Article Cite This: Ind. Eng. Chem. Res. XXXX, XXX, XXX−XXX
pubs.acs.org/IECR
Degradation of Amine Solvents in a CO2 Capture Plant at Lab-Scale: Experiments and Modeling Serena Delgado, Benoît Valentin, Domitille Bontemps, and Olivier Authier* EDF R&D Lab Chatou, 6 quai Watier, 78400 Chatou, France ABSTRACT: Carbon dioxide (CO2) concentrations in the atmosphere have increased significantly over the past century. Many methods have been devised to reduce CO2 industrial emissions, e.g., CO2 postcombustion absorption by amine-based solvents. Solvent degradation losses are very critical in this process, due to economic and environmental issues. The two main degradation pathways of amine-based aqueous solutions in the presence of CO2 are oxidative and thermal degradation. In this work, a lab-scale pilot plant has been set up to carry out degradation experiments during continuous and dynamic cycles of absorption and stripping with three different amine solvents: MEA (monoethanolamine) used as benchmark solvent for CO2 capture, a blend of 1MPZ (1-methylpiperazine) and PZ (piperazine), and a blend of MDEA (methyldiethanolamine) and MEA. The experimental data have been used to assess the performance of CO2 absorption over time and experimental conditions. The variation of CO2 fraction at the gas outlet of the reactor has been used as an indicator of solvent degradation. To simulate the behavior of the plant at different experimental conditions and with each solvent, a dynamic model has been developed, on the basis of the validation of a fast reaction regime. It reproduces accurately the pilot plant’s behavior during the absorption and stripping phases. Among the solvents’ physical properties, the effect of viscosity appears to be the most critical for the CO2 absorption efficiency. Kinetics of solvent degradation has finally been optimized to match experimental observations. Of the three solvents studied, 1MPZ/PZ is the most stable, whereas MEA and MDEA/MEA have quite similar degradation rates. of CO2 are oxidative and thermal degradation.9 Oxidative degradation takes place mainly at conditions found in the absorber, but dissolved oxygen may also be present in other parts of the plant. Oxidative degradation is observed in the presence of O2 in the solution. Indeed, flue gases from power plants contain a fraction of O2 (typically 6 vol %10) due to the excess of O2 required for complete combustion in the boiler. Thermal degradation takes place mainly in the stripper and reboiler, especially in the presence of CO2 and when the process temperature is the highest. In literature studies, thermal and oxidative degradation experiments are conducted in various open and closed test benches which are hard to compare. With regards to oxidative degradation, Léonard et al.,11 Supap et al.,10 Bello and Idem,12 and Pinto et al.13 have developed MEA degradation rate expressions. Several assumptions used to determine the kinetic parameters are the same for all studies. First, the experimental conditions are often more extreme than in industrial CO2 capture units, in terms of pressure within the reactor, volume fraction of O2 in the gas phase, and temperature. The degradation of MEA can then be measured over a shorter period than in real plants, from 1 to 3 weeks.
1. INTRODUCTION Due to increasing concern over carbon dioxide (CO2) emissions and their effect on the environment and climate change, many methods have been devised to reduce them over the recent decades. The most common technique currently undergoing research and development is CO2 capture and storage. Postcombustion CO2 capture by chemical absorption in amine solvent is as of today the most mature technology in the power industry.1 The CO2 capture process is mainly composed of three major pieces of equipment: (1) an absorber where CO2 is separated from the flue gas by absorption in an amine solvent; (2) a stripper where CO2 is desorbed from the solvent by heat provided in a reboiler; (3) an economizer where the sensible heat of the CO2-lean stream is recovered by heating the CO2-rich stream.2 Some of the critical issues of this process are solvent degradation that generates losses in process efficiency and additional operating costs to treat degradation products and replace degraded solvent inventory.3 Moreover, some degradation products are volatile and corrosive and can cause environmental and material issues. Understanding the mechanisms of solvent degradation is therefore important to be able to better control them.4−8 The solvents used in chemical absorption processes are mainly aqueous amine solutions such as monoethanolamine (MEA) which is a widely available chemical solvent. The two main degradation pathways of MEA solutions in the presence © XXXX American Chemical Society
Received: December 22, 2017 Revised: March 2, 2018 Accepted: April 5, 2018
A
DOI: 10.1021/acs.iecr.7b05225 Ind. Eng. Chem. Res. XXXX, XXX, XXX−XXX
Article
Industrial & Engineering Chemistry Research
Figure 1. Diagram of the CO2 capture pilot plant at lab-scale.
