Carbon Dioxide Capture and Acid Gas Injection Scrivener Publishing 100 Cummings Center, Suite 541J Beverly, MA 01915-6106 Publishers at Scrivener Martin Scrivener (martin@scrivenerpublishing.com) Phillip Carmical (pcarmical@scrivenerpublishing.com) Carbon Dioxide Capture and Acid Gas Injection Edited by Ying Wu, John J. Carroll and Weiyao Zhu This edition first published 2017 by John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, USA and Scrivener Publishing LLC, 100 Cummings Center, Suite 541J, Beverly, MA 01915, USA © 2017 Scrivener Publishing LLC For more information about Scrivener publications please visit www.scrivenerpublishing.com. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, except as permitted by law. Advice on how to obtain permission to reuse material from this title is available at http://www.wiley.com/go/permissions. Wiley Global Headquarters 111 River Street, Hoboken, NJ 07030, USA For details of our global editorial offices, customer services, and more information about Wiley products visit us at www.wiley.com. Limit of Liability/Disclaimer of Warranty While the publisher and authors have used their best efforts in preparing this work, they make no representations or warranties with respect to the accuracy or completeness of the contents of this work and specifically disclaim all warranties, including without limitation any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created or extended by sales representatives, written sales materials, or promotional statements for this work. The fact that an organization, website, or product is referred to in this work as a citation and/or potential source of further information does not mean that the publisher and authors endorse the information or services the organization, website, or product may provide or recommendations it may make. This work is sold with the understanding that the publisher is not engaged in rendering professional services. The advice and strategies contained herein may not be suitable for your situation. You should consult with a specialist where appropriate. Neither the publisher nor authors shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages. Further, readers should be aware that websites listed in this work may have changed or disappeared between when this work was written and when it is read. Library of Congress Cataloging-in-Publication Data ISBN 978-1-118-93866-9 Cover image: Gas Drilling Machine | Cylonphoto | Dreamstime.com | Gas Storage Spheres | Sasin Tipchai | Dreamstime.com | Na tural Gas Plant | Jevtic | Dreamstime.com Cover design by Kris Hackerott Set in size of 11pt and Minion Pro by Exeter Premedia Services Private Ltd., Chennai, India Printed in 10 9 8 7 6 5 4 3 2 1 Contents Prefacexiii 1Enthalpies of Carbon Dioxide-Methane and Carbon Dioxide-Nitrogen Mixtures: Comparison with Thermodynamic Models 1 Erin L. Roberts and John J. Carroll 1.1 Introduction 1 1.2 Enthalpy 2 1.3 Literature Review 2 1.3.1 Carbon Dioxide-Methane 4 1.3.2 Carbon Dioxide-Nitrogen 4 1.4 Calculations 5 1.4.1 Benedict-Webb-Rubin 6 1.4.2 Lee-Kesler 12 1.4.3 Soave-Redlich-Kwong 17 1.4.4 Peng-Robinson 23 1.4.5 AQUAlibrium 28 1.5 Discussion 33 1.6 Conclusion 36 References37 2Enthalpies of Hydrogen Sulfide-Methane Mixture: Comparison with Thermodynamic Models Erin L. Roberts and John J. Carroll 2.1 Introduction 2.2 Enthalpy 2.3 Literature Review 2.4 Calculations 2.4.1 Lee-Kesler 2.4.2 Benedict-Webb-Rubin 2.4.3 Soave-Redlich-Kwong 39 39 40 40 41 41 43 43 v vi Contents 2.4.4 Redlich-Kwong 47 2.4.5 Peng-Robinson 47 2.4.6 AQUAlibrium 50 2.5 Discussion 50 2.6 Conclusion 52 References54 3Phase Behavior and Reaction Thermodynamics 55 Involving Dense-Phase CO2 Impurities J.A. Commodore, C.E. Deering and R.A. Marriott 3.1 Introduction 55 3.2 Experimental 57 3.3 Results and Discussion 58 3.3.1 Phase Behavior Studies of SO2 Dissolved in 58 Dense CO2 Fluid 3.3.2 The Densimetric Properties of CS2 and 60 CO2 Mixtures References61 4Sulfur Recovery in High Density CO2 Fluid 63 S. Lee and R.A. Marriott 4.1 Introduction 64 4.2 Literature Review 64 4.3 Methodology 65 4.4 Results and Discussion 66 4.5 Conclusion and Future Directions 67 References68 5Carbon Capture Performance of Seven Novel Immidazolium and Pyridinium Based Ionic Liquids Mohamed Zoubeik, Mohanned Mohamedali and Amr Henni 5.1 Introduction 5.2 Experimental Work 5.2.1 Materials 5.2.2 Density Measurement 5.2.3 Solubility Measurement 5.3 Modeling 5.3.1 Calculation of Henry’s Law Constants 5.3.2 Critical Properties Calculations 5.3.3 Peng Robinson EoS 71 71 73 73 73 73 76 76 76 76 Contents vii 5.4 Results and Discussion 77 5.4.1 Density 77 5.4.2 Critical Properties 77 5.4.3 CO2 Solubility 78 5.4.4 The Effect of Changing the Cation 81 5.4.5 The Effect of Changing the Anion 84 5.4.6 Henry’s Law Constant, Enthalpy and Entropy Calculations85 5.4.7 Thermodynamic Modeling of CO2 Solubility 86 5.5 Conclusion 87 Acknowledgements88 References88 6Vitrisol a 100% Selective Process for H2S Removal in the Presence of CO291 W.N. Wermink, N. Ramachandran, and G.F. Versteeg 6.1 Introduction 92 6.2 Case Definition 94 6.3 “Amine-Treated” Cases by PPS 95 6.3.1 Introduction to PPS 95 6.3.2 Process Description 96 6.3.3 PFD 97 6.3.4 Results 97 6.3.4.1 Case 1 97 6.3.4.2 Case 2 97 6.4 Vitrisol Process Extended with Regeneration of Active Component99 6.4.1 Technology Description 99 6.4.2 Parameters Determining the Process Boundary Conditions99 6.4.3 Absorption Section 101 6.4.4 Regeneration Section 102 6.4.5 Sulphur Recovery Section 104 6.4.6 CO2-Absorber105 6.4.7 PFD 105 6.5 Results 105 6.6 Discussion 110 6.6.1 Comparison of Amine Treating Solutions to Vitrisol 6.6.2 Enhanced H2S Removal of Barnett Shale Gas (case 2) 112 viii Contents 6.7 Conclusions 113 6.8 Notation 115 References115 Appendix 6-A: H&M Balance of Case 1 (British Columbia shale) of the Amine Process 117 Appendix 6-B H&M Balance of Case 2a (Barnett shale) of the Amine Process with Stripper Promoter 119 Appendix 6-C H&M Balance of Case 3 (Barnett shale) of the Amine Process (MEA) 121 Appendix 6-D: H&M Balance of Case 1 (British 123 Columbia shale) of the Vitrisol process Appendix 6-E H&M Balance of Case 2 (Barnett shale) 125 of the Vitrisol Process 7 New Amine Based Solvents for Acid Gas Removal 127 Yohann Coulier, Elise El Ahmar, Jean-Yves Coxam, Elise Provost, Didier Dalmazzone, Patrice Paricaud, Christophe Coquelet and Karine Ballerat-Busserolles 7.1 Introduction 128 7.2 Chemicals and Materials 131 7.3 Liquid-Liquid Equilibria 131 7.3.1 LLE in {methylpiperidines – H2O} and {methylpiperidines – H2O – CO2}131 7.3.2 Liquid-Liquid Equilibria of Ternary 135 Systems {Amine – H2O – Glycol} 7.3.3 Liquid-Liquid Equilibria of the Quaternary Systems {CO2 – NMPD – TEG – H2O}136 7.4 Densities and Heat Capacities of Ternary 137 Systems {NMPD – H2O – Glycol} 7.4.1 Densities 137 7.4.2 Specific Heat Capacities 137 7.5 Vapor-Liquid Equilibria of Ternary Systems {NMPD – TEG – H2O – CO2}139 7.6 Enthalpies of Solution 140 7.7 Discussion and Conclusion 143 Acknowledgments143 References144 Contents ix 8Improved Solvents for CO2 Capture by Molecular Simulation Methodology 147 William R. Smith 8.1 Introduction 147 8.2 Physical and Chemical Models 149 8.3 Molecular-Level Models and Algorithms for Thermodynamic Property Predictions 150 8.4 Molecular-Level Models and Methodology for MEA–H2O–CO2153 8.4.1 Extensions to Other Alkanolamine Solvents and Their Mixtures 155 Acknowledgements157 References157 9Strategies for Minimizing Hydrocarbon Contamination in Amine Acid Gas for Reinjection 161 Mike Sheilan, Ben Spooner and David Engel 9.1 Introduction 162 9.2 Amine Sweetening Process 162 9.3 Hydrocarbons in Amine 164 9.4 Effect of Hydrocarbons on the Acid Gas Reinjection System 166 9.5 Effect of Hydrocarbons on the Amine Plant 167 9.6 Minimizing Hydrocarbon Content in Amine Acid Gas 171 9.6.1 Option 1. Optimization of the Amine Plant Operation 171 9.6.2 Option 2. Amine Flash Tanks 176 9.6.3 Option 3. Rich Amine Liquid Coalescers 178 9.6.4 Option 4. Use of Skimming Devices 180 9.6.5 Option 5. Technological Solutions 182 References183 10Modeling of Transient Pressure Response for CO2 Flooding Process by Incorporating Convection and Diffusion Driven Mass Transfer Jianli Li and Gang Zhao 10.1 Introduction 10.2 Model Development 10.2.1 Pressure Diffusion 10.2.2 Mass Transfer 10.2.3 Solutions 185 186 187 187 188 190 x Contents 10.3 Results and Discussion 191 10.3.1 Flow Regimes 191 10.3.2 Effect of Mass Transfer 192 10.3.3 Sensitivity Analysis 195 10.3.3.1 CO2 Bank 195 10.3.3.2 Reservoir Outer Boundary 196 10.4 Conclusions 196 Acknowledgments197 References197 11 Well Modeling Aspects of CO2 Sequestration 199 Liaqat Ali and Russell E. Bentley 11.1 Introduction 199 11.2 Delivery Conditions 200 11.3 Reservoir and Completion Data 201 11.4 Inflow Performance Relationship (IPR) and Injectivity Index 201 11.5 Equation of State (EOS) 202 11.6 Vertical Flow Performance (VFP) Curves 205 11.7 Impact of the Well Deviation on CO2 Injection 208 11.8 Implication of Bottom Hole Temperature (BHT) on Reservoir 209 11.9 Impact of CO2 Phase Change 213 11.10 Injection Rates, Facility Design Constraints and Number of Wells Required 214 11.11 Wellhead Temperature Effect on VFP Curves 214 11.12 Effect of Impurities in CO2 on VFP Curves 216 11.13 Concluding Remarks 217 Conversion Factors 218 References218 12Effects of Acid Gas Reinjection on Enhanced Natural Gas Recovery and Carbon Dioxide Geological Storage: Investigation of the Right Bank of the Amu Darya River Qi Li, Xiaying Li, Zhiyong Niu, Dongqin Kuang, Jianli Ma, Xuehao Liu, Yankun Sun and Xiaochun Li 12.1 Introduction 12.2 The Amu Darya Right Bank Gas Reservoirs in Turkmenistan 221 222 223 Contents xi 12.3 Model Development 223 12.3.1 State equation 224 12.3.1.1 Introduction of Traditional PR State Equation 224 12.3.1.2 Modifications for the Vapor-Aqueous System 224 12.3.2 Salinity 225 12.3.3 Diffusion 226 12.3.3.1 Diffusion Coefficients 226 12.3.3.2 The Cross-Phase Diffusion Coefficients226 12.4 Simulation Model 227 12.4.1 Model Parameters 227 12.4.2 Grid-Sensitive Research of the Model 227 12.4.3 The Development and Exploitation Mode 230 12.5 Results and Discussion 230 12.5.1 Reservoir Pressure 230 12.5.2 Gas Sequestration 232 12.5.3 Production 235 12.5.4 Recovery Ratio and Recovery Percentage 238 12.6 Conclusions 239 12.7 Acknowledgments 240 References241 Index245 Preface The Sixth International Acid Gas Injection Symposium (AGIS VI) was held in Houston, Texas, in September 2016. As with previous Symposia, the focus of AGIS VI was the injection of acid gas (CO2, H2S, and mixtures of these components) for the purposes of disposal or for enhanced oil and/or gas recovery. This book contains select papers from the Symposium in Houston. The capture of carbon dioxide from flue gas and its disposal into a subsurface geological formation remains a viable option for the clean use of hydrocarbon fuels. The related technology is acid gas injection. Here the H2S and CO2 are removed from raw natural gas. This volume contains papers directly related to these two topics ranging from the physical properties of the gas mixtures, evaluation of new and existing solvents, and subsurface engineering aspects of the process. Furthermore, contributors came from Canada, Europe, and China, as well as from the host country, the United States. And this is reflected in the papers in this volume. On a very sad note, Marco Satyro passed away on September 8, 2016, just prior to the Symposium. Marco was a good friend of AGIS being an active member of the Technical Committee for many years. He contributed many papers and encouraged many others to participate. At the first AGIS he presented the paper “The Performance of State of the Art Industrial Thermodynamic Models for the Correlation and Prediction of Acid Gas Solubility in Water” and this paper appeared in the first volume of the Advances in Natural Gas Engineering. He also was the coauthor of several other contributions to the Series and they are listed below. This volume is dedicated to the memory of Dr. Satyro. References – papers of M.A. Satyro from the Advances in Natural Gas Engineering series. M.A. Satyro, and J. van der Lee, “The Performance of State of the Art Industrial Thermodynamic Models for the Correlation and Prediction of Acid Gas Solubility in Water”, pp. 21–34, Acid Gas Injection and Related Technologies, Y. Wu and J.J. Carroll (eds.), Scrivener Publishing (2011). xiii xiv Preface H. Motahhari, M.A. Satyro, and H.W. Yarranton, “Acid Gas Viscosity Modeling with the Expanded Fluid Viscosity Correlation”, pp. 41–52, Carbon Dioxide Sequestration and Related Technologies, (2011), Y. Wu, J.J. Carroll, and Z. Du (eds.), Scrivener Publishing (2011). J. van der Lee, J.J. Carroll, and M.A. Satyro, “A Look at Solid CO2 Formation in Several High CO2 Concentration Depressuring Scenarios”, pp. 117–128, Sour Gas and Related Technologies, Y. Wu, J.J. Carroll, and W. Zhu (eds), Scrivener Publishing (2012). M.A. Satyro, and J.J. Carroll, “Phase Equilibrium in the Systems Hydrogen Sulfide + Methanol and Carbon Dioxide + Methanol”, pp. 99–109, Gas Injection for Disposal and Enhanced Recovery, Y. Wu, J.J. Carroll, and Q. Li (eds.), Scrivener Publishing (2014). A.R.J. Arendsen, G.F. Versteeg, J. van der Lee,R. Cota, and M.A. Satyro, “Comparison of the Design of CO2-capture Processes using Equilibrium and Rate Based Models”, pp. 155–174, Gas Injection for Disposal and Enhanced Recovery, Y. Wu, J.J. Carroll, and Q. Li (eds.), Scrivener Publishing (2014). M.A. Satyro and H.W. Yarranton, “A Simple Model for the Calculation of Electrolyte Mixture Viscosities”, pp. 95–104, Acid Gas Extraction for Disposal and Related Topics, Y. Wu, J.J. Carroll, and W. Zhu (eds.), Scrivener Publishing (2016). 1 Enthalpies of Carbon Dioxide-Methane and Carbon Dioxide-Nitrogen Mixtures: Comparison with Thermodynamic Models Erin L. Roberts and John J. Carroll Gas Liquids Engineering, Calgary, Alberta, Canada Abstract The physical properties of acid-gas injection streams are important for use in design considerations of the acid-gas scheme. One such property is the enthalpy of the stream. As carbon dioxide is rarely pure, with methane and nitrogen being common impurities in the stream, the effect of these impurities on the enthalpy is also important to consider. This study compares experimentally determined excess enthalpies and enthalpy departures from literature to the enthalpy predictions of five different models, Benedict-Webb-Rubin, Lee-Kesler, Soave-Redlich-Kwong, and Peng-Robinson from VMGSim, as well as AQUAlibrium software. The mixtures studied are carbon dioxide-methane, as well as carbon dioxide- nitrogen mixtures at a wide range of compositions. The Soave-Redlich-Kwong model gave the most accurate predictions for both the excess enthalpies and enthalpy departures, with Lee-Kesler frequently giving the least accurate predictions for the mixtures. 1.1 Introduction An increase in demand of natural gas has led producers to pursue poorer quality reservoirs. These contain higher levels of carbon dioxide that then must be responsibly disposed. Regulations prevent the flaring of the acidgas mixtures, therefore requiring an alternate means of disposal. One such method is the injection of acid gas into subsurface reservoirs. Ying Wu, John J. Carroll and Weiyao Zhu (eds.) CO2 Capture and Acid Gas Injection, (1–38) 2017 © Scrivener Publishing LLC 1 2 Carbon Dioxide Capture and Acid Gas Injection An understanding of the physical properties of the stream is essential in the design of the acid-gas injection scheme. The enthalpy of the stream is required in the design of the compressor for injection. Common impurities in the carbon dioxide include methane and nitrogen; therefore the effect of these impurities on the enthalpy of carbon dioxide is required for design. This paper investigates the accuracy of five different thermodynamic models for predicting such mixtures. Four different equations of state, Benedict-Webb-Rubin (BWR), Lee-Kesler (LK), Soave-Redlich-Kwong (SRK), Peng-Robinson (1978) were used with VMGSim software, as well as the AQUAlibrium model. BWR and LK are multi-constant equations, and SRK and PR78 are cubic equations of state. The AQUAlibrium model uses a variation of Peng-Robinson. 1.2 Enthalpy The enthalpy of mixtures can be determined in a number of ways. One method is to use excess enthalpy (enthalpy of mixing). Excess enthalpy is defined as HE Hm i xi H i (1.1) where: HE – Excess enthalpy Hm – Enthalpy of mixture Hi – Enthalpy of component i xi – mol fraction of component i Alternatively, the enthalpy of the mixture can be represented as an enthalpy departure, a difference between the enthalpy at a given pressure, and the enthalpy at a reference pressure while keeping the temperature constant. Enthalpies can be expressed in J/mol, or for greater relevance to acidgas injection design, can be expressed in HP/MMSCFD. The conversion between units is 1 HP/MMSCFD to 53.86 J/mol. 1.3 Literature Review A review of literature was performed to compile experimental data for the enthalpy of carbon dioxide-methane mixtures as well as carbon dioxidenitrogen mixtures. Table 1.1 summarizes the relevant data used in this study. 1. 2. 3. 4. 5. 6. Lee & Mather(1972) Barry et ale (1982) Ng & Mather(1976) Peterson & Wilson (1974) Lee & Mather(1970) Hejmadi et ale (1971) 31,40 3.5,6.5 0.1-0.9 1-12 40 0.5 0.7-13.8 -46-149 0.2-0.7 0.145, 0.423 3-13.7 0-90 0.1-0.9 0.1-0.9 1-11 0.5-4.6 Composition(mol% CO) Pressure (MPa) 20,32,40 10-80 Temperature(DC) Nitrogen Nitrogen Methane Methane Methane Methane Impurity Table 1.1 Summaryof experimentaldata ofenthalpyof carbondioxide mixtures. • Excess enthalpies • All vapourenthalpies • 27 data points • Excess enthalpies • All vapourenthalpies • 108 data points • Enthalpydeparture • Liquid and vapourdensities • 46 data points • Enthalpydeparture • Liquid and vapourdensities • 42 data points • Excess enthalpies • All vapourenthalpies • 60 data points • Excess enthalpies • All vapourenthalpies • 646 data points Comments 6 5 4 3 2 1 Ref. Enthalpies of Carbon Dioxide-Methane 4 CARBONDIOXIDECAPTUREANDACID GAS INJECTION 1.3.1 CarbonDioxide-Methane The most extensive study performed for enthalpies of carbon dioxidemethane mixtures was performedby Lee & Mather (1972). Their study consisted of mol fractions of 0.1-0.9, taken at intervals of 0.1, for a total of 9 different mol fractions. Measurementsof excess enthalpywere reportedat 8 different temperatures from10-80 °C, with ranges of pressure of 1.0-4.4 MPafor 10°C, 1.0- 5.07 for 20 °C, 1.0-11.1 for 40°C, and 1.0-10.1 for 32 °C, 50°C, 60 °C, 70°C, and80 °C. In total,648 datapoints were reported.Two typographicalerrors were found in the data set; they are not included in the numerical error analysis but are represented in the figures. Anothersmaller study was performedby Barry et ale (1982), for excess enthalpies of carbon dioxide-methanemixtures. Datawas taken at three different temperatures,20°C, 32 °C, and 40°C. Seven different pressures were used, rangingfrom 0.51 MPato 4.6 MPa,with pressureofover 2 MPa onlybeing measuredfor 40°C. The mol fractionsmeasuredwere nottaken in increments, instead were taken at a wide variety of fractions ranging from 0.1 to 0.9. Two other studies were done using enthalpy departures by Ng & Mather(1976) and Peterson & Wilson (1974). Ng & Mather(1976) used pressures of 3-13.7 MPa, and temperaturesof 0-90 °C for mol fractions of 0.145 and 0.423. They used the ideal gas enthalpyas a reference point to measure the enthalpydeparture.Peterson & Wilson (1974) only measured equimolarmixtures of carbondioxide and methanewith pressures from 0.7-13.8 MPa and temperaturesof 255.4 K-422 K. The reference enthalpyused was measuredat a pressure of 0.138 MPa. These twostudies were the only ones that measured both liquid and vapor enthalpies, insteadofjust vapor. 1.3.2 CarbonDioxide-Nitrogen Lee & Mather(1970) and Hejmadi et ale (1971) studiedthe excess enthalpies of carbondioxide-nitrogenmixtures. Lee & Mather(1970) looked at mole fractionsfrom 0.1-0.9 at intervalsof0.1. Pressuresfrom 1.01 MPato 12.16 MPawere used, atonly a single temperatureof40°C. Hejmadi et ale (1971) used only two differenttemperaturesof 31°C and 40 °C, andtwo differentpressuresof3.5 MPaand6.5 MPa. Theyused mole fractionof nitrogenfrom 0.2-0.7. Enthalpies of Carbon Dioxide-Methane 1.4 5 Calculations The experimental enthalpies were compared to calculated enthalpies using BWR, LK, SRK, and PR78 thermodynamic models from VMGSim software, as well as using AQUAlibrium software. The six different mixtures (four with methane, two with nitrogen) as summarized in Table 1.1 were evaluated. Four error functions for both the excess enthalpies and the enthalpy departures were used to analyze the accuracy of the prediction of each method. For the excess enthalpies, the absolute average difference (AAD) was defined as; 1 NP AAD E E H exp H calc (1.2) where: NP – number of points HEexp – experimental excess enthalpy HEcalc – calculated excess enthalpy and the average difference (AD) was defined as: AD 1 NP E E H exp H calc (1.3) The absolute average error (AAE) in excess enthalpies was defined as: AAE 1 NP E E H exp H calc E H calc 100% (1.4) and the average error (AE) was defined as: AE 1 NP E E H exp H calc E H calc 100% (1.5) For enthalpy departures, the absolute average difference AAD 1 NP (H o H )exp (H o H )calc (1.6) 6 Carbon Dioxide Capture and Acid Gas Injection where H° – enthalpy of mixture at reference pressure H – enthalpy of mixture at measured pressure and the average difference was defined as: AD 1 NP (H o H )exp (H o H )calc The absolute average error for enthalpy departure was defined as: AAE 1 NP (H o H )exp (H o H )calc (H o H )calc 100% (1.7) (1.8) and the average error was defined as: AE 1.4.1 1 NP (H o H )exp (H o H )calc (H o H )calc 100% (1.9) Benedict-Webb-Rubin For the Lee & Mather (1972) methane data of excess enthalpies, the AAD was 78.1 J/mol and the AD was 2.6 J/mol. The AAE was 19.0% and the AE was -14.6%. The maximum difference was 2113.2 Jlmol occurring at 8.11 MPa and a mole fraction of 0.2. The maximum error was 131.7% at the same conditions as the maximum difference. At lower pressures, the enthalpies were overestimated, and at the higher pressures they were underestimated. The greatest deviations occurred when there was a rapid change in enthalpy with pressure. This occurred at around 7–10 MPa for the 32°C and 40 °C temperatures.There was also avery large difference between the calculated and experimental enthalpy for the 10.13 MPa isobar at 50 °C. Figures 1.1 through 1.8 show the experimental and calculated enthalpies for the different temperatures. The Barry et al. (1982) methane data of excess enthalpies had an AAD of9.1 Jlmol, an AD of-8.3 Jlmol, an AAEof 14.2% and an AEof -11.0%. The maximumdifference was 46.5 Jlmol at 4.6 MPa,40°C and 0.351 mole fractionmethane.The maximumerrorwas 42.5% at 0.53 MPa,32°C and 0.63 mole fraction methane The deviations are smaller due to the lower pressure range of the data. The Lee & Mather (1970) nitrogen data of excess enthalpies taken at 40°C had similar results as the Lee& Mather (1972) methane data for the 40 °C data,with the greatest difference occurring at 9.12 MPa. The Enthalpies of Carbon Dioxide-Methane 600 500 Excess enthalpy (J/mol) 4.36 400 1.01 MPa 2.03 MPa 3.04 MPa 4.05 MPa 4.36 MPa 4.05 300 3.04 200 2.03 100 1.01 0 0 0.2 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.1 Experimental and calculated enthalpies at 10 °C using BWR (Lee & Mather, 1972). 700 600 Excess enthalpy (J/mol) 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 5.07 500 400 300 4.05 200 3.04 100 2.03 1.01 0 0 0.2 0.4 0.6 Mol fraction methane (J/mol) 0.8 1 Figure 1.2 Experimental and calculated enthalpies at 20 °C using BWR (Lee & Mather, 1972). 7 8 Carbon Dioxide Capture and Acid Gas Injection 4,500 8.61 4,000 10.1 3,500 Excess enthalpy (J/mol) 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 8.61 MPa 9.12 MPa 10.13 MPa 9.12 3,000 2,500 2,000 8.11 1,500 1,000 7.09 500 6.08 5.07 0 0 4.05 3.0 2.0 1.01 0.4 0.6 Mol fraction methane 0.2 1 0.8 Figure 1.3 Experimental and calculated enthalpies at 32 °C using BWR (Lee & Mather, 1972). 3,500 2,500 Excess enthalpy (J/mol) 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 11.15 MPa 11.15 3,000 10.1 2,000 9.12 1,500 8.11 1,000 7.09 500 0 0 0.2 6.08 5.07 4.05 3.04 2.02 1.01 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.4 Experimentaland calculated enthalpies at40°C using BWR (Lee& Mather, 1972). Enthalpies of Carbon Dioxide-Methane 2,000 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 1,800 Excess enthalpy (J/mol) 1,600 1,400 10.13 1,200 9.12 1,000 800 8.11 600 7.09 400 6.08 200 5.07 4.05 2.02 0 0 0.2 3.04 1.01 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.5 Experimental and calculated enthalpies at 50 °C using BWR (Lee & Mather, 1972). 1,200 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 1,000 Excess enthalpy (J/mol) 10.13 800 9.12 600 8.11 7.09 400 6.08 5.07 200 4.05 3.04 0 0 0.2 2.02 1.01 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.6 Experimental and calculated enthalpies at 60 °C using BWR (Lee & Mather, 1972). 9 10 Carbon Dioxide Capture and Acid Gas Injection 800 600 Excess enthalpy (J/mol) 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 10.13 700 9.12 500 8.11 400 7.09 300 6.08 5.07 200 4.05 100 0 3.04 0 0.2 2.02 1.01 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.7 Experimental and calculated enthalpies at 70 °C using BWR (Lee & Mather, 1972). 600 10.13 Excess enthalpy (J/mol) 500 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 9.12 400 8.11 7.09 300 6.08 200 5.07 4.05 100 3.04 2.02 1.01 0 0 0.2 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.8 Experimental and calculated enthalpies at 80 °C using BWR (Lee & Mather, 1972). Enthalpies of Carbon Dioxide-Methane 11 AAD was 151.1 J/mol, the AD was 58.7 J/mol, the AAE was 15.0% and the AE was –0.7%. The maximum difference was 969.8 J/mol at 9.1 MPa, and 0.1 mole fraction nitrogen. The maximum error was 70% at the same conditions as the maximum difference. Figure 1.9 shows the calculated and experimentalenthalpiesfor the BWRmodel at 40°C. The Hejmadi et al. (1971) nitrogen of excess enthalpies data had an AAD of 26.1 J/mol, and AD of –11.0 J/mol, an AAE of 9.5% and an AE of –7.9%. The maximum difference was 90.8 J/mol at 6.5 MPa, 31 °C, and 0.239 mole fraction nitrogen. The maximum error was 14.1% at 3.4 MPa,40°C and 0.67 mole fraction nitrogen. As with the Barry et al. (1982) methane data, the lower deviations are likely due to the lower pressure range used in the measurements, as the highest pressure used was 6.5 MPa and the greatest deviations typically occurred around 7–10 MPa for temperatures in the 30-40 -c range. For the Peterson & Wilson (1974) methane data for enthalpy departures, the AAD was 56.4Jlmol, the AD was 26.2Jlmol, the AAE was 3.7% andthe AE was 1.4%. Twopointswere omittedfrom the errorcalculations 4,000 1.01 MPa 2.03 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 11.15 MPa 12.16 MPa 3,500 11.1 Excess enthalpy (J/mol) 3,000 12.1 10.1 2,500 2,000 9.1 1,500 8.11 1,000 7.09 6.08 5.07 4.05 500 0 0 0.2 3.04 1.01 2.02 0.4 0.6 Mol fraction nitrogen (–) 0.8 1 Figure 1.9 Experimentaland calculatedenthalpiesat 40°C using BWR (Lee & Mather, 1970). 12 Carbon Dioxide Capture and Acid Gas Injection due to BWR predicting a vapor/liquid mix. The Ng & Mather (1976) methane data for enthalpy departures had an AAD of 192.3 J/mol, an AD of 182.2 J/mol, an AAE of 3.8% and an AE of 3.0% 1.4.2 Lee-Kesler The Lee & Mather (1972) methane data for excess enthalpies had an AAD of 46.7 Jlmol, an AD of -43.2 [Zmol, an AAE of 20.1%, and an AE of –19.7%. Figures 1.10 through 1.17 show the experimental and calculated enthalpies for the 8 different temperatures. The greatest differences typically occurred at the highest pressure and at low methane mole fractions for all temperatures. The maximum difference was 505.5 J/mol occurring at 50 °C, 10.1 MPa and 0.1 mol fraction methane. The greatest errors always occurred at a mole fraction of 0.1 and a pressure of 1.01 MPa. The maximum error was 98.0% occurring at 80 °C. For almost all data points, LK overestimated the enthalpies. The only conditions where they were underestimated was at high methane mole fraction and high pressures. For the Barry et al. (1970) methane data of excess enthalpies the AAD was 12.0 Jlmol, the AD was -11.4 Jlmol, the AAE was 19.0%and the AE 600 4.36 500 Excess enthalpy (J/mol) 4.05 1.01 MPa 2.03 MPa 3.04 MPa 4.05 MPa 4.36 MPa 400 300 3.04 200 2.03 100 1.01 0 0 0.2 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.10 Experimental and calculated enthalpies at 10 °C using LK (Lee & Mather, 1972). Enthalpies of Carbon Dioxide-Methane 700 5.07 600 Excess enthalpy (J/mol) 500 1.01 MPa 2.03 MPa 3.04 MPa 4.05 MPa 5.07 MPa 400 4.05 300 3.04 200 2.03 100 1.01 0 0 0.2 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.11 Experimental and calculated enthalpies at 20 °C using LK (Lee & Mather, 1972). 4,500 4,000 8.11 8.61 Excess enthalpy (J/mol) 3,500 9.12 10.13 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 8.61 MPa 9.12 MPa 10.13 MPa 3,000 2,500 2,000 1,500 7.09 1,000 6.08 5.07 500 0 0 0.2 4.05 1.01 2.02 3.04 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.12 Experimental and calculated enthalpies at 32 °C using LK (Lee & Mather, 1972). 13 Carbon Dioxide Capture and Acid Gas Injection 3,500 3,000 9.12 2,500 Excess enthalpy (J/mol) 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 11.15 MPa 10.13 11.15 2,000 1,500 8.11 1,000 7.09 500 0 0 0.2 6.08 5.07 4.05 3.04 2.02 1.01 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.13 Experimentaland calculatedenthalpiesat 40°C using LK (Lee& Mather, 1972). 