Available online at www.sciencedirect.com
Energy Procedia
Energy Procedia 4 (2011) 1520–1525
Energy Procedia 00 (2010) 000–000
www.elsevier.com/locate/procedia www.elsevier.com/locate/XXX
GHGT-10
Characterisation of CO2 absorption in various solvents for PCC applications by Raman Spectroscopy Monika Vogta1*, Christoph Paselb, Dieter Bathena,b b
a IUTA e.V., Bliersheimer Str. 60, 47229 Duisburg, Germany Chair of Thermal Process Technology, University Duisburg-Essen, Lotharstr. 1, 46057 Duisburg, Germany
Elsevier use only: Received date here; revised date here; accepted date here
Abstract Results of investigations on inline solvent analytics by Raman spectroscopy during gas scrubbing of CO2 from power plant flue gases are presented. Superior aim is the optimisation of the gas treating process concerning CO2 capture rate and energy demand.
c© 2010 ⃝ 2011 Elsevier Published byAll Elsevier Ltd. Open access under CC BY-NC-ND license. Ltd. rights reserved keywords: PCC; CO2 solvent loading; inline analytics; Raman spectroscopy
1. Motivation The focus of ongoing R&D activities in post combustion capture (PCC) is on the reduction of energy consumption in solvent regeneration and the verification of operating performance under typical conditions of a power plant. How a CO2 scrubber has to be operated efficiently will be investigated in test programs at several pilot plants in the following years [1, 2, 3, 4]. With regard to process optimisation, an effective solvent analysis considering characteristic parameters such as the actual loading of the solvent during absorption and regeneration is vitally important. A continuous control of the capture efficiency e.g. in the petroleum industry is performed today via a balance of the gaseous phase. This method does not allow the instantaneous causal research of decreasing CO2 capture efficiency. To counteract decreasing CO2 capture the operators often rely on empirical knowledge by enhancing the solvent flow ratio, application of additives, partial exchange of the solvent or elevating the regenerator temperature [5, 6]. All these actions result in higher operating costs. Caused by the complexity of the scrubber operation and its high integration into the power plant process, the process control via observing the CO2 concentration in raw and cleaned gas is defective. Furthermore the larger dimensions of the planned scrubbers will induce a major delay in data acquisition. By investigating the solvent quality at different, process relevant positions in the solvent circuit with a suitable inline analysis an optimisation of the scrubber operation concerning energy and ressources demand would be possible. This is advantageous during
* Corresponding author. Tel.: +49-2065-418-175; fax: +49-2065-418-211 E-mail address:
[email protected].
doi:10.1016/j.egypro.2011.02.020
2
M. Vogt etVogt/ al. / Energy EnergyProcedia Procedia004 (2010) (2011)000–000 1520–1525
1521
adapting the scrubber to changes of the power plant like part and full load operation. Also changes of the solvent by deterioration and degradation can be detected and dealt with effectively, which will increase the availability of the equipment. An inline analysis of the solvent is not possible up to now, although several research groups are working on this topic. NMR-, IR- and Raman spectroscopy are generally applicable for that purpose. [7] describes a laboratory study with fourier transform infrared spectroscopy (FT-IR) coupled to an ATR-probe (attenuated total reflection) on the reaction monitoring of CO2 absorption in mono ethanol amine (MEA), 2-amino2-methyl-1-propanol (AMP) and methyl diethanol amine (MDEA) which showed the general applicability. The Chair of Thermodynamics at the University of Kaiserslautern applies high resolution nuclear magnetic resonance (NMR) spectroscopy to investigate the reaction kinetics of CO2 and several solvents. These investigations mainly aim at the reaction process [8, 9] and the detection of the chemical structure of degradation products [10]. The Raman spectroscopy can be implemented as inline solvent analysis during CO2 gas scrubbing due to the fact that a sample preparation is not necessary and the water content in the solvent does not influence the spectra. The low intensity of the water band as well as its clear separation from the fingerprint range in the Raman spectrum is advantageous for this application. The Raman spectrum is characteristic for each molecule and the intensity of Raman scattering is directly proportional to the concentration of the respective analyte. These properties of Raman spectroscopy allow the identification of specific analytes in a multi component mixture. The characteristic distribution of intensities depends only slightly from environmental conditions like solvent pressure and temperature. A further benefit is the fact, that nearly each chemical compound has at least one Raman active bond. The Raman spectra are recorded within a few seconds with the possibility of improved detection limits by enhancing the measuring time. Since the exciting laser beam as well as the resulting Raman radiation can be transferred lossless via a fibre optical cable, the installation of Raman spectroscopy in an industrial gas scrubber may be attractive. On this account IUTA and the Chair of Thermal Process Engineering of the University of Duisburg-Essen are working on Raman spectroscopy for quantitative determination of the amine content, the CO2 loading and the amount of solvent specific reaction products. The feasibility to determine the concentration of selected alkanolamines and the solvent specific reaction products formed during CO2 capture will be shown. 2. Experiment All experiments were performed with the RamanRxn1 analyzer of Kaiser Optical Systems SARL, equipped with a 785-nm Invictus™ laser and fiber-coupled MR probe fitted with a 1/2" immersion optic. Spectra were acquired within seconds using 400mW of laser power. The investigations concentrated on aqueous solutions of mono ethanol amine (MEA), 2-(2-Aminoethoxy) ethanol (Diglycolamine® DGA) or methyldiethanolamine (MDEA). For all experimental investigations the equilibrium apparatus (see Figure 1) at the Chair of Thermal Process Engineering at the University of Duisburg-Essen was used. A thermo and pressure stable immersion optic has been installed in the reaction vessel of the equilibrium apparatus. Inside the reaction vessel a certain amount of solvent is filled and the absorption of CO2 takes place until thermodynamic equilibrium under the specific process conditions is reached. By varying the operation parameters (pressure and temperature) typical conditions in a CO2 absorber can be simulated. The mass transfer is enhanced by stirring the liquid during gas feeding via a frit. The amount of CO2 fed into the reaction vessel is calculated via a mass flow controller and is used to calculate the solvent loading. After reaching equilibrium at different CO2loadings, the spectra for the calibration procedure are recorded. For each of the solvents the whole concentration range of the expected solvent loadings under real operation conditions is considered. For each loading 20 spectra are recorded of which 15 are used for the calibration procedure while the remaining spectra are used for the comparison of measurement and calculation.
