. All rights reserved.

Fuel 234 (2018) 1518–1527

M. Sheng et al.

Nomenclature α β CAmine DCO2 G GI kL k2 kAmH kDETA,P kDETA,S kH2O

KGav

CO2 loading, mol mol−1 2 )/2 /9.8) high gravity level (=(2πN /60)2 (rin2 + rout −1 concentration of amine solvent, mol L diffusion coefficient of CO2 in liquid, m2 s−1 gas flow rate, m3 h−1 inert gas molar flow rate, kmol m2 h−1 liquid-phase mass transfer coefficient, m s−1 second order reaction-rate constant, m3 kmol s−1 third-order reaction rate constant contributed by amine, m3 kmol s−1 third-order reaction rate constant contributed by primary amino group of DETA, m3 kmol s−1 third-order reaction rate constant contributed by secondary amino group of DETA, m3 kmol s−1 third-order reaction rate constant contributed by H2O,

L Lf G/L N Ns rin rout SN SR t T yin yout

m3 kmol s−1 overall volumetric mass-transfer coefficient, kmol m−3 kPa−1 h−1 solvent flow rate, L h−1 solvent flux rate, m3 m−2 h−1 gas-liquid ratio, L·L−1 rotation speed of RPB, rpm· number of packing layers in RPB inner radius of the packing, cm inner radius of the packing, cm surface renewable rate of liquid elements, s−1 the ratio of kL,N to kL,1000rpm average liquid retention time, s temperature, °C inlet CO2 concentration, % outlet CO2 concentration, %

Fig. 1. Contribution of world’s global primary energy consumption in 2016 by fuel (a) and region (b) according to BP energy outlook 2017 [1].

multiple-phase fields like acid gas absorption [9,19–22], VOC Stripping, [23] distillation [24], and bromination [25], and waste water treatment [26,27]. Due to the strong centrifugal force created by highspeed packing, liquid is spilt or spread into tiny liquid elements, including films, droplets and threads, and these will inevitably provide a large gas-liquid contact area. Moreover, the fast coalescence and redispersion of liquid elements between packing layers leads to a high surface renewable rate and thus an intensification on liquid-phase mass transfer process according to Higbie’s penetration theory [9]. Jassim et al. [20] adapted MEA for CO2 capture in RPB and found RPB possesses an advantage of size reduction as compared to conventional packed column. Simulated works and experimental studies by Qian et al. [9], Joel et al. [28], and Yu et al. [29], also confirm that RPB can obtain better mass transfer efficiency with an obvious advantages of size reduction and energy reduction as compared to conventional masstransfer device, implying RPB can be a promising device for CO2 capture. Contacting time between gas and liquid in RPB is much shorter than conventional packed column because of the very short liquid retention time in RPB (normally < 1 s), [30] which requires highly efficient solvent with a fast reaction rate with CO2 to match the short retention time

based solvents, like piperazine (PZ) [7,8], N-methyldiethanolamine (MDEA) [9,10], 2-amino-2-methyl-1-propanol (AMP) [10] and Diethylenetriamine (DETA) [11], to overcome the limitations of MEA. DETA is a linear polyamine and contains three amino groups (see Fig. 2), which exhibits the high reaction rate and high capacity for CO2 absorption. Hartono et al. [11] investigated the kinetics of DETA-CO2 system and the results show that DETA has a much higher reaction rate with CO2 than that of MEA. Fu et al. [12] compared the CO2 capture performance between aqueous DETA and aqueous MEA in a packed column with Dixon rings, and they found that DETA has a much higher overall gas-phase volumetric mass-transfer coefficient (KGav) than that of MEA. Zhang et al. [13] measured the regeneration heat duty for CO2 desorption, and found that DETA consumes less heat duty than MEA for regenerating a certain amount of CO2. The results of a pilot-plant test work, which carried out in a packed column with Sulzer DX500 packing by Gao et al. [14], agrees well with the results obtained by Fu et al. and Zhang et al. These studies show DETA has much higher reaction rate and CO2 capacity, but lower regeneration heat duty than MEA, which means DETA is a promising solvent for CO2 capture. CO2 capture by chemical absorption method belongs to a gas–liquid mass transfer process accompanied with chemical reactions, and the mass transfer resistance mainly exists in liquid phase [12]. Therefore, besides the efforts made in developing new solvents, researchers should pay more attention to intensify the gas-liquid mass transfer of CO2 via developing new type device or using additives like nanofluids and catalysts [15–18]. Rotating packed bed (RPB), as a high-efficiency process intensification (PI) device, has been successfully applied in