2,000 1,800 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 10.13 1,600 Excess enthalpy (J/mol) 1,400 9.12 1,200 1,000 800 8.11 600 7.09 6.08 400 200 0 0 0.2 5.07 4.05 3.04 2.02 1.01 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.14 Experimental and calculated enthalpies at 50 °C using LK (Lee & Mather, 1972). Enthalpies of Carbon Dioxide-Methane 1,200 10.13 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 1,000 9.12 Excess enthalpy (J/mol) 800 8.11 600 7.09 400 6.08 5.07 200 4.05 3.04 0 1.01 0 0.2 2.02 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.15 Experimental and calculated enthalpies at 60 °C using LK (Lee & Mather, 1972). 800 10.13 700 Excess enthalpy (J/mol) 600 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 9.12 500 8.11 400 7.09 300 6.08 5.07 200 4.05 3.04 100 2.02 0 1.01 0 0.2 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.16 Experimental and calculated enthalpies at 70 °C using LK (Lee & Mather, 1972). 15 16 Carbon Dioxide Capture and Acid Gas Injection 600 10.1 Excess enthalpy (J/mol) 500 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 9.1 400 8.1 7.0 300 6.0 200 5.0 4.0 100 3.0 2.0 0 1.0 0 0.2 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.17 Experimental and calculated enthalpies at 80 °C using LK (Lee & Mather, 1972). was -16.6%. Themaximumdifference of 67.0 J/mol occurredat 4.6 MPa, 40°C, and a methane mole fraction of 0.649. The maximum error was 48.3% at 40 "C, 0.52 MPa,and 0.252 mole fractionmethane. The Lee & Mather (1970) nitrogen data of excess enthalpies had an AAD of231 l/mol, an AD of-226.4, an AAEof27.7% andan AE of -27.2%. The maximum difference of 718.9 J/mol occurred at 12.16 MPa, and a methane mole fraction of 0.2. The maximum errorwas 55.4% at 1.01 MPa,40°C, 0.1 mole fraction nitrogen. Figure 1.18 shows the calculated and experimentalenthalpiesfor the LKmodel at 40°C. The excess enthalpy data for nitrogen from Hejmadi et al. (1971) had an AD of –153.7 and an AE of –50.8%. All data points were overestimated by LK, resulting in an AAD and AAE of the same magnitude as the AD and AE. The maximum difference was 342.4 J/mol at 6.5 MPa, 31°C, 0.31 mole fraction nitrogen. The maximum error was 72.3% at 3.4 MPa, 31 °C, 0.228 mole fraction nitrogen. Enthalpies of Carbon Dioxide-Methane 17 4,000 12.1 11.1 3,500 1.01 MPa 2.03 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 11.15 MPa 12.16 MPa 9.1 10.1 Excess enthalpy (J/mol) 3,000 2,500 8.11 2,000 1,500 7.09 1,000 6.08 5.07 4.05 500 0 0 0.2 3.04 2.02 1.01 0.4 0.6 Mol fraction nitrogen (–) 0.8 1 Figure 1.18 Experimentaland calculatedenthalpiesat 40°C using LK (Lee& Mather, 1970). The Ng & Mather (1976) enthalpy departure data for methane had an AAD of 152.8Jlmol, an AD of-151.37 Jlmol, an AAE of 4.0% and an AE of -4.1 %. The only point where LKunderestimatedthe enthalpywas at 3 MPa,10°C and amethanemole fractionof 0.145. ThePeterson& Wilson (1974) enthalpydeparturedata had an AAD of 149.3l/mol, and AD of -145.9 Jlmol, and AAE of5.8% and an AE of -3.6%. The greatest errorand -45°C °C and –20 °C. difference occurredat temperaturesof 1.4.3 Soave-Redlich-Kwong The Lee & Mather (1972) excess enthalpy methane data had an AAD Figures 1.19 through 1.26 show the experimental and calculated excess enthalpies as predicted by SRK for the different temperatures. The SRK underestimated the excess enthalpies for the majority of the data points. The greatest differences generally occurred at low methane mole fractions was 98% occurring at a 1.01 MPa, 0.1 mole fraction methane and 50 °C. 18 Carbon Dioxide Capture and Acid Gas Injection 600 Excess enthalpy (J/mol) 500 400 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 4.36 MPa 4.36 300 4.05 200 3.04 100 2.03 1.01 0 0 0.2 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.19 Experimental and calculated enthalpies at 10 °C using SRK (Lee & Mather, 1972). 700 600 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa Excess enthalpy (J/mol) 500 5.07 400 300 4.05 200 3.04 100 2.03 0 1.01 0 0.2 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.20 Experimental and calculated enthalpies at 20 °C using SRK (Lee & Mather, 1972). Enthalpies of Carbon Dioxide-Methane 19 4,500 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 8.61 MPa 9.12 MPa 10.13 MPa 4,000 Excess enthalpy (J/mol) 3,500 8.11 9.12 8.61 3,000 10.13 2,500 2,000 1,500 7.09 1,000 6.08 500 5.07 0 0 4.05 3.04 2.02 1.01 0.4 0.6 Mol fraction methane 0.2 0.8 1 Figure 1.21 Experimental and calculated enthalpies at 32 °C using SRK (Lee & Mather, 1972). 3,500 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 11.15 MPa 3,000 Excess enthalpy (J/mol) 2,500 10.13 11.15 9.12 2,000 1,500 8.11 1,000 7.09 500 4.05 0 0 0.2 6.08 5.07 3.04 2.02 1.01 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.22 Experimentalandcalculatedenthalpiesat 40°C using SRK (Lee& Mather, 1972). 20 Carbon Dioxide Capture and Acid Gas Injection 2,000 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 1,800 Excess enthalpy (J/mol) 1,600 10.13 1,400 1,200 9.12 1,000 800 8.11 600 7.09 400 6.08 5.07 4.05 200 0 0 0.2 3.04 2.02 1.01 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.23 Experimental and calculated enthalpies at 50 °C using SRK (Lee & Mather, 1972). 1,200 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 1,000 Excess enthalpy (J/mol) 10.13 800 9.12 600 8.11 400 7.09 6.08 200 0 5.07 0 0.2 4.05 3.04 2.02 1.01 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.24 Experimental and calculated enthalpies at 60 °C using SRK (Lee & Mather, 1972). Enthalpies of Carbon Dioxide-Methane 800 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 700 10.13 Excess enthalpy (J/mol) 600 500 9.12 400 8.11 300 7.09 6.08 200 5.07 4.05 3.04 100 0 0 0.2 2.02 1.01 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.25 Experimental and calculated enthalpies at 70 °C using SRK (Lee & Mather, 1972). 600 Excess enthalpy (J/mol) 500 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 10.13 9.12 400 8.11 300 7.09 200 6.08 5.07 4.05 100 0 3.04 2.02 1.01 0 0.2 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.26 Experimental and calculated enthalpies at 80 °C using SRK (Lee & Mather, 1972). 21 22 Carbon Dioxide Capture and Acid Gas Injection The Barry et al. (1982) methane data of excess enthalpy had an AAD of 9.4 Jlmol an AD of 9.4 Jlmol an AAE of 22.9% and an AE of 22.8%. The maximum difference of 22.3 Jlmol occurredat 4.6 MPa (the highest pressure used), 40°C, and a methane mole fraction of 0.646. The maxi-mum errorof 54.8% occurredat 1.15 MPa,20°C and 0.883 mole fraction methane. The Lee & Mather (1970) nitrogen excess enthalpy data had an AAD of 69.7 Jlmol, an AD of 61.3 Jlmol, an AAE of 12.4%, and an AE of 11.5%. Figure 1.27 shows the experimental and calculated enthalpies for 40°C using SRK. The maximum difference was 348.7 Jlmol occurring at the same conditions as the Lee & Mather (1972) methane data. The maximum error was 36.0% occurring at 9.12 MPa and 0.9 mol fraction nitrogen. For the Hejmadi et al. (1971) excess enthalpy nitrogen data, the AD was 45.2 Jlmol and the AE was 14.0%. Allthe enthalpieswere underestimated 4,000 1.01 MPa 2.03 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 11.15 MPa 12.16 MPa 3,500 12.16 3,000 Excess enthalpy (J/mol) 11.15 9.12 10.13 2,500 2,000 1,500 8.11 1,000 7.09 500 0 0 0.2 6.08 5.07 4.05 1.01 3.04 2.02 0.4 0.6 Mol fraction nitrogen (–) 0.8 1 Figure 1.27 Experimentaland calculatedenthalpiesat 40°C using SRK (Lee& Mather, 1970). Enthalpies of Carbon Dioxide-Methane 23 by SRK, therefore the ADD and AAE were the same values as the AD and AE. The maximum difference was 131.1 J/mol and occurred at 6.5 MPa, 31 °C, and 0.725 mole fraction nitrogen. The maximum error was 19.0% occurringat 3.4 MPa,31°C and 0.729 mol fractionnitrogen. For the enthalpy departure data for methane, the Ng & Mather (1976) data had an AAD of 56.0 J/mol, an AD of –12.8 J/mol, an AAE of 2.5% and an AEof 1.2%. The Peterson & Wilson (1974) datahadan AADof98.1, an AD of -97.3 Jlmol, an AAEof3.7% and an AEof -4.8%. 1.4.4 Peng-Robinson The Lee & Mather (1972) excess enthalpy methane data using PengRobinsonhad an AAD of 40.6 Jlmol, an AD of 36.3 Jlmol, a AAE of 9.9% and an AE of 8.4%. Figures 1.28 through 1.35 show the calculated and experimental enthalpies for Peng-Robinson. The maximum difference was 504.7 Jlmol occurringat a methane mole fraction of 0.1, 10.13 MPa and 600 Excess enthalpy (J/mol) 500 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 4.36 MPa 400 4.36 4.05 300 200 3.04 100 2.03 1.01 0 0 0.2 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.28 Experimental and calculated enthalpies at 10 °C using PR (Lee & Mather, 1972). Carbon Dioxide Capture and Acid Gas Injection 700 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 600 Excess enthalpy (J/mol) 500 5.07 400 300 4.05 200 3.04 100 2.03 0 1.01 0 0.2 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.29 Experimental and calculated enthalpies at 20 °C using PR (Lee & Mather, 1972). 4,500 4,000 Excess enthalpy (J/mol) 3,500 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 8.61 MPa 9.12 MPa 10.13 MPa 8.11 9.12 8.61 10.13 3,000 2,500 2,000 1,500 1,000 7.09 500 6.08 5.07 4.05 0.2 0 0 3.04 2.02 1.01 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.30 Experimental and calculated enthalpies at 30 °C using PR (Lee & Mather, 1972). Enthalpies of Carbon Dioxide-Methane 3,500 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 11.15 MPa 3,000 Excess enthalpy (J/mol) 2,500 11.15 10.13 9.12 2,000 1,500 8.11 1,000 7.09 500 0 6.08 0 5.07 4.05 3.04 2.02 1.01 0.2 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.31 Experimentalandcalculatedenthalpiesat 40°C using PR (Lee& Mather, 1972). 2,000 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 1,800 1,600 Excess enthalpy (J/mol) 1,400 10.13 1,200 1,000 9.12 800 8.11 600 400 7.09 200 6.08 5.07 4.05 3.04 00 0.2 2.02 1.01 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.32 Experimental and calculated enthalpies at 50 °C using PR (Lee & Mather, 1972). 25 26 Carbon Dioxide Capture and Acid Gas Injection 1,200 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 1,000 Excess enthalpy (J/mol) 10.13 800 9.12 600 8.11 400 7.09 6.08 200 0 5.07 0 0.2 4.05 3.04 2.02 1.01 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.33 Experimental and calculated enthalpies at 60 °C using PR (Lee & Mather, 1972). 800 700 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 10.13 Excess enthalpy (J/mol) 600 9.12 500 400 8.11 300 7.09 6.08 200 5.07 4.05 100 0 3.04 1.01 0 0.2 2.02 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.34 Experimental and calculated enthalpies at 70 °C using PR (Lee & Mather, 1972). Enthalpies of Carbon Dioxide-Methane 27 Excess enthalpy (J/mol) 600 500 10.13 400 9.12 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 8.11 300 7.09 200 6.08 5.07 100 0 4.05 3.04 2.02 1.01 0 0.2 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.35 Experimental and calculated enthalpies at 80 °C using PR (Lee & Mather, 1972). 40 "C. The maximumerror of 39.6% occurredat 0.1 mole fraction methane, 10.13 MPa, and 50 °C. The average errors in enthalpies decreased as the temperatures increased, as well as with increasing pressure. The Barry et al. (1982) methane excess enthalpy data had an AAD of 9.6, and AD of 9.5, an AAE of 23.2 and an AE of 23.2. The maximum difference was 22.3 J/mol at 4.6 MPa and 0.686 mole fraction methane.The maximum error was 54.6% at 0.521 MPa, 20 "C, and 0.816 mole fraction methane. For the Lee & Mather (1970) excess enthalpy data for nitrogen, the AAD was 79.4 J/mol, the AD was 73.3 J/mol, the AAE was 13.3%andthe AE was 12.6%. The maximum difference, of 378.6 J/mol, and the maximum error of 38.9% occurred at pressures of 10.13 MPa and 9.12 MPa respectively. Figure 1.36 shows the experimentalandcalculatedenthalpiesfor the 40 "C nitrogen mixture. The Hejmadi et al. (1971) excess enthalpy data for nitrogen had an AD of45.6 J/mol andan AEof 14.3%. All the datapointswere underestimated by PR78, therefore the ADD and AE were the same as the AD and AE. The maximum difference of 127.9 J/mol occurred at 6.5 MPa, 31 °C, and a 28 Carbon Dioxide Capture and Acid Gas Injection 4,000 1.01 MPa 2.03 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 11.15 MPa 12.16 MPa 3,500 12.16 Excess enthalpy (J/mol) 3,000 2,500 11.1 10.13 9.12 2,000 1,500 8.11 1,000 7.09 500 0 0 0.2 6.08 5.07 4.05 1.01 3.04 2.02 0.4 0.6 Mol fraction nitrogen (–) 0.8 1 Figure 1.36 Experimentaland calculatedenthalpiesat 40°C using PR (Lee& Mather, 1970). nitrogenmole fractionof0.239. The maximumerrorwas 19.5% at 3.4 MPa, 31 °C, and a mole fraction of 0.729 nitrogen. For the enthalpy departure data, the Ng & Mather (1976) had an AAD of 110.3 J/mol, an AD of -82.0 J/mol, a AAE of 5.2% and an AEof -4.4%. The Peterson&Wilson (1974) datahad an AAD of 160.2 J/mo!, an AD of 160.2 J/mol, and AAE of 9.7% and an AE of –9.7%. 1.4.5 AQUAlibrium The Lee & Mather (1972) excess enthalpy methane data had an AAD of 39.2 J/mol, an AD of 34.7 J/mol, an AAE of 9.8% and an AE of 8.3%. Figures 1.37 through 1.44 show the experimentaland calculatedenthal-- pies for the different temperatures using AQUAlibrium. The maximum difference was 512.2 J/mol at 11.15 MPa, 40°C, and 0.1 mole fraction methane.The maximumerrorwas 41.8 at 10.13 MPa,50°C, and0.5 mole fraction methane. The difference in enthalpies decreased as the temperatures increased. For the Barry et al. (1982) methane data, the AAD was 9.6 J/mol, the AD was 9.1 J/mol, the AAE was 23.4% and the AE was 22.1%. Themaximum Enthalpies of Carbon Dioxide-Methane 29 600 Excess enthalpy (J/mol) 500 400 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 4.36 MPa 4.36 4.05 300 200 3.04 100 2.03 1.01 0 0 0.2 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.37 Experimental and calculated enthalpies at 10 °C using AQUAlibrium (Lee & Mather, 1972). 700 600 Excess enthalpy (J/mol) 500 5.07 400 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 300 4.05 200 3.04 100 0 2.03 0 0.2 1.01 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.38 Experimental and calculated enthalpies at 20 °C using AQUAlibrium (Lee & Mather, 1972). 30 Carbon Dioxide Capture and Acid Gas Injection 4,500 4,000 Excess enthalpy (J/mol) 3,500 3,000 8.11 8.61 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 8.61 MPa 9.12 MPa 10.13 MPa 9.12 10.13 2,500 2,000 1,500 7.09 1,000 500 00 6.08 5.07 4.05 3.04 2.02 1.01 0.2 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.39 Experimental and calculated enthalpies at 30 °C using AQUAlibrium (Lee & Mather, 1972). 3,500 3,000 Excess enthalpy (J/mol) 2,500 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 11.15 MPa 10.13 11.15 9.12 2,000 1,500 8.11 1,000 7.09 500 0 0 6.08 0.2 5.07 4.05 1.01 3.04 2.02 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.40 Experimentaland calculatedenthalpiesat 40°C using AQUAlibrium(Lee & Mather, 1972). Enthalpies of Carbon Dioxide-Methane 31 2,000 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 1,800 1,600 10.13 Excess enthalpy (J/mol) 1,400 1,200 9.12 1,000 800 8.11 600 7.09 400 6.08 200 0 5.07 0 0.2 4.05 3.04 2.02 1.01 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.41 Experimental and calculated enthalpies at 50 °C using AQUAlibrium (Lee & Mather, 1972). 1,200 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa Excess enthalpy (J/mol) 1,000 10.13 800 9.12 600 8.11 400 7.09 6.08 200 0 5.07 0 0.2 3.04 1.01 4.05 2.02 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.42 Experimental and calculated enthalpies at 60 °C using AQUAlibrium (Lee & Mather, 1972). 32 Carbon Dioxide Capture and Acid Gas Injection 800 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 700 10.13 Excess enthalpy (J/mol) 600 500 9.12 400 8.11 300 7.09 6.08 200 5.07 4.05 3.04 100 0 0 0.2 1.01 2.02 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.43 Experimental and calculated enthalpies at 70 °C using AQUAlibrium (Lee & Mather, 1972). Excess enthalpy (J/mol) 600 500 10.13 400 9.12 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 8.11 300 7.09 200 6.08 5.07 4.05 100 0 0 0.2 3.04 2.02 1.01 0.4 0.6 Mol fraction methane 0.8 1 Figure 1.44 Experimental and calculated enthalpies at 80 °C using AQUAlibrium (Lee & Mather, 1972). Enthalpies of Carbon Dioxide-Methane 33 difference was 22.3Jlmol at 4.6 MPa,40°C, and0.686 mole fractionmeth-ane. The maximum error was 54.6% at 0.52 MPa,20°C, and 0.477 mole fraction methane. The Lee & Mather (1970) nitrogen data had an AAD of 62.2, and AD of 62.2, an AAE of 13.0%,a nd an AE of 11.9%. Figure 1.45 shows theexperi-mentalandcalculatedenthalpiesfor the 40°C mixture.The maximumdif-ference was 368.6 J/mol at 10.13 MPa and 0.1 mole fraction nitrogen. The maximum error was 38.0% at 1.01 MPa and 0.9 mole fraction nitrogen. The Hejmadi et al. (1971) data for nitrogen had an AAD and AD of 44.1 Jlmol and an AAEand AE of 14.1%. The maximum difference was 119.2 J/mol at 6.5 MPa, 31 °C, and 0.239 mole fraction nitrogen. The maximum errorwas 19.5% at 3.4 MPa,31°C, and0.729 mole fractionnitrogen. The Ng & Mather (1976) enthalpy departure data for methane had an AAD of 116.0 Jlmol, an ADof -94.2 J/mo!, an AAEof 5.4% and an AE of 4.8%. The Peterson & Wilson (1974) enthalpydeparturedatahad an AAD of 163.7 J/mol, an AD of –163.7, an AAE of 10.0% and an AE of 10.0%. 1.5 Discussion Table 1.1 and Table 1.2 show the AAD and AAE for all excess enthalpy data for all thermodynamic models, as well as a weighted average, based on the number of data points used, of all mixtures for each model. For the excess enthalpy data the SRK model provided the best overall AAE. The AQUAlibrium model provided the best overall AAD with SRK obtaining similar results. Overall, SRK, PR78 and AQUAlibrium all achieved similar results, and predicted better than both LK and BWR, with LK being the less accurate of the two. Table 1.2 Absolute average difference in excess enthalpies for methane and nitrogen mixtures using the different models. BWR LK SRK PR78 AQUA 78.1 46.7 38.6 40.6 39.2 9.1 12.0 9.4 9.5 9.6 Lee & Mather (1970)- Nitrogen 151.1 231.0 69.7 79.4 62.2 Hejmadi et al. (1971)- Nitrogen 26.1 153.7 45.2 45.6 44.1 Weighted Average 81.0 71.3 40.8 43.5 40.2 Lee & Mather (1972)- Methane Barry et al. (1982)- Methane 34 Carbon Dioxide Capture and Acid Gas Injection For the AAE, both the Lee & Mather (1970), (1972) data sets were best predicted by SRK, while the Hejmadi et al. (1971) and Barry et al. (1982) data were best predicted by BWR. The Hejmadi et al. (1971) and Barry et al. (1982) data were taken at much smaller ranges of pressure, with maximums of 6.5 and 4.6 MParespectively, comparedto a maximumpressure of 12.16 MPa and 11.15 of the Lee & Mather (1970), (1972) data, respec- tively. In Figures1.3 and 1.4, showingthe predictionsfor BWR at32°C and 40 "C respectively, it can beseen thatwhen the enthalpyis changingrapidly with increasing pressure, the BWR model provides very poor predictions. This caused the greater error in the Lee and Mather (1970), (1972) data compared to the Hejmadi et al. (1971) and Barry et al. (1982) data, where pressures where rapid enthalpy change was happening were not measured. Over wide ranges of pressures, the SRK model provided the more accurate predictions. For the AAD, the same trend was found as for the AAE, except for the optimal model for the Lee & Mather (1970) data being the AQUAlibrium model. Compared to the Lee & Mather (1972) data, the 1970 data was only measured at single temperatureof 40°C. When comparing Figure 1.22 showing SRK at 40°C for Lee & Mather (1972) to Figure 1.45 showing 4,000 1.01 MPa 2.02 MPa 3.04 MPa 4.05 MPa 5.07 MPa 6.08 MPa 7.09 MPa 8.11 MPa 9.12 MPa 10.13 MPa 11.15 MPa 12.16 MPa 3,500 Excess enthalpy (J/mol) 3,000 2,500 11.15 10.13 12.16 9.12 2,000 1,500 8.11 1,000 7.09 500 0 0 0.2 6.08 5.07 4.05 3.04 1.01 2.02 0.4 0.6 Mol fraction nitrogen (–) 0.8 1 Figure 1.45 Experimentaland calculatedenthalpiesat 40°C using AQUAlibrium(Lee & Mather, 1970). Enthalpies of Carbon Dioxide-Methane 35 AQUA for Lee & Mather (1970), the AQUAlibrium model predicts the excess enthalpy much more accurately, specifically at pressures of 9.12 MPa. However, when comparingFigure 1.45 to Figure 1.40,b oth of which are AQUAlibriumat 40°C, with Figure 1.45being for nitrogenandFigure 1.40 being for methane, the nitrogen mixture is much more accurate, whereas Figure 1.45 closely resembles Figure 1.22. Therefore, it is likelythat the optimal model for the Lee & Mather (1970) was AQUAlibrium due to the impurity being nitrogen rather than methane. The Lee & Mather (1972) data allows for a comparison of the accuracy of each model at temperatures from 10 °C to 80 °C. For all models except for LK,32°C and 40 °C gave the largest differences. For LK, thelargest differences were at 32 °C and 50 °C. For SRK, PR78 and AQUAlibrium, 60 °C, 70 °C, and 80 °C gave smaller differences than the low temperatures of 10 °C, and 20 °C. For BWR, the opposite trend occurred with the low temperatureshaving smaller differences. For LK,1 0°C, 20 °C, 40°C, and 60–80 °C all had similar differences. Tables 1.3and 1.4 show the AADand AAE for theenthalpydeparture data for all 5 thermodynamic models as well as a weighted average of the Table 1.3 Absolute average error in excess enthalpies for methane and nitrogen mixtures using the different models. BWR LK SRK PR78 AQUA Lee & Mather (1972)- Methane 19.0 20.1 9.7 9.9 9.8 Barry et al. (1982)- Methane 14.3 19.1 22.9 23.2 23.4 Lee & Mather (1970)- Nitrogen 15.0 27.7 12.4 13.3 13.0 Hejmadi et al. (1971)- Nitrogen 9.5 50.8 14.0 14.3 14.1 17.9 22.0 11.1 11.4 11.3 Weighted Average Table 1.4 Absolute average difference in enthalpy departure for methane ­mixtures using the different models. BWR LK SRK PR78 AQUA Ng & Mather (1976) 3.8 4.1 2.5 5.2 5.4 Peterson & Wilson (1974) 3.7 5.8 4.5 9.7 10.0 Weighted Average 3.8 5.0 3.5 7.5 7.8 36 Carbon Dioxide Capture and Acid Gas Injection Table 1.5 Absolute average error in enthalpy departure for methane mixtures using the different models. Ng & Mather (1976) Peterson & Wilson (1974) Weighted Average BWR LK SRK PR78 AQUA 192.3 152.8 56.0 110.3 116.0 56.4 149.3 98.1 160.2 163.7 121.6 151.0 77.9 136.2 140.8 two mixtures for all models. As with the excess enthalpy data, the SRK model performed well, with SRK giving the smallest overall AAE and AAD. However, unlike the excess enthalpies, the PR78 and AQUAlibrium models were much worse than SRK. For the AAD, PR78, AQUAlibrium, BWR and LK all gave similar values, of which were considerably greater than SRK. For the AAE, SRK had the smallest value, closely followed by BWR, with PR78 and AQUAlibrium having the highest AAE. Using both the AAE and AAD as criteria, Ng & Mather (1976) was best predicted by SRK,while Peterson & Wilson (1974) was best pre-dicted by BWR. The Ng & Mather (1976) data may have been better predicted by SRK compared to BWR because BWR was less accurate at predicting high pressure enthalpies, and while both papers had similar pressure ranges, the Ng & Mather (1976) paper had more data at higher pressures. Additionally, the method used to calculate the enthalpy departures varied between the two methods, with Peterson & Wilson (1974) using a reference enthalpyof 0.14 MPa, and Ng & Mather(1976) uses the ideal gas enthalpy as the reference enthalpy. The difference in calculation methods may affect the AAE and AAD comparison between data sets. 1.6 Conclusion The excess enthalpy data and the enthalpy departure data were overall predicted most accurately by the SRK model, with an exception of AQUAlibrium giving the smallest AAD for excess enthalpies. For the excess enthalpy data, the AQUAlibrium, SRK, and PR78 models all produced similar results, with BWR and LK giving much higher errors, with LK giving the higher of the two. For the enthalpy departure data, when using AAE as the criteria, LK, BWR and SRK all gave similar results, with PR78 and AQUA giving much greater errors. When looking at AAD as the Enthalpies of Carbon Dioxide-Methane 37 criteria, SRK was by far the best, with the other four giving comparative results. For the excess enthalpy data, the optimal methods of SRK for AAE and AQUAlibrium for AAD, gave average errors of 11.1% and average differences of40.2 J/mo!. For use in acid gasinjection,a difference of40.2 Jlmol equates to 0.75 HP/ MMSCFD. For enthalpy departures, SRK gave an overall average error of 3.5% and an average difference of 77.9 J/mol, or 1.5 HP/ MMSCFD. A difference of 0.75- 1.5 HP/MMSCFD is an acceptable margin of error when considering the design of a compressor; however, the maximum difference for the SRKmodel was 504.4 Jlmol, or 9.4 HP/MMSCFD under certain conditions, which may pose a problem in the compression of the stream. References 1. Barry, A., Kallaguine, S., and R. Ramalho, “Direct Determination of Enthalpy of Mixing for the Binary Gaseous System Methane-Carbon Dioxide by an Isothermal Flow Calorimeter,” J. Chem. Eng. Data 27, 258-264, 1982. 2. Hejmadi, A.V., Katz, D.L., and J.E. Powers, “Experimental Determination of the Enthalpy of Mixing of N2 + CO2 Under Pressure,” J. Chem. Thermo., 3,483-496, 1971. 3. Lee, J.I., and Mather, A.E., “The Excess Enthalpy of Gaseous Mixtures of Nitrogen and Carbon Dioxide,” J. Chem. Thermo., 2, 881–895, 1970. Lee, J.I., and Mather, A.E., “The Excess Enthalpy of Gaseous Mixtures of Carbon Dioxide with Methane,” Can. J. Chem. Eng., 50, 95–100, 1972. 5. Ng, H.J., and Mather, A.E., “Isothermal Joule-Thomson Coefficients in Mixtures of Methane and Carbon Dioxide” J. Chem. Eng. Data, 21,291-294,1976. 6. Peterson, J.M., and Wilson, G.M., “Enthalpy and Phase Boundary Measurements on Carbon Dioxide and Mixtures of Carbon Dioxide with Methane, Ethane and HydrogenSulfide:' BrighamYoung University, Provo, Utah, 1974. 2 Enthalpies of Hydrogen SulfideMethane Mixture: Comparison with Thermodynamic Models Erin L. Roberts and John J. Carroll Gas Liquids Engineering, Calgary, AB, Canada Abstract In the design of an acid gas injection scheme, the physical properties of the stream are required to ensure successful injection into the subsurface reservoir. Common impurities of a carbon dioxide acid gas injection stream are hydrogen sulfide and methane. The excess enthalpies of these hydrogen sulfide-methane mixtures are important in determining the compressor specifications in the acid gas injection design to ensure proper injection to the subsurface reservoir. This study compares the experimental data of excess enthalpies of a hydrogen sulfide-methane mixture to the calculated excess enthalpies of six different thermodynamic models, Lee Kesler, Benedict-Webb-Rubin, Soave-Redlich-Kwong, Redlich-Kwong, Peng-Robinson, and AQUAlibrium. All models were found to have considerable error when predicting excess enthalpies. The best model was Lee-Kesler with average absolute errors of 22.5% and absolute average differences of 22.4 J/mol. 2.1 Introduction Stricter regulations placed on the natural gas industry around carbon dioxide emissions have led to the disposal of the carbon dioxide by acid gas injection to be a more favorable option. Common impurities in the carbon dioxide stream are methane and hydrogen sulfide. Traditionally, the hydrogen sulfide was converted to elemental sulfur by the Claus process. However, due to a decrease in demand of sulfur, injection into subsurface reservoirs has become the more economical procedure for many gas plants. Ying Wu, John J. Carroll and Weiyao Zhu (eds.) CO2 Capture and Acid Gas Injection, (39–54) 2017 © Scrivener Publishing LLC 39 40 Carbon Dioxide Capture and Acid Gas Injection The excess enthalpies of these hydrogen sulfide-methane mixtures are required for the design of the acid gas injection scheme. This study uses six different thermodynamic models to predict the excess enthalpy of a hydrogen sulfide-methane mixture. Five different equations of state, Benedict-Webb-Rubin (BWR), Lee-Kesler (LK), Soave-Redlich-Kwong (SRK), Redlich-Kwong (RK), and Peng-Robinson (1978) were used with VMGSim software, as well the AQUAlibrium model. BWR and LK are multi-constant equations, and SRK, RK and PR78 are cubic equations of state. The AQUAlibrium model uses a variation of Peng-Robinson. 2.2 Enthalpy Typically the enthalpy is expressed as a molar enthalpy, measured in J/mol, or a mass enthalpy, measure in J/g. Other common units used for enthalpies include BTU/lb or BTU/lbmol. The enthalpy of mixtures can be determined in a number of ways. One method is to use excess enthalpy (enthalpy of mixing). Excess enthalpy is defined as HE Hm i xi H i (2.1) where: HE – Excess enthalpy Hm – Enthalpy of mixture Hi – Enthalpy of component i xi – mol fraction of component i 2.3 Literature Review A review of the literature was performed to find all experimental data for the enthalpies of the binary system of hydrogen sulfide and methane. Only one experimental data set was found for this binary system, performed by Barry et al. (1982). The enthalpy data was in the range of 0.18 to 0.85 mol fraction hydrogen sulfide, taken at nominal pressures of 0.507 MPa, 1.013 MPa, and 1.52 MPa, and at nominal temperatures of 293.15 K, 305.15 K and 313.15 K. Only the nominal temperatures and pressures were reported in the data. Another experimental data set was found, also by Barry et al. (1983), but used a ternary system of carbon dioxide, hydrogen Enthalpies of Hydrogen Sulfide-Methane Mixture 41 sulfide and methane. The data was at the same nominal temperatures and pressures, and at a wide range of mol fractions. 2.4 Calculations The experimental enthalpies from Barry et al. (1982) was compared to calculated enthalpies calculated using Lee-Kesler, Benedict-Webb-Rubin, Soave-Redlich-Kwong, Redlich-Kwong and Peng-Robinson thermodynamic models from VMGSim as well as AQUAlibrium software. Four different error functions were used to compare the different thermodynamic models. The absolute average difference in excess enthalpies (AAD) was defined as; 1 NP AAD E E H exp H calc (2.2) NP– number of points HEexp– experimental excess enthalpy HEcalc– calculated excess enthalpy and the average difference (AD) was defined as: where: AD 1 NP E E H exp H calc (2.3) The absolute average error (AAE) in excess enthalpies was defined as: AAE 1 NP E E H exp H calc E H calc 100% (2.4) and the average error (AE) was defined as: AE 2.4.1 1 NP E E H exp H calc E H calc (2.5) 100% Lee-Kesler The Lee-Kesler model gave an AAD of 22.5 J/mol, an AD of 22.4 J/mol with a maximum enthalpy difference of 60.3 J/mol occurring at 1.52 MPa and 42 Carbon Dioxide Capture and Acid Gas Injection 293.15 K. The AAE was 25.2%, and the AE was 24.8% with a maximum error of 43.5% at 1.52 MPa and 293.15 K. Figures 2.1 through 2.3 show the experimental excess enthalpies and calculated excess enthalpies using LK for 293.15 K, 305.15 K, and 313.15 K. The errors and differences greatly increased with increasing mole fraction of methane in the mixture, with errors averaging around 15% for mole fractions from 0.2 to 0.3 methane, 200 0.507 MPa 1.013 MPa 1.52 MPa 180 Excess enthalpy (J/mol) 160 1.52 140 120 100 1.013 80 60 40 0.507 20 0 0 0.2 0.4 0.6 0.8 Mol fraction hydrogen sulfide 1 Figure 2.1 Experimental and calculated enthalpies at 293.15 K using Lee-Kesler (Barry et al., 1982). 