1522
M. Vogt al. / Energy Procedia 4 (2011) 1520–1525 Vogt et / Energy Procedia 00 (2010) 000–000
3
Figure1: process flow diagram of the equilibrium apparatus 3. Results In preliminary experiments Raman spectra of MEA, DGA and MDEA were recorded at several concentration levels. By comparing the spectra at the different concentration levels an identification of relevant bands for the particular solvent was achieved. Also the identification of the solvent characteristic bands in mixtures of different solvents (e.g. 15 weight % MEA and 35 weight % MDEA in water) was successful. To obtain spectra of protonated MEA (MEAH+) strong acidic aqueous solutions of MEA were measured. The spectra of the alkaline MEA solution and the protonated MEA solution showed no significant differences. Therefore it is not possible to distinguish protonated and deprotonated MEA. The Raman spectra of 30 weight % MEA solutions at 25 °C, 40 °C and 60 °C at different CO2-loadings were recorded. The spectra showed only minor differences so that the temperature dependency of Raman spectroscopy can be neglected in this temperature range. By continuous recording of spectra during CO2 absorption and thermal desorption the applicability of Raman spectroscopy for the qualitative monitoring of the reaction process was proven. As an example Figure 2 shows the spectra at the loading of 40 weight % DGA with CO2 at 25 °C with the spectra range of 300 to 1.700 cm-1 used for quantification. The Raman spectra were recorded during the absorption process of an unloaded solution up to a limit of 0.4 mol CO2 / mol amine using an immersion probe. During the absorption e.g. the intensity of the bands at 400, 1200, 1350 and 1600 cm-1 decreased. Without any further specification these bands can be allocated to the amine. In
1523
M. Vogt etVogt/ al. / Energy EnergyProcedia Procedia004 (2010) (2011)000–000 1520–1525
4
contradiction the intensity of the bands 1000, 1150, 1400 and 1700 cm-1 increased so they are allocated to the reaction products (mainly carbamate). Also the qualitative reaction monitoring of MDEA with Raman spectroscopy is possible, although no carbamate is formed. Instead, in tertiary amines carbonate is formed already at low CO2 loadings.
reaction product NH2
reaction product
reaction product
reaction product NH2
NH2 NH2
400
600
800
1000 1200 Raman Shift (1/cm)
1400
1600
Figure 2: reaction monitoring with Raman spectroscopy in 40 weight % DGA For the primary amines MEA (30 weight %) and DGA® (30 weight %) a reliable calibration up to solvent loadings of 0.5 mole CO2/mole amine has been achieved. Multivariate linear regression PLS 1 (partial least squares, one principal component) has been applied. For the calibration the whole spectral range between 300 and 1700 cm-1 was regarded without any data processing. The deviation is mainly due to experimental errors in the range of 2 %. The multivariate calibration for MEA e.g. with 68 samples and validation with 12 test samples resulted in a coefficient of correlation of 99.9 % with a mean prediction error of 2.2 % for the loading. The mean blank value is 0.01 mol CO2 / mol amine while the standard deviation accounts for 0.003 mol CO2 / mol amine. This results in a detection limit of 0.02 mol CO2 / mol amine. This high accuracy of the linear regression method becomes comprehensible when concentration gradients of the species as a function of CO2 loading are plotted. These concentration gradients were calculated using the electrolyte-NRTL model (see Figure 3). For both primary amines MEA and DGA the figure illustrates a nearly linear dependency of amine, protonated amine and carbamate concentration up to a loading of 0.5 mole CO2 / mole amine. At higher loadings in the range of 0.5 to 0.7 mole CO2/mole amine the reaction CO2 + carbamate + 2 H2O ҡ amineH+ + 2 HCO3(1) becomes dominant. By this reaction mechanism more CO2 can be chemically bound. As carbamate is consumed, the carbamate concentration reaches a maximum at approximately 0.5 mole CO2/mole amine. At loadings higher than 0.7 mole CO2/mole amine also the formation of CO2 (aq) becomes relevant (see Figure 3). In technical absorption processes loadings of 0.4 up to 0.45 mole CO2 / mol amine can be reached. The provisional calibration for the primary amines allows already to determine the solvent loading. Furthermore it is possible to exceed the range up to loadings of 0.6 mole CO2/mole amine for MEA, as the multivariate calibration model still delivers satisfying results.