Fig. 2. Structure of DETA. 1519

Fuel 234 (2018) 1518–1527

M. Sheng et al.

and high mass transfer efficiency in RPB. Fig. 3 presents the comparisons of reaction rate between CO2 and different solvents, including amine-based solvents, ammonia (NH3) and potassium glycinate (PG). PZ and its derivatives (i.e. 1-(2-aminoethyl)piperazine (AEPZ) and N(2-hydrox-ethyl)piperazine (N-HEPZ)) possess the highest reaction rate with CO2, followed by polyamines, like ethylenediamine (EDA), 2-((2Aminoethyl)amino)ethanol (AEEA) and DETA. Previous studies by Yu et al. [32] and Sheng et al. [21] exhibit that adding 5–15 wt% of PZ into aqueous DETA solution can improve CO2 removal efficiency in RPB. However, the effect of PZ concentration on CO2 removal efficiency was less significant than that of others operation conditions. Moreover, PZ has a poor water-solubility, and Freeman and Rochelle recommended a concentrated PZ (40 wt%) solution with a relatively high CO2 loading to avoid the risk of solid precipitation under low temperature [7,8]. Given these conclusions as well as the cost of absorbent, adding a small amount of PZ and its derivative into aqueous DETA solution may not be a very economic choice, and an individual DETA solution may still match the short retention time and high mass transfer efficiency in RPB due to its high reaction rate with CO2. In addition, the investigation of aqueous DETA solution as solvent for CO2 capture in RPB is inadequate. Therefore, combining RPB’s intensification in mass transfer process and solvent’s high reaction rate with CO2 can not only enhance the CO2 capture performance but also reduce the size of absorber. In this work, an investigation on the absorption of CO2 into aqueous DETA solution in RPB, which adopts DETA as absorbent and RPB as absorber to intensify the CO2 capture process, was presented to explore the dependences of CO2 removal efficiency on various operation conditions. The comparisons of KGav between RPB and packed columns from literature were also presented. In addition, the modelling study on KGav, including empirical model and Back-Propagation Neural Network (BPNN) model, was deduced for the further design of the scale-up of this process. This work could provide a potential solution for improving CO2 capture performance as well as promoting the commercialization of post-combustion amine-based CO2 capture technology.

CO2 + DETA +

{

DETA DETAH+ ↔ DETCOO− + ⎧ + H2 O ⎨ ⎩ H3 O

(4)

−rCO2 - DETA = (kDETA,P CDETA + kDETA,S CDETA + kH2O CH2O ) CDETA CCO2 (5)

−3869.3 ⎞; kDETA,S T ⎠ −4092.8 ⎞ = 4.6294 × 108 exp ⎛ ⎝ T ⎠

kDETA,P = 7.4692 × 109 exp ⎛ ⎝

(6)

where rCO2-DETA is the overall reaction between CO2 and DETA; kDETA,P is third-order reaction rate constant contributed by primary amino group of DETA; and kDETA,S is third-order reaction rate constant contributed by secondary amino group of DETA; kH2O is third-order reaction rate constant contributed by H2O. The value of kDETA,P calculated by Eq. (6) is 9.2 times that of kMEA at 40 °C as shown in Fig. 3, which shows a high reaction rate between CO2 and DETA. 3. Experimental section 3.1. Materials DETA (purity ≥ 98.0%) was purchased from Tianjin Fuchen Chemical Reagents Factory, and aqueous solution with a certain concentration was prepared with deionized water. CO2 gas (purity ≥ 99.9%) was supplied by Beijing Ruyuanruquan Technology Co. Ltd, and the air was generated by an oil free air compressor (TYW-1, Suzhou Tongyi Electrical and Mechanical Co. Ltd). Standard 1.0N H2SO4 solution was purchased from National Analytic Center, China. Concentrated H2SO4 (> 95 wt%) was purchased from Beijing Chemical Works to prepare dilute H2SO4 solution (∼1 mol L−1), which was used for CO2 loading measurement as described in Section 3.2. All chemicals were used without further purification. 3.2. Experimental procedure

2. Reactions during absorption of CO2 into DETA solvent

The experimental setup is shown in Fig. 5. Stainless wire mess was

To the best of our knowledge, the only available kinetics studies of CO2 absorption into aqueous DETA solution published in literature were reported by Hartono et al. [11,33]. According to their study, the reaction between CO2 and DETA can be described by zwitterion mechanism and termolecular mechanism, and further, the latter may be more accurate:

CO2 + 2H2 O↔ H3 O + + HCO−3

(1)

CO2 + OH−↔HCO−3

(2)

+ ⎧ DETAH ⎧ DETA H2 O ⎪ H3 O+ ⎪ ⎪ ⎪ − H2 O CO2 + DETA + OH ↔ DETACOO− + ⎨ HCO− ⎨ 3 H2 CO3 ⎪ ⎪ ⎪ CO32 − ⎪ HCO−3 ⎩ ⎩

(3)

Normally, the reaction between hydroxyl ion and CO2 can be neglected due to its low concentration [34]. NMR study on reaction products from another work of Hartono et al. [35] shows that the concentrations of HCO−3 and CO32 - are very low as compared to DETA concentration, and they are generated when CO2 loading reaches as high as 1.38 (See Fig. 4). Normally, the high loading has far exceeded the optimal operating range in view of the regeneration heat duty of CO2 according to Zhang et al. [13] and Gao et al. [14] Therefore, the effects of CO32 - and HCO−3 in Eq. (3) can be neglected, and the Eq. (3) can be simplified to be Eq. (4). Because DETA has three amino groups, the overall reaction rate can be considered to be a combination of CO2 reaction with primary and secondary amino groups which are expressed as Eqs. (5) and (6).