180 0.507 MPa 1.013 MPa 1.52 MPa 160 Excess enthalpy (J/mol) 140 1.52 120 100 80 1.013 60 40 0.507 20 0 0 0.2 0.4 0.6 0.8 Mol fraction hydrogen sulfide 1 Figure 2.2 Experimental and calculated enthalpies at 305.15 K using Lee-Kesler (Barry et al., 1982). Enthalpies of Hydrogen Sulfide-Methane Mixture 43 140 0.507 MPa 1.013 MPa 1.52 MPa 120 Excess enthalpy (J/mol) 1.52 100 80 1.013 60 40 0.507 20 0 0 0.2 0.4 0.6 0.8 Mol fraction hydrogen sulfide 1 Figure 2.3 Experimental and calculated enthalpies at 313.15 K using Lee-Kesler (Barry et al., 1982). and errors of around 40% for methane mole fractions of 0.8. The errors and differences generally increased with increasing temperature and increasing pressure, but only by 1–2% between the lowest and highest values. 2.4.2 Benedict-Webb-Rubin The BWR model gave an AAD and an AD of 43.9 J/mol with a maximum difference of 96.5 J/mol at 1.52 MPa and 293.15 K. The AAE and AE was 49.9% with a maximum error of 57.6% at 1.52 MPa and 293.15. Figures 2.4 through 2.6 show the experimental excess enthalpies and calculated enthalpies using BWR for the three different temperatures. Unlike the Lee-Kesler model, there was no significant change in error with increasing mole fraction of methane, with all mole fractions having errors around 50%. There was generally a small increase in error and difference with increasing pressure and temperature, but only by about 1–3% between the highest and lowest values. 2.4.3 Soave-Redlich-Kwong The SRK model gave an AAD and AD of 49.6 J/mol with a maximum of 110.6 J/mol at a pressure of 1.52 MPa and a temperature of 293.15 K. 44 Carbon Dioxide Capture and Acid Gas Injection 200 0.507 MPa 1.013 MPa 1.52 MPa 180 Excess enthalpy (J/mol) 160 140 120 100 1.52 80 60 1.013 40 20 0 0.507 0 0.2 0.4 0.6 0.8 Mol fraction hydrogen sulfide 1 Figure 2.4 Experimental and calculated enthalpies at 293.15 K using Benedict-WebbRubin (Barry et al., 1982). 180 0.507 MPa 1.013 MPa 1.52 MPa 160 Excess enthalpy (J/mol) 140 120 100 1.52 80 60 1.013 40 0.507 20 0 0 0.2 0.4 0.6 0.8 Mol fraction hydrogen sulfide 1 Figure 2.5 Experimental and calculated enthalpies at 305.15 K using Benedict-WebbRubin (Barry et al., 1982). Enthalpies of Hydrogen Sulfide-Methane Mixture 45 140 0.507 MPa 1.013 MPa 1.52 MPa Excess eanthalpy (J/mol) 120 100 80 1.52 60 1.013 40 20 0 0 0.507 0.2 0.4 0.6 0.8 Mol fraction hydrogen sulfide 1 Figure 2.6 Experimental and calculated enthalpies at 313.15 K using Benedict-WebbRubin (Barry et al., 1982). 200 0.507 MPa 1.013 MPa 1.52 MPa 180 Excess enthalpy (J/mol) 160 140 120 100 1.52 80 60 1.013 40 0.507 20 0 0 0.2 0.4 0.6 0.8 Mol fraction fydrogen sulfide 1 Figure 2.7 Experimental and calculated enthalpies at 293.15 K using Soave-RedlichKwong (Barry et al., 1982). The AAE and AE were both 57.2% with a maximum error of 65.1% at 0.507 MPa and 293.15 K. Figures 2.7 through 2.9 show the experimental excess enthalpies and calculated excess enthalpies using SRK for the three different temperatures. The mole fraction of methane in the mixture had little effect on the errors and difference, as with the BWR model. There was also very little difference in errors with changing pressure. However, 46 Carbon Dioxide Capture and Acid Gas Injection 180 0.507 MPa 1.013 MPa 1.52 MPa 160 Excess enthalpy (J/mol) 140 120 100 80 1.52 60 1.013 40 20 0 0.507 0 0.2 0.4 0.6 0.8 Mol faraction hydrogen sulfide 1 Figure 2.8 Experimental and calculated enthalpies at 305.15 K using Soave-RedlichKwong (Barry et al., 1982). 140 0.507 MPa 1.013 MPa 1.52 MPa Excess enthalpy (J/mol) 120 100 80 1.52 60 40 1.013 20 0 0.507 0 0.2 0.4 0.6 0.8 Mol fraction hydrogen sulfide 1 Figure 2.9 Experimental and calculated enthalpies at 313.15 K using Soave-RedlichKwong (Barry et al., 1982). Enthalpies of Hydrogen Sulfide-Methane Mixture 47 the errors and differences decreased slightly with increasing temperatures, which was different than the LK and BWR models, though the difference in errors was only about 3% between the lowest and highest values. 2.4.4 Redlich-Kwong The RK model gave an AAD and AD of 62.1 J/mol with a maximum difference of 135.6 J/mol at 1.52 MPa and 293.15 K. The AAE and AE were both 72.1% with a maximum of 77.8% at 0.507 MPa and 293.15 K. Figures 2.10 through 2.12 show the experimental excess enthalpies and the excess enthalpies calculated by RK for the three different temperatures. There was a significant effect of methane mole fraction on the errors and differences, with methane mole fractions around 0.2 having errors of 68% and methane mole fractions around 0.8 having errors of around 78%. Unlike the LK, BWR, and SRK model, for the RK model, errors decreased with increasing pressure, but only by about 3% between the highest and lowest values. There was very little difference in errors with changing temperature. 2.4.5 Peng-Robinson The AAD and AD for the PR78 model was 50.5 J/mol with a maximum difference of 112.4 J/mol at 1.52 MPa and 293.15 K. The AAE and AE was 200 0.507 MPa 1.013 MPa 1.52 MPa 180 Excess enthalpy (J/mol) 160 140 120 100 80 60 1.52 40 1.01 20 0 0.507 0 0.2 0.4 0.6 Mol fraction hydrogen sulfide 0.8 1 Figure 2.10 Experimental and calculated enthalpies at 293.15 K using Redlich-Kwong (Barry et al., 1982). 48 Carbon Dioxide Capture and Acid Gas Injection 180 0.507 MPa 1.013 MPa 1.52 MPa 160 Excess enthalpy (J/mol) 140 120 100 80 60 1.52 40 1.013 20 0.507 0 0 0.2 0.4 0.6 0.8 Mol fraction hydrogen sulfide 1 Figure 2.11 Experimental and calculated enthalpies at 305.15 K using Redlich-Kwong (Barry et al., 1982). 140 0.507 MPa 1.013 MPa 1.52 MPa Excess enthalpy (J/mol) 120 100 80 60 1.52 40 1.013 20 0.507 0 0 0.2 0.4 0.6 0.8 Mol fraction hydrogen sulfide 1 Figure 2.12 Experimental and calculated enthalpies at 313.15 K using Redlich-Kwong (Barry et al., 1982). Enthalpies of Hydrogen Sulfide-Methane Mixture 49 58.2% with a maximum of 66% at 0.507 MPa, and 293.15 K. Figures 2.13 through 2.15 show the experimental excess enthalpies and calculated excess enthalpies using PR78 for the three different temperatures. The errors and differences decreased by about 3% between the lowest temperature and the highest temperature. Pressure did not affect the errors and differences, 200 0.507 MPa 1.013 MPa 1.52 MPa 180 Excess enthalpy (J/mol) 160 140 120 100 80 1.52 60 1.013 40 20 0.507 0 0 0.2 0.4 0.6 0.8 Mol fraction hydrogen sulfide 1 Figure 2.13 Experimental and calculated enthalpies at 293.15 K using Peng-Robinson (Barry et al., 1982). 180 0.507 MPa 1.013 MPa 1.52 MPa 160 Excess enthalpy (J/mol) 140 120 100 80 1.52 60 1.013 40 20 0 0.507 0 0.2 0.4 0.6 0.8 Mol fraction hydrogen sulfide 1 Figure 2.14 Experimental and calculated enthalpies at 305.15 K using Peng-Robinson (Barry et al., 1982). 50 Carbon Dioxide Capture and Acid Gas Injection 140 0.507 MPa 1.013 MPa 1.52 MPa Excess enthalpy (J/mol) 120 100 80 1.52 60 40 1.013 20 0 0.507 0 0.2 0.4 0.6 0.8 Mol fraction hydrogen sulfide 1 Figure 2.15 Experimental and calculated enthalpies at 313.15 K using Peng-Robinson (Barry et al., 1982). and the methane mole fraction had only a small affect, with slightly greater errors and differences with increasing methane mole fraction. 2.4.6 AQUAlibrium The AQUAlibrium model had very similar results to the Peng-Robinson model with an AAD and AD of 49.2 with a maximum difference of 109.9 J/mol at 1.52 MPa, and 293.15 K. The AAE and AE were 56.9% with a maximum error of 64.9% at 0.507 MPa and 293.15 K. Figures 2.15 through 2.18 show the experimental excess enthalpies and calculated excess enthalpies using AQUAlibrium for the three different temperatures. As with the Peng-Robinson model, the errors and difference decreased by about 3% between the lowest and highest temperatures, and pressure and mole fraction did not have a significant effect. 2.5 Discussion For predicting enthalpies, the six models all had considerable errors and differences. The best model was LK, which predicted excess enthalpies Enthalpies of Hydrogen Sulfide-Methane Mixture 51 200 0.507 MPa 1.013 MPa 1.52 MPa 180 Excess enthalpy (J/mol) 160 140 120 100 80 60 1.52 1.013 40 20 0 0 0.507 0.2 0.4 0.6 0.8 Mol fraction hydrogen sulfide 1 Figure 2.16 Experimental and calculated enthalpies at 293.15 K using AQUAlibrium (Barry et al., 1982). 180 160 0.507 MPa 1.013 MPa 1.52 MPa Excess enthalpy (J/mol) 140 120 100 80 1.52 60 40 1.01 20 0.507 0 0 0.2 0.4 0.6 0.8 Mol fraction hydrogen sulfide 1 Figure 2.17 Experimental and calculated enthalpies at 305.15 K using AQUAlibrium (Barry et al., 1982). within 25% or 22.5 J/mol, on average. The RK model had the highest errors and differences with an average of 72.1% or 62.1 J/mol. The other four models, BWR, SRK, PR78, and AQUAlibrium all had absolute average errors around 50-60% and differences of 55-65 J/mol. Both the LK and BWR models increased in error in increasing temperature. However, the SRK, PR78 and AQUAlibrium models all decreased 52 Carbon Dioxide Capture and Acid Gas Injection 140 0.507 MPa 1.013 MPa 1.52 MPa Excess enthalpy (J/mol) 120 100 80 1.52 60 40 1.013 20 0 0.507 0 0.2 0.4 0.6 0.8 Mol fraction hydrogen sulfide 1 Figure 2.18 Experimental and calculated enthalpies at 313.15 K using AQUAlibrium (Barry et al., 1982). in error with increasing temperature. A change in temperature had little effect on the errors for the RK model. Similar to the trend with temperature, the LK and BWR models both increased in error with an increase in pressure. The RK model decreased in error with an increase in pressure. A change in pressure had little effect on the SRK, PR78, and AQUAlibrium models. For all models, an increase in mole fraction of methane increased the error and difference. However, this effect was almost negligible in the BWR, SRK, PR78 and AQUAlibrium models. A significant increase in error with increasing mole fraction of methane occurred with the LK and BWR models. LK was found to be the best model, with an AAE of 25%, predicting values 25–50% more accurate than the other five models. The errors were greatly affected by the methane mole fraction in the mixture, as seen in Figure 2.19. The errors varied from 2% at very low fractions to almost 45% at very high fractions. If only data consisting of less than 0.3 mole fraction methane is considered, the error of the LK model is reduced to 15%, or 16.1 J/mol on average. 2.6 Conclusion Table 2.1 shows the various error for all enthalpy data for all thermodynamic models. The six different models, LK, BWR, SRK, RK, PR78 and AQUAlibrium all lacked the ability to provide accurate predictions of the Enthalpies of Hydrogen Sulfide-Methane Mixture 53 50 Absolute excess enthalpy error (%) 45 40 35 30 25 20 15 293.15 K 305.15 K 313.15 10 5 0 0 0.2 0.4 0.6 Mole fraction methane 0.8 1 Figure 2.19 Effect of mole fraction of methane on the error in excess enthalpy prediction for Lee-Kelser at 293.15 K, 305.15 K, and 313.15 K. Table 2.1 AAD, AAE, maximum difference and maximum error using the six different thermodynamic models for the hydrogen sulfide-methane mixture Model AAD (J/mol) LK 22.5 BWR SRK Max difference (J/mol) AAE (%) Max error (%) 60.3 25.2 43.5 43.9 96.5 49.9 57.6 49.6 110.6 57.2 65.1 RK 62.1 135.6 72.1 77.8 PR78 50.5 112.4 58.2 66.0 AQUA 49.2 109.9 56.9 64.9 experimental excess enthalpies of Barry et al. (1982). The best model was LK, with an AAE of 25%, and an AAD of 22.5 J/mol. The RK model produced the least accurate prediction with an AAE of 72.1% and an AAD of 62.1 J/mol. The pressure and temperature had a small effect on many of the models, with LK and BWR increasing in error with both increasing temperature and pressure. SRK, PR78 and AQUAlibrium decreased in error with increasing temperature and the RK model decreased in error with 54 Carbon Dioxide Capture and Acid Gas Injection increasing pressure. All models had an increase in error with an increase in mole fraction; however, this effect was only significant for the LK and RK models. The most accurate predictions were achieved by LK. When only small methane mole fraction of less than 0.3 are considered, the average absolute error is reduced to 15%, with an absolute average difference of 16.1 J/mol. In this analysis, only one data source was analyzed, as this was the only experimental data set readily available. Further analysis of the models with additional data sets is required to determine the validity of the models for predicting excess enthalpies of hydrogen sulfide-methane mixtures. References 1. Barry, A., Kallaguine, S., and R. Ramalho, “Excess Enthalpies of the Binary System Methane-Hydrogen Sulfide by Flow Calorimetry”, J. Chem. Eng. Data, 27, 436–439, 1982. 2. Barry, A., Kallaguine, S., and R. Ramalho, “Ternary System Methane-Carbon Dioxide-Hydrogen Sulfide. Excess Enthalpy Data by Flow Calorimetry”, J. Chem. Eng. Data, 28, 375–381, 1983. 3 Phase Behavior and Reaction Thermodynamics Involving Dense-Phase CO2 Impurities J.A. Commodore, C.E. Deering and R.A. Marriott Department of Chemistry, University of Calgary, Calgary, Alberta, Canada Abstract High-density CO2 streams destined for subsurface reinjection contain multiple impurities which can change the phase behavior, density or reaction chemistry. Industrial streams include those aimed at Carbon Capture and Sequestration, Enhanced Oil Recovery or sulfur/carbon management by Acid Gas Injection. The purification, compression and injection processes for these streams involve fluids over a large range of temperature (0–150 °C) and pressure (0.1–35 MPa). While many chemical activity models for calculating complex high-pressure equilibria have been reported for aqueous systems, CO2 rich systems have received very little attention. Building on our work with H2O, H2S and COS in CO2, this new study focuses on the volumetric effect of dissolving CS2 or SO2 into a high-pressure CO2 fluid at conditions up to p = 35 MPa and temperatures ranging from T = 50 to 125 °C. We describe our densimetric experiments and how those measurements allow us to fully describe the fugacities of CS2 or SO2. These mixing coefficients obtained and resulting fugacities can be directly incorporated into Gibbs Energy Minimization routines, for calculation of high-pressure phase behavior and exploration of chemical reactivity. 3.1 Introduction High-pressure CO2 streams can contain a variety of minor chemical species with potential to react to change phase behavior and/or chemical composition. Whether the intent of an injectate stream is Carbon Capture and Ying Wu, John J. Carroll and Weiyao Zhu (eds.) CO2 Capture and Acid Gas Injection, (55–62) 2017 © Scrivener Publishing LLC 55 56 Carbon Dioxide Capture and Acid Gas Injection Storage (CCS), Enhanced Oil Recovery (EOR) and/or conventional Acid Gas Injection (AGI), our group has been interested in understanding how the chemical equilibria can change under compression and injectate conditions, i.e., beyond the critical conditions for CO2 fluids (Tc = 30.94 °C and pc = 7.38 MPa). While we do not develop marketable simulation tools, our recent research has been aimed at density measurements to provide reference quality mixing parameters for the benchmarking of Gibbs Energy Minimization based simulators. Impurities of interest include hydrogen sulfide (H2S), sulfur dioxide (SO2), carbonyl sulfide (COS), oxygen (O2) and carbon disulfide (CS2) among others. Our approach has been to measure densities with an in-house built densimeter [1]. By measuring the change in density caused by dissolving a small amount of impurity, we then calibrate mixing parameters for reduced Helmholtz Energy reference equations. All measurements have been completed in the single phase region, where density changes are converted to apparent molar volumes. Because apparent molar volumes are a type of excess property, these properties are used to optimize mixing parameters without the optimization being affected by imperfections in the pure-component equations of state (EOS), i.e., apparent molar volumes are more sensitive to the intermolecular interactions versus the bulk density change. In recent examples of this work, we have published results for H2O and COS in CO2, which have allowed us to explore high-pressure water dewpoint and COS hydrolysis equilibria [1, 2]: COS + H2O H2S + CO2,(3.1) Although the previous parameters were only calibrated with our volumetric measurements at a single low concentration, the resulting symmetric mixing coefficients were shown to better predict the phase behavior of H2O + CO2 and COS + CO2, thus providing an external validation of our measurements and optimization approach. In this work we describe the measurements for the volumetric studies of CS2 and SO2 in dense CO2 fluids up to p = 35 MPa and temperatures up to T = 125 °C. Here CS2 can enter a CO2 injectate stream from incomplete combustion (flue gas) or hydrocarbon production, whereas SO2 can only enter an injectate stream through CCS or flue gas. The calibrated parameters from this work showed a significant improvement over the estimated parameters when compared to measured dew and bubble points pressures for SO2 + CO2 systems [3–5]. CS2 is different from the previous impurity studies, because there are currently no high-accuracy Helmholtz energy equations for CS2. Thus, we report an alternative method for calculating fugacity, through a Fluctuation Solution Theory correlation. Phase Behavior and Reaction Thermodynamics 57 3.2 Experimental The densities of impurities in dense CO2 fluid were obtained using a vibrating tube densimeter (VTD) which was described in the previous work of Deering et al. and shown to accurately measure density to an estimated uncertainty of ±0.07 kg m–3 over a wide range of temperature and pressure [1]. With a VTD, the density of the fluid of interest can be related to a reference fluid whose density is accurately known over the conditions of interest by measuring the period of oscillation of the vibrating tube. The fundamental equation relating the period of oscillation for both the fluid of interest and reference fluid is given by equation 2, ρ – ρo = KT(τ2 – τo2),(3.2) where ρ and ρo are the fluid of interest and reference densities, respectively; τ and τo are the vibrating tube’s period of oscillation of the fluid of interest and reference fluid; KT is the isothermal pressure dependent calibration constant (calibrated with a second well characterized fluid/calibration fluid). Degassed H2O and N2 were used as the calibration and reference fluids in this work. The density of the reference fluid (N2) was calculated using Span et al. [4] and the density for the calibration fluid (H2O) was calculated using Wagner and Pruß [5]. Mixtures were gravimetrically prepared in an evacuated 500 cm3 stainless steel vessels and agitated on a rocking table for two weeks for homogeneity. Each binary mixture was analyzed by a gas chromatograph (GC) to verify the composition and/or identify any other impurities. The purity from supplier was deemed sufficient and was used without any further purification, see Table 3.1. The prepared mixtures were transferred to a syringe pump (Teledyne-ISCO 260D) which can control pressure with a precision of ±0.005 MPa. Table 3.1 Chemical name, purities, source and analysis method. Chemical name Source Analysis method Purity mol % Carbon Dioxide Praxair Inc. GC-TCD/FID >99.9995 Carbon Disulfide Praxair Inc. GC-TCD/FID >99.9 Sulfur dioxide Praxair Inc GC-TCD/FID >99.98 Carbonyl sulfide Praxair Inc. GC-TCD/FID >99.9 Nitrogen Praxair Inc. GC-TCD/FID >99.998 Water In-house, EMD Millipore Resistivity 18 MΩ∙cm–1 58 Carbon Dioxide Capture and Acid Gas Injection Table 3.2 Impurity concentrations in CO2 and the conditions studied. T/°C p/MPa x/mol% Reference COS 49.76–120.10 2.5–35 2.737 Deering et al. [2] H2O 50–125 2.5–35 0.280 Deering et al. [1] SO2 50.92–126.84 2.5–35 1.042 This work CS2 50.94–127.35 2.5–35 1.011 This work Solute Density measurements were completed by charging the fluid from the syringe pump into the vibrating tube to measure the period of the fluid isothermally across the pressure range of interest. Table 3.2 shows the concentrations of solutes in dense CO2 (1) phase used in this work and the previous studies. 3.3 Results and Discussion 3.3.1 Phase Behavior Studies of SO2 Dissolved in Dense CO2 Fluid The relative density measurements of the SO2(2) + CO2(1) system and pure CO2 from this work were combined to calculate apparent molar volumes – (Vϕ,2) of the SO2: – Vϕ,2/(cm3∙mol–1) = M2/ρ – 1000Δρ/(mρρ1),(3.3) where M2 is the molar mass of SO2, ρ and ρ1 are the densities of the mixture and the pure CO2 respectively and m is the molality of the mixture. The apparent molar volumes were then used to optimize the symmetric parameters found within the multi-fluid EOS through minimizing the sum of a weighted sum of the squares difference between the volumes from equation 3 and those calculated. The optimization began with the estimated parameters of Kunz et al. [8] and the weighting was applied as the reciprocal of the squared uncertainties of the apparent molar volumes. The calculation of the mixture volumes using multi-fluid EOS requires the combination of a pure fluid EOS and mixture contribution from the composition dependent reducing functions for supercritical density and temperature (δ and τ). The description of the δ and τ functions which contain the binary parameters to be optimized uses the formulation by Kunz et al. [8]: Phase Behavior and Reaction Thermodynamics 59 N N i j xi x j v ,ij v ,ij xi xj 1 xj 8 v ,ij xi 1 1 1/3 c ,i 1/3 c, j 3 (3.4) and N i N j xi x j T ,ij T ,ij xi T ,ij xi xj xj Tc ,iTc , j , (3.5) where Tc and ρc are the critical temperature and densities for the pure fluids, N is the number of components, x is the mole fraction of pure components in the mixture. βv,ij and γv,ij in equation 4 and βT,ij and γT,ij in equation 5 are used to fit symmetric and asymmetric portions of the mixing behavior. In this work, only the symmetric parameters (γv,ij and γT,ij) were required to adequately calculate the apparent molar volumes. The Helmholtz free energy EOSs used to describe the pure fluid contribution are reported in Table 3.3. The optimized symmetric parameters obtained in this work were used to calculate the dew and bubble point pressures for comparison to the measured vapor-liquid equilibrium data for SO2(2) + CO2(1) systems (see Figure 3.1) [3–5]. The results from the calculation with the optimized parameter showed a better agreement with the literature vapor-liquid equilibrium data and a significant improvement over the original estimated parameters by Kunz et al. [8]. We note that this improved agreement is based only on our apparent molar volumes which were measured at a single concentration of 1.042 mol% SO2, again showing the utility in the volumetric data for optimizing mixing parameters. Table 3.3 The Helmholtz EOS model used to describe the pure component fluid. Component Formula Pure fluid EOS used Carbon dioxide CO2 Reference equation of state by Span and Wagner [9] Sulfur disulfide SO2 Short fundamental equation of state by Lemmon and Span [10] 60 Carbon Dioxide Capture and Acid Gas Injection 12 SO2(2) + CO2(1) 10 p/MPa 8 6 4 60.06 °C 2 –10 °C 0 0 0.2 0.4 0.6 Mole fraction CO2(1) 0.8 1 Figure 3.1 The p-x diagram for SO2(2) + CO2(1) at T = –10 and 60.06 °C. , Lachet et al. [3]; ,Caubet et al. [5]; ---, estimated binary mixing parameters combined with highaccuracy equations-of-state —, optimized binary mixing parameters from this work combined with high accuracy reduced Helmholtz EOS. 3.3.2 The Densimetric Properties of CS2 and CO2 Mixtures The relative density data for the CS2(3) + CO2(1) mixture was used in calculating the apparent molar volume for CS2 dissolved in dense CO2 fluid. Because no reduced Helmholtz energy equation of state for pure CS2 was found in the literature, the calculated apparent molar volumes were used to optimize coefficients within a Fluctuation Solution Theory based correlation equation;11 cm3mol V3 1 V1o o T ,1RT a13 b13 exp c13 / V1o 1 V1o , (3.6) –o where V1 is the molar volume of the pure solvent (CO2), κoT,1, is the isothermal compressibility of the pure solvent and a13, b13, c13 are adjustable parameters. The obtained adjustable parameters within the model show a ~ 50 cm3 mol–1 volume change upon increase in pressure to a dense CO2 region. No available other literature density data were found for comparison. The data are shown in Figure 3.2. The adjustable parameters from equation 3 can then be used to calculate fugacities for CS2 without employing an equation of state for CS2: ln 3 o 1 a13 b13 V1o b13 exp c13 V1o 1 c13 , (3.7) Phase Behavior and Reaction Thermodynamics 61 1000 CS2(3) + CO2(1) 127.353 °C – V3 /(cm3 mol–1) 600 101.632 °C 200 76.280 °C –200 50.941 °C –600 0 5 10 15 20 p/MPa 25 30 35 40 Figure 3.2 The apparent molar volume for CS2(3) dissolved in dense phase CO2(1) investigated for p ≤ 35MPa, ( ), denotes experimental data from this work. where 3∞ and 1o are infinite dilution fugacity coefficients of solute and solvent respectively. The above equation, in combination with the previous mixing parameters, can be utilized in Gibbs Free Energy minimization models for calculating chemical equilibria over a wide range of temperatures and pressures. Future studies will include H2S and O2 in CO2, in addition to H2S rich fluids. References 1. Deering, C. E., Cairns, E. C., McIsaac, J. D., Read, A. S., and Marriott, R. A. The partial molar volumes for water dissolved in high-pressure carbon dioxide from 318.28 K to 369.40 K and pressures to 35 MPa. The Journal of Chemical Thermodynamics 93, 337–346, 2015. 2. Deering, C. E., Saunders, M. J., Commodore, J. A., and Marriott, R. A. The Volumetric Properties of Carbonyl Sulfide and Carbon Dioxide Mixtures from T = 322 to 393 K and p = 2.5 to 35 MPa: Application to COS Hydrolysis in Subsurface Injectate Streams. Journal of Chemical and Engineering Data 61, 1341–1347, 2016. 3. Lachet, V., Bruin, T. de, Ungerer, P., Coquelet, C., Valtz, A., Hasanov, V., Lockwood, F., and Richon, D. Thermodynamic behavior of the CO2+SO2 mixture: Experimental and Monte Carlo simulation studies. Energy Procedia 1, 1641–1647, 2009. 62 Carbon Dioxide Capture and Acid Gas Injection 4. Blümcke, A., Ueber die Bestimmung der specifischen Gewichte und Dampfspannungen einiger Gemische von schwefliger Säure und Kohlensäure. Ann. Phys. Leipzig 270, 10–21, 1888. 5. Caubet, F., The liquifaction of gas mixtures. Z. Kompr. Fluess. Gase 8, 65, 1904 6. Span, R. A Reference Equation of State for the Thermodynamic Properties of Nitrogen for Temperatures from 63.151 to 1000 K and Pressures to 2200 MPa. Journal of Physical and Chemical Reference Data 29, 1361, 2000. 7. Wagner, W., and Pruß, A. The IAPWS formulation 1995 for the thermodynamic properties of ordinary water substance for general and scientific use. Journal of Physical and Chemical Reference Data 31, 387–535, 2002. 8. Kunz, O., and Wagner, W. The GERG-2008 Wide-Range Equation of State for Natural Gases and Other Mixtures: An Expansion of GERG-2004. J. Chem. Eng. Data. 57, 3032–3091, 2012. 9. Span, R., and Wagner, W. A New Equation of State for Carbon Dioxide Covering the Fluid Region from the Triple-Point Temperature to 1100K at Pressures up to 800 MPa. J. Phys. Chem. Ref. Data. 25, 1509–1595, 1996 10. Lemmon, E. W., and Span, R. Short Fundamental Equations of State for 20 Industrial Fluids. J. Chem. Eng. Data. 51, 785–850, 2006. 11. O’Connell, J. P., Sharygin, A. V., and Wood, R. H. Infinite Dilution Partial Molar Volumes of Aqueous Solutes over Wide Ranges of Conditions. Industrial & Engineering Chemistry Research 35, 2808–2812, 1996. 