developing an apparent degradation rate expression. The two studies have the same assumptions, use similar experimental conditions, and lead to very close results. Few studies at lab-scale have developed a degradation rate expression for oxidative and thermal degradation for several amine solvents. In this work, a lab-scale pilot plant has been set up to carry out degradation experiments during 300 cycles of absorption and stripping, i.e., around 300 h of continuous operation, with three promising aqueous amine solvents: MEA (primary amine) used as benchmark solvent for CO2 capture, a blend of 1MPZ (1-methylpiperazine, cyclic amine) and PZ (piperazine, cyclic amine),22 and a blend of MDEA (methyldiethanolamine, tertiary amine) and MEA. The experimental data have been used to assess the performance of CO2 absorption versus time and experimental conditions. A dynamic model has been developed to simulate the main phenomena involved in the experiments under different conditions and with the three solvents, i.e., flows, mass transfer, thermodynamics, and kinetics of solvent absorption. Kinetics of solvent degradation has been finally optimized to match experimental results.
The authors assume that the accelerated degradation of MEA at lab-scale is representative of plant-scale conditions. Second, the thermal decomposition of MEA in the absence of CO2 and O2 is often neglected at first approximation because the temperature never comes close to 200 °C. Similarly, NOx- and SOxinduced degradation are considered negligible. Lastly, the influence of degradation products on oxidative degradation is neglected. However, the studies do not agree on the limiting step of the oxidation kinetics and the concentration of dissolved O2 in the solution. One key factor is the limiting step of the oxidation phenomenon. On the one hand, Goff and Rochelle14 put forward a mass transfer effect in MEA oxidative degradation. Therefore, experiments need to include a gas circulation. The apparatus must then be a semiopen batch reactor or an open batch reactor to be representative. Léonard et al.,11 Vevelstad et al.,9 Chandan et al.,15 Sexton and Rochelle,16 and Voice and Rochelle17 all used semiopen batch systems in their kinetic experiments. On the other hand, Supap et al.18 argue that there is no mass transfer limitation at the beginning of the reaction. Therefore, their study has been performed in closed vessels with no circulation of liquid and gas and their model is based on initial degradation rates. Other authors make the same assumption and choice in experimental apparatus, e.g., Zhao et al.19 and Bello and Idem.12 Both approaches lead to variations in experimental conditions. The monitored values are also different: on the one hand, an overall degradation rate is calculated, and on another hand, the initial degradation rate is calculated. Another major difference between the studies is the correlation chosen to calculate the concentration of dissolved O2 in the solution. Léonard et al.11 use a Henry’s constant for O2 in pure water, whereas Supap et al.,10 Bello and Idem,12 and Pinto et al.13 use a correlation from Rooney and Daniels20 developed for an aqueous solution of MEA. With regards to thermal degradation of MEA with CO2, the method is close to the oxidative degradation studies. A model that considers a detailed degradation mechanism has been developed by Davis.21 Instead of just one apparent kinetic law for MEA degradation, this study assumes elementary reactions and the number of parameters to fit is accordingly larger; however, the kinetic orders of the different species are determined by the assumption of elementary reactions. Léonard et al.11 assess the validity of this model by conducting experiments and
2. MATERIALS AND METHODS The lab-scale pilot plant23 is designed to reproduce, during continuous long periods, the dynamic cycling of the solvent between the absorber and the stripper columns with controlled fast heating up and cooling down of the solvent (Figure 1). All piping is made of hastelloy to uncouple oxidation and corrosion phenomena. The reactor is designed as a semibatch vertical bubble column (glass cylinder) acting as both an absorber and a stripper. The solvent stream runs countercurrent to the gaseous stream that is injected by a porous sparger at the bottom of the reactor. The hydrodynamics in the column corresponds to heterogeneous flow regime (superficial gas velocity lower than 0.2 m s−1). The reactor height is 350 mm with an internal diameter of 60 mm. The liquid volume in reactor is around 0.6 L. This bubble column is characterized by high volumetric liquid phase mass transfer coefficient (around 0.02 s−1 in the air−water system under ambient conditions) and very efficient operability at lab-scale (simple construction and little maintenance).24 The disadvantage of high pressure drop in bubble columns is not critical at lab-scale. Depending on absorption and stripping B