1524
M. Vogt al. / Energy Procedia 4 (2011) 1520–1525 Vogt et / Energy Procedia 00 (2010) 000–000
5
7,0
molality [mol/kg water]
6,0
5,0 MEAH+
HCO3-
4,0 MEA 3,0
Carbam at
2,0
1,0 CO2 (aq)
0,0 0,0
0,2
0,4
0,6
0,8
1,0
loading [m ol CO2/m ol MEA]
Figure 3: species during loading of 30 weight % MEA at 25 °C (thermodynamic equilibrium, electrolyte-NRTL model) [AspenPlus™ CO2 Capture Process Models, 11] As tertiary amines (like MDEA) or sterically hindered amines do not form carbamate, another reaction mechanism is vital. Up to loadings of 0.8 mole CO2/mole amine the following two reactions are dominant. CO2 + MDEA ҡ MEAH+ + HCO3-
(2)
HCO3- + H2O ҡ CO32- + H3O+ (3) Again a linear dependency of amine concentration and the summarised concentration of carbonate and hydrocarbonate on loading exists. Hence a calibration with linear regression for Raman spectroscopy is expected to be successful also for those amines, which do not form carbamate. 4. Outlook The potential of Raman spectroscopy for inline concentration determination of various characteristic reactants and products of CO2 absorption in several solvents will be further investigated. It is foreseen to extend the solvents to other alkanolamines, sterically hindered amines, amino acid salts and piperazine. Furthermore the detection of typical markers to qualify the solvent degradation will be examined. The determination of free amine and other reaction products, which are not allocated to a Raman band, was done by modelling the thermodynamic equilibrium. This approach is limited, if more complex or unknown reaction processes are dominant or the influence of degradation products becomes significant. To deal with these constraints the calibration procedure will be optimised by using innovative multivariate methods. After finalising the calibration at the equilibrium apparatus the immersion optic will be attached to the absorber and regenerator column of the IUTA e.V. pilot plant. The test programme at the IUTA plant aims at the verification of the calibration methods under pilot plant conditions and the suitability of the Raman spectroscopy as inline solvent analytics to improve the efficiency of a gas scrubber.
6
M. Vogt etVogt/ al. / Energy EnergyProcedia Procedia004 (2010) (2011)000–000 1520–1525
1525
5. List of References [1] Zero Emissions Platform: Zep’s comprehensive list of announced large-scale European CCS demonstration projects, URL: http://www.zeroemissionsplatform.eu/projects.html [2] De Koeijer, G., 2009: European CO2 Test Centre Mongstad – Testing, Verification and Demonstration of PostCombustion Technologies, Vortrag GHGT-9, Washington 2008, Energy Procedia 1 (2009) [3] Moser, P. Schmidt, S., Sieder, G., Garcia, H., Ciattaglia, I., Klein, H., 2009.: Enabling Post Combustion Capture Optimization – The Pilot Plant Project at Niederaussem, Vortrag GHGT-9, Washington 2008, Energy Procedia 1 (2009) [4] Jockenhoevel, T., Schneider, R., Rode, H., 2009. Development of an Economic Post-Combustion Carbon Capture Process, Vortrag GHGT-9, Washington 2008, Energy Procedia 1 (2009) [5] Jenkins, J. L., Haws, R., 2002. Understanding gas treating fundamentals, PTQ Winter 2001/2002, 61-71 [6] Amine Best Practice Group, 2007. Amine Basic Practices Guidelines. http://refiningonline.com/abpg_kb/ABPG5.pdf [7] Jackson, P., Robinson, K., Puxty, P. and Attalla, M.: In situ Fourier Transform-Infrared (FT-IR) analysis of carbon dioxide absorption and desorption in amine solutions, presentation GHGT-9, Washington 2008, Energy Procedia 1 (2009) [8] Hoch, S.: http://thermo.mv.uni-kl.de/index.php?id=60 [9] Yang, Q., Bown, M., Ali, A., Winkler, D., Puxty, G., Attalla, M.: A Carbon-13 NMR Study of Carbon dioxide Absorption and Desorption with Aqueous Amine Solutions, presentation GHGT-9, Washington 2008, Energy Procedia 1 (2009) [10] Strazisar, B. R., Anderson, R.R. and White, C. M.: Degradation Pathways for Monoethanolamine in a CO2 Capture Facility, Energy Fuels 2003, 1034-1039 [11] Austgen, D.M., Rochelle, G.T., Peng, X., Chen, C., 1989. Model of vapor-liquid equilibria for aqueous acid gas-alkanolamine systems using the electrolyte-NRTL equation. Industrial and Engineering Chemistry Research, 28, 1060–1073.