4 Sulfur Recovery in High Density CO2 Fluid S. Lee and R.A. Marriott Department of Chemistry, University of Calgary, Calgary, Alberta, Canada Abstract After purification of natural gas through aqueous amines, several sources result in low-quality low-pressure H2S acid gas mixtures (<1% H2S in CO2), where this lowlevel of H2S cannot sustain the front-end-furnace in a traditional Claus conversion process for commercial sulfur recovery. Other available methods to recover low-level H2S normally produce poor quality non-commercial sulfur or rely on disposable scavenging systems. Alternatively, several newer cryogenic separation processes have been demonstrated, where the resulting CO2 rich acid gas liquids do not require recompression. Despite the more favorable efficiency of these processes, the high-pressure CO2 fluids either need to be reinjected into zones which can safely contain H2S, purified by complex multi-stage cryogenic separation or H2S would need to be removed after the cryogenic separation. In this study, we propose the direct oxidation of H2S to elemental sulfur in a dense-phase highpressure CO2 fluid. Through such a process, CO2 is purified and high-quality elemental sulfur is produced for commercial acid production. This study describes some initial experimental measurement for sulfur solubility in CO2. Based on the high-pressure experimental measurements, a solubility model for sulfur in CO2 is offered, which has allowed us to estimate the conditions at which high-pressure sulfur recovery may be possible. Our current work is introduced, where we are now focusing on the catalytic conversion of H2S at the high-pressure conditions provided through this work. The high-pressure catalytic conditions will be tested along feed fluid impurities such as methyl thiol (CH3SH), carbonyl sulfide (COS), and carbon disulfide (CS2). Ying Wu, John J. Carroll and Weiyao Zhu (eds.) CO2 Capture and Acid Gas Injection, (63–70) 2017 © Scrivener Publishing LLC 63 64 Carbon Dioxide Capture and Acid Gas Injection 4.1 Introduction Several unconventional natural gas streams contain a significant amount of low-quality acid gas, e.g., Horn River in N.E. BC (10% CO2 and 500 ppm H2S) [1]. The acid gas components must be removed from sour gas to produce sweet gas for environmental and safety reasons [2] and to meet pipeline requirements [3]. If CO2 is sold as a feed for other chemical processes, enhanced oil recovery (EOR) or injected into storage/disposal reservoirs, H2S will need to be removed. The primary available methods to remove H2S include (i) scavengers, (ii) conventional aqueous amine separation and the Claus process, or (iii) low-level oxidation (LO-CAT and others) [4]. Non-regenerable scavengers are a less economical choice when the volume of natural gas is large, and for poor quality acid gas fluids, the overall concentration of H2S is not large enough to maintain Claus furnace temperature. Liquid redox processes often encounter problems such as absorber plugging, bacterial contamination and poor-quality sulfur. This poor-­ quality sulfur is often landfilled due to impurity [5]; therefore, the commercial value can become an environmental liability. Finally, the purified CO2 using these methods is discharged at low pressure, where CO2 would need to be compressed at significant cost [6]. Alternative methods for the separation of high-pressure (HP) acid gas from sour natural gas include the cryogenic processes [7–12]. The cryogenic processes discharge the acid gas as a HP liquid that can be injected into reservoirs for geo-sequestration or EOR without any further compression (or minimal compression). Further distillation to remove the H2S is costly and complicated [13], and there are currently no commercial methods to oxidize low-level H2S to elemental sulfur in the HP CO2 fluid. Creating a novel solution to convert the H2S into S8 while maintaining the densephase CO2, would not only allow a producer to capture the economic benefits of sulfur recovery, but also to conserve a HP CO2 stream that would require minimal to no compression or purification to meet requirements for sale (Figure 4.1). The first step in determining whether the oxidation of H2S in a dense CO2 fluid is feasible is to explore if the solubility of sulfur in CO2 is sufficient. 4.2 Literature Review Our previous sulfur solubility model [14] can be used to predict sulfur vapor pressure or sulfur dew-point in CO2; however, the model was calibrated for sour gases and was not fit specifically for the purpose of HP Sulfur Recovery in High Density CO2 Fluid 65 CH4 to sales Cryogenic separation O2 Catalytic reactor H2S/CO2 (l or sc) + minor impurities H2S CO2 COS CH4 Separator CO2 (l) S8 (s) Figure 4.1 Simplified schematic of the proposed sulfur recovery process. CO2. Limited data were available in the literature for S8 solubility in CO2, with no solubility data for T > 120 °C. Thus, measuring the solubility of elemental sulfur in dense CO2 was necessary and was the initial focus of this work. We note that the solubility of sulfur in pure CO2 data from other reports [15, 16] were less consistent, which had been previously recognized by Dowling et al. [14] and Serin et al. [17]. In addition, the solubilities calculated using the existing model by Dowling et al. [14] were consistently less than the literature data for p > 15 MPa. These deviations were likely a result of the model being calibrated for sour gas mixtures; therefore, a better fit-for-purpose model was needed for pure CO2 and more accurate solubility calculations in the HP region where the best conditions for catalysis are estimated to be based on theoretical equilibrium conversions [18, 19]. Larger solubilities are desired for larger inlet H2S concentrations, as reaching the sulfur dew-point could lead to deposition on the catalyst and decreased conversion rates. 4.3 Methodology The solubility of sulfur in CO2 was measured using a custom-built HP saturation column followed by cold and chemical traps shown in Figure 4.2. 66 Carbon Dioxide Capture and Acid Gas Injection Data logging computer Gas meter HP transducer Platinum resistance thermometer Teledyne Isco260D syringe pump Glass wool U-glass traps Sulfur equilibrium vessel Poppet valve Thermostated zone Figure 4.2 Schematic of in-house built sulfur solubility measurement apparatus. The column loaded with sulfur was pressurized with CO2 and was shut in to reach equilibrium. Sulfur saturated CO2 was slowly released to ambient pressure and temperature. The released fluid flowed through two glass-wool u-traps before reaching a flow meter, for measuring the net discharged CO2. Sulfur deposited on the transfer lines and the traps were quantified using gas chromatography [20]. The total amount of deposited sulfur was related to the net volume of CO2 that had exited the saturation vessel to give the solubility of S8 in HP CO2. 4.4 Results and Discussion Sulfur solubility in CO2 was measured at pressures of p = 10 and 20 MPa and temperatures between T = 50 to 151 °C. The new experimental data were consistent with Serin et al. [17]. While the experimental temperatures were not the same, the literature data at T = 60 and 90 °C were aligned between the experimental data of T = 50 and 100 °C. The experimental results indicated that the solubility of sulfur increased with pressure (beyond p = 5 MPa) and temperature. The effect of pressure on sulfur solubility can be explained by the increase of the CO2 fluid density. The effect of temperature on sulfur solubility is due to the sulfur vapor pressure increase. Early indications from the modeling using a Mesmer-type equation did not allow for sufficient elemental sulfur to stay dissolved in CO2 Sulfur Recovery in High Density CO2 Fluid 67 50 10 p/MPa 40 Catalytic conditions 30 Separation conditions 20 (0.1% H2S feed) 10 0 0.001 0.0001 0 50 0.1 0.01 100 [S8] = 1.432 g/m3 150 T/°C 200 250 300 Figure 4.3 Sulfur dew-point based on the Mesmer CO2 model. ([S8] < 1.432 g m–3, which corresponds to [H2S] < 1000 ppm, which is not great for an inlet threshold) for the initially chosen catalytic conditions T < 150 °C. Revised conditions at increased T and p, where the elemental sulfur will remain dissolved in the dense-phase CO2 if 1.432 g m–3 < [S8] < 10.0 g m–3 are still promising. This concentration range corresponds to 1000 ppm < [H2S] < 0.7%, which allows for a larger threshold for H2S inlet concentration (Figure 4.3). 4.5 Conclusion and Future Directions The Mesmer model showed that a non-sub-dew-point process may be possible at a higher temperature than initially considered. However, increasing T will decrease the total possible conversion of H2S to S8. Increasing temperature to a point where the catalytic condition is above the expected sulfur dew-point, but still low enough temperature for a sufficient conversion of H2S, will now be the focus of this project in association with experimental catalysis. The parameters from the solubility models can be used for Gibbs Energy Minimization routines to calculate the equilibrium conversion in real fluid conditions. Upon selection of favorable pressures and temperatures, the catalytic conversion of H2S to elemental sulfur in dense CO2 over alumina and titania will be studied using a custom heterogeneous catalysis system. A further study will need to find temperature ranges that work for different inlet H2S concentrations as well as the catalytic efficiency under the effect of feed fluid impurities. 68 Carbon Dioxide Capture and Acid Gas Injection References 1. BC Oil and Gas Commission. Hydrocarbon and By-Product Reserves Report January-December 2012. http://www.bcogc.ca/node/11111/download (accessed Oct 17, 2016). 2. Tyndall, K.; McIntush, K.; Lundeen, J.; Fisher, K.; Beitler, C., When is CO2 more hazardous than H2S. Hydrocarb Process 90 (1), 45–48, 2011. 3. TransCanada, Gas Quality Specifications TransCanada and other pipelines. 2010. 4. Dalrymple, D. A.; Trofe, T. W.; Evans, J. M., Liquid redox sulfur recovery options, costs, and environmental considerations. Environmental Progress 8 (4), 217–222, 1989. 5. Primack, H. S.; Reedy, D. E.; Kin, F. R., Method of stabilizing chelated polyvalent metal solutions. Google Patents: 1984. 6. Campbell, J. M., Gas conditioning and processing. Campbell Petroleum Series: Vol. 4, 1982. 7. Kelley, B. T.; Valencia, J. A.; Northrop, P. S.; Mart, C. J., Controlled Freeze Zone for developing sour gas reserves. Energy Procedia 4, 824–829, 2011. 8. Lallemand, F.; Lecomte, F.; Streicher, C., Highly Sour Gas Processing: H2S Bulk Removal With the Sprex Process. In International Petroleum Technology Conference, International Petroleum Technology Conference: Doha, Qatar, 2005. 9. Terrien, P.; Dubettier, R.; Leclerc, M.; Meunier, V. In Engineering of Air Separation and Cryocap units for large size plants, Oxyfuel Combustion Conference, Ponferrada, Spain, 9–13 Sep, 2013; Ponferrada, Spain, 2013. 10. Holmes, A. S.; Ryan, J. M.; Price, B. C.; Styring, R. E., Process Improves Acid Gas Separation. Hydrocarb Process 61 (5), 131–136, 1982. 11. Ryan, J.; Schaffert, F., CO2 Recovery by the Ryan-Holmes Process. Chemical Engineering Progress 80 (10), 53–56, 1984. 12. ZareNezhad, B.; Hosseinpour, N., An extractive distillation technique for producing CO2 enriched injection gas in enhanced oil recovery (EOR) fields. Energ Convers. Manage. 50 (6), 1491–1496, 2009. 13. Guvelioglu, G. H.; Higginbotham, P.; Palamara, J. E.; Arora, G.; Mamorsh, D. L.; Fisher, K. S., H2S Removal from CO2 by Distillation. In Laurance Reid Gas Conditioning Conference, Norman, Oklahoma, 2015. 14. Dowling, N. I.; Marriott, R. A.; Primak, A.; Manley, S., The Kinetics of H2S Oxidation by Trace O2 and Prediction of Sulfur Deposition in Acid Gas Compression Systems. In Sour Gas and Related Technologies, John Wiley & Sons, Inc.: pp. 183–214, 2012. 15. Gu, M. X.; Li, Q.; Zhou, S. Y.; Chen, W. D.; Guo, T. M., Experimental and Modeling Studies on the Phase-Behavior of High H2S-Content Natural-Gas Mixtures. Fluid Phase Equilibr. 82, 173–182, 1993. Sulfur Recovery in High Density CO2 Fluid 69 16. Kennedy, H. T.; Wieland, D. R., Equilibrium in the Methane-Carbon DioxideHydrogen Sulfide-Sulfur System. T. Am. I. Min. Met. Eng. 219 (7), 166–169, 1960. 17. Serin, J. P.; Jay, S.; Cezac, P.; Contamine, F.; Mercadier, J.; Arrabie, C.; LegrosAdrian, J. M., Experimental studies of solubility of elemental sulphur in supercritical carbon dioxide. J. Supercrit. Fluid 53 (1–3), 12–16, 2010. 18. Gamson, B.; Elkins, R., Sulfur from hydrogen sulfide. Chemical Engineering Progress 49 (4), 203–215, 1953. 19. Energy, A. A.; Resources, N.; Paskall, H. G.; Research, W.; Ltd, D., Capability of the Modified-claus Process : a Final Report to the Department of Energy and Natural Resources of the Province of Alberta. Western Research & Development: 1979. 20. Clark, P. D.; Lesage, K. L., Quantitative-Determination of Elemental Sulfur in Hydrocarbons, Soils, and Other Materials. J. Chromatogr. Sci. 27 (5), 259–261, 1989. 5 Carbon Capture Performance of Seven Novel Immidazolium and Pyridinium Based Ionic Liquids Mohamed Zoubeik, Mohanned Mohamedali and Amr Henni Acid Gas Removal Laboratory, Clean Energy Technologies Research Institute (CETRi), University of Regina, Regina, SK, Canada Abstract The objective of this study is to compare the solubility of carbon dioxide (CO2) in seven ionic liquids, namely 1,2,3-Tris(diethylamino) cyclopropenylium dicyanamide, 1-Ethyl-3-methylimidazolium L-(+)- lactate, 3-Methyl-1-propylpyridinium bis [(trifluoromethyl) sulfonyl]imide, Ethyldimethyl propylammonium bis(trifluoromethyl sulfonyl)imide, 1,2,3-Tris(diethylamino) cyclopropenylium bis(trifluoromethanesulfonyl)imide, 1-(4-Sulfobutyl) -3-methylimidazolium Bis(trifluoro methanesulfonyl)imide, 1-(4-Sulfobutyl)-3-methylimidazolium hydrogen sulfate. Solubility measurements were performed at 313.15, 323.15 and 333.15 K, up to 20,000 mbar. CO2 solubility decreased in the following order: [TCD][TF2N] > [PMPY][TF2N] > [EMMP][TF2N] > [emim][LACTATE] > [TCD][DCN] > [(CH2)4SO3HMIm][TF2N] > [(CH2)4SO3HMIm] [HSO4]. The three ionic liquids, [TCD][TF2N], [PMPY][TF2N] and [EMMP][TF2N], show great potential for CO2 capture. Reported values of Henry’s law constants, enthalpies and entropies of absorption for CO2 were compared. The Peng-Robinson Equation of state, with quadratic mixing rules, was capable of correlating all data satisfactorily for all the ionic liquid systems. 5.1 Introduction Climate change is considered one of the greatest environmental challenges facing our civilization to date. The anthropogenic emissions of carbon dioxide (CO2) represent the greatest contribution to global warming and climate Ying Wu, John J. Carroll and Weiyao Zhu (eds.) CO2 Capture and Acid Gas Injection, (71–90) 2017 © Scrivener Publishing LLC 71 72 Carbon Dioxide Capture and Acid Gas Injection change. Human activities, especially those aimed at energy production, are the principal sources of CO2 emissions, of which fossil fuel combustion represents the vast majority. Carbon dioxide is the main greenhouse gas and its concentration in the atmosphere reached 400 ppm in 2013 [1, 2]. Therefore, the deployment of environmentally benign, energy efficient and economic CO2 capture technologies is becoming an important research topic [3]. There has been continuous improvement in the field of capture technologies for CO2; however, they are all associated with major drawbacks. Currently, the use of absorption using chemical solvents, predominantly aqueous amine solutions for CO2 separation, is considered the only commercially available technology for the capture of CO2 from flue gases. However, amine absorption systems are considered expensive due to the high energy required in the regeneration step, in addition to amine degradation during the thermal regeneration process [4]. Amongst potential solvents for CO2 capture, ionic liquids (ILs), as nonvolatile solvents, have been given much attention and are regarded as promising candidates [5]. Ionic liquids are salts composed of cations and anions and mainly exist as liquid at room temperature. They possess some enviable characteristics such as low vapor pressure, high thermal and chemical stability rendering them as potential alternatives to the energy intensive amine scrubbing ­process to achieve an environmentally and economically viable CO2 separation [6]. In addition to their unique properties, ILs could be systematically synthesized and tailored to fine-tune their final properties toward more efficient gas separation characteristics with a proper combination of the cations and anions counterparts [7]. One of the most commonly studied ILs in the literature are the imidazolium based ILs along with other sulfonium, ammonium, and phosphonium based solvents [7]. The availability of a wide range of choices of cations and anions allows for more options for the optimization of the design and synthesis of new ILs to ultimately improve the CO2 solubility and reduce the energy required in the regeneration step. A combined experimental and theoretical approach was adopted by Yan et al. to understand the structureproperty relationship due to the addition of various functional moieties such as aromatics, aliphatics and silane based groups attached to the triazolium based ILs [8]. The study reported on effective methods for the prediction of the modified ILs properties [8]. Anions substitution of ILs has gained considerable attention in the past decades as a method to modify the final properties of ILs. For instance, phosphonium hydroxide reacting with substituted phenols resulted in ILs possessing new properties. Both the viscosity and decomposition temperature of the ILs were significantly impacted by the variations of the substituent group attached to the anion part [9]. Several review papers are available in the literature to summarize the numerous studies performed on using ILs for CO2 capture, and special Carbon Capture Performance of Seven Novel 73 attention was dedicated to understanding the effects of varying the anion, cation, substituent groups on the cation and anions [10–12]. In this work, we discuss the CO2 solubility in seven promising ILs measured in our lab using a gravimetric microbalance (IGA). The selection of the ionic liquids was based on the fact that the results will allow us to better understand the effect of varying the anion and cation parts, for instance, the bis(trifluoromethylsulfonyl) imide anion, which is common among some Ils, was reported to have high affinity toward CO2 due to its high fluorination content. Furthermore, we also aimed at studying the effects of changing the cation to compare different types of chemical functionality on the cation with the same anion. The Peng-Robinson equation of state, with a quadratic mixing rule, was used to correlate the experimental solubility data. Henry’s law constants, the entropy and enthalpy of the absorption process were also derived and reported. 5.2 5.2.1 Experimental Work Materials Ionic liquids used in this work were purchased from Sigma-Aldrich, io-litec and Solvionic, as reported in Table 5.1, with their acronyms and molecular weights. Research grade carbon dioxide (CO2) was purchased from Praxair, with a purity of 99.99 wt.%. 5.2.2 Density Measurement The densities of ionic liquids used in this research were measured at different temperatures using an Anton Paar DMA 4500 digital density meter. The device allows for precision within 0.00001 g∙cm−3 and the uncertainty of the measurements was estimated to be 0.00005 g∙cm−3. The apparatus consists of a glass U-tube with a PT100 platinum resistance thermometer with an uncertainty of 0.01 K. The density meter was calibrated with air and bi-distilled water. The U-tube was carefully cleaned and dried for 30 min at 353.15 K before injecting the ionic liquids. Approximately 2 mL of a sample were slowly injected inside the glass U-tube of the apparatus. When the desired temperature was reached, the density was measured. The average of at least three measurements was reported. 5.2.3 Solubility Measurement Solubility was measured using an Intelligent Gravimetric Analyzer (IGA 003) from Hiden Analytical (Figure 5.1). The gravimetric microbalance Shorthand name [TDC] [DCN] [EMIM] [LACTATE] [PMPY] [TF2N] [EMMP] [TF2N] [TDC] [TF2N] [(CH2)4SO3HMIm][TF2N] [(CH2)4SO3HMIm][HSO4] Ionic liquid 1,2,3-Tris (diethylamino) cyclopropenylium dicyanamide 1-Ethyl-3-methylimidazolium L-(+)-lactate 3-Methyl-1-propylpyridinium bis[(trifluoromethyl)sulfonyl]imide Ethyldimethylpropyl-ammonium bis(trifluoro methylsulfonyl)imide 1,2,3-Tris(diethylamino) cyclopropenylium bis(trifluoromethanesulfonyl)imide 1-(4-Sulfobutyl)-3-methylimidazolium bis(trifluoromethanesulfonyl)imide 1-(4-Sulfobutyl)-3-methylimidazolium hydrogen sulfate Table 5.1 Characteristics of ionic liquids used in this work. Structures 316.35 499.43 532.56 396.37 416.40 200.23 318.46 Molecular weight 74 Carbon Dioxide Capture and Acid Gas Injection Carbon Capture Performance of Seven Novel 75 IGA P Red = CO2 Black = vacuum Blue = water MFC-A MFC-B Reactor and sample cell CO2 gas cylinder T Water bath Vacuum pump Figure. 5.1 Schematic of the gravimetric microbalance. contains a sample bucket where the liquid is placed inside a pressurevessel that is able to operate up to 20 bar and 500 °C. For each experiment, a small amount of ionic liquid samples in the range of 60 to 90 mg liquid sample was loaded into the sample container. Once the sample was loaded, the chamber was sealed. After stability was attained, the temperature was set at the degassing temperature of 348 K using an external water jacket. The sample was then dehydrated and degassed by completely evacuating the reactor using a diaphragm pump until the pressure reached 20 mbar, followed by a turbo pump (Pfeiffer) to achieve a vacuum of about 10 mbar. The degassing step was continued for about 10 hours to remove all traces of water and other volatile contaminants until a stable weight was achieved for about one hour, at which point the final weight was recorded. Temperature was then set at the absorption temperature using a water bath (Polyscience) with accuracy of 0.1 K. Temperature was measured with a type K platinum thermocouple (±0.1 K). When the set temperature was reached, at the desired pressure value, parameters related to the mass relaxation behavior were recorded by the IGASwin software. The absorption process was then initiated by allowing CO2 via a mass flow controller (MFC) to reach a pre-set ­pressure inside the microbalance chamber. Any real-time weight change upon absorption was automatically recorded. Pressure and temperature were kept constant until equilibrium was reached. Then, the 76 Carbon Dioxide Capture and Acid Gas Injection pressure was raised to the second data point of the isotherm, and this process was repeated for all other pressure measurements. A sufficient time of about 4 hours was given to reach equilibrium and allow for weight stability. 5.3 5.3.1 Modeling Calculation of Henry’s Law Constants Henry’s law constant is calculated based on the definition given below using the fugacity data obtained from PR-EOS, at near dilution conditions Hi T , P lim xi 0 fi L xi (5.1) fiL represents the fugacity of the gas dissolved in the liquid phase. Applying the equilibrium condition that the chemical potential is equal in the gas and liquid streams, the following equation of Henry’s law can be deduced: Pi H i (T )xi (5.2) where, P is the partial pressure of the gas and H (T) is Henry’s law ­constant. i i In conclusion it related the equilibrium solubility with the partial pressure of the gas [13]. 5.3.2 Critical Properties Calculations The classical Lydersen-Joback-Reid modified method [14] was used to predict the critical properties of the ionic liquids. 5.3.3 Peng Robinson EoS The equilibrium CO2 solubility was correlated using Peng-Robinson equation of state (PR-EoS) [14]. The PR-EOS parameters are obtained by the following equations: P RT a(T ) V b V V b b(V b) (5.3) Carbon Capture Performance of Seven Novel 77 ai i 0.45724 bi i mi 5.4 5.4.1 T 0.07780 R2Tci 2 Pci 2 (5.4) RTci Pci 1 mi 1 Tri 0.5 (5.5) 2 (5.6) 0.37464 1.54226wi 0.26992wi 2 a xi x j ai a j b xi x j 0. 5 bi b j 2 (1 kij ) (5.7) (5.8) 1 (5.9) (1 Iij ) Results and Discussion Density Figure 5.2 shows the experimental measurements of the density for the seven ionic liquids. The trend in the experimental density decreased linearly with increasing temperature, with [(CH2)4SO3HMIm] [TF2N] ­showing the highest density whereas the lowest density was noted for [TDC][DCN] as shown in Figure 5.2. The density-temperature data were modeled using a linear relationship for all the ionic liquid as shown in Table 5.2. The average absolute deviations (AADs) between the linear model predictions and the experimental data was found to be satisfactory as reported in Table 5.2. 5.4.2 Critical Properties The predicted properties such as the critical points, normal boiling points, and the acentric factors of the seven ionic liquids investigated in this report are presented in Table 5.3 below. 78 Carbon Dioxide Capture and Acid Gas Injection 2.4 2.2 2.0 1.8 [TDC ][DCN] [PMPY] [TF2N] [EMMP] [TF2N] [TDC] [TF2N] [EMIM][LACATATE] [(CH2)4SO3HMIm][TF2N] [(CH2)4SO3HMIm][HSO4] 1.6 1.4 1.2 1.0 0.8 280 300 320 T(K) 340 360 Figure 5.2 Experimental density of pure ionic liquids. Table 5.2 Temperature-dependent density correlations for the studied ionic liquids. Ionic liquids Density (g/cm3) [TCD][CN] ρ (g/cm3) = 1.10193–0.0006 × [T(C)] 0.05 [EMIM] [LATATE] ρ (g/cm3) = 1.1601–0.0007 × [T(C)] 0.02 [TCD][Tf2N] ρ (g/cm3) = 1.2947–0.0009 × [T(C)] 0.17 [EMMP][TF2N] ρ (g/cm3) = 1.4236–0.0009 × [T(C)] 0.06 [PMPY][TF2N] ρ (g/cm3) = 1.1416–0.0009 * [T(C)] 0.06 [(CH2)4SO3HMIm][TF2N] ρ (g/cm3) = 1.6016–0.0009 * [T(C)] 0.09 [(CH2)4SO3HMIm][HSO4] ρ (g/cm3) = 1.4533–0.0006 * [T(C)] 0.14 5.4.3 AAD (%) CO2 Solubility The accuracy of measuring the solubility of CO2 using the IGA was first verified by measuring the solubility in [bmim][PF6] at 323.15 K and comparing the results with previously published results by Shiflet [15] and 396.37 416.36 532.56 318.5 200.23 499.43 316.4 [PMPY][TF2N] [TDC][TF2N] [TDC][DCN] [EMIM][LACTATE] [(CH2)4SO3HMIm][TF2N]] [(CH2)4SO3HMIm][HSO4] MW (g/mol) [EMMP][TF2N] Ionic liquids Table 5.3 Critical properties of ionic liquids. 1017.6 1097.6 693.4 858.6 938.1 839.8 715.4 Tb (K) 1433.0 1612.8 912.7 1073.7 1255.7 1234.2 1038.7 Tc (K) 25.88 32.7 28.24 16.15 18.03 27.55 25.88 Pc (bar) 744.8 1070.1 620.1 1115.9 1394.0 964.7 955.5 Vc (cm3/mol) 0.8437 0.377 0.9702 1.0726 0.5876 0.3070 0.3334 ω 0.3602 0.2615 0.2260 0.2019 0.2407 0.2591 0.2863 ZC Carbon Capture Performance of Seven Novel 79 80 Carbon Dioxide Capture and Acid Gas Injection Anthony [16]. The AAD% of the measured solubility and those reported in the literature [15, 16] were 4 and 12%, respectively. The reported CO2 solubility was at (313.15, 323.15 and 333.15) K with pressures up to 20 bar, and the results are presented in Figures 5.3, 5.4, and 5.5, respectively. 40 CO2 mole fraction (%) 35 30 25 20 15 10 5 0 0 2000 4000 [TCD][DCN] 6000 8000 10000 12000 14000 16000 18000 20000 Pressure (mbar) [PMPY][TF2N] [EMMP][TF2N] [TCD][TF2N] [(CH2)4SO3HMIm][TF2N] [EMIM][LACTATE] [(CH2)4SO3HMIm][HSO4] Figure 5.3 CO2 solubility in seven promising ionic liquids at temperature of 313.15 K [17, 18]. 35 CO2 mole fraction (%) 30 25 20 15 10 5 0 0 2000 4000 [PMPY][TF2N] 6000 8000 10000 12000 14000 Pressure (mbar) [EMMP][TF2N] [TCD][TF2N] 16000 18000 20000 [EMIM][LACTATE] Figure 5.4 CO2 solubility in seven promising ionic liquids at 323.15 K [17, 18]. Carbon Capture Performance of Seven Novel 81 30 CO2 mole fraction (%) 25 20 15 10 5 0 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000 Pressure (mbar) [PMPY][TF2N] [EMMP][TF2N] [TCD][TF2N] [EMIM][LACTATE] [TCD][DCN] Figure 5.5 CO2 solubility in seven promising ionic liquids studied in this work at 333.15 K [17, 18]. 5.4.4 The Effect of Changing the Cation As stated earlier, changing the anion structure of the IL significantly impacts the CO2 capture abilities of the IL; the cation also can considerably influence the properties of the resultant IL [19]. We have the opportunity to examine a set of ILs, with four ILs sharing the same bis(trifluoromethyl sulfonyl)imide anion ([Tf2N]) and four different cations: 3-methyl-1-­ propylpyridinium, ethyldimethylpropylammonium, 1,2,3-tris(diethylamino) cyclopropenylium, and 1-(4-sulfobutyl)-3-methylimidazolium bis(trifluoro methane sulfonyl)imide. The CO2 solubility in decreasing order is as follows: [TDC][TF2N]>[PMPY][TF2N]>=[EMMP] [TF2N]> [(CH2)4SO3HMIm][TF2N]] as shown in Figure 5.6. It is evident that [TDC] cation has the highest CO2 solubility when compared to the other three cations. As apparent from the chemical structure shown in Table 5.1, the [TDC] cation contains three nitrogen (amine) atoms each surrounded by two ethyl radicals, [PMPY] has one and [EMPP] one. [TDC] is expected therefore to have higher capacity to interact with CO2. Looking closely at the structure of [(CH2)4SO3HMIm] [TF2N] with two nitrogen atoms in the ring seem to reduce the accessibility of CO2 to the absorption sites on these nitrogen atoms which are most stable in the ring structure than with an ethyl group as in [TDC] leading to the lowest solubility shown in [(CH2)4SO3HMIm][TF2N] cation. Another reason for the higher solubility of [TDC] cation is attributed to its lowest density as compared to the 82 Carbon Dioxide Capture and Acid Gas Injection 20000 18000 Pressure (M bar) 16000 14000 12000 10000 8000 6000 [PMPY ][Tf2N] [EMMP][Tf2N] [TCD][Tf2N] [(CH2)4SO3HMIm][TF2N] 4000 2000 0 0 10 20 30 Mole fraction of CO2 in ionic liquids (%) 40 Figure 5.6 Comparison of solubility in four ionic liquids with the same anion to illustrate the effect of the cation at 313 K. other cations (see Figure 5.2), which corresponds to a higher free volume within the structure of the [TDC][Tf2N] than in the other bulkier ILs, as reported in the literature [19, 20]. Note also that the four cations reported in this study are different in their basic structure; for instance [TDC] is a propenylium based cation, [PMPY] is a pyridinium based cation, [EMMP] is an ammonium based cation and [(CH2)4SO3HMIm][TF2N] is an imidazolium based cation. It is clearly observed that the cation with the highest number of nitrogen atoms possess strong basic strength, and that the propenylium based cation is a stronger base than the imidazolium based IL. Although the four cations have different performance in terms of ­equilibrium CO2 capacity, this has still less overall impact on the properties of ILs as opposed to the influence of the anion, as will be demonstrated in the following section. Another comparison has been carried out between [TDC][DCN] and data for [bmim][DCN] published by Aki et al. in order to compare the performance of [TDC] cation with the widely used [bmim], which is an imidazolium based cation as shown in Figure 5.7 [21]. Again, [bmim] has two nitrogen atoms; however, [TDC] has three, leading to an increased basicity. One of the widely used anions in the field of CO2 capture using ionic liquids is [Tf2N], which received considerable attention and has shown good performance due to the presence of two fluoroalkyl groups in its structure. We have also compared ILs with [Tf2N] anions with some ILs published in literature to get insights on the effects of the cation with [Tf2N] as an anion as shown in Figure 5.8. This again shows the good performance of [TDC] Carbon Capture Performance of Seven Novel 83 100 [BMIM][DCN] (Aki et al. 2004) [TDC][DCN](this work) Pressure (bar) 80 CH3 N+ 60 N CN – N CN CH3 40 H3C H3C CH3 N 20 +N C N CH3 – NΞC–N–CΞN CH3 CH3 0 0.0 0.1 0.2 0.3 0.4 Mole fraction of CO2 in ILs 0.5 0.6 Figure 5.7 Comparison of CO2 solubility in [TDC] and [bmim] cations with same anion at 313 K [18, 21]. 