DOI: 10.1021/acs.iecr.7b05225 Ind. Eng. Chem. Res. XXXX, XXX, XXX−XXX
Article
Industrial & Engineering Chemistry Research Table 1. Experimental Conditions for Each Solventa
phases, temperature in the reactor can be controlled by a heat exchanger between 40 and 125 °C (maximum deviation of around 0.9 °C) and pressure can be assigned between 1.0 and 4.0 bar (precision of ±0.05 bar). During the absorption phase, the temperature is kept constant (around 40 °C) by the heat exchanger and the pressure is around 1 bar. The CO2 capture rate in the reactor varies continuously from a maximum value (i.e., CO2-lean solvent) at the beginning to a minimum value (i.e., CO2-rich solvent) at the end of absorption. During the stripping phase, a pressure of 4 bar is applied by N2 injection in the reactor (transition time between phases: around 1 min). Then, the solvent is heated up by the heat exchanger to around 120 °C linearly. The temperature is maintained at this level during around 6 s before decreasing again linearly to 40 °C. The solvent heating and cooling rates (around 5 °C min−1) as well as the absorption time (from 35 to 60 min) are adapted according to the solvent composition. The temperature within the liquid phase is continuously measured in the reactor. The pressure is also continuously measured at the condensate tank beneath the condenser. To check the efficiency of liquid mixing under ambient flow conditions, a tracer study has also been performed in water phase by pulsing with tracer in the system (i.e., 1.5 mL of NaCl saturated solution injected just above the liquid level in the reactor). For that purpose, the system was operated with liquid phase recycling and the tracer concentration was measured at the reactor outlet as a function of time. That analysis led to modeling, in first approximation, the liquid phase loop as two stirred reactors: one for the column and the other for the exchanger (total liquid residence time of around 16 s). Before its injection, the synthetic gas mixture (i.e., CO2 and N2 supplied by Air Liquide and air which is filtered, cleaned, and dried) is controlled in flow rate and composition (precisions of ±0.6 L min−1 for N2 and of ±0.12 L min−1 for CO2 and air). The gas is saturated at a regulated temperature with distilled water (18.2 mΩ) before entering the reactor, in order to minimize losses by evaporation and maintain the water balance in the reactor. A mist eliminator recovers droplets from the product vapor. The level of the liquid phase in the reactor is measured during transition phases, when no gas injection occurs, in order to set the humidification temperature at the reactor inlet. After contact with the gas in the reactor, the solvent circulates in a tubular heat exchanger before recycling at the top of the reactor (flow rate: 5 L min−1). The heat exchanger is supplied with mineral oil to enable a rapid heating or cooling of the solvent. At the outlet of the reactor, the gas is cooled down to 5 °C in the condenser in order to wash it and to recirculate the condensates (water−solvent mixture), thus minimizing the losses to the atmosphere due to mechanical entrainment and evaporation. A fraction of gas is then analyzed with gas analyzers (Environnement MIR 9000, precision of ±2%; Horiba PG-250, precision of ±0.2%) for monitoring continuously (i.e., every 10 s) CO2 and O2 volume fractions. The gas residence time from the reactor to the analyzer is no more than a few seconds. After final washing, the exhaust gas is carried away. The solvent loading factor (expressed in mol CO2 per mol amine) is assessed by periodic sampling in the liquid phase and total inorganic carbon analysis (Shimadzu TOC-L CSH). The 50× diluted sample is acidified with 30 wt % phosphoric acid causing inorganic carbon to be evolved as CO2. The amount of CO2 emitted is then measured with an infrared analyzer. Furthermore, the analysis of degradation products like heat stable salts which stay in the liquid phase and the volatile degradation products like NH3 are presented in other studies.25
case (A)
case (B)
case (C)
MEA
1MPZ/PZ
MDEA/MEA
solvent composition
phase N2 flow rate (L/min) CO2 flow rate (L/min) air flow rate (L/min) temperature (°C) pressure (bar) time (min) a
MEA: 30%w; H2O: 70%w
1MPZ: 30%w; PZ: 10%w; H2O: 60%w
MDEA: 25%w; MEA: 5%w; H2O: 70%w
A 21.5 4.2 4.3 50 1.013 35
A 22.1 3.6 4.3 42 1.013 40
A 21.2 4.5 4.3 40 1.013 60
S 45 0 0 120 3.9 28
S 45 0 0 123 3.9 38
S 45 0 0 123 3.9 36
A: absorption phase; S: stripping phase; flow rate: 20°C and 1 atm.