Fig. 3. Comparison of reaction rate constants of various solvents: (a) Secondorder reaction rate constant k2; (b) Third-order reaction rate constant kAmH. (drawn by Sheng on the basis of a literature review [31]). 1520

Fuel 234 (2018) 1518–1527

M. Sheng et al.

Fig. 4. Products generated from reaction between CO2 and DETA. (drawn according to reference [35]).

The CO2 removal efficiency was calculated by

packed in the RPB, and the diameter of packing wire was 0.25 mm. The outer diameter, inner diameter and axial height of packing were 15 cm, 5 cm and 5.3 cm, respectively. Prior to each experiment, the RPB was pre-heat and pre-wet through circulating hot solvent, and at the same, the gaseous mixture of CO2 and air with a certain proportion was introduced into RPB. Normally, this procedure took about 15 min to reach a steady state. When temperature and inlet CO2 centration reached the pre-set values, a fresh lean solvent with a certain temperature was pumped into RPB, and then the liquid contacted with the gaseous mixture counter-currently to capture CO2. All experiments were carried out under atmosphere pressure, and all data were recorded when the system reached a steady state. CO2 concentration in gaseous mixture was monitored by infrared CO2 analyzer (GXH-3010F, Beijing Huayun Analytical Instrument Institution), and CO2 loading in solvent was analyzed by measuring the volume of released CO2 from solvent through adding an excess amount of dilute H2SO4 solution [36]. Amine concentration was analyzed by potentiometric titration method with standard 1.0 N H2SO4 solution, and the operation was achieved by an Automatic Potentiometric Titrator (ZDJ-4B, INESA Scientific Instrument Co., Ltd).

y (1−y in ) ⎤ η = ⎡1− out × 100% ⎢ y in (1−y out ) ⎥ ⎣ ⎦

(7)

where yin and yout are CO2 concentration at gas inlet and gas outlet, respectively. 4. Results and discussion 4.1. Effects of solvent concentration and CO2 loading Solvent concentration and CO2 loading are vital parameters for the design of amine-based PCC, and Fig. 6 shows the dependences of CO2 removal efficiency on solvent concentration and CO2 loading in RPB. It can be seen that CO2 removal efficiency obviously decreased as CO2 loading increased, and this change was more evident when CO2 loading exceeded the range of 0.4–0.5. Increasing DETA concentration favored the CO2 removal, but it had a limited effect on CO2 removal efficiency when CO2 loading was lower

Fig. 5. Experimental setup for the CO2 capture in an RPB. 1521

Fuel 234 (2018) 1518–1527

M. Sheng et al.

Fig. 6. Dependences of CO2 removal efficiency on solvent concentration and CO2 loading in RPB. Table 1 Viscosity of aqueous DETA solution. DETA concentration

a

25 wt% 30 wt%a 10 wt%b 20 wt%b 30 wt%b 40 wt%b a b

Viscosity/cP T = 30 °C

T = 40 °C

T = 50 °C

2.8413 3.7357 1.211 2.011 3.572 7.310

2.101 2.6839 0.970 1.540 2.582 4.859

1.6153 2.0119 0.799 1.221 1.95 3.302

Obtained from Ref. [37]. Obtained from Ref. [38].

Fig. 8. Effect of rotation speed on liquid retention time and SR.

Fig. 7. Dependence of CO2 removal efficiency on rotation speed in RPB.

than 0.4. Although high DETA concentration will accelerate the reaction between CO2 and DETA, the solvent viscosity, increasing as DETA concentration increases as shown in Table 1, will hinder the diffusion of CO2 molecule in the liquid and then impede CO2 absorption. The

Fig. 9. Dependence of CO2 removal efficiency on solvent flow rate in RPB.

1522

Fuel 234 (2018) 1518–1527

M. Sheng et al.

Fig. 13. Comparison of KGav between packed columns and RPB under similar gas-liquid ratio.

Fig. 10. Dependence of CO2 removal efficiency on gas flow rate in RPB.