0.30 P = 13 bar P = 12 bar P = 14.79 bar P = 16 bar Mole fraction of CO2 0.25 0.20 0.15 0.10 0.05 [C ho m m im ][T Ak f2 N] ][T im m [b [h lin e] [T f2 N] M ul do on et al .( 20 07 ie ) t f2 al N] . (2 Ak 00 [E M ie 4) M [C ta P 6H l. ( ][T 4F 20 F2 [P 9m 04 N] M [C ) im PY (th 8H ] ] i [T [T 4F s w f2 F2 13 or N] N] m k) M im ( t ul hi ][T d s o w f2 on or N] k) et M ul al . (2 do on 00 [T 7) et DC al [o ][T . (2 m F2 00 im N] 7) ][T ( t f2 hi N] sw Ak or ie k) ta l. ( 20 04 ) 0.00 Figure 5.8 Comparison of CO2 solubility at 60 °C with different cations paired with [TF2N] anion at 333.15 K and about 12 to 14.97 bar. 84 Carbon Dioxide Capture and Acid Gas Injection 20000 18000 16000 Pressure (M bar) 14000 12000 10000 8000 [PMPY][Tf2N] [EMMP][Tf2N] [TCD][Tf2N] [(CH2)4SO3HMIm][TF2N] [TCD][DCN] [EMIM][LACTATE] [(CH2)4SO3HMIm][HSO4] 6000 4000 2000 0 0 10 20 30 Mole fraction of CO2 in ionic liquids (%) 40 50 Figure 5.9 Comparison of the reported solubility data of CO2 in the seven ionic liquids at 313.15 K. as a promising cation for CO2 absorption at low pressure. As for [omim], an imidazolium based cation, with very long alkyl chain length (C8H17), it showed good performance (Figure 5.9) as compared to other imidazolium based ILs as discussed by Aki and coworkers [21]. Finally, as presented in Figure 5.9, CO2 solubility decreases as follows: [TCD][TF2N]>[PMPY][TF2N]>[EMMP][TF2N]>[emim] [LACTATE]>[TCD][DCN]> [(CH2)4SO3HMIm][TF2N]>[(CH2)4SO3HM Im][HSO4]. As mentioned earlier, at low pressure, [emim][LACTATE] has the highest solubility, but behaves like [emim] [Ac] and other solvents that have a chemical interactions with CO2. The low solubility of [(CH2)4SO3HMIm] [HSO4] can be explained by the possibility of a loss of solvent due to a reverse reaction accelerated by the high vacuum and temperature used in the initial steps in the operation the microbalance. 5.4.5 The Effect of Changing the Anion Evidence from experimental solubility measurements, and theoretical molecular computations indicate that CO2 solubility in ILs is primarily dependent on the anion side of the IL [22, 23]. The most common cations Carbon Capture Performance of Seven Novel 85 investigated are [bmim] and [emim] with varying anions such as [Tf2N], [PF6], [NO3] and [DCN]. The use of 1,2,3-tris(diethylamino)cyclopropenylium [TDC] as a cation with two different anions [Tf2N] and [DCN] is discussed in this study. [Tf2N] based IL has shown better performance than [DCN] based ionic liquids due to the presence of multiple fluoro groups in [Tf2N] confirming the previous findings in literature [21]. This trend is also observed in the case of changing the anion from [HSO4] to [Tf2N] with the same cation [(CH2)4SO3HMIm]. It was found that the [Tf2N] anion had 4 times higher CO2 solubility than the [HSO4] anion [24]. The high CO2 solubility, as seen particularly with the [TF2N] anion, is ascribed to the fluoroalkyl groups in [Tf2N], which are known to be highly reactive with CO2 [21]. This might be attributed to the favorable interactions between the negative fluorine ions and the positively charged carbon in CO2 [21, 25]. Furthermore, comparing solubility data for [emim][LACTATE], reported by our group, with [emim][FAP] data reported by Althuluth et al. [26] and [emim][TF2N] as reported by Schilderman et al. [25] shows that the [FAP] anion has higher CO2 solubility at high pressures, most likely due to the presence of a large number of fluorine atoms [FAP]. However the CO2 uptake shown by the [LACTATE] cation is more pronounced at low pressure, which could be attributed to the possible reaction between the [Lactate] anion and CO2 during the absorption process, similar to the [acetate] based anion as reported by Shiflett and Yokozeki [27] with the added advantage of being more environmentally friendly due to the lower number of fluoro groups involved. The ionic liquid, [emim] [LACTATE], has a different solubility isotherm than all other ionic liquids, possessing a noticeable high CO2 solubility at low pressures, which could probably mean that [emim] [LACTATE] has both physical and chemical interactions with CO2, similar to other carboxylic anions such as [emim] [pivalate], [emim][Ac] and [emim] [benzoate] [28]. 5.4.6 Henry’s Law Constant, Enthalpy and Entropy Calculations Henry’s law constants for CO2 in the ILs are given in Table 5.4. The experimental solubility data were fitted to a polynomial and then Henry’s law constants were found by taking the slopes at low pressures. The ionic liquid with the lowest Henry’s law constant is [TDC][TF2N], which has shown the highest CO2 equilibrium capacity indicating an inversely proportional relationship between the temperature and solubility. Enthalpy and entropy values for CO2, in the studied ILs, are also reported in Table 5.4. 86 Carbon Dioxide Capture and Acid Gas Injection Table 5.4 Henry’s law constants and enthalpies and entropies of absorption for CO2 in the studied ionic liquids [17, 18]. H (bar) Ionic liquids 313.15 K 323.15 K 333.15 K ∆h (kJ/mol) ∆s (J/mol∙K) [Emmp][TF2N] 36.1 53.0 61.0 –22.8 –69.4 [PMPY][TF2N] 43.7 52.1 60.1 –13.8 –42.9 [TDC][TF2N] 37.2 43.4 49.3 –12.2 –37.8 [TDC][DCN] 57.2 66.3 77.3 –13.0 –40.4 [EMIM][LACTATE] 46.2 54.4 64.8 –14.6 –45.7 [(CH2)4SO3HMIm] [TF2N] 58.8 70.9 – –15.7 –49.5 [(CH2)4SO3HMIm] [HSO4] 274 301.4 – –8 –25.2 The highest heat of absorption is observed with [Emmp][TF2N] followed by [(CH2)4SO3HMIm][TF2N] and [EMIM][LACTATE] indicating strong interactions with CO2. The negative values for entropy show a higher degree of ordering as CO2 dissolved in these ILs [15]. 5.4.7 Thermodynamic Modeling of CO2 Solubility Several thermodynamic models have been proposed for modeling the equilibrium solubility of CO2 in ionic liquids. The Peng-Robinson (PR) EoS was used for the correlation of the data. The regression of the experimental data to the models was performed to obtain the interaction parameters between CO2 and the ionic liquid at different temperatures. The average absolute deviations in percentage (AAD%) between the model estimations and the experimental data, were obtained for all the ionic liquids at the three different temperatures. AAD% 100 N Pi exp Pi calc Pi exp (5.10) where, N represents the number of equilibrium data points at each temperature, Pexp and Pcalc are the experimental equilibrium pressure and the calculated pressure, respectively. Table 5.5 summarizes the AAD% obtained correlating the different IL systems averaged for the three temperatures. Carbon Capture Performance of Seven Novel 87 Table 5.5 Standard deviations PR-EoS for the ionic liquids + CO2 system. Ionic liquids 5.5 AAD% [EMMP][TF2N] 2.2 [PMPY][TF2N] 2.3 [TDC][TF2N] 3.1 [TDC][DCN] 2.0 [EMIM][LACATE] 0.5 [(CH2)4SO3HMIm][TF2N]] 1.2 [(CH2)4SO3HMIm][HSO4] 1.4 Conclusion In this study, the CO2 solubility in seven novel ionic liquids is compared to the best ionic liquid in the literature. CO2 solubility decreased in the following order: [TCD][TF2N] > [PMPY][TF2N] > [EMMP][TF2N] > [emim][LACTATE] > [TCD][DCN] > [(CH2)4SO3HMIm][TF2N] > [(CH2)4SO3HMIm] [HSO4]. [EMIM][LACATE] showed the high capacity for CO2 but both the solubility curve shapes and the difficulty of the EoS to correlate the data suggest that the interaction with CO2 is much more than just a simple physical absorption. [EMMP][TF2N] seems promising as it showed similar solubility trends to some ionic liquids that are well known for their high solubility, such as [hmim][TF2N]. Both ILs are similar and show high capacity for CO2 absorption due to a high degree of fluorination. However, high CO2 capacities were not found in the case of [TCD][TF2N], [PMPY][TF2N], [EMMP][TF2N] when compared to [bmim][Ac], for example, as this IL was shown to react chemically with CO2 to form a chemical intermediary product which is responsible for its high CO2 solubility. The most promising ionic liquid among the seven ionic liquids investigated is [TCD][TF2N], which is a propenylium based ionic liquid paired with the well-known [Tf2N] anion, in addition to the three nitrogen atoms in the [TCD] structure with two ethyl groups attached to each nitrogen. The effect of fluorination and presence of S=O groups on the TF2N anion act synergistically to increase the CO2 solubility by increasing the CO2-philicity of molecules due to Lewis base-Lewis acid interactions with the carbon 88 Carbon Dioxide Capture and Acid Gas Injection atom from CO2. However, the use of fluorinated ionic liquids is not without its environmental and health drawbacks when used in high concentrations. Acknowledgements The authors acknowledge the support of the Acid Gas Removal Laboratory at the University of Regina as well as the FGSR at the University of Regina. In addition, the first author would like to thank the CBIE (Canadian Bureau of International Education) and the Libyan government for a graduate scholarship. References 1. G.M. Crawley, Fossil Fuels: Current Status and Future Directions, World Scientific, 2016. 2. IPCC, Climate Change 2014–Impacts, Adaptation and Vulnerability: Regional Aspects, Cambridge University Press, 2014. 3. N.J. Rosenberg, W.E. Easterling III, P.R. Crosson, J. Darmstadter, Greenhouse warming: Abatement and adaptation, Routledge, 2016. 4. G.T. Rochelle, Amine scrubbing for CO2 capture, Science, 325, pp.1652–1654, 2009. 5. S. Babamohammadi, A. Shamiri, M.K. Aroua, A review of CO2 capture by absorption in ionic liquid-based solvents, Reviews in Chemical Engineering, 31, pp. 383–412, 2015. 6. X. Zhang, X. Zhang, H. Dong, Z. Zhao, S. Zhang, Y. Huang, Carbon capture with ionic liquids: overview and progress, Energy & Environmental Science, 5, pp. 6668–6681, 2012. 7. M. Ramdin, T.W. de Loos, T.J. Vlugt, State-of-the-art of CO2 capture with ionic liquids, Industrial & Engineering Chemistry Research, 51, pp. 8149–8177, 2012. 8. F. Yan, M. Lartey, K. Damodaran, E. Albenze, R.L. Thompson, J. Kim, M. Haranczyk, H.B. Nulwala, D.R. Luebke, B. Smit, Understanding the effect of side groups in ionic liquids on carbon-capture properties: a combined experimental and theoretical effort, Physical Chemistry Chemical Physics, 15, pp. 3264–3272, 2013. 9. C. Wang, H. Luo, H. Li, X. Zhu, B. Yu, S. Dai, Tuning the physicochemical properties of diverse phenolic ionic liquids for equimolar CO2 capture by the substituent on the anion, Chemistry–A European Journal, 18, pp. 2153–2160, 2012. 10. M. Hasib-ur-Rahman, M. Siaj, F. Larachi, Ionic liquids for CO2 capture— development and progress, Chemical Engineering and Processing: Process Intensification, 49, pp. 313–322, 2010. Carbon Capture Performance of Seven Novel 89 11. E. Torralba-Calleja, J. Skinner, D. Gutiérrez-Tauste, CO2 capture in ionic liquids: a review of solubilities and experimental methods, Journal of Chemistry, 2013, pp. 1–16, 2013. 12. S.D. Kenarsari, D. Yang, G. Jiang, S. Zhang, J. Wang, A.G. Russell, Q. Wei, M. Fan, Review of recent advances in carbon dioxide separation and capture, Rsc Advances, 3, pp. 22739–22773, 2013. 13. J.M. Prausnitz, R.N. Lichtenthaler, E.G. de Azevedo, Molecular thermodynamics of fluid-phase equilibria, Pearson Education, 1998. 14. D.-Y. Peng, D.B. Robinson, A new two-constant equation of state, Industrial & Engineering Chemistry Fundamentals, 15, pp. 59–64, 1976. 15. M.B. Shiflett, A. Yokozeki, Solubilities and diffusivities of carbon dioxide in ionic liquids:[bmim][PF6] and [bmim][BF4], Industrial & Engineering Chemistry Research, 44, pp. 4453–4464, 2005. 16. J.L. Anthony, E.J. Maginn, J.F. Brennecke, Solubilities and thermodynamic properties of gases in the ionic liquid 1-n-butyl-3-methylimidazolium hexafluorophosphate, The Journal of Physical Chemistry B, 106, pp. 7315–7320, 2002. 17. M. Zoubeik, M. Mohamedali, A. Henni, Experimental solubility and thermodynamic modeling of CO2 in four new imidazolium and pyridinium-based ionic liquids, Fluid Phase Equilibria, 419, pp. 67–74, 2016. 18. M. Zoubeik, A. Henni, Experimental and thermodynamic study of CO2 solubility in promising [TF2N and DCN] ionic liquids, Fluid Phase Equilibria, 376, pp. 22–30. 2014. 19. H. Tokuda, K. Hayamizu, K. Ishii, M.A.B.H. Susan, M. Watanabe, Physicochemical properties and structures of room temperature ionic liquids. 2. Variation of alkyl chain length in imidazolium cation, The Journal of Physical Chemistry B, 109, pp. 6103–6110, 2005. 20. M.J. Muldoon, S.N. Aki, J.L. Anderson, J.K. Dixon, J.F. Brennecke, Improving carbon dioxide solubility in ionic liquids, The Journal of Physical Chemistry B, 111, pp. 9001–9009, 2007. 21. S.N. Aki, B.R. Mellein, E.M. Saurer, J.F. Brennecke, High-pressure phase behavior of carbon dioxide with imidazolium-based ionic liquids, The Journal of Physical Chemistry B, 108, pp. 20355–20365, 2004. 22. J.L. Anthony, J.L. Anderson, E.J. Maginn, J.F. Brennecke, Anion effects on gas solubility in ionic liquids, The Journal of Physical Chemistry B, 109, pp. 6366–6374, 2005. 23. S. Seo, M. Quiroz-Guzman, M.A. DeSilva, T.B. Lee, Y. Huang, B.F. Goodrich, W.F. Schneider, J.F. Brennecke, Chemically tunable ionic liquids with aprotic heterocyclic anion (AHA) for CO2 capture, The Journal of Physical Chemistry B, 118, pp. 5740–5751, 2014. 24. C. Cadena, J.L. Anthony, J.K. Shah, T.I. Morrow, J.F. Brennecke, E.J. Maginn, Why is CO2 so soluble in imidazolium-based ionic liquids?, Journal of the American Chemical Society, 126 pp. 5300–5308, 2004. 90 Carbon Dioxide Capture and Acid Gas Injection 25. A.M. Schilderman, S. Raeissi, C.J. Peters, Solubility of carbon dioxide in the ionic liquid 1-ethyl-3-methylimidazolium bis (trifluoromethylsulfonyl) imide, Fluid Phase Equilibria, 260, pp. 19–22, 2007. 26. M. Althuluth, M.T. Mota-Martinez, M.C. Kroon, C.J. Peters, Solubility of carbon dioxide in the ionic liquid 1-ethyl-3-methylimidazolium tris (pentafluoroethyl) trifluorophosphate, Journal of Chemical & Engineering Data, 57, pp. 3422–3425, 2012. 27. M.B. Shiflett, A. Yokozeki, Phase behavior of carbon dioxide in ionic liq­ uids:[emim][acetate],[emim][trifluoroacetate], and [emim][acetate]+[emim] [trifluoroacetate] mixtures, Journal of Chemical & Engineer­ing Data, 54, pp. 108–114, 2008. 28. J. Blath, N. Deubler, T. Hirth, T. Schiestel, Chemisorption of carbon dioxide in imidazolium based ionic liquids with carboxylic anions, Chemical Engineering Journal, 181, pp. 152–158, 2012. 29. M.H. Al-Rashed, K.H. Alkhaldi, M.S. Al-Tuwaim, M.S. Fandary, A.S. Al-Jimaz, Extraction of butylbenzene from dodecane using hexafluorophosphate-based ionic liquids: Effect of cation change, Journal of Chemical & Engineering Data, 57, pp. 2907–2914, 2012. 6 Vitrisol® a 100% Selective Process for H2S Removal in the Presence of CO2 W.N. Wermink, N. Ramachandran, and G.F. Versteeg PROCEDE Gastreating, Enschede, The Netherlands Abstract Over recent years PROCEDE developed a solvent called Vitrisol that is 100% selective for H2S removal from industrial gases in the presence of CO2. Examples of possible applications are the removal of H2S from biogas, FPSO and associated gas. Vitrisol is able to remove in one stage more than 99.9+% of the H2S present in the gas phase and has the typical characteristics of very selective H2S scavengers. However, a major difference of Vitrisol compared to the traditional scavengers is that Vitrisol can be completely regenerated, resulting in a solvent with fully restored activity and crystalline sulphur. For the absorption process required for the removal of H2S, there is no real process conditional constraint and the operating pressures and temperatures can vary at least between 0.1–10 MPa and 283–363 K, respectively. The Vitrisol regeneration process takes place at temperatures below 373 K. The process pressure can vary from atmospheric up to 0.5–1 MPa. The Vitrisol process can be described with the following overall reaction equation: H2S + 0.5O2 H2O + So(s) In the Vitrisol process no vast amounts of energy are required for the regeneration of the solvent; therefore this process has an extremely low energy footprint. In the present contribution the performance of Vitrisol will be demonstrated for applications in shale gas production with typical compositions of 10–1000 ppmv of H2S and CO2 of about 1–10 vol.%. The Vitrisol process is also compared to a standard amine treating process designed for selective H2S removal. From the results it can be concluded that significant reductions can be achieved by using the Vitrisol process for operational costs as depicted in the energy consumptions Ying Wu, John J. Carroll and Weiyao Zhu (eds.) CO2 Capture and Acid Gas Injection, (91–126) 2017 © Scrivener Publishing LLC 91 92 Carbon Dioxide Capture and Acid Gas Injection of the overall process. As the costs of energy (and cooling) are extremely location dependent, no attempt was made to quantify the capital savings. Also it must be noted that contrary to the amine processes for the Vitrisol process, no additional treatment of the off-gas stream is required as the H2S is directly converted to crystalline sulphur and CO2 that can be emitted to the environment. Moreover, this study also illustrates clearly that it is advantageous to first remove H2S from a gas stream containing both H2S and CO2 prior to CO2 removal to reduce operational costs. 6.1 Introduction Hydrogen sulphide (H2S) is a highly toxic and corrosive gas. Removal of H2S from acidic gas streams, such as natural gas, industrial gas or biogas, is important for safety, health, environmental and economic reasons. Several regenerative and non-regenerative H2S removal processes are readily available, which are economically viable only for specific gas compositions and gas flow rates. Apart from non-regenerative H2S removal by the use of, e.g., adsorbents, all the regenerative aqueous liquid redox desulphurization processes (e.g., THIOPAQ, LO-CAT, SulFerox) capture CO2 to varying extents besides H2S. The conventional method of removing H2S from natural gas is using an amine process. Subsequently, the H2S in the stripper gas is converted to elemental sulphur by a consecutive Claus process. For natural gas fields, usually containing more CO2 than H2S, this will result in an inlet acid gas stream for the Claus process that is low in H2S and high in CO2 content. The inlet gas stream should contain at least 20 mol% of H2S to be able to produce a stable flame in a Claus furnace. Modification of the Claus process is needed between 20 and 50 mol% H2S in the inlet acid gas stream. Above 50 mol% H2S content no modification of the Claus process is required [1, 2]. Moreover, owing to the co-absorption of CO2 the regeneration costs of the amine process are substantially increased. The Vitrisol process [3] is a recently developed selective desulphurization process based on the removal of H2S by precipitation with copper sulphate (CuSO4) in an aqueous, acidic solution. Copper sulphide (CuS) and sulphuric acid are formed in the gas treating process [4–6]: H2S(g) + Cu2+ + SO42– + 2H2O + CuS(s) + 2H3O+ + SO42–(6.1) The Vitrisol process is able to remove H2S from acidic gas streams without the co-absorption of CO2 [5, 7]. Because the precipitation reaction Vitrisol a 100% Selective Process for H2S Removal 93 occurs rapidly, the removal of H2S is limited by mass transfer in the gas phase. A Vitrisol pilot absorber was built to remove H2S from biogas, obtain representative samples of CuS and to verify design rules. Operational boundary conditions were determined with respect to continuous operation in the absorber and batch-wise operation of the absorption liquid. The current status of the Vitrisol process is scavenger-like application. Cu2+, the active compound in the absorption liquid, becomes depleted during H2S removal. It must be noted, however, that nowadays copper is an expensive commodity; therefore increasing amounts of H2S lead to increasing operational costs. In order to reduce the operational costs for large amounts of H2S and/or large-scale applications, a regeneration step was developed to replenish Cu2+. The regeneration step is based on an operation encountered in copper ore processing, i.e., the dissolution of CuS with ferric sulphate (Fe2(SO4)3) [8, 9]. Copper sulphate, elemental sulfur (S°) and ferrous sulphate (FeSO4) are produced in this process: Fe(SO4)3 + CuS(s) to: 2FeSO4 + CuSO4 + S°(s) (6.2) Ferrous sulphate can be reoxidized to ferric sulphate with O2 according 4FeSO4 + 2H2SO4 + O2 2Fe2(SO4)3 + 2H2O(6.3) Resulting in the overall net reaction for the removal of H2S: H2S + 0.5O2 S°(s) + H2O(6.4) For the development of the regeneration process, the reaction behavior of the parallel reactions occurring during the dissolution of CuS, i.e., Reactions 2 and 3, respectively, were investigated. Wermink and Versteeg [7, 10] studied the oxidation of ferrous ions in acidic sulphate solutions (Reaction 3), and proposed kinetic equations derived by using both the data obtained for the initial reaction rates and the experimentally determined Fe2+ concentration profiles, respectively. Furthermore, Wermink and Versteeg [11] investigated the behavior of the oxidation of ferrous ions in acidic sulphate solution, in the presence of Cu2+. It was concluded that Cu2+ enhanced the oxidation rate of Fe2+; however, some of the experiments were affected by the rate of mass transfer of oxygen. Besides Fe2+ oxidation, Wermink and Versteeg [7, 12] studied the dissolution reaction of CuS with Fe3+ (Reaction 2). Representative samples of 94 Carbon Dioxide Capture and Acid Gas Injection CuS, obtained from VitrisoI pilot absorber operations [13], were used in the study. It was concluded that an increase in temperature increased the rate of dissolution. Full conversion of CuS could be obtained, independent of temperature. From the above-mentioned investigations it was concluded that relatively mild conditions are required for the regeneration process, i.e., temperatures below 373 K and at pressures ranging from atmospheric to 1 MPa. The conditions required in the regeneration process are case dependent, e.g., on the amount of H2S to be removed. In order to demonstrate the applicability of the Vitrisol technology, two (conceptual) process designs of the Vitrisol process have been compared to standard amine treating processes of cases previously published by Weiland and Hatcher [14]. 6.2 Case Definition The cases used to evaluate the process designs of the Vitrisol process and standard amine treating processes are examples of shale gas from British Columbia and an example of one of the gas plants built to process gas from fields in the Barnett shale, as previously published by Weiland and Hatcher [14] (see Table 6.1). In the simulations of the present study, the gases were considered to be saturated with water. Table 6.1 Case Specifications. Case 1 Case 2 Gas British Columbia shale Barnett shale H2S (ppmv) 26 750 CO2 (vol%) 1.1 2.5 CH4 (vol%) balance balance T (°C) 31.8 32.2 P (MPa) 3.10 6.62 Flow (MMSCFD) 90 330 H2S removed (kg/h) 3.35 417 Vitrisol a 100% Selective Process for H2S Removal 95 In both cases the gases should be treated to pipeline quality, i.e., 4 ppmv H2S and below 2 vol.% of CO2, according to Weiland and Hatcher [14]. For Case 1, the only compound required to be removed is H2S, because CO2 is already below pipeline specifications. Therefore it is desirable to select a process with the lowest possible removal of CO2. In Case 2, both H2S and CO2 need to be treated to reach pipeline specifications. The Vitrisol process is not able to remove CO2; therefore additional processing is required for Case 2 to obtain on spec gas. For this purpose also an amine treating process is chosen. 6.3 6.3.1 “Amine-Treated” Cases by PPS Introduction to PPS Alkanolamines have been widely used for more than 80 years in the gas treating industry, i.e., petrochemicals, refineries, natural gas processing. Recently, formulated amines that comprise a promoter have been incorporated in gas treating and in large-scale post-combustion CO2 capture. The acid gas treating industry mainly consists of processes where one or more gaseous components are transferred from the gas phase to the liquid phase followed by a chemical reaction. Due to the complexity of the absorption processes, modeling them requires very precise knowledge of reaction ­kinetics, mass transfer, thermodynamics and physical ­properties. In ­addition to the development of rigorous models that account for the aforesaid phenomena, it is also important to incorporate the correct description of vapor-liquid and liquid phase chemical equilibria including the speciation of the various components. A steady-state rate based flowsheeting software for the simulation of the acid gas treating processes has been developed by Procede.15 The flowsheeting tool has models that can do the design, optimization and analysis of acid gas treating processes including both pre- and post-combustion CO2 capture, respectively. The process simulator consists of a user-friendly graphical user interface (GUI) and a powerful numerical solver that can handle rigorous simultaneous solutions of thermodynamics, kinetics, rate-based mass transfer equations (also known as rate-based model) and supports all unit operations involved in gas treating such as absorbers, strippers, flash drums, heaters, pumps, compressors, mixers and splitters as well as work flow tools such as automatic water and solvent makeup calculators. 96 Carbon Dioxide Capture and Acid Gas Injection The Procede Process Simulator (PPS) has extensive, carefully evaluated databases of thermodynamic parameters, interaction coefficients, kinetics that have been optimized to accurately predict vapor liquid equilibrium (VLE), thermodynamic and physical properties and kinetically enhanced mass transfer (both approximate and rigorous) for amine- and mixture of amines-based capture processes. PPS is able to describe complete gas treating processes involving complex flow schemes with multiple recycle loops. Both absorber and stripper can be modeled as rate-based columns. For optimal predictions of column performances, the program includes databases of various commercially available tray types and a large collection of both dumped and structured packings; several mass transfer and hydrodynamic correlations from open literature are implemented. PPS has capabilities where users can include detailed characterization of proprietary amines, mixtures of amines, mixtures of amine and physical solvents obtained from experiments used for the development of new gas treating processes. 6.3.2 Process Description The pipeline specifications of shale gas treating is to remove H2S to <4 ppmv and CO2 down to 2.0 vol.%. As both acidic gases will be absorbed simultaneously, usually excessive amounts of CO2 are removed. The major challenge is to treat the stripper gas, which also has considerable amounts of CO2 which might be of inferior quality for a Claus plant.14 It is important to note that the H2S absorption process is usually mainly gas-phase resistance controlled and CO2 absorption is liquid-phase resistance controlled. Traditionally N-methyldiethanolamine (MDEA) has been the solvent of choice in terms of cost and effectiveness. The gas treating plant consisted a/o of an absorber, stripper, lean-rich heat exchanger, lean solvent cooler, reboiler, condenser and pumps. The absorber and stripper were modeled using the Higbiepenetration model for mass transfer. Structured packings were used in the absorber and the stripper and the absorber section is operated counter currently. Aqueous 50 wt.% MDEA solution was selected as the solvent, for Case 2a and a stripper promoter was added. For Case 1, the solvent circulations rate was 20 m3/h and for Case 2 and Case 2a, it was 100 m3/h. The dimensions of the absorber and stripper columns are presented in Table 6.2; structured packing Sulzer MellaPak Plus with a geometric surface of a = 250 m2/m3 was used. For the calculations of the mass transfer parameters the correlations of Brunazzi and Paglianti [16] for the specific area, Bravo et al. [17] for the liquid phase mass transfer coefficient and Bravo et al. [18] for the gas phase mass transfer coefficient, respectively. Vitrisol a 100% Selective Process for H2S Removal 97 Table 6.2 Case specifications. Case 1 Case 2 Case 2a British Columbia shale Barnett shale Barnett shale Absorber Packing Sulzer MellaPak Plus Sulzer MellaPak Plus Sulzer MellaPak Plus Height (m) 12 25 25 Diameter (m) 1 1.35 1.35 Sulzer MellaPak Plus Sulzer MellaPak Plus Sulzer MellaPak Plus 3 10 10 1.22 1.5 1.5 Stripper Packing Height (m) Diameter (m) 6.3.3 PFD 6.3.4 Results 6.3.4.1 Case 1 For the process conditions given in Table 6.1 and dimensions given in Table 6.2 the H2S-specification could be met. As a consequence of the H2S removal also CO2 was (partly) removed from 1.10 vol.% to 0.98 vol.%. The stream with a flow of 128 Nm3/h leaving the stripper (stream 6 in Figure 6.1) consists of 97 vol.% CO2 and 1.75 vol.% H2S and the stream that needs further processing. The H&M balances of Case 1 are given in Appendix A. Next, the utility consumptions were calculated. For pumps, an e­ fficiency of 80% was assumed. Heating can be performed with low-pressure steam. Cooling is performed with either air (above 40 °C) or water (below 40 °C). For the heat integration lean-rich heat exchanger (E-101), a temperature approach 5 °C was assumed. Though MDEA has a low volatility, some MDEA make-up is required. The utility consumptions are given in Table 6.3. 6.3.4.2 Case 2 For the process conditions given in Table 6.1 and dimensions given in ­Table 6.2, both the H2S-and CO2-specification could not be met. Therefore 98 Carbon Dioxide Capture and Acid Gas Injection 4 Gas outlet CO2 6 10 Formulator-1 16 5 13 7 8 C-102 P-101 15 C-101 14 E-101 1 Gas inlet Flash gas 9 12 11 18 V-101 3 Pump-1 Figure 6.1 PFD of the amine process with regeneration. Table 6.3 Utility consumptions of cases 1 and 2 of the amine process. Case 1 Case 2a British Columbia shale Barnett shale Electrical power (kW) 22.4 239 Heating duty (kW) 569a 5191a Cooling duty (kW) 371b, 0c 2892b, 0c 0.081 0.86 MDEA (kg/h) Reboiler duty Cooling duty above 40 °C c Cooling duty below 40 °C a b a stripping promoter has been added. In this particular case, H3PO4 with a concentration of 0.3 wt.