The protocol of each campaign is detailed in Table 1. Three amine solvents diluted in distilled water have been characterized: MEA (purity of 99%, Alfa Aesar), blend of 1MPZ (purity of 99%, Merck) and PZ (purity of 99%, Merck), and blend of MDEA (purity of 98%, Alfa Aesar) and MEA. The campaigns on 1MPZ/PZ have been performed three times to confirm the repeatability of the results (standard deviation of ±0.033 on maximum CO2 gas fraction at the outlet of the reactor).
3. MODELING The dynamic model is based on a systemic approach to account for gas and liquid flows inside the reactor and the liquid loop. A mass balance of CO2 is achieved by modeling the mass transfer of CO2, the thermodynamic equilibrium of CO2 between the two phases, and absorption kinetics in liquid phase. Additionally, solvent degradation kinetics is considered in the system as an apparent kinetic rate expression. 3.1. Gas−Liquid Transfer. The flow of the liquid phase in a bubble column is usually described by an ideal mixing model or by the plug flow model with dispersion.24,26 At first approximation, the chemical system is represented in Figure 2 by a simple network of three stirred reactors. The CO2 transfer between gas and liquid takes place in the reactor, and the mass transfer between the two phases is represented by the doublefilm model of Whitman.27 As part of this model, the absorption flux can then be expressed as i
G G Φ = k Ga(cCO − cCO ) 2 2
(1)
CO2 absorption by amine can be represented by the following basic equation where A is the amine and P the products of absorption. CO2 + νA → P (2) where ν is 2 for MEA, 1.5 for 1MPZ/PZ, and 2.2 for MDEA/ MEA. The coefficients used for the reactions between amines and CO2 (in mol of CO2 per mol of total amine) are assessed from the thermodynamic equilibrium under the conditions of absorption, according to the reactor temperature and CO2 partial pressure at reactor inlet. The chemical reactions taking place in the liquid phase tend to accelerate the CO2 mass transfer.28 Thus, an enhancement factor defined as the ratio between the absorption flux with chemical reaction and the physical absorption flux in a similar flow without reaction is used. The absorption flux can be rewritten as i
L L Φ = EkLa(cCO − cCO ) 2 2
C
(3) DOI: 10.1021/acs.iecr.7b05225 Ind. Eng. Chem. Res. XXXX, XXX, XXX−XXX
Article
Industrial & Engineering Chemistry Research
Figure 2. Gas−liquid system and mass transfer according to the double-film model.
Table 2. Parameters for the Gas−Liquid Equilibrium of CO2 in Amine Solvents and Coefficients of Antoine Law for Each Solvent parameter
MEA
1MPZ/PZ
1.7 11.2
1.3 28.3
1.0 215.4
K1
− 9013.5 T + 26.8
− 6258.4 T + 19.1
− 9870.9 T + 25.9
K2
exp(− 3498.0 T + 8.5)
exp(− 569.3 T + 2.9)
exp(− 593.6 T + 4.7)
K3
4333.2
1430.0
174.0
a b c
4.98 1557.48 −89.62
T − 9.49
DCO2,L (5)
rov = kovcAcCO2
i ln(pCO ) = A ln α + K1 + 2
(6)
∑j (kovjcj)DCO2,L kL
(7)
When the reaction takes place mainly within the liquid film, the gas−liquid system is in a fast reaction regime as follows:29 3 < Ha