30 wt% DETA, especially when CO2 loading was low as shown in Fig. 6. 4.2. Effect of rotation speed In the RPB, liquid is split into very fine liquid elements, including films, droplets and threads, by high-speed rotating packing, and high rotation speed favors the formation of fine liquid elements. Visual studies in RPB show that average thickness of liquid film is about 10−5 m [30] and average diameter of liquid droplet is about 10 4 m, [39] resulting in a large gas-liquid contact area and thus enhancing the mass transfer process. Fig. 7 shows that CO2 removal efficiency significantly increased with increasing rotation speed, which highly benefits from the formation of fine liquid elements and large gas-liquid contact area. Previous studies have demonstrated that effective gas-liquid contact area (ae) is proportional to rotation speed, i.e... n is in the range of 0.24–0.68 [40]. As aforementioned, liquid has a short retention time (t) in the RPB. Qian et al. [9] believed that intensification of RPB on gas-liquid mass transfer process mainly benefits from the high surface renewable rate (SN) and short life time of liquid elements. Liquid retention time and surface renewable rate of liquid elements can be calculated by Eqs. (8) and (9), respectively [9]. SR is defined as the ratio of liquid-side mass transfer coefficients (kL,N) to that obtained under 1000 rpm as Eq. (10), which could intuitively reflect the intensification performance according to Danckwerts’s surface renewable theory [41]. The calculated results have been plotted in Fig. 8. It can be seen that the value of SR significantly increases due to the drop in liquid retention time when rotation speed increases, meaning the mass transfer process, i.e. CO2 absorption process, can be enhanced significantly.

Fig. 11. Dependence of CO2 removal efficiency on temperature in RPB.

−0.5448

t=

Fig. 12. Dependence of CO2 removal efficiency on inlet CO2 concentration in RPB.

2 2 rout−rin ⎡ ⎛ 2πN ⎞2 (rin + rout ) ⎤ ⎥ 0.02107L0.2279 ⎢ ⎝ 60 ⎠ 2 ⎣ ⎦

SN =

Ns t

SR =

kL, N = kL,1000rpm

(8)

(9)

DCO2 SN DCO2 S1000rpm

=

t1000rpm tN

(10)

where kL is liquid-phase mass transfer coefficient; rin and rout are inner radius and outer radius of packing, respectively; Ns is the number of packing layer in RPB. However, when rotation is too high, CO2 may not be completely absorbed by the solvent in a very short contact time before it exits from the RPB, which weakens the CO2 removal process. Therefore, a slower

viscosity of 20 wt% DETA is about 60% of that of 30 wt% DETA under the temperature in a typical range of 30–50 °C [37,38]. Thus, low viscosity favored 20 wt% DETA to obtain removal efficiency as good as

1523

Fuel 234 (2018) 1518–1527

M. Sheng et al.

Table 2 Detailed comparison of removal performance between packed column and RPB. Packed column

Packed column

Packed column

RPB

References Packing type

Fu et al. [12] ∅ 3 × 3 Dixon ring

Gao et al. [42] Sulzer DX

Fu et al. [14] BX 500

This work Stainless wire mesh

Packing dimension Diameter (cm) Height (cm) Volume (cm3)

2.4 140 633

2.8 120 739

15 200 35,343

5 (ID), 15 (OD) 5.3 (axial height) 833

DETA, MEA 2 0.179–0.215 30 1.2–4.2 0.32–0.49 4.7–7.2 0.32–1.2

DETA

DETA, MEA 2 for DETA 6 for MEA 0 40 5–40 1.7–6.3 20.4–72.5 1.6–3.5

DETA

Operation conditions Solvent Solvent concentration (mol L−1) CO2 loading (mol mol−1) Temperature (°C) Solvent flow rate (L h−1) Gas flow rate (m3 h−1) Gas retention time (s) KGav (kmol m−3 kPa−1 h−1) a

2–3 0.184–0.826 30–50 1.2–3.0 0.32–0.59 4.52–8.22 0.55–1.35

1.98–3.01a 0.15–0.52 30–50 11–32 2.0–3.5 0.86–1.5 1.12–7.44

Determined by titration method.

Fig. 14. Relationship between operating conditions and KGav.

growth rate in CO2 removal efficiency was observed when rotation speed exceeded 1000 rpm as shown in Fig. 7. 4.3. Effect of solvent flow rate Fig. 9 shows the dependence of CO2 removal efficiency on solvent flow rate in RPB. It can be seen that the removal efficiency increased as the solvent flow rate increased, and this tendency became more evident when CO2 loading increased. Previous studies have confirmed that CO2 absorption into aqueous DETA solution is a liquid-film control process [12,14]. Increasing solvent flow rate not only enlarges the effective gasliquid contact area, but also provides more free amine molecules. Additional, surface renewable rate of liquid also increases and then enhances the mass transfer process according to Eqs. (8)–(10). These factors enhance the mass transfer process and then lead to a higher CO2 removal efficiency. 4.4. Effect of gas flow rate

Fig. 15. Literature data of CO2 solubility in 30 wt% DETA solvent.