% was used. It turned out that meeting the H2S-specification was the process limiting step, in the simulations an ­outlet concentration was attained of 3.92 ppmv. The CO2-specification of 2.0 vol.% was easily realized, the simulations gave an outlet concentration Vitrisol a 100% Selective Process for H2S Removal 99 of 1.83 vol.%, so about 35% more CO2 was removed that was demanded. The stream (gas) with a flow of 2175 Nm3/h leaving the stripper consists of 90.1 vol.% CO2 and 9.9 vol.% H2S, a gas outlet stream that certainly demands further processing. The H&M balances of Case 2a are given in Appendix B. The utility consumptions are also given in Table 6.3. It can be concluded that amine gas treating processes can be used for the upgrading of Cases 1 and 2 to meet the desired specifications. However, the obtained off-gases need further processing. The composition of the offgases is such that treating with a Claus-operation is not feasible. 6.4 Vitrisol Process Extended with Regeneration of Active Component First an in-depth description of the Vitrisol process will be provided, which is necessary to understand process designs, the H&M balances and equipment designs for Cases 1 and 2, respectively. 6.4.1 Technology Description For medium- and large-scale H2S removal operations, as e.g., the described shale gas Cases in Table 6.1, the Vitrisol process will be operated with a regeneration section in order to minimize operational costs (i.e., the consumption of copper). The Vitrisol process with regeneration basically consists of three steps, i.e., the absorption section, the regeneration section, and the sulphur recovery section, respectively. 6.4.2 Parameters Determining the Process Boundary Conditions The parameters that have a substantial impact on the process design of the Vitrisol process with regeneration because of a/o issues with stable operation are: 1. The wt.% of copper sulphide (CuS) in the aqueous absorption liquid exiting the absorber, 2. The concentration of copper sulphate (CuSO4) in the aqueous absorption liquid exiting the absorber. From experiments performed with the previously mentioned Vitrisol pilot absorber13 it was concluded that the rheological and hydraulic 100 Carbon Dioxide Capture and Acid Gas Injection behavior of the aqueous solution abruptly changed above concentrations of 3 wt.% to 5 wt.% of CuS. Furthermore, it was observed that the addition of Fe2+ to the absorption liquid (containing CuSO4 and H2SO4) did not affect the H2S removal efficiency in the absorber. In the conceptual design exit concentrations of CuS of 0.25 wt.% and 1.0 wt.% were chosen for Cases 1 and 2 respectively. Flexibility and turndown options are introduced into the design with respect to operation by not designing the process near the maximum CuS concentration. E.g.: 1. Sudden spikes in H2S concentration will not affect the operability of the process significantly 2. A consistent, temporary increase in H2S concentration of the gas in time will not require significant alteration of the installed hardware (a phenomena often encountered with, e.g., shale gas) From the work by Ter Maat et al. [4–6] it can be concluded that the reaction rate of H2S with Cu2+ is instantaneous compared to mass transfer rate, even at very low concentrations of Cu2+. Therefore, the removal of H2S is not a function of the concentration of Cu2+ in the absorption liquid. Moreover, from the work by Wermink and Versteeg [7, 10, 11] regarding Fe2+ oxidation experiments performed in the presence of Cu2+ it ­followed that the presence of Cu2+ increased the conversion rate of Fe2+ with O2 to Fe3+ substantially compared to Fe2+ oxidation experiments ­performed ­without Cu2+. From the Fe2+ oxidation experiments performed in the presence of Cu2+ it was concluded that the presence of sulphuric acid (H2SO4) did not affect the conversion rate of Fe2+ to Fe3+, and that an increase in Fe2+ concentration increased the conversion rate of Fe2+ to Fe3+ more significantly than an increase in Cu2+ concentration. From the experimental work it was concluded that the dissolution of CuS, in the presence of only Fe2+, Cu2+, H2SO4, and O2 (no Fe3+ initially), ­followed an initial conversion rate of CuS equal to the conversion rate of Fe2+ with O2 in the presence of Cu2+ up to roughly 80% conversion. Therefore it was concluded that: 1. Up to a CuS conversion of roughly 80% the oxidation of Fe2+ is the rate determining step in a batch reactor 2. It is desirable to increase the Fe2+ concentration instead of the Cu2+ concentration to obtain a higher conversion rate of Fe2+ in the oxidation step of the process Vitrisol a 100% Selective Process for H2S Removal 101 Concentrations of CuSO4, FeSO4 and H2SO4 of 0.05 M, 0.75 M and 0.1 M were chosen, respectively, for the absorption liquid exiting the absorber. 6.4.3 Absorption Section The absorption of H2S is completely limited by mass transfer in the gas phase and therefore first order with respect to the removal of H2S. This implies that most of the CuS solids are precipitating in the first section of a G/L contactor. Therefore in the selection of the absorber type a combination has been chosen of a packed bed and bubble column. Deep removal is realized in the packed bed section while the gas-liquid disengagement zone below the packed bed section is designed as a bubble column. This bubble column section also acts as a liquid storage vessel. The bubble column section was designed based on: 1. The specific interfacial areas of the bubble column for Cases 1 and 2 were determined to be a = 190 m2/m3 and a = 250 m2/m3, respectively, see Oyevaar [19]. 2. The gas hold-ups of the bubble column for Cases 1 and 2 were determined to be = 0.35 and = 0.37, respectively, see Oyevaar [19]. 3. The gas phase mass transfer coefficient was calculated using the approach of Colombet [20]; however, from this work it is not evident how the effect of process pressure must be taken into account as the gas phase diffusion coefficient is inversely proportional to the pressure. 4. The gas phase mass transfer coefficient was corrected for the pressure in two manners, i.e., according to the film theory and Higbie’s penetration theory, respectively. According to film theory, the gas phase diffusion coefficient is inversely proportional with pressure, whereas the gas phase diffusion coefficient is inversely proportional with the root of pressure according to penetration theory. 5. The actual superficial velocity of the gas in the bubble column was set to vs,G = 0.15 m.s–1. Although the absorption of H2S into the Vitrisol liquid is irreversible and therefore no real preference exists for counter-current operation, this mode has been selected in the present designs. 102 Carbon Dioxide Capture and Acid Gas Injection The absorber column was designed based on: 1. As packing 1-inch Pall rings were selected. The effective interfacial area was calculated according to the correlations of Billet and Schultes [12]. 2. The gas phase mass transfer coefficient was calculated according to the correlations of Onda et al. [22]. 3. The absorber operation was designed to reach H2S pipeline specification independent of the bubble column operation. In Case 2, the pressure of the Vitrisol liquid exiting the absorber is flashed to the pressure in the oxidizer reactor. The flash vessel is designed for a residence time of half an hour of the liquid. The flash vessel is stirred to maintain a homogeneous solid-liquid solution. In Case 1 the flash vessel is omitted, because CH4 losses are negligible. Prior to entry in the oxidizer, heat is exchanged between the Vitrisol liquid leaving the absorption section and regenerated Vitrisol liquid entering this section. 6.4.4 Regeneration Section From the work by Wermink and Versteeg12 it was concluded that at all the temperatures investigated, i.e., temperatures ranging from 25 °C to 90 °C, a conversion of CuS of approximately 100% could be obtained. Moreover, with temperature the rate of dissolution of CuS increased. For the components Fe3+ and H2SO4 both zero order dependencies were observed. Based on CuS dissolution experiments at 90 °C, in the presence of CuS, FeSO4 and O2, it was concluded that the first 80% conversion of CuS occurred at a rate identical to the initial conversion rate of Fe2+ with O2 in the presence of Cu2+. The latter 20% could be fully converted, but required an additional residence time of approximately one hour. Furthermore, at a level of 80% conversion of CuS sufficient amounts of Fe3+ were produced to convert the remaining 20% of the CuS. This was independently confirmed experimentally in which the O2 supply was stopped at a conversion level of 80%. Therefore it was concluded that continuous air flow was not required to fully dissolve the CuS and retain the Cu2+. From oxidation experiments of Fe2+ with oxygen in the presence of Cu2+ it was concluded that some of the performed experiments were affected by mass transfer of O2 in the liquid.11 Therefore intrinsic kinetics could not be determined exactly. As the laboratory scale reactor was designed to have very high mass transfer rates, it can be concluded that for process scale Vitrisol a 100% Selective Process for H2S Removal 103 types, with substantially lower kL and a values, the mass transfer of O2 to the liquid is the rate determining step of the CuS dissolution. Therefore the design of this unit is based on the absorption rate of O2. The dissolution process of CuS will be carried out in three steps; i.e., an oxidation step, an extraction step and a candle filter operation, respectively. In the oxidation step a continuous pressurized air flow is fed to a gasliquid contactor with high-intensity stirring at a temperature of 90 °C and a pressure of 1 MPa. The heat generated by the compressor can be fully integrated in the Vitrisol process. An excess amount of O2 is present to ensure a stable O2 partial pressure. For compressors, pumps and stirrers an efficiency of 80% was assumed. Build-up of water can occur in the process, as the overall reaction is given by: H2S + 0.5S2 H2O + So(S) Preferably, the water produced in the regeneration section is evaporated in the absorber. However, for Case 1 the liquid flow is relatively low compared to the gas flow, resulting in a negligible temperature increase of the gas and therefore no water evaporation. Consequently, for Case 1 the compressor was designed to provide a larger flow to both ensure a stable O2 pressure and evaporate sufficient amounts of H2O. For Case 2 the process has been designed to evaporate the excess water in the absorber; therefore the compressor is designed to only ensure a stable O2 pressure. The basis of design of the oxidizer is the relation by Van ‘t Riet23 derived for kLa in stirred vessels with ionic aqueous solutions. A power to volume ratio of the stirrer of 2000 W/m3 is selected as design parameter, resulting in a kLa of 0.15 s–1 and 0.22 s–1 for Cases 1 and 2 respectively. The solubility of O2 in the Vitrisol solution was determined with an adapted version of the model by Weisenberger and Schumpe [11, 24]. Subsequently the Vitrisol liquid enters the extraction step, i.e., a liquid-liquid contactor with high-intensity stirring also at a tWr is dissolved and the dissolution of CuS proceeds in the absence of air. The amount of xylene entering the extractor is equal to 90% of the maximum solubility of sulphur in xylene between a temperature of 50 °C and 90 °C to be able to dissolve an amount of sulphur equal to the sulphur present in the CuS formed in the absorber. The residence time in the liquid-liquid contactor of τ = 22.5 min was determined from experiments dissolving sulphur in xylene. A residence time of approximately 1 h for full conversion of CuS is not required, because the extraction step is operated continuously instead of batch-wise, resulting in increased conversion rates of CuS. A candle filter operation is placed subsequent to the extraction step 104 Carbon Dioxide Capture and Acid Gas Injection to prevent possible slip of unconverted CuS to the crystallization section, because modes of operation could be possible that result in a conversion of CuS of below 100%. Subsequent to the candle filter operation the solids-free Vitrisol liquid and xylene enter a settler to separate both phases. A residence time of 15 min is selected to separate both phases. The Vitrisol liquid is returned to the absorber and xylene, partially saturated with dissolved sulphur, enters the crystallizer in the sulphur recovery section at a temperature of 90 °C. The dissolved xylene amount in the Vitrisol liquid is determined from the solubility of p-xylene in water.25 Xylene, dissolved in the Vitrisol liquid, is stripped in the absorber. Xylene concentrations in the gas leaving the absorber in Cases 1 and 2 are determined to be 2.0 ppmv and 16.6 ppmv, respectively. 6.4.5 Sulphur Recovery Section In the crystallizer the xylene with the dissolved sulphur is decreased in temperature from 90 °C to 50 °C, resulting in the formation of sulphur crystals. Sulphur solubilities and dissolution times in various organic solvents, e.g., toluene, p-xylene, m-xylene and o-xylene, were experi­mentally determined and used in the design of the crystallizer. Afterwards, the xylene containing dissolved sulphur as well as crystallized sulphur is treated in a vacuum belt filter to separate the sulphur crystals from the liquid. The dry matter content of the sulphur crystals was assumed to be 70 wt.%. Xylene with dissolved sulphur is returned to the extraction step, sulphur cake is further processed in a sulphur melting operation. Sulphur cake is heated till the boiling point of xylene in the sulphur melting operation, i.e., a temperature of 140 °C. Xylene evaporates and is returned to the extraction step after condensation. 99+% liquid Sulphur is stored in a storage tank. A make-up stream of xylene is fed to the extraction step because of xylene losses in the absorber. It should be mentioned that the sulphur recovery section could be performed differently, when continuous removal of sulphur is not required, and the amount of sulphur to be removed is rather low (like e.g., Case 1). A saturation vessel could be included in the process design, i.e., a relatively large stirred vessel containing xylene with a sulphur storage capacity equal to a couple of weeks to months of operation. A possible mode of operation would be to remove (part of) the saturated xylene, and crystallizing the sulphur from the saturated xylene in vessel(s) in contact to the surroundings. Subsequently, the lean xylene can be returned to the Vitrisol process, and the solid sulphur removed. In this type of process the use of continuous cooling and sulphur removal is not required. Vitrisol a 100% Selective Process for H2S Removal 105 6.4.6 CO2-Absorber For Case 2 the CO2 concentration is too high to meet the pipeline specifications (see Table 6.1). Therefore an additional CO2 removal technology must be applied. For the CO2 removal step also an amine-based technology is chosen. It must be noted, however, that after the Vitrisol process the H2S concentration is already 4 ppmv (or lower as will be discussed below) and therefore intrinsically no need exists to use a selective amine as MDEA. Moreover, the gas produced by the stripper is 99+% pure CO2 which can be reused or directly vented to the environment depending on the local regulations. Initially, an aqueous 50 wt.% MDEA (Case 2b) solvent has been selected for the removal of the CO2 in order to meet the specifications. Also, simulations by PPS have been carried out for an aqueous 30 wt.% MEA (Case 3) solution. For both Cases 2b and 3a, the solvent circulation rate was 55 m3/h, respectively. 6.4.7 PFD The PFD of the Vitrisol process with regeneration for Case 1 is shown in Figure 6.2. The PFD of the Vitrisol process with regeneration for Case 2 is shown in Figure 6.3a. The PFD of the CO2 removal process for Case 2 is shown in Figure 6.3b. 6.5 Results The H&M balances of the Vitrisol process for Cases 1 and 2 are given in Appendices D and E, respectively. The utility consumptions are given in Table 6.4. Because the Vitrisol process can be operated at relatively mild conditions, heating can be performed with low-pressure steam. Cooling can be performed with cooling water and/or air, because cooling duties above a temperature of 40 °C are required. Xylene make-up is required because xylene losses arise from xylene exiting the process in the absorber. Xylene exiting the absorber is not considered to be a loss, because xylene is a component frequently encountered in natural gas. Xylene concentrations in the gas leaving the absorber in Cases 1 and 2 are determined to be 2.0 ppmv and 16.6 ppmv, respectively. The utility consumptions for the removal of CO2 from the Vitrisol treated gas stream of Case 2 are given in Table 6.5b as Cases 2b and 3, 1 Gas inlet C-101 T-103 3 E-101 P-105 4 21 E-102 7 P-xylene 5 K-101 R-101 M-101 6 14 E-103 T-102 9 Air 10 Figure 6.2 PFD of the Vitrisol process with regeneration for Case 1. 2 Gas outlet Chemical makeup 8 P-104 11 19 E-105 V-101 M-102 Air 20 S-101 P-101 P-103 12 13 18 S-102 24 P-106 17 15 T-101 23 S-105 22 P-102 16 99+% sulphur E-106 E-107 S-103 S-104 E-104 106 Carbon Dioxide Capture and Acid Gas Injection 1 Gas inlet 6 4 23 E-102 9 16 R-101 M-102 P-xylene 7 K-101 8 9 E-103 T-102 Air 10 8 P-104 13 21 E-105 V-102 M-103 Air 22 20 S-101 P-101 P-103 14 15 Figure 6.3a PFD of the Vitrisol process with regeneration for Case 2 (H2S removal section). V-101 5 E-101 P-105 Flash gas 3 C-101 T-103 M-101 2 Gas outlet Chemical make up S-102 26 P-106 19 17 T-101 25 S-105 24 P-102 16 99+% sulphur E-106 E-107 S-103 S-104 E-104 Vitrisol a 100% Selective Process for H2S Removal 107 1 4 C-101 10 12 Flash gas 9 Formulator-1 Figure 6.3b PFD of the CO2 removal process for Case 2. Gas inlet Gas outlet 3 V-101 11 8 Pump-1 P-101 16 E-101 7 18 15 14 C-102 5 13 6 CO2 108 Carbon Dioxide Capture and Acid Gas Injection Vitrisol a 100% Selective Process for H2S Removal 109 Table 6.4 Utility consumptions of Cases 1 and 2 of the Vitrisol process. Case 1 Case 2 British Columbia shale Barnett shale Electrical power (kW) 11.4 784 Heating duty (kW) 18.9 520 Cooling duty (kW) a 1.4 178a 0.94 28.3 Xylene (kg/h) Cooling duty above 40 °C a Table 6.5a Absorber and stripper dimensions. Case 2b Case 3 Barnett shale Barnett shale Absorber Packing Sulzer MellaPak Plus Sulzer MellaPak Plus 25 17 1.35 1.38 Sulzer MellaPak Plus Sulzer MellaPak Plus Height (m) 10 10 Diameter (m) 1.5 1.5 Height (m) Diameter (m) Stripper Packing Table 6.5b Utility Consumptions of Cases 2b and 3 of the Amine Process. Electrical power (kW) Barnett shale Barnett shale 129 132 2140 2060a Cooling duty (kW) 607b, 0c 626b, 0c 0.79d 0.50e Reboiler duty Cooling duty above 40 °C c Cooling duty below 40 °C d MDEA e MEA b Case 3 Heating duty (kW) Amine (kg/h) a Case 2b a 110 Carbon Dioxide Capture and Acid Gas Injection respectively. From Tables 6.3 and 6.5b it can be observed that the utility consumption, especially the reboiler duty, has considerably reduced when H2S is removed upstream of the amine plant. Equipment specifications are given in Table 6.5a. From the simulations no real preference for either two amines can be made based on operational costs. The dimensions of the MEA-based are slightly smaller compared to MDEA. In the comparison the data obtained for MDEA will be used. 6.6 6.6.1 Discussion Comparison of Amine Treating Solutions to Vitrisol The total utility consumptions for processing British Columbia shale to pipeline quality with standard amine treating solutions and Vitrisol are given in Table 6.6. From Table 6.6 it can be concluded that significant reductions with respect to utility consumptions can be achieved when Vitrisol is used to remove H2S instead of a standard amine process. Furthermore, the H2S Table 6.6 Utility consumptions of cases 1 and 2 of the amine process. Standard amine process Vitrisol® process 22.4 Heating duty (kW) 569a Cooling duty (kW) 371 , 0 b 11.4 18.9 c 1.4b, 0c 0.081 Xylene (kg/h) 0.94 Xyleneabsorber gas (ppmv) 2.0 CO2,stripper gas (kg/h) 239 H2Sstripper gas (kg/h) 3.35 H2Sstripper gas (ppmv) 1.65 × 104 Reboiler duty the cooling duty above 40 °C c the cooling duty below 40 °C b Case 1 Electrical power (kW) MDEA (kg/h) a Case 1 Vitrisol a 100% Selective Process for H2S Removal 111 Table 6.7 Utility consumptions of cases 1 and 2 of the Amine process. Electrical power (kW) Case 2a Case 2 Standard amine process Vitrisol process + MDEA process 239 784 + 129 Heating duty (kW) 5191a 520 + 2140 Cooling duty (kW) 2892b, 0c 178b + 607b, 0c 0.86 0.79 MDEA (kg/h) Xylene (kg/h) 28.3 CO2,stripper gas (kg/h) 4889 3634 H2Sstripper gas (kg/h) 417 1.79 H2Sstripper gas (ppmv) 9.24 × 104 625 Reboiler duty Cooling duty above 40 °C c Cooling duty below 40 °C a b is oxidized in the Vitrisol process, nullifying additional H2S removal ­ perations downstream. E.g., the stripper gas in the amine process still o contains 1.65 × 104 ppmv of H2S, as is shown in Table 6.6. Xylene make-up is not considered to be a loss, because xylene is a component frequently encountered in natural gas. The total utility consumptions for processing Barnett shale to ­pipeline quality with standard amine treating solutions and a combination of Vitrisol with a standard amine treating solution are given in Table 6.7. From Table 6.7 it can be concluded that significant reductions with respect to utility consumptions can be achieved when Vitrisol is used to remove H2S prior to CO2 removal with a standard amine process. Furthermore, though the H2S is oxidized in the Vitrisol process, an additional H2S removal operation is required downstream to remove H2S from the amine stripper gas. As described in Table 7, the quantity of H2S to be removed from the stripper gas varies significantly between Barnett shale treated by a standard amine process for both H2S and CO2 removal and the combination of the Vitrisol process for H2S removal and standard amine process for CO2 removal. 112 Carbon Dioxide Capture and Acid Gas Injection 6.6.2 Enhanced H2S Removal of Barnett Shale Gas (case 2) As explained before, the Barnett shale gas coming from the Vitrisol process is required to be treated by a standard amine process to remove CO2. However, the feed gas for the amine absorber, containing 4 ppmv of H2S, will result in an increased H2S concentration coming from the amine stripper. If the H2S content of the feed gas for the amine absorber would be decreased below 4 ppmv, the possibility exists to produce a stripper gas which can be directly vented to the environment without additional H2S removal. The only requirement would be to provide additional packing in the Vitrisol absorber to enhance the removal of H2S. Figure 6.4 shows the amount of H2S in the gas leaving the stripper for varying of amounts of H2S in the feed gas. Figure 6.5 shows the amount of H2S in the gas entering the amine absorber and the gas leaving the amine stripper for varying packing heights in the Vitrisol absorber. From Figures 6.4 and 6.5 it can be concluded that it is advantageous to remove more H2S upstream with the Vitrisol process than is required for 2.0 700 1.8 600 H2S out (kg/h) H2S out (ppm) 500 H2Sout (kg/h) 1.4 1.2 400 1.0 300 0.8 0.6 200 0.4 100 0.2 0.0 0 0 1 2 3 4 H2Sfeed (ppm) Figure 6.4 H2S content of the gas exiting the stripper as a function of the H2S concentration in the feed gas. 5 H2Sout (ppm) 1.6 Vitrisol a 100% Selective Process for H2S Removal 113 10 1000 H2S feed H2S out 1 100 2 4 6 8 10 0.1 10 0.01 1 0.001 H2Sout (ppm) H2Sfeed (ppm) 0 0.1 Hpacking vitrisol absorber (m) Figure 6.5 H2S content of gas entering the amine absorber (feed) and of gas exiting the amine stripper (out) as a function of packing height in the Vitrisol absorber. the gas specifications as this enables an easy handling downstream of the amine stripper gas, e.g., venting CO2 to the environment. From Figure 6.5 it can be concluded that a packing height of 3.4 m is required to meet the 4 ppmv specifications, however, this will lead to a H2S concentration in the CO2 off-gas from the amine stripper of 625 ppmv. In order to arrive at a concentration of H2S of about 1 ppmv additionally about 4 m packing is required (about 4 m3). 6.7 Conclusions Over recent years PROCEDE developed a solvent called Vitrisol that is 100% selective for H2S removal from industrial gases in the presence of CO2. Vitrisol is able to remove in one stage more than 99.9+% of the H2S present in the gas phase and has the typical characteristics of very selective H2S scavengers. However, a major difference of Vitrisol compared 114 Carbon Dioxide Capture and Acid Gas Injection to the traditional scavengers is that Vitrisol can be completely regenerated, resulting in a solvent with fully restored activity and crystalline sulphur. For the absorption process required for the removal of H2S, there is no real process conditional constraint and the operating pressures and temperatures can vary at least between 0.1–10 MPa and 283–363 K, respectively. The Vitrisol regeneration process takes place at temperatures below 373 K. The process pressure can vary from atmospheric up to 0.5–1 MPa. The Vitrisol process can be described with the following overall reaction equation: H2S + 0.5 O2 H2O + S°(s) In the Vitrisol process no vast amounts of energy are required for the regeneration of the solvent; therefore this process has an extremely low energy footprint. In the present contribution the performance of Vitrisol is demonstrated for two applications in shale gas production, as previously described by Weiland and Hatcher, i.e., British Columbia gas (Case 1) and Barnett gas (Case 2) respectively. The Vitrisol process is also compared to a standard amine treating process designed for selective H2S removal. From the results it can be concluded that significant reductions can be achieved by using the Vitrisol process for the operational costs as depicted in the energy consumptions of the overall process. E.g., in Case 1 the total energy consumption of the Vitrisol process is 32 kW, whereas the total energy consumption of the standard amine treating process is 962 kW. In Case 2 the total energy consumption of the Vitrisol process in combination with a standard amine treating process is 4358 kW, whereas the total energy consumption of the standard amine treating process is 8322 kW. Moreover, no additional treatment of the off-gas stream is required for the process with Vitrisol upstream of the amine treating unit, as the H2S is directly converted to crystalline sulphur. Depending on the local governmental regulation the off-gas could be directly vented to the environment. This study also illustrates clearly that it is advantageous to first remove H2S from a gas stream containing both H2S and CO2 prior to CO2 removal to reduce operational costs. As the costs of energy (and cooling) are extremely location dependent, no attempt was made to quantify the capital savings. Vitrisol a 100% Selective Process for H2S Removal 115 6.8 a D e H k P T vs Notation interfacial area diameter hold-up height mass transfer coefficient pressure temperature superficial velocity [m2/m3] [m] [-] [m] [m.s–1] [Pa] [°C, K] [m.s–1] Subscripts and superscripts G L gas liquid References 1. Mcintyre, G. & Lyddon, L.; “Claus sulphur recovery options,” Petroleum Technology Quarterly Spring, p. 57–61, 1997. 2. Perry, D. et al.; “Better acid gas enrichment,” Sulphur, vol. 326, p. 38–42, 2010. 3. Versteeg, G.F. and Ter Maat, H.; “Method and system for selective removal of contamination from gas flows,” WO9855209, publication date 10-12-1998. 4. Maat, H. ter, “The development of a novel, selective desulfurization process,” PhD thesis, University of Twente, 2006. 5. Ter Maat, H. et al.; “The removal of hydrogen sulfide from gas streams using an aqueous metal sulfate absorbent. Part I. The absorption of hydrogen sulfide in metal sulfate solutions,” Separation and Purification Technology, vol. 43, p. 183–197, 2005. 6. Ter Maat, H. et al.; “Theoretical and experimental study of the absorption rate of H2S in CuSO4 solutions. The effect of enhancement of mass transfer by a precipitation reaction,” Trans IChemE, Part A, Chemical Engineering Research and Design, vol. 85(A1), p. 100–108, 2007. 7. Wermink, W.N. and Versteeg, G.F.; “The oxidation of Fe(II) in acidic sulphate solutions at elevated pressures. Part I. Kinetics above 1 M H2SO4,” paper submitted. 8. Peacey, J. et al.; “Copper hydrometallurgy – current status, preliminary economics, future direction and positioning versus smelting,” Trans. Nonferrous Met. Soc. China, vol. 14, no. 3, p. 560–568, 2004. 9. Dutrizac, J.E. and MacDonald, R.J.C.; “The kinetics of dissolution of covellite in acidified ferric sulphate solutions,” Canadian Metallurgical Quarterly, vol. 13, no. 3, p. 423–433, 1974. 116 Carbon Dioxide Capture and Acid Gas Injection 10. Wermink, W.N. and Versteeg, G.F.; “The oxidation of Fe(II) in acidic sulphate solutions at elevated pressures. Part II. Influence of H2SO4 and Fe(III),” paper submitted. 11. Wermink, W.N. et al.; “The oxidation of Fe(II) with Cu(II) in acidic sulphate solutions at elevated pressures,” paper submitted. 12. Wermink, W.N. and Versteeg, G.F.; “The dissolution of CuS with Fe(III) in acidic sulphate solutions,” paper submitted. 13. Ter Maat et al., “Vitrisol a new, 100% selective technology for the removal of H2S,” Proc. 8th International Conference of Sour Oil & Gas Advanced Technology (SOGAT), Abu Dhabi, UAE, 2012. 14. Weiland, R.H. and Hatcher, N.A.; “Overcome challenges in treating shale gas,” Hydrocarbon Processing, vol. 91, no. 1, p. 45–48, 2012. 15. van Elk, E.P. et al.; “A new flowsheeting tool for flue gas treating,” Energy Procedia, vol. 1, no. 1, p. 1481–1488, 2009. 16. Brunazzi, E. and Paglianti, A.; “Liquid-film mass-transfer coefficient in a column equipped with structured packings,” Ind. Eng. Chem. Res., vol. 36, p. 3792–3799, 1997. 17. Bravo, J.L. et al.; “A comprehensive model for the performance of columns containing structured packings,” IChemE Symposium Series 128, p. A439–A457, 1992. 18. Bravo, J.L. et al.; “Mass transfer in gauze packings,” Hydrocarbon Processing, vol. 64, no. 1, p. 91–95, 1985. 19. Oyevaar, M.H. and Westerterp, K.R.; “Interfacial areas and gas hold-ups in gas-liquid contactors at elevated pressures from 0.1 to 8.0 MPa,” Chemical Engineering Science, vol. 46, no. 5/6, p. 1217–1231, 1991. 20. Colombet, D. et al.; “Mass or heat transfer inside a spherical gas bubble at low to moderate Reynolds number,” International Journal of Heat and Mass Transfer, vol. 67, p. 1096– 1105, 2013. 21. Billet, R. and Schultes, M.; “Prediction of mass transfer columns with dumped and arranged packings,” Trans. AchemE, vol. 77, Part A, p. 498–504, 1999. 22. Onda, K. et al.; “Mass transfer coefficients between gas and liquid phases in packed columns,” Journal of Chemical Engineering of Japan 1, p. 56–62, 1968. 23. Van ‘t Riet, K.; “Review of measuring methods and results in nonviscous gasliquid mass transfer in stirred vessels,” Ind. Eng. Chem. Process Des. Dev., vol. 18, no. 3, p. 357–364, 1979. 24. Weisenberger, S. and Schumpe, A.; “Estimation of gas solubilities in salt solutions at temperatures from 273 K to 363 K,” AIChE Journal, vol. 42, no. 1, p. 298–300, 1996. 25. Knauss, K.G. and Copenhaver, S.A.; “The solubility of p-xylene in water as a function of temperature and pressure and calculated thermodynamic quantities,” Geochimica et Cosmochimica Acta, vol. 59, no. 12, p. 2443–2448, 1995. Vitrisol a 100% Selective Process for H2S Removal 117 Appendix 6-A: H&M Balance of Case 1 (British Columbia shale) of the Amine Process Summary Inlet Streams Stream # – 1 Stream Name – Stream From Block – Gas Inlet Pressure (absolute) bar 31.0000 Temperature °C 31.8000 Vapour fraction (molar) – 1.0000 Flow (molar) kmol/hr 4481.4208 Flow (mass) kg/hr 7.3292E+04 Flow (Volume) m3/hr 3439.7550 Carbon dioxide mol% 1.0977 Hydrogen sulfide mol% 2.5946E-03 Water mol% 0.2078 MDEA mol% 0.0000 Nitrogen mol% 0.0000 Methane mol% 98.6919 Ethane mol% 0.0000 Propane mol% 0.0000 n-Butane mol% 0.0000 n-Pentane mol% 0.0000 118 Carbon Dioxide Capture and Acid Gas Injection Outlet Streams Stream # – 4 6 9 Stream Name – Stream Stream Stream To Block – Gas Outlet CO2 Flash Gas Pressure (absolute) bar 31.0000 1.5000 1.7500 Temperature °C 36.0907 45.0000 32.2132 Vapour fraction (molar) – 1.0000 1.0000 1.0000 Flow (molar) kmol/hr 475.5915 5.9776 0.7981 Flow (mass) kg/hr 7.3045E+04 250.7511 12.9991 Flow (Volume) m3/hr 3494.9809 104.6327 11.5333 Carbon dioxide mol% 0.9773 91.1097 0.6938 Hydrogen sulfide mol% 3.9259E-04 1.6469 2.3215E-02 Water mol% 0.2201 6.4914 2.3448 MDEA mol% 1.4902E-05 1.2999E–08 7.2078E-05 Nitrogen mol% 0.0000 0.0000 0.0000 Methane mol% 98.8022 0.7520 96.9381 Ethane mol% 0.0000 0.0000 0.0000 Propane mol% 0.0000 0.0000 0.0000 n-Butane mol% 0.0000 0.0000 0.0000 n-Pentane mol% 0.0000 0.0000 0.0000 Vitrisol a 100% Selective Process for H2S Removal 119 Appendix 6-B H&M Balance of Case 2a (Barnett shale) of the Amine Process with Stripper Promoter Summary Inlet Streams Stream # – 1 Stream Name – Stream From Block – Gas Inlet Pressure (absolute) MPa 6.6200 Temperature °C 32.2000 Vapour fraction (molar) – 1.0000 Flow (molar) kmol/hr 1.6432E+04 Flow (mass) kg/hr 2.7536E+05 Flow (Volume) m3/hr 5534.3376 Carbon dioxide ppmv 2.4964E+04 Hydrogen sulfide ppmv 748.9186 Water ppmv 1441.8694 MDEA ppmv 0.0000 H3PO4 ppmv 0.0000 Nitrogen ppmv 0.0000 Methane ppmv 9.7285E+05 Ethane ppmv 0.0000 Propane ppmv 0.0000 n-Butane ppmv 0.0000 n-Pentane ppmv 0.0000 120 Carbon Dioxide Capture and Acid Gas Injection Outlet Streams Stream # – 4 6 9 Stream Name – Stream Stream Stream To Block – Gas Outlet CO2 Flash Gas Pressure (absolute) MPa 6.6200 0.1500 0.1750 Temperature °C 43.5532 45.0000 33.2011 1.0000 1.0000 1.0000 Vapour fraction (molar) – Flow (molar) kmol/hr 1.6313E+04 132.0003 6.2467 Flow (mass) kg/hr 2.7013E+05 5461.7403 113.2937 Flow (Volume) m3/hr 5806.9475 90.5366 Carbon dioxide ppmv 1.8309E+04 8.4166E+05 6.8778E+04 Hydrogen sulfide ppmv 3.9253 9.2417E+04 6841.0857 Water ppmv 2109.6663 6.4919E+04 2.3575E+04 MDEA ppmv 0.4414 1.2144E–02 0.7804 H3PO4 ppmv 0.0000 0.0000 0.0000 Nitrogen ppmv 0.00000 0.0000 0.0000 Methane ppmv 9.7958E+05 10008.3982 9.0081E+05 Ethane ppmv 0.0000 0.0000 0.0000 Propane ppmv 0.0000 0.0000 0.0000 n-Butane ppmv 0.0000 0.0000 0.0000 n-Pentane ppmv 0.0000 0.0000 0.0000 2309.9574 Vitrisol a 100% Selective Process for H2S Removal 121 Appendix 6-C H&M Balance of Case 3 (Barnett shale) of the Amine Process (MEA) Summary Inlet Streams Stream # – 1 Stream Name – Stream From Block – Gas Inlet Pressure (absolute) bar 66.2000 Temperature °C 32.2000 Vapour fraction (molar) – 1.0000 Flow (molar) kmol/hr 1.6432E+04 Flow (mass) kg/s 76.4265 Flow (Volume) m /hr 5535.7196 Carbon dioxide mol% 2.4964 Water mol% 0.1441 MEA mol% 0.0000 Methane mol% 97.3595 3 Outlet Streams Stream # – 4 6 Stream Name – Stream Stream To Block – Gas Inlet CO2 outlet Pressure (absolute) bar 66.2000 1.5000 Temperature °C 32.2000 40.0000 Vapour fraction (molar) – 1.0000 1.000 Flow (molar) kmol/hr 1.6432E+04 77.7385 Flow (mass) kg/s 76.4265 0.9216 Flow (Volume) m3/hr 5535.7196 1339.1137 Carbon dioxide mol% 2.4964 94.8935 Water mol% 0.1441 4.9996 MEA mol% 0.0000 1.3785E-08 Methane mol% 97.3595 0.1069 122 Carbon Dioxide Capture and Acid Gas Injection Formulated Streams Stream # – 18 Stream Name – Stream From Block – Formulator-1 Pressure (absolute) bar 66.2000 Temperature °C 35.9375 Vapour fraction (molar) – 0.0000 Flow (molar) kmol/hr 2507.0022 Flow (mass) kg/s 16.4953 Flow (Volume) m /hr 55.0007 Carbon dioxide mol% 3.9649 Water mol% 85.2589 MEA mol% 10.7762 Methane mol% 2.2505E-22 3 11 35 29 413 70957 2165 kg/hr kg/hr kg/hr kg/hr kg/hr kg/hr kg/hr kg/hr kg/hr kg/hr kg/hr kg/hr H2SO4 CuSO4 FeSO4 Fe2(SO4)3 H2S CuS S CH4 CO2 O2 N2 Xylene 4.0 168 264 0.9 2165 70955 9.4 0.61 45 168 n.a. n.a. 1.62 9.4 413 45 36 3308 n.a. 0.9 413 29 45 3265 1.62 1.8 3308 n.a. 0.25 kg/hr 0 [ppm] 0.25 H2O 90 L+S [H2S] 0 L 37.5 0 L+S 31.8 10 3.79 3.76 [wt.%] G 32.2 31 3.62 3.76 Conc.solids 31.8 31 3.62 3.76 7 8 9 10 35 11 407 42 36 0.5 n.a. 0 G 15 1.01 39 0.047 9 0.5 n.a. 0 G 90 10 5.0 0.047 35 9 2.5 3308 n.a. 0.11 L+S 90 10 3.79 3.80 35 413 45 36 2 n.a. 0 G 90 10 5.4 0.049 Compressed Compressed air Vitrisol air Water Air 6 G [°C] 31 3440 5 LP HT CuS rich Vitrisol Air 4 Phase Temperature 31 3440 [act m3/h] [bara] Flow Pressure 73.3 [ton/h] Flow 73.3 HP CuS rich HP lean Vitrisol Vitrisol NG outlet NG inlet 3 2 Stream number 1 n.a. L 90 1.01 0 0 Vitrisol & xylene 11 2 n.a. 0 G 90 10 54 0.049 12 55 5.1 3308 0 L 3.9 3.9 Appendix 6-D: H&M Balance of Case 1 (British Columbia shale) of the Vitrisol process Vitrisol a 100% Selective Process for H2S Removal 123 [ppm] kg/hr kg/hr kg/hr kg/hr kg/hr kg/hr kg/hr kg/hr kg/hr kg/hr kg/hr kg/hr kg/hr [H2S] H2O H2SO4 CuSO4 FeSO4 Fe2(SO4)3 H2S CuS S CH4 CO2 O2 N2 Xylene 55 5.1 413 45 36 3308 n.a. 0 90 L [°C] Temperature 1 3.9 [wt.%] [bara] Pressure Conc.solids [act m3/h] Flow 0.94 413 45 36 3308 n.a. 0 L 90 31 3.8 3.8 Vitrisol 3.9 14 13 Vitrisol & xylene Phase [ton/h] Flow Stream number 54 5.1 n.a. 0 L 90 1 0.07 0.059 Xylene 15 54 5.1 n.a. 5.5 L+S 50 1 0.07 0.059 Xylene 16 1.3 3.2 n.a. 70 S+L 50 1 3.09E-03 4.44E-03 Sulphur cake 17 19 20 21 22 52 2.0 n.a. 0 L 50 1 0.06 0.054 2.2 n.a. 0 L 15 10 2.56E-03 2.22E-03 55 2.0 n.a. 0 L 90 10 0.068 0.057 n.a. 0 L 0 0 1.3 3.2 n.a. 0 G 140 1 0.41 0.0013 Xylene Vitrisol Xylene make-up Xylene make-up Xylene 18 23 Xylene 24 n.a. 0 L 140 1 1.3 n.a. 0 L 140 1 1.75E-03 1.71E-03 3.15E-03 1.28E-03 Liquid sulphur 124 Carbon Dioxide Capture and Acid Gas Injection 275.4 [ton/h] [act m3/h] [bara] Flow Flow Pressure G 0 [wt.%] [ppm] Phase Conc.solids [H2S] kg/hr kg/hr kg/hr kg/hr kg/hr kg/hr kg/hr kg/hr CuS S CH4 CO2 O2 N2 Xylene 18050 28.3 18050 256470 256346 2.24 kg/hr Fe2(SO4)3 H2S 419 884 kg/hr 124 1170 12619 kg/hr FeSO4 99998 CuSO4 427 2287 427 kg/hr kg/hr n.a. 1 L+S 32.3 66.2 110.8 117.0 H2SO4 4 0 G 35.4 66.2 5524 3 4 5 6 7 28.3 12619 2838 1086 100074 n.a. 0 L 38.0 66.2 110.8 116.6 107.1 0.75 n.a. 0 G 32.1 10 13.4 0.108 16.7 1170 12619 884 2287 99997 n.a. 1 L+S 32.1 10 110.7 116.9 16.7 1170 12619 884 2287 99997 n.a. 1 L+S 90 10 113.4 116.9 HP CuS LP CuS LP HT rich HP lean rich CuS rich Vitrisol Vitrisol Flash gas Vitrisol Vitrisol H2O 749 32.2 Temperature [°C] 66.2 5534 NG outlet NG outlet NG inlet 274.8 2 1 Stream number 3195 979 45.0 n.a. 0 G 15 1.01 3513 4.28 Air 8 10 11 12 3195 979 45.0 n.a. 0 G 90 10 448.6 4.28 317 225 11903 2462 1086 100072 n.a. 0.46 L+S 90 10 113.5 117.0 3195 783 16.7 942 191 n.a. 0 G 90 10 457.8 4.24 n.a. 0 L 0 0 Compressed Compressed air Vitrisol air Water 9 Appendix 6-E H&M Balance of Case 2 (Barnett shale) of the Vitrisol Process 3195 783 16.7 191 n.a. G 90 1.01 4518 4.24 Air 13 Vitrisol a 100% Selective Process for H2S Removal 125 14 [°C] Temperature kg/hr kg/hr kg/hr kg/hr CO2 O2 N2 Xylene kg/hr CuS kg/hr kg/hr H2S kg/hr kg/hr Fe2(SO4)3 CH4 kg/hr FeSO4 S 2838 12619 kg/hr CuSO4 6708 639 1086 100072 kg/hr kg/hr H2SO4 n.a. H2O [H2S] 0 [wt.%] [ppm] Conc.solids L Phase 90 10 121.9 [act m /h] [bara] 3 Flow 124.0 Vitrisol & xylene [ton/h] Pressure Flow Stream number 15 16 17 6708 639 12619 2838 1086 100072 n.a. 0 L 90 1 121.9 124.0 28.3 12619 2838 1086 100072 n.a. 0 L 90 66.2 113.6 116.6 6680 639 n.a. 0 L 90 1 8.35 7.32 Vitrisol & xylene Vitrisol Xylene 6680 639 n.a. 5.5 L+S 50 1 8.19 7.32 Xylene 18 19 160 393 n.a. 70 S+L 50 1 0.38 0.55 Sulphur cake 21 6520 246 n.a. 0 L 50 1 7.80 6.77 188 n.a. 0 L 15 10 0.22 0.19 Xylene Xylene make-up 20 23 24 25 26 6708 246 n.a. 0 L 90 10 8.39 6.95 n.a. 0 L 0 0 160 n.a. 0 G 140 1 51.0 0.16 393 n.a. 0 L 140 1 0.22 0.39 160 n.a. 0 L 140 1 0.21 0.16 Vitrisol Liquid Xylene make-up Xylene sulphur Xylene 22 126 Carbon Dioxide Capture and Acid Gas Injection 7 New Amine Based Solvents for Acid Gas Removal Yohann Coulier1,2,3, Elise El Ahmar3, Jean-Yves Coxam1, 2, Elise Provost4, Didier Dalmazzone4, Patrice Paricaud4, Christophe Coquelet3 and Karine Ballerat-Busserolles1,2,3 Clermont Université, Université Blaise Pascal, Institut de Chimie de ClermontFerrand, Clermont-Ferrand, France 2 CNRS, UMR 6296, Institut de Chimie de Clermont-Ferrand, Clermont-Ferrand, France 3 MINES ParisTech, PSL – Research University, CTP – Centre of Thermodynamics of Processes, Fontainebleau, France 4 UCP, ENSTA ParisTech, Université Paris-Saclay, Palaiseau cedex, France. 1 Abstract Treatment and separation of multicomponent gases using absorption/desorption cycles in aqueous solutions is a very well-known and efficient method, used for natural gas and biogas purification or greenhouse gas mitigation. More specifically, aqueous solutions of amine are used with efficiency for CO2 removal from gas mixtures. However, the large energetic cost involved with carbon capture ­processes is a critical downside. To address this issue, a new class of amine is considered to decrease the cost of the regeneration: the demixing amines. These amines present a lower critical solution temperature that can be used with benefit in post-­ combustion processes. Precedent studies conducted on methylpiperidines in water have shown the difficulty of obtaining “ideal” absorbent systems. The aim of this chapter is to evaluate the influence of a physical absorbent, namely triethylene glycol, on the thermodynamic properties (such as liquid-liquid equilibria, vaporliquid equilibria, heat capacities, densities and heat of solutions) of aqueous solutions of demixing amine in order to design new operation units for carbon capture process. Ying Wu, John J. Carroll and Weiyao Zhu (eds.) CO2 Capture and Acid Gas Injection, (127–146) 2017 © Scrivener Publishing LLC 127 128 Carbon Dioxide Capture and Acid Gas Injection 7.1 Introduction Chemical absorption of acid gases by amine based solvents has found applications in a wide variety of industries including gas processing and the removal of CO2 from synthesis gas in the production of hydrogen or ammonia. Other applications of this technology are the purification of biogas [1] or CO2 removal from post-combustion gases in power plants [2]. The principle of the technology applied to carbon capture processes is based on selective absorption/desorption cycles of CO2 in aqueous absorbents. A schematic view of the process is shown in Figure 7.1. Although the process is well adapted and extensively used for natural gas treatment, the composition of the industrial effluents leads to a loss of energetic efficiency. CO2storage or valorisation Lean solvent purety: 99% Stripper Treated fume CO2 Elimination > 90% Heat exchanger Absorber Industrial effluent p = 1 atm/T = 40 °C 3–30% CO2 T = 100–140 °C Reboilling CO2 Rich solvent (a) CO2 CO2 lean amine Treated gas Water + amine Decanter (b) Water + amine + CO2 Water Stripper Flue gas Absorber Water + CO2 Figure 7.1 Schematic representation of the CO2 separation process [3]; (a) classical alkanolamine based absorbents; (b) demixing solvents. New Amine Based Solvents for Acid Gas Removal 129 The major problem is the cost of the regeneration step that requires a lot of energy to be efficient. It is thus necessary to adapt this process in order to reduce the energetic cost of the desorption step. Demixing solvents were proposed as an option for CO2 capture to reduce the energy consumption involved in the regeneration of the absorbent [4]. These new absorbent solutions are constituted of amines that are partially miscible with water, under specific conditions of temperature and gas loading [5]. As shown in Figure 7.1b, in the absorber, the aqueous solution of amine remains monophasic and a large quantity of CO2 is absorbed similarly to the process using monoethanolamine (MEA). By increasing the temperature in the decanter, the solution separates in two liquid phases, one amine phase containing almost no CO2, and one aqueous phase containing chemically and physically absorbed CO2. Since the solubility of the gas in the aqueous phase is smaller than in the original monophasic solution due to composition and temperature changes, the CO2 in excess desorbs from the solution while the remaining CO2 is contained in the water rich phase. As a result, only this part of the solvent is heated in the regeneration step of the separation process. The excess CO2 from the decanter and the separated CO2 from the stripper are then compressed and transported for being used or stored in safe conditions. In order to apply such a process, two important parameters have to be considered: The liquid-liquid phase separation should only occur in the decanter and has to be avoided in the absorber. For that purpose, the temperature of phase separation needs to be bigger than the maximum temperature in the absorption column. The amine rich phase should contain as little water as p ­ ossible. The process is efficient if most of the CO2 not released in the decanter remains dissolved in the aqueous phase. Thus, the amine phase to be directly recycled in the absorber. The DACOOTA project presented by Ballerat-Busserolles et al. [6] and Fandino et al. [7] deals with the understanding of thermodynamic equilibria in {amine + H2O} and {CO2 + amine + H2O} systems which exhibit partial miscibility with water. This project is simultaneously supported by the French National Agency of Research (ANR, [ANR-12-IS09-0001]) and the Natural Sciences and Engineering Research Council of Canada (NSERC). The goal of this research project is to elucidate the structure-property relationships for the potential amines under investigation, determine phase diagrams with or without dissolved CO2, develop thermodynamic models, 130 Carbon Dioxide Capture and Acid Gas Injection and evaluate the capabilities of the selected solvents for CO2 absorption. In this project, methods to determine liquid-liquid equilibria (LLE) in mixtures containing a well-controlled quantity of gas dissolved were developed in order to elucidate part of the questions concerning this process. In recent years, the addition of a physical solvent in aqueous solutions of amines was considered to optimize some steps of the process [8]. For example, in order to prevent equipment corrosion in processes of CO2 capture with aqueous amines solutions, the use of a co-solvent such as glycol has already been explored [9]. Benefits due to the replacement of a part of the water by a physical solvent are the reduction of the specific heat capacity of the absorbent, together with the decrease of amine degradation and the reduction of evaporation, lowering the cost of the separation process. In order to design new operation units for CO2 removal or to evaluate the retrofits of existing processes, it is important to investigate the thermophysical properties of the new demixing solvents containing physical co-solvents. This includes phase equilibrium measurements (vapor-liquid and liquid-liquid equilibria), as well as the study of transport and energetic properties. The knowledge of these thermophysical properties will allow the evaluation of the impact of addition of physical solvent on CO2 mass transfer. Moreover, CO2 gas stream is not pure and contain other chemicals such as N2, Ar, NOx, and SO2 in the case of post-combustion capture process or H2 and SO2 in case of pre-combustion process, and H2S and mercaptans in case of gas processing or biogas purification. The impacts of these other chemicals on the thermophysical properties and phase diagram need also to be investigated. In this work, the thermodynamic properties of a new ­ demixing solvent composed of an aqueous solution of piperidines, namely N-methylpiperidine (NMPD) or 2-methylpiperidine (2MPD), and a physical solvent, triethylene glycol (TEG), are reported. Relying on the thermodynamic representation of the process [6], the benefit of adding a co-solvent were investigated as follows: For the decantation step, the liquid-liquid equilibria of {Amine – H2O – TEG} systems with dissolved CO2 were studied. For transport properties in the lines and energy cost of the heating, densities and heat capacities of solutions were investigated at different temperatures. For solvent recycling and evaporation concerns, vapor-­ liquid equilibra (VLE) measurements for different CO2 loadings were performed on {Amine – H2O – TEG} systems. New Amine Based Solvents for Acid Gas Removal 131 For energetic aspects of absorption and regeneration, the enthalpies of solution of CO2 in {Amine – H2O – TEG} were determined. A comparative and comprehensive study to determine the positive effects coming from the addition of a physical solvent on the demixing solvent is proposed for all the investigated properties. 7.2 Chemicals and Materials N-methylpiperidine, 2-methylpiperidine, and triethylene glycol were used without further purification. Water was distilled and degassed before use (resistivity 18.2 MW·cm). Solutions were prepared by mass; uncertainty in mass fraction (w) is estimated to be less than ± 10-4. The solutions were stored in glass bottles in an opaque cabinet to prevent any photo-degradation. Suppliers, purities and CAS numbers of all chemicals used in this study are given in Table 7.1. 7.3 Liquid-Liquid Equilibria 7.3.1 LLE in {methylpiperidines – H2O} and {methylpiperidines – H2O – CO2} The LLE of the binary systems {NMPD – H2O} and {2MPD – H2O} have previously been studied by Coulier et al. [10] and Stephenson et al. [11]. An experimental technique recently developed by Coulier et al. [12] allows the determination of liquid-liquid equilibria with controlled quantities of Table 7.1 Suppliers, CAS numbers and stated purities (mass fraction w) of chemicals used in this study. CAS Number w Sigma-Aldrich 626-67-5 99.9% a 2-methylpiperidine (2MPD) Sigma-Aldrich 109-05-7 98.3% Triethylene glycol (TEG) Sigma-Aldrich 112-27-6 99.0% Carbon dioxide (CO2) Air Products 124-38-9 99.995% Chemical Suppliers N-methylpiperidine (NMPD) racemate a 132 Carbon Dioxide Capture and Acid Gas Injection dissolved CO2. The LLE data were measured by Coulier et al. [12] using the cloud point method. It consists of determining the temperature at which a second liquid phase appears or disappears in a liquid system. For solutions containing dissolved CO2, two different apparatuses using the visual determination of the temperature of phase separation were set up depending on the range of temperatures investigated. The first apparatus is a visual phase equilibrium cell SPM20 from Thar instruments. The equipment features a high-pressure chamber provided with pressure and temperatures sensors and a thick sapphire window that allows the visualization of the cloud point through a camera connected to a computer. The second cell, supplied by CTP Mines ParisTech, is fully made of sapphire, allowing the visualization of the entire sample, instead of a limited zone. This cell is immerged in a silicon oil cooling bath to extend measurements to temperatures below 273 K. The detailed characteristics of both ­apparatuses are given in Table 7.2. Aqueous solutions of amine loaded with controlled quantities of CO2 are prepared in a custom-made flow mixing cell. The overall experimental arrangement of the two systems is depicted in Figure 7.2. The mixing cell Table 7.2 Characteristics of the visual cells used for cloud point measurements. Equilibium cell Sapphire cell T (K) Room T – 393 270–393 Control of T Heat tape Thermostatic bath p (MPa) 1–400 1–80 Control of p Buffer volume Buffer volume Inner volume (mL) 10–20 adjustable 5 Visualization of the sample sapphire window Full sample Mixing cell High p pump Water + amine Equilibrium cell P High p pump CO2 Figure 7.2 Overall experimental setup of liquid-liquid equilibrium cells for solutions containing dissolved gas. New Amine Based Solvents for Acid Gas Removal 133 is built with the same structure as the one developed at ICCF for enthalpies of solution measurements [13]. The mixing point consists of a Y piece, where two 1/16" stainless steel tubes are soldered on the top branches of the Y, while a unique tube containing the final mixture goes out from the bottom branch of the mixing point. The two fluids, CO2 and the aqueous amine solution, are injected into the mixing cell supplied by two ISCO model 100 DM high-pressure syringe pumps. As the syringe pumps deliver constant volumetric flow rates, they were regulated at a constant temperature of 298.15 K using a thermostatic bath in order to calculate accurately the composition of the aqueous solutions containing dissolved gas. The system pressure is maintained constant at 0.02 MPa using a buffer volume of 1 dm3 equipped with a back pressure regulator and placed at the end of the flow line. The gas loading α (mol CO2 / mol amine) of the mixture leaving the mixing unit was determined using the molar flow rates delivered by the two syringe pumps (Eq 7.1). nCO2 namine (7.1) where ṅCO andṅamine are the molar flow rates of CO2 and aqueous solution 2 of amine respectively. To calculate the molar flow rates, the densities of the aqueous solution of amine and CO2 are needed at the experimental conditions of temperature and pressure. The densities of the solution as a function of the pressure were measured using an Anton Paar densimeter DMA HP. The densities of CO2 were calculated using the equation of state from Span and Wagner [14]. Details on the calculation of the loading charge and its uncertainty are found in Arcis et al. paper [13]. The relative uncertainty on loading charge using this method is estimated to be less than 4%. The same devices are used to measure temperature of phase separation for solutions without dissolved gas. In that case, the solutions are directly injected in the visual cell, without using the mixing cell prior to the entrance of the visual cell. The procedure for the cloud point determination is the same independently of the system measured (visual isochoric method). Once the cell is entirely filled with the homogeneous solution (without any vapor phase), it is isolated from the pumps. Then, the temperature in the cell is increased at a definite scanning rate (0.2 to 1 K/min) to find the tightest possible temperature interval in which the second phase appears. During this procedure, the cell is still connected to the buffer volume to avoid pressure increasing due to thermal expansion. The change in turbidity is detected visually. The uncertainty on the temperature of the cloud point was estimated from 380 380 360 360 340 340 T/K T/K 134 Carbon Dioxide Capture and Acid Gas Injection 320 300 300 280 0.0 (a) 320 0.2 0.4 xa 0.6 280 0.0 0.8 (b) 0.2 0.4 xa 0.6 0.8 Figure 7.3 Phase diagram, temperature versus mole fraction, for ternary mixtures of (a), {CO2 –NMPD – H2O} and (b), {CO2 –2MPD – H2O}, at constant loading charges: opened circle, α = 0 [10, 11] and filled circle, α = 0.2. Solid lines are smooth fitting lines. reproducibility tests and is less than u(T) = 2K, while uncertainty on such temperature determination for one experiment is u(T) = 0.5 K. The phase diagrams of the binary systems {NMPD – H2O} and {2MPD – H2O} were previously determined [10, 11] and the lower critical solution temperatures were found to be 318 K for xNMPD = 0.07 and 339 K for x2MPD = 0.05, respectively. Concerning the liquid-liquid phase diagrams of the binary systems illustrated in Figure 7.3, the behavior of the two methylpiperidines with water is very different. For example at 353 K, without CO2, the water rich phase of the {NMPD – H2O} system is poor in amine (xNMPD = 0.005) and the water content of the amine rich phase is rather small (xw = 0.2). While at the same temperature, the water rich phase of the {2MPD – H2O} system is rather poor in amine (x2MPD = 0.017) but the amine rich phase is highly rich in water (xw = 0.82). Without CO2, the phase diagrams of the binary systems show that using NMPD instead of 2MPD is more favorable for the demixing process. At a constant gas loading charge of 0.2, the temperatures of phase separation decrease significantly with the addition of NMPD and reach 280 K for a composition of amine solution xNMPD = 0.11. Measurements were not feasible for more concentrated solutions, solutions due to the limits of temperatures of our techniques (270 K–393 K). With 2MPD, the phase diagram with dissolved CO2 is similar to the one without CO2 up to x2MPD = 0.046. We do not observe any significant change of the lower critical end point. Moreover, a significant shrinkage of the immiscibility gap is observed. Finally, we can also notice that the “amine phase” is very rich in water. Those differences are mainly due to different chemical reactions occurring in the solution in the presence of CO2 [12]. New Amine Based Solvents for Acid Gas Removal 135 Considering those phase diagrams, none of these amines can reach the requirements of the proposed process with CO2. The ideal system considering these methylpiperidines would be a compromise between the large phase diagram of NMPD and the temperature of phase separation obtained with 2MPD. 7.3.2 Liquid-Liquid Equilibria of Ternary Systems {Amine – H2O – Glycol} The addition of a physical solvent, triethylene glycol (TEG) was considered to increase the temperatures of phase separation of the mixtures, without changing the shape of the curve. A test study was then realized in the ternary liquid system {(N- or 2-)MPD – H2O – TEG} to verify the influence of the TEG on the LLE. The visual technique previously described was used to evaluate the influence of the glycol on the LLE at atmospheric pressure. For that purpose, increasing amounts of TEG were added to aqueous solutions of NMPD and 2MPD with a starting amine composition wa = 0.2. The temperatures of phase separation for both systems are presented in Figure 7.4. In an aqueous solution of 2MPD (w2MPD = 0.2), the addition of small amounts of TEG leads to a sharp increase of the temperatures of phase splitting, limiting the amount of TEG to wTEG = 0.075 due to the temperature 370 360 T/K 350 340 330 320 310 0.00 0.05 0.10 0.15 0.20 0.25 0.30 wTEG Figure 7.4 LLE of the ternary systems: – TEG}. , {NMPD – H2O – TEG} and , {2MPD – H2O 136 Carbon Dioxide Capture and Acid Gas Injection range of the technique. For the ternary system {NMPD – H2O – TEG}, phase separation temperatures are also rising while adding TEG. Nevertheless, these temperatures stay low enough with a reasonable amount of physical solvent to be undertaken in the demixing process. 7.3.3 Liquid-Liquid Equilibria of the Quaternary Systems {CO2 – NMPD – TEG – H2O} The influence of CO2 on the phase diagram was then evaluated in mixtures containing NMPD and TEG. The liquid-liquid equilibrium data were determined at 0.5 MPa for two mixtures, {NMPD (20) – TEG (20) –H2O (60)} and {NMPD (20) – TEG (30) –H2O (50)}. Numbers in brackets denote the weight percent of each mixture component. Figure 7.5 compares the phase diagrams of these two systems as a function of CO2 loading charge with the one without TEG determined by Coulier et al. [12]. As shown in section 7.3.2, adding TEG to an aqueous solution of NMPD yields to an increase of the temperatures of phase separation. The shape of the LLE curves investigated with TEG is similar to the one obtained by Coulier et al. [12] without TEG. The main difference concerns the temperature of the lower critical end point which increases while adding TEG. However, it is a very valuable benefit for the process with demixing solvent since temperatures of phase separation can be controlled by the quantity of physical solvent. 373 353 T/K 333 313 293 273 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 /(mol CO2/mol amine) Figure 7.5 Phase diagram, temperature versus loading charge, for quaternary mixtures of {CO2 – NMPD – TEG – H2O}: , {CO2 – NMPD (20) – TEG (0) – H2O (80)}; : {CO2 – NMPD (20) – TEG (20) – H2O (60)} and , {CO2 – NMPD (20) – TEG (30) – H2O (50)}. Numbers in brackets denote the weight percent of each mixture component. Dash lines are smooth fitting lines. New Amine Based Solvents for Acid Gas Removal 137 As the LLE regions are well controlled by adding TEG, measurements of the other thermodynamic properties of the mixtures were carried out, to provide additional information in case of a process development and to compare the capacity of such solvents with ones without TEG. From the previous results, NMPD appears to be the most promising amine for this application. Thermodynamic properties such as densities and heat capacities of mixtures containing this amine were determined. 