Fig. 10 shows the dependence of CO2 removal efficiency on gas flow rate in RPB. The dependence of removal efficiency on gas flow rate was opposite to that of solvent flow rate, and increasing gas flow rate was unfavorable for CO2 absorption. This tendency became more evident 1524

Fuel 234 (2018) 1518–1527

M. Sheng et al.

Fig. 16. The structure of 4-layer BPNN model used in this work.

larger volatilization loss, which are unfavorable to CO2 absorption. Therefore, too high temperature than 60 °C is not a very good choice. The results obtained in this work were similar to the works of Fu et al. [12]. 4.6. Effect of CO2 concentration Fig. 12 shows the dependence of CO2 removal efficiency on inlet CO2 concentration in RPB. Because the absorption process is mainly controlled by liquid-film mass transfer process, reducing gas-film resistance through increasing CO2 concentration has a limited effect on CO2 removal efficiency. Moreover, increasing CO2 concentration needs more free amine molecules to absorb the excessive CO2 to keep a constant removal efficiency. Therefore, CO2 removal efficiency decreased with increasing CO2 concentration as shown in Fig. 12, especially under high CO2 loading. 4.7. Comparisons with conventional packed column Fig. 17. Comparison between predicated KGav and experimental data.

Packed column (PC) is a commonly-used device for amine-based PCC, and three mass transfer studies of CO2 absorption into aqueous DETA solution have been reported in literature, including two laboratory-scale studies and one pilot-scale test. The comparison of CO2 removal performance between PC and RPB is plotted in Fig. 13, and the detail information is list in Table 2. The overall volumetric masstransfer coefficient (KGav) in RPB was calculated by

when CO2 loading increased and liquid flow rate and rotation speed decreased. Although increasing gas flow rate can reduce the gas-film mass transfer resistance, it has a limited effect on CO2 transfer from gas to liquid because this process is a liquid-film control process. Moreover, increasing gas flow rate reduces the gas retention time in RPB, which reduces the contact time between gas and liquid. Therefore, CO2 removal efficiency decreases when gas flow rate increases.

KG a v = 4.5. Effect of temperature

GI ⎡ yCO2 ,in (1−yCO2 ,out ) × ln 2 −rin2 ) ⎢ yCO2 ,out (1−yCO2 ,in ) πPH (rout ⎣ yCO2 ,in yCO2,out ⎞ ⎤ + ⎜⎛ − ⎟ − − 1 y 1 yCO2,out ⎠ ⎥ CO2 ,in ⎝ ⎦

Fig. 11 shows the dependence of CO2 removal efficiency on temperature in RPB. It can be seen that the removal efficiency increased with elevating temperature, and this trend was more evident when CO2 increased. Temperature is a vital factor, which determines the reaction rate between CO2 and DETA, and elevating temperature accelerates the reaction rate according to Eq. (6). Also, elevating temperature can largely reduce the solvent viscosity as shown in Table 1, which is favorable to the diffusion of CO2 and amine molecules in liquid phase. However, elevating temperature results in a lower CO2 solubility and

(11)

From Fig. 13, RPB obtained higher values of KGav than PCs under the similar gas-liquid ratio, and KGav obtained by DETA solvent was much higher than that obtained by MEA solvent. In this comparison, the size of RPB in this work was close to that of PC used by Fu et al. [12], but much smaller than that of PC used by Gao et al. [14] as show in Table2. The gas retention time in RPB and PC in Fu et al’s work were 0.86 s and 5.4 s under mainly-studied gas flow rate of 3.5 m3 h−1 and 1525

Fuel 234 (2018) 1518–1527

M. Sheng et al.

0.42 m3 h−1, respectively. And the value was as high as 32.7 s in PC of Gao et al with a gas flow rate of 3.9 m3 h−1. In their study, the gas flow rate was investigated in the range of 1.7–6.3 m3 h−1, which leaded to a very long gas retention time from 20.4 to 72.5 s. Another work by Fu et al. [42], also investigated the mass transfer performance in a Sulzer DX packing column. Unfortunately, no specific but foggy operation conditions were given in their work. Thus, the comparison result is not plotted in Fig. 13, but listed in Table 2. It can be seen that RPB obtained higher mass transfer performance with a similar size, which is consistent with results shown in Fig. 13. From these comparison results, one can see that the gas retention time in RPB can be controlled below 1.5 s due to its intensification on mass transfer process, which means that RPB can be used for larger flue gas treatment as compared to PC with a similar device size.

approaches [42,47]. It has been applied in the fields of chemical engineering and shows a competitive performance as compared to conventional approaches [42,47,48]. Back-Propagation Neural Network (BPNN) model, on the basis of the steepest gradient descent method with attempt to minimize the error between input and output data, is one of the most classic and commonly-used ANN models [42,47]. The Fig. 16 shows a schematic structure of a 4-layer BPNN model developed in this work. A total of 218 data sets were obtained from experimental work. 164 data sets (75%) were used for training the net and the residual 54 data sets (25%) were used for testing the net. From our analysis, the suitable parameters were obtained as shown in Fig. 16. The comparison result is plotted in Fig. 17, and the predicted values of KGav by BPNN model agree well with experimental data with a satisfactory AARD (average absolute relative deviation) of 7.85%. However, BPNN model is a black box, and thus, no mathematic explanation can be obtained to elucidate the relationships between input operating conditions and output value of KGav.

5. Model development 5.1. Empirical correlation model KGav is a vital parameter for the design of RPB, and a deep understanding of KGav can help designer make an accurate assessment of RPB for a specific reactive absorption system. For a mass transfer process accompanied by chemical reactions, many factors, including operating conditions, mass transfer parameters, physicochemical properties and hydraulics properties, simultaneously determine the ultimate process performance. However, all mass transfer parameters, physicochemical properties and hydraulics properties can be determined by a function of operating conditions as shown in Fig. 14. Therefore, many empirical correlations of KGav as a function of the operating parameters have been developed for packed column [43], and the majority of them have a similar form as

K G α v ∝ Lfm

6. Conclusions In this work, a rotating packed bed (RPB) was adopted as a PI device to improving CO2 removal performance into aqueous solution of diethylenetriamine (DETA) from simulated flue gas. The dependences of CO2 removal efficiency on various operating conditions were investigated. Results shows CO2 loading has an important effect on removal efficiency. When CO2 loading is low than 0.15, changing other operating conditions have a limited effect. When CO2 loading is higher than 0.4, increasing DETA concentration is beneficial to CO2 removal. As rotation speed and solvent flow rate increase, CO2 removal efficiency increases due to the increase in effective gas-liquid contact area and surface renewable rate of liquid elements in RPB. Increasing gas flow rate and CO2 concentration lead to a reduction in CO2 removal efficiency, and removal efficiency increases with elevating temperature. Comparison results between RPB and packed column (PC) in the literature demonstrate that the gas retention time in RPB can be controlled below 1.5 s due to its intensification on mass transfer process, which means RPB can fulfill larger flue gas treatment task than PCs with a similar size, and DETA solvent exhibits a much higher removal performance than MEA solvent. Therefore, this PI technology, which combines the advantages of DETA solvent and RPB, is expected to be a competitive candidate for improving CO2 capture. A BPNN model was developed for predicting the value of overall volumetric mass-transfer coefficient (KGav), and the predicted values agree well with experimental data with a satisfactory AARD (average absolute relative derivation) of 7.85%. Additionally, in view of the large deviation of experimental data of CO2 solubility in literature, more experimental work is need to develop a precise empirical model or theoretical model of KGav.

(α eq−α ) CAmine n PCO 2

(12)

where Lf is liquid flow flux; PCO2 is CO2 partial pressure; αeq is CO2 loading in equilibrium with PCO2; exponent m is in the typical range of 0.18–0.67, mainly depending on solvents and packing types, and exponent n is normally equals 1. Fu et al. [12] proposed a correlation for DETA solvent in a Dixonring packed column as

K G α v = L f0.67 GI0.08 ⎡0.075 × ⎢ ⎣

(α eq−α ) CDETA PCO2

+ 0.142⎤ ⎥ ⎦

(13)

We attempted to adopt Eq. (14) or other similar correlation forms to correlate KGav with operating conditions. But unfortunately, we failed to give a precise correlation because of an obvious difference in CO2 solubility data in literature [44–46]. Additionally, no satisfied results were obtained even if αeq was fix at a maximum theoretical value of 1.5. The difference in CO2 solubility in 30 wt% DETA is plotted in Fig. 15, and other comparison results under various conditions can be seen in Table S1 and Figs. S1 and S2 (see Supplementary material). It can be seen that there is a large deviation in literature data, and more experimental work is need to develop a precise empirical model or theoretical model of KGav.

(α eq−α ) CDETA K G α v = F× Lfm1 GIm2 βm3e m4 × T ⎡ + m5⎤ ⎥ ⎢ PCO2 ⎦ ⎣

Acknowledgements One of the authors, Miaopeng Sheng, would like to thank Ms. Shuying Wu from our group for her kind help and suggestions. The financial support from the National Key R&D Program of China (No. 2017YFB0603300), the Fundamental Research Funds for the Central Universities (No. JD1706) and the National Natural Science Foundation of China (No. U1607114) were gratefully acknowledged.

(14)

where F, m1–m5 are parameters regressed from experimental data. 5.2. Back-Propagation Neural Network (BPNN) model

Appendix A. Supplementary data

Artificial neural network (ANN) model, with the excellent ability for deal with the uncertainties of linear or nonlinear relationships among noisy data, is a powerful tool to settle the problems that are difficult for conventional process simulation tools and empirical correlation

Supplementary data associated with this article can be found, in the online version, at https://doi.org/10.1016/j.fuel.2018.07.136. 1526

Fuel 234 (2018) 1518–1527

M. Sheng et al.

References

bed reactor. Chem Eng J 2014;240:503–8. [26] Modak JB, Bhowal A, Datta S. Extraction of dye from aqueous solution in rotating packed bed. J Hazard Mater 2016;304:337–42. [27] Mosleh S, Rahimi M, Ghaedi M, et al. Photocatalytic degradation of binary mixture of toxic dyes by HKUST-1 MOF and HKUST-1-SBA-15 in a rotating packed bed reactor under blue LED illumination: central composite design optimization. RSC Adv 2016;6:17204–14. [28] Joel AS, Wang M, Ramshaw C, et al. Process analysis of intensified absorber for post-combustion CO2 capture through modelling and simulation. Int J Greenhouse Gas Control 2014;21:91–100. [29] Yu CH, Chen MT, Chen H, et al. Effects of process configurations for combination of rotating packed bed and packed bed on CO2 capture. Appl Energy 2016;175:269–76. [30] Guo K, Guo F, Feng Y, et al. Synchronous visual and RTD study on liquid flow in rotating packed-bed contactor. Chem Eng Sci 2000;55:1699–706. [31] Sheng MP. Study on Mixed Absorption Solutions and Technologies for CO2 Absorption [Master Thesis]. Beijing, China: Beijing University of Chemical Technology; 2015. (in Chinese). [32] Yu C-H, Cheng H-H, Tan C-S. CO2 capture by alkanolamine solutions containing diethylenetriamine and piperazine in a rotating packed bed. Int J Greenhouse Gas Control 2012;9:136–47. [33] Hartono A, Svendsen HF. Kinetics reaction of primary and secondary amine group in aqueous solution of diethylenetriamine (DETA) with carbon dioxide. Energy Procedia 2009;1:853–9. [34] Aboudheir A, Tontiwachwuthikul P, Chakma A, et al. Kinetics of the reactive absorption of carbon dioxide in high CO2-loaded, concentrated aqueous monoethanolamine solutions. Chem Eng Sci 2003;58:5195–210. [35] Hartono A, da Silva EF, Grasdalen H, et al. Qualitative determination of species in DETA− H2O− CO2 system using 13C NMR spectra. Ind Eng Chem Res 2007;46:249–54. [36] Tontiwachwuthikul P, Meisen A, Lim CJ. CO2 absorption by NaOH, monoethanolamine and 2-amino-2-methyl-1-propanol solutions in a packed column. Chem Eng Sci 1992;47:381–90. [37] Hartono A, Svendsen HF. Density, viscosity, and excess properties of aqueous solution of diethylenetriamine (DETA). J Chem Thermodyn 2009;41:973–9. [38] Liao CP, Leron RB, Li MH. Mutual diffusion coefficients, density, and viscosity of aqueous solutions of new polyamine CO2 absorbents. Fluid Phase Equilib 2014;363:180–8. [39] Zhang J. An Experimental and Simulation Study on Liquid Flowing and Mass Transfer in RPB [PhD. Thesis]. Beijing, China: Beijing University of Chemical Technology; 1996.(in Chinese). [40] Neumann K, Hunold S, Beer MD, et al. Mass transfer studies in a pilot scale RPB with different packing diameters. Ind Eng Chem Res 2018;57:2257–66. [41] Danckwerts PV. Gas-liquid reactions. New York: Mcgraw-hill Book Co.; 1970. [42] Fu K, Chen G, Sema T, et al. Experimental study on mass transfer and prediction using artificial neural network for CO2 absorption into aqueous DETA. Chem Eng Sci 2013;100:195–202. [43] Afkhamipour M, Mofarahi M. Review on the mass transfer performance of CO2 absorption by amine-based solvents in low-and high-pressure absorption packed columns. RSC Adv 2017;7:17857–72. [44] Puxty G, Rowland R, Allport A, et al. Carbon dioxide postcombustion capture: a novel screening study of the carbon dioxide absorption performance of 76 amines. Environ Sci Technol 2009;43:6427–33. [45] Chang Y-C, Leron RB, Li M-H. Equilibrium solubility of carbon dioxide in aqueous solutions of (diethylenetriamine+piperazine). J Chem Thermodyn 2013;64:106–13. [46] Kim YE, Moon SJ, Yoon YI, et al. Heat of absorption and absorption capacity of CO2 in aqueous solutions of amine containing multiple amino groups. Sep Purif Technol 2014;122:112–8. [47] Chen G, Luo X, Zhang H, et al. Artificial neural network models for the prediction of CO2 solubility in aqueous amine solutions. Int J Greenhouse Gas Control 2015;39:174–84. [48] Wu S, Zhang LL, Sun B, et al. Mass Transfer Performance for CO2 Absorption by 2((2-aminoethyl)amino) ethanol Solution in Rotating Packed Bed. Energy Fuels 2017;31.

[1] BP energy. BP energy outlook 2017. Available online: , 2017 (accessed 4.2.2018). [2] BoalSwarm, Sierra Club and Greenpeace. Boom and Bust-Tracking Global Coal Power Plants. Available online: , 2017 (accessed 4.2.2018). [3] Yuan J, Wang Y, Zhang W, et al. Will recent boom in coal power lead to a bust in China? A micro-economic analysis. Energy Policy 2017;108:645–56. [4] Schäffer A, Brechtel K, Scheffknecht G. Comparative study on differently concentrated aqueous solutions of MEA and TETA for CO2 capture from flue gases. Fuel 2012;101:148–53. [5] Stec M, Tatarczuk A, Więcław-Solny L, et al. Pilot plant results for advanced CO2 capture process using amine scrubbing at the Jaworzno II Power Plant in Poland. Fuel 2015;151:50–6. [6] Nittaya T, Douglas PL, Croiset E, et al. Dynamic modelling and control of MEA absorption processes for CO2 capture from power plants. Fuel 2014;116:672–91. [7] Rochelle G, Chen E, Freeman S, et al. Aqueous piperazine as the new standard for CO2 capture technology. Chem Eng J 2011;171:725–33. [8] Freeman SA, Dugas R, Van Wagener DH, et al. Carbon dioxide capture with concentrated, aqueous piperazine. Int J Greenhouse Gas Control 2010;4:119–24. [9] Qian Z, Xu L, Cao H, et al. Modeling study on absorption of CO2 by aqueous solutions of N-methyldiethanolamine in rotating packed bed. Ind Eng Chem Res 2009;48:9261–7. [10] Wang Z, Fang M, Pan Y, et al. Amine-based absorbents selection for CO2 membrane vacuum regeneration technology by combined absorption–desorption analysis. Chem Eng Sci 2013;93:238–49. [11] Hartono A, da Silva EF, Svendsen HF. Kinetics of carbon dioxide absorption in aqueous solution of diethylenetriamine (DETA). Chem Eng Sci 2009;64:3205–13. [12] Fu K, Sema T, Liang Z, et al. Investigation of mass-transfer performance for CO2 absorption into diethylenetriamine (DETA) in a randomly packed column. Ind Eng Chem Res 2012;51:12058–64. [13] Zhang X, Fu K, Liang Z, et al. Experimental studies of regeneration heat duty for CO2 desorption from diethylenetriamine (DETA) solution in a stripper column packed with Dixon ring random packing. Fuel 2014;136:261–7. [14] Gao J, Yin J, Zhu F, et al. Postcombustion CO2 capture using diethylenetriamine (DETA) solvent in a pilot-plant test bed compared to monoethanolamine (MEA) solvent. Environ Prog Sustainable Energy 2017;36:1131–8. [15] Wang M, Joel AS, Ramshaw C, et al. Process intensification for post-combustion CO2 capture with chemical absorption: a critical review. Appl Energy 2015;158:275–91. [16] Liang Z, Idem R, Tontiwachwuthikul P, et al. Experimental study on the solvent regeneration of a CO2-loaded MEA solution using single and hybrid solid acid catalysts. AIChE J 2016;62:753–65. [17] Wang T, Yu W, Liu F, et al. Enhanced CO2 absorption and desorption by monoethanolamine (MEA)-based nanoparticle suspensions. Ind Eng Chem Res 2016;55:7830–8. [18] Cui M, Chen S, Qi T, et al. Investigation of CO2 capture in nonaqueous ethylethanolamine solution mixed with porous solids. J. Chem. Eng Data 2018. [19] Lin C-C, Liu W-T, Tan C-S. Removal of carbon dioxide by absorption in a rotating packed bed. Ind Eng Chem Res 2003;42:2381–6. [20] Jassim MS, Rochelle G, Eimer D, et al. Carbon dioxide absorption and desorption in aqueous monoethanolamine solutions in a rotating packed bed. Ind Eng Chem Res 2007;46:2823–33. [21] Sheng M, Sun B, Zhang F, et al. Mass-transfer characteristics of the CO2 absorption process in a rotating packed bed. Energy Fuels 2016;30:4215–20. [22] Zou H, Sheng M, Sun X, et al. Removal of hydrogen sulfide from coke oven gas by catalytic oxidative absorption in a rotating packed bed. Fuel 2017;204:47–53. [23] Gudena K, Rangaiah GP, Lakshminarayanan S. Optimal design of a rotating packed bed for VOC stripping from contaminated groundwater. Ind Eng Chem Res 2012;51:835–47. [24] Sudhoff D, Leimbrink M, Schleinitz M, et al. Modelling, design and flexibility analysis of rotating packed beds for distillation. Chem Eng Res Des 2015;94:72–89. [25] Wang W, Zou H-K, Chu G-W, et al. Bromination of butyl rubber in rotating packed

1527