7.4 Densities and Heat Capacities of Ternary Systems {NMPD – H2O – Glycol} Densities and heat capacities are two essential thermodynamic properties that need to be measured to optimize separation processes. Indeed the densities drive part of the transport properties of the solutions, and heat capacities control the energetic cost resulting from heating during the process. Densities are also needed for any calculations of molar properties from volumetric measurements (solution composition in LLE or enthalpies of solutions containing CO2, heat capacities…). 7.4.1 Densities The densities of the ternary solutions {NMPD – H2O – TEG} were measured at atmospheric pressure using an Anton Paar DMA 5000 density meter and the experimental procedure is given by Coquelet et al. [15]. Compositions in amine, water and TEG and the ranges of investigated temperatures are resumed in Table 7.3. The range of temperature studied is limited by the LLE as measurements have to be realized for homogeneous one phase solutions. The range of studied temperature is then larger when adding TEG, as explained in section 7.3.2. The influence of glycol on the densities of aqueous solution of NMPD is shown in Figure 7.6. An increase of the densities is observed with the addition of TEG at all studied temperatures. The density of solutions decreases also when the temperature is increased. The curves are mostly shifted to the highest values of densities when TEG is added to the solution. 7.4.2 Specific Heat Capacities The specific heat capacities of aqueous amine solutions were determined by using a differential scanning microcalorimeter (microSC) from SETARAM, France, equipped with liquid Cp cells of 1 mL inner volumes. The detection 138 Carbon Dioxide Capture and Acid Gas Injection Table 7.3 Composition of the ternary systems {NMPD – H2O – TEG} and ­temperature range investigated in the density study. NMP TEG H2O T wt % K 20 20 60 283–333 20 30 50 283–343 30 20 50 283–338 30 30 40 283–343 20 80 0 283–343 30 70 0 283–343 0 100 0 283–343 100 0 0 283–343 20 0 80 283–313 30 0 70 283–313 1.08 1.06 /g.cm–3 1.04 1.02 1.00 0.98 0.96 280 290 300 310 320 330 340 350 T/K Figure 7.6 Densities of the ternary systems {NMPD (w %) – TEG (w %) – H2O (w%)}. , {NMPD (20) – TEG (0) – H2O (80)}; , {NMPD (20) – TEG (20) – H2O (60)}; , {NMPD (20) – TEG (30) – H2O (50)}; , {NMPD (20) – TEG (80) – H2O (0)}. Numbers in brackets denote the weight percent of each mixture component. Dash lines are smooth fitting. New Amine Based Solvents for Acid Gas Removal 139 5.0 4.5 Cp/J.g–1.K–1 4.0 3.5 3.0 2.5 2.0 283 293 303 313 323 333 T/K Figure 7.7 Specific Heat Capacities as a function of temperature for the ternary systems {NMPD (w %) – TEG (w %) – H2O (w %)}. , {NMPD (20) – TEG (0) – H2O (80)}; , {NMPD (20) – TEG (20) – H2O (60)}; , {NMPD (20) – TEG (30) – H2O (50)}; , {NMPD (20) – TEG (80) – H2O (0)}. Numbers in brackets denote the weight percent of each mixture component. Dash lines are smooth fitting lines. is based on the Calvet principle. The experimental procedure is given by Coulier et al. [16]. First, a blank experiment is performed by filling both the sample and reference cells with nitrogen (N2). Then, the sample cell is filled with the studied mixture while the reference cell is filled with N2. An experimental run is made of a 20 min isothermal step at 278.15 K followed by temperature scanning (0.5 K·min–1) up to 333.15 K. Experiments were carried out at constant pressure (0.1 MPa) in both the sample and reference cells. The influence of the physical solvent on the specific heat capacities is shown in Figure 7.7. As expected, TEG reduces the heat capacities of the absorbent solution. The heat capacity is close to 2 J.g–1.K–1 when water is replaced by glycol as a solvent for the NMPD. This decrease is highly valuable for process design as the cost for heating the mixtures is drastically reduced with TEG. 7.5 Vapor-Liquid Equilibria of Ternary Systems {NMPD – TEG – H2O – CO2} A specific description of the experimental device used in this work to measure VLE data has been reported by Zhang et al. [17]. Shortly, the 140 Carbon Dioxide Capture and Acid Gas Injection technique of measurements is based on the “static-analytic” method described by Laugier and Richon [18] and experimental procedure is fully described in Coquelet and Richon [19]. With this apparatus both the liquid and vapor phases can be sampled under pressure using ROLSI capillary samplers [20, 21]. The equilibrium cell is immersed in a thermo-regulated liquid bath. In order to ensure accurate temperature measurements in the equilibrium cell and to check for thermal gradients, the temperature is measured at the top and bottom flanges through two 100 Ω platinum resistance thermometer probes. A variable-speed stirrer inside the cell accelerates the mass transfer between phases and reduces the time needed to achieve equilibrium. Pressures are measured by three pressure transducers of which the maximum absolute pressures are 0.35 bar, 1 bar and 10 bar, respectively. Sample analysis is carried out by a gas chromatograph equipped with a thermal conductivity detector (TCD). After calibration the uncertainty on CO2 composition in liquid phase is lower than 0.04. Before measuring VLE, the equilibrium cell and its loading lines were first evacuated. About 30 mL of the mixture {NMPD (14) – TEG (17) – H2O (69)} was introduced via a press at room temperature. The solution was then heated to 313 K. Meanwhile, an adequate stirring was maintained inside the cell. Phase equilibrium was assumed to be achieved while temperature and pressure readings stabilized for at least 30 min. The first pressure measurement gave the vapor pressure of the mixture investigated. Carbon dioxide was then loaded from a gas tank with controlled temperature and pressure. For each equilibrium condition, at least six samples of the liquid phase were withdrawn and analyzed to ensure composition repeatability within ±1%. CO2 was then further introduced to measure the next equilibrium condition. The solubility of CO2 in a solution of {NMPD (14) – TEG (17) – H2O (69)} was determined at 313 K. Experiments were conducted for different CO2 loading charges (α), up to the saturation of the absorbent solution and are illustrated in Figure 7.8. 7.6 Enthalpies of Solution The experimental setup used in this study has been carefully reported elsewhere [13]. Briefly, the enthalpy of solution of CO2 in the ternary system {NMPD – H2O – TEG} was measured by using a custom-made flow-­mixing cell adapted to a Setaram BT2.15 heat conduction differential calorimeter. Experiments were carried out at constant temperature and pressure. The New Amine Based Solvents for Acid Gas Removal 141 7 6 p/bar 5 4 3 2 1 0 0.0 0.2 0.4 0.6 0.8 1.0 /(mol CO2/mol amine) Figure 7.8 Equilibrium pressure as a function of CO2 loading charge for the system {NMPD (14) – TEG (17) – H2O (69)} at 313 K. Numbers in brackets denote the weight percent of each mixture component. two fluids to be mixed (CO2 and ternary solution) were injected into the flow lines by two high-pressure syringe pumps, thermo-regulated at near ambient temperature. Experiments were carried out at different loadings α (moles CO2/mol amine). The gas loading charge is determined as described in the previous section. Enthalpies of solution of CO2 in solutions of {NMPD – H2O – TEG} were measured at 313 K at pressure of 1 MPa, for two absorbent mixtures (wNMPD = 0.20, wTEG = 0.20). Experiments were conducted for different loading charges (α), up to the saturation of the absorbent solution. As an example, experimental enthalpies measured for {NMPD – H2O – TEG} and expressed in kJ.mol−1 of CO2 (Figure 7.9a) and of NMPD (Figure 7.9b) have been plotted versus loading charge α (mol of CO2/mol of amine). In Figure 7.9a, the enthalpies of solution for CO2 are exothermic and equivalent, up to a loading charge of 0.5. The average enthalpy values ΔsolHav, for α < 0.5 is found to be –71.5 kJ.mol−1. These values were not determined for the binary system {NMPD – H2O} with wa = 0.2 because phase separations would occur while adding CO2 in this experimental condition of temperature (Figure 7.5). In Figure 7.9b, experimental enthalpies of solution expressed in kJ.mol−1 of NMPD show two different domains. In the first domain (0 < α < 1), ΔsolH increases linearly with the loading charge. The value of the slope in this domain is equal to ΔsolHav obtained previously 142 Carbon Dioxide Capture and Acid Gas Injection 80 70 50 40 30 – –1 solH/K.J.mol 60 20 10 0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 1.4 1.6 1.8 /(mol CO2/mol amine) (a) 80 70 60 – –1 solH/K.J.mol 50 40 30 20 10 0 0.0 (b) 0.2 0.4 0.6 0.8 1.0 1.2 /(mol CO2/mol amine) Figure 7.9 Enthalpy of solution (−ΔsolH) versus CO2 loading charge for an aqueous solution {NMPD (20) – TEG (20) – H2O (60)}at T = 313 K and p= 1.0 MPa. (a) ΔsolH/ (kJ. mol−1 of CO2), straight lines show the average values for the enthalpies of solution at low loadings (α < 0.5); (b) ΔsolnH/(kJ.mol−1 of NMPD). (Figure 7.9a). The second domain where the enthalpy of solution stays constant is characteristic of a saturated solution. The intersection between unsaturated (enthalpy increase) and saturated (plateau) domains yields the solubility limit (s). The experimental solubility limit of CO2 in the ternary system {NMPD – H2O – TEG} was graphically determined at 313 K and 1.0 MPa New Amine Based Solvents for Acid Gas Removal 143 7.7 Discussion and Conclusion A part of the thermodynamic properties required (phase diagram, heat capacity, heat of absorption, density) for the design of the new operation units for CO2 removal have been determined in this study. These preliminary results show the substantial benefits of the addition of a physical solvent to an aqueous solution of amine. Among these advantages, the specific heat capacities of liquid phases can be lowered allowing energy savings, and the temperatures of the phase separation in the presence of CO2 can be controlled. Nevertheless, additional thermodynamic data (such as Henry’s law constant) are needed to complete this study, in order to develop a reliable thermodynamic model that takes into account the mechanism of ­reaction of CO2 with the amine and the formation of electrolytes species. For that purpose, an original device developed by Provost et al. [22] allowing the simultaneous measurements of the pressure and liquid phase composition, as a function of time will be used. The liquid phase composition is evaluated through the analysis Fourier Transform InfraRed (FTIR) spectrum, recorded in situ with an Attenuated Total Reflection (ATR) accessory. In addition to this, thermodynamics models developed by Paricaud et al. [23] will be applied to these systems and could run within process simulation software. The first step will be the development of a thermodynamic model for electrolyte solutions, which considers the most important chemical species in the aqueous and amine solutions, and is able to predict the liquid-liquid immiscibility as well as the chemical and phase equilibria. This model will be used to describe both the phase equilibria and energetic properties such as heat capacities. It will be implemented into a code that is compatible with the CAPE-OPEN interface of Simulis Thermodynamics and prosim plus software developed by the Prosim company. Through this interface, we will be able to simulate the main three elements of the separation process (decanter and absorption and desorption columns) in either the Prosim plus or Aspen one environment, and estimate the cost and energy requirement for the CO2 capture. Acknowledgments This research work was part of collaboration between the Centre Thermodynamics of Processes (CTP) of MINES ParisTech and the Institute of Chemistry of Clermont-Ferrand (ICCF). The recommendations and concerns from Dr. John Carroll (Gas Liquid Engineering) concerning industrial applications of demixing amines are an inspiration of 144 Carbon Dioxide Capture and Acid Gas Injection this work. Constructive advice and considerable expertise given by Pascal Théveneau and Alain Valtz (CTP) are gratefully acknowledged. Financial support allowing the acquisition of the microcalorimeter for heat capacities measurements from Contrat d’Objectif Partagé CNRS-UBP-FederRégion Auvergne is also acknowledged. References 1. Z hao, Q., Leonhardt, E., MacConnell, C., Frear, C., and Chen, S., Purification technologies for biogas generated by anaerobic digestion. p. 1–24, 2010. 2. Rochelle, G.T., Amine Scrubbing for CO2 Capture. Science, 325(5948), p. 1652–1654, 2009. 3. Raynal, L., Bouillon, P.-A., Gomez, A., and Broutin, P., From MEA to demixing solvents and future steps, a roadmap for lowering the cost of post-combustion carbon capture. Chemical Engineering Journal, 171(3), p. 742–752, 2011. 4. Raynal, L., Briot, P., Dreillard, M., Broutin, P., Mangiaracina, A., Drioli, B.S., Politi, M., La Marca, C., Mertens, J., Thielens, M.-L., Laborie, G., and Normand, L., Evaluation of the DMX Process for Industrial Pilot Demonstration – Methodology and Results. Energy Procedia, 63, p. 6298–6309, 2014. 5. Raynal, L., Alix, P., Bouillon, P.-A., Gomez, A., le Febvre de Nailly, M., Jacquin, M., Kittel, J., di Lella, A., Mougin, P., and Trapy, J., The DMX process: An original solution for lowering the cost of post-combustion carbon capture. Energy Procedia, 4(0), p. 779–786, 2011. 6. Ballerat-Busserolles, K., Lowe, A.R., Coulier, Y., and Coxam, J.Y., Calorimetry in Aqueous Solutions of Demixing Amines for Processes in CO2 Capture, in Acid Gas Extraction for Disposal and Related Topics. John Wiley & Sons, Inc. p. 69–80, 2016. 7. Fandiño, O., Yacyshyn, M., Cox, J.S., and Tremaine, P.R., Speciation in LiquidLiquid Phase-Separating Solutions of Aqueous Amines for Carbon Capture Applications by Raman Spectroscopy, in Acid Gas Extraction for Disposal and Related Topics. John Wiley & Sons, Inc. p. 81–94, 2016. 8. Dash, S.K. and Bandyopadhyay, S.S., Studies on the effect of addition of piperazine and sulfolane into aqueous solution of N-methyldiethanolamine for CO2 capture and VLE modelling using eNRTL equation. International Journal of Greenhouse Gas Control, 44, p. 227–237, 2016. 9. Zheng, C., Tan, J., Wang, Y.J., and Luo, G.S., CO2 Solubility in a Mixture Absorption System of 2-Amino-2-methyl-1-propanol with Ethylene Glycol. Industrial & Engineering Chemistry Research, 52(34), p. 12247–12252, 2013. 10. Coulier, Y., Ballerat-Busserolles, K., Rodier, L., and Coxam, J.Y., Temperatures of liquid–liquid separation and excess molar volumes of {N-methylpiperidine– water} and {2-methylpiperidine–water} systems. Fluid Phase Equilibria, 296(2), p. 206–212, 2010. New Amine Based Solvents for Acid Gas Removal 145 11. Stephenson, R.M., Mutual solubility of water and pyridine derivatives. Journal of Chemical & Engineering Data, 38(3), p. 428–431, 1993. 12. Coulier, Y., Lowe, A.R., Moreau, A., Ballerat-Busserolles, K., and Coxam, J.-Y., Liquid-Liquid Equilibria in {Amine - H2O - CO2} Mixtures: New Methods for Key Data. Fluid Phase Equilibria, 2016. accepted. 13. Arcis, H., Ballerat-Busserolles, K., Rodier, L., and Coxam, J.-Y., Enthalpy of Solution of Carbon Dioxide in Aqueous Solutions of Monoethanolamine at Temperatures of 322.5 K and 372.9 K and Pressures up to 5 MPa. Journal of Chemical & Engineering Data, 56(8), p. 3351–3362, 2011. 14. Span, R. and Wagner, W., A New Equation of State for Carbon Dioxide Covering the Fluid Region from the Triple‐Point Temperature to 1100 K at Pressures up to 800 MPa. Journal of Physical and Chemical Reference Data, 25(6), p. 1509–1596, 1996. 15. Coquelet, C., Valtz, A., Courtial, 7., and Richon, D., Volumetric properties of the water + 3-(dimethylamino) propylamine (DMAPA) mixture at atmospheric pressure from 283.15 to 353.15 K. Thermochimica Acta, 448(1), p. 47–51, 2006. 16. Coulier, Y., Ballerat-Busserolles, K., Mesones, J., Lowe, A., and Coxam, J.-Y., Excess Molar Enthalpies and Heat Capacities of {2-Methylpiperidine– Water} and {N-Methylpiperidine–Water} Systems of Low to Moderate Amine Compositions. Journal of Chemical & Engineering Data, 60(6), p. 1563–1571, 2015. 17. Zhang, F., Théveneau, P., El Ahmar, E., Canet, 7., Soo, C.-B., and Coquelet, C., An improved static–analytic apparatus for vapor–liquid equilibrium (PTxy) measurement using modified in-situ samplers. Fluid Phase Equilibria, 409, p. 425–433, 2016. 18. Laugier, S. and Richon, D., New apparatus to perform fast determinations of mixture vapor–liquid equilibria up to 10 MPa and 423 K. Review of Scientific Instruments, 57(3), p. 469–472, 1986. 19. Coquelet, C. and Richon, D., Chapter 14 Solubility of BTEX and Acid Gases in Alkanolamine Solutions in Relation to the Environment, in Developments and Applications in Solubility. The Royal Society of Chemistry. p. 219–235, 2007. 20. Guilbot, P., Valtz, A., Legendre, H., and Richon, D., Rapid on-line samplerinjector: a reliable tool for HT-HP sampling and on-line GC analysis. Analusis, 28(5), p. 426–431, 2000. 21. Théveneau, P. and Legendre, H., Dispositif pour prélever des micro-échantillons d’un fluide à l’état liquide contenu dans un containeur. 2014, French Patent Application No. 1460309. 22. Diab, F., Provost, E., Laloué, N., Alix, P., Souchon, V., Delpoux, O., and Fürst, W., Quantitative analysis of the liquid phase by FT-IR spectroscopy in the system CO2/diethanolamine (DEA)/H2O. Fluid Phase Equilibria, 325, p. 90–99, 2012. 23. Fukumoto, A., Sales Silva, L.P., Paricaud, P., Dalmazzone, D., and Fürst, W., Modeling of the dissociation conditions of H2 + CO2 semiclathrate hydrate formed with TBAB, TBAC, TBAF, TBPB, and TBNO3 salts. Application to CO2 capture from syngas. International Journal of Hydrogen Energy, 40(30), p. 9254–9266, 2015. 8 Improved Solvents for CO Capture by Molecular Simulation Methodology 2 William R. Smith Department of Mathematics and Statistics and Dept. of Chemistry, University of Guelph, Guelph, Canada, and Faculty of Science, University of Ontario Institute of Technology, Oshawa, Canada Abstract The goal of this paper is to describe a strategy for implementing molecular simulation methodology to model CO2 capture systems by combining advanced molecular-level modeling and experimental measurement methodologies, with the goal of discovering new alkanolamine solvents that yield improved CO2 solubility. The project has been funded for a 3-year period ending December 2018 by the Natural Sciences and Engineering Research Council of Canada under its International Strategic Partnership Program. A collaborating partner, funded by l’Agence Nationale de la Recherche of France, is a team at Blaise Pascal University, Clermont-Ferrand, France. The University of Guelph is primarily responsible for the theoretically based work described in this paper, and the French team is primarily responsible for related experimental research. The research of both groups will be tightly integrated to achieve the project goals. 8.1 Introduction Although alternatives exist for carbon capture in other contexts, we focus here on the relatively mature industrial methodology of CO2 capture from a post-combustion gas stream that can be effected by its preferential absorption in a liquid solvent. The use of aqueous alkanolamine as a solvent to absorb CO2 was patented in 1930 [17], and is expected to be the dominant technology for US coal-fired plants by 2030 [4]. The amine is subsequently regenerated by stripping with steam and condensation of the Ying Wu, John J. Carroll and Weiyao Zhu (eds.) CO2 Capture and Acid Gas Injection, (147–160) 2017 © Scrivener Publishing LLC 147 148 Carbon Dioxide Capture and Acid Gas Injection T T A b s o r b e r P P S t r i p p e r C o m p r e s s o r Reboiler T Figure 8.1 CO2 capture by amine scrubbing. (Reproduced with permission from the PhD thesis of Micka¨el Simond [19].) H2O from the stripper vapor, leaving pure CO2 that can be compressed for purposes such as enhanced oil recovery or for geological sequestration. The process is represented in Figure 8.1. It was evaluated in 1991 [18] for commercial use in coal-fired power plants and found to have unacceptably high energy requirements and costs, motivating the need for improved methodologies. Developing solvents with improved solubility and energy properties would decrease the cost, but the number of potential solvent alternatives is vast. Some have been developed commercially and demonstration facilities have been built [7] (including the Shell–CanSolv facility at Estevan Saskatchewan, opened in October 2014), but one of the impediments in the quest to discover improved solvents is the considerable time and cost entailed in the design and construction of experiments to test alternative solvents. Although macroscopic thermodynamic models for predicting CO2 solubility in candidate solvents have been and continue to be developed (e.g. [20, 21]), this is a challenging task, since the solvents are very nonideal, and the process complexity is well beyond that of normal VLE modeling due to the solvent’s chemical reactions with CO2. The predominantly empirical basis of these models precludes confidence in the accuracy of their predictions beyond the ranges of experimental conditions from which their parameters have been determined. In addition, the typically large number of parameters in such models requires large (and costly to obtain) sets of experimental data for their determination for each proposed solvent. Improved Solvents for CO2 Capture 149 A potentially more cost-effective way to screen potential solvent candidates and to guide experiments is to develop more fundamentally based and highly accurate mathematical models containing smaller parameter sets to predict the properties of the systems involved, and this strategy is at the core of the project. The models and algorithms based on molecularlevel simulation technology proposed herein can also supplement the macroscopic models and provide a means to extend their predictions to a much wider range of operating conditions. Such models also allow the prediction of multiple thermodynamic and transport properties using the same parameter set, furthering their advantage over the macroscopic approach, which requires different models for each property and corresponding additional parameter sets that must be determined from experimental data. Significant progress in computational technology has now made possible the relatively low-cost implementation of increasingly sophisticated molecular-based models and algorithms like those described in this paper. 8.2 Physical and Chemical Models In general, one may describe the thermodynamic properties of solutions like those considered here that are characterized by strong intermolecular interactions, by means of a physical model [22, 23], a chemical model [24–26], or a combination of the two approaches for different system components. A physical model assumes the existence of only the original system species, whereas a chemical model assumes the existence of additional species accompanied by chemical reactions among them. Although the additional species are sometimes justified by spectroscopic evidence [27], their introduction often has an empirical basis. Another distinguishing model characteristic concerns whether it employs a macroscopic thermodynamic description via either an equation of state (EOS) and/or a chemical potential model, or a molecular-level description, by which take here to mean a mathematical model of the molecular-level interactions in terms of a force field (FF). A simple example of a macroscopic chemical model is one for H2O that assumes the existence of the distinct species {H3O+, OH–, H2O} and the accompanying chemical reaction 2H2O = H3O+ + OH– (8.1) On the other hand, a macroscopic physical model for the system such as SAFT (for a recent review, see [20]) employs an EOS for H2O that mimics the reaction’s effects by means of a relatively simple molecular picture 150 Carbon Dioxide Capture and Acid Gas Injection that incorporates orientationally dependent association sites on the water molecules that result in the formation of hydrogen bonds. Similarly, ­ a molecular-level physical model for H2O employs a detailed FF for the H2O molecule with strong orientationally dependent attractive forces; TIP4P [28] and SPC/E FF [29] are typical water FFs often used, although there are many others (see the review of Guillot [30]). However, when compounds chemically react, rather than form relatively weak association bonds, models of the system must take this into account as a separate feature. Macroscopic chemical models have considerable flexibility in fitting experimental data, since the introduction of each new species introduces corresponding additional parameters that can be adjusted to fit the data; however, this flexibility comes at the cost of requiring extensive experimental data to determine the increased number of parameters required to describe their empirical temperature and pressure dependence. Macroscopic chemical models include NRTL [31] and UNIQUAC [32, 33], augmented by semi-empirical Pitzer chemical potential expressions [34], the electrolyte EOS of Fuürst and Renon [35] and the CPA (cubic-plus-association) [36, 37] EOS. Physical models have the advantage of generally requiring fewer parameters, but unless they adequately reflect the underlying intermolecular interactions, their predictions may not be sufficiently accurate. The SAFT EOS (see, for example, [20, 38, 39]) is an example of a physical model that has both both macroscopic and molecular-level characteristics. It has become routinely used by chemical engineers for various applications and as a component of chemical process simulation software packages such as AspenPlus [40] and PROSIM [41]. The practical utility of the SAFT approach lies in the fact that although its underlying FF is a relatively crude approximation, its thermodynamic consequences can be expressed in analytical form. (Its historical origin is related to the simple Smith-Nezbeda model [42] of associating hard-sphere fluids [43].) It is important to note that SAFT parameters depend only on the interactions between molecular sites and are independent of temperature. Notwithstanding, SAFT retains a somewhat empirical flavor, in view of its approximate molecular picture. 8.3 Molecular-Level Models and Algorithms for Thermodynamic Property Predictions Molecular-level models describe each species molecule by means of a mathematical model of its structure and its interactions with other system molecules, and in a sense are a molecular-level analogue of a macroscopic physical model. Chemical reactions among the molecules of the Improved Solvents for CO2 Capture 151 system can be implemented in conjunction with such models by means of the Reaction Ensemble Monte Carlo algorithm [10, 11], discussed below. Molecular-level models have the following general advantages over macroscopic models: a. Their fundamental basis enables them to be used for predictions beyond the range of thermodynamic conditions in which the parameters have been obtained by fitting to experimental data; b. They are capable of systematic improvement, by developing improved descriptions of the molecular interactions; c. They have many fewer parameters than is the case for macroscopic models for a given level of accuracy; for example (apart from the SAFT model), although temperature- and pressure-dependent parameters are typically required for macroscopic models, this dependence is automatically accounted for in the mathematical structure of molecular-level models. This results in a reduced need for experimental data to fit the parameters; d. They require only a single (relatively small) parameter set to predict (given appropriate algorithms) all thermodynamic and transport properties of a substance; in contrast, different macroscopic models (and corresponding experimental data to fit their parameters) are often required for different properties. The primary disadvantage of molecular-level models is their computational complexity, making it infeasible for them to be embedded in software such as chemical process simulators. Nonetheless, the time and cost required for computing data points using molecular-level models is orders of magnitude less than the cost of experiments to obtain the much larger data sets required to fit the larger parameter sets of macroscopic models. Provided they are sufficiently accurate, this also makes molecular-based models a valuable tool for generating pseudo-experimental data, to which empirical macroscopic models can be fitted for use in such simulation software. Although molecular-level models promise increased predictive capability, their use has been historically impeded not only by their computational complexity, but also by the lack of suitable algorithms to directly and efficiently implement certain thermodynamic properties and processes of industrial interest. The problem of computational complexity has been greatly alleviated by advances in computer technology, but the latter problem has continued to exist. 152 Carbon Dioxide Capture and Acid Gas Injection Both Molecular Dynamics and Monte Carlo methods have been used for over 40 years for molecular-level thermodynamic property predictions (see, for example, [44, 45]), and each technique has advantages for different types of problems. Monte Carlo methods are generally more suitable for calculating properties and simulating processes related to chemical potentials, which are arguably the most important solution properties underlying reaction and phase equilibria. However, although methods based on Monte Carlo algorithms have a long history of use for phase equilibrium calculations since the development in 1987 of the Gibbs Ensemble Monte Carlo (GEMC) algorithm [46], their use for chemically reacting systems is more recent. This has been enabled by the development of the Reaction Ensemble Monte Carlo (REMC) algorithm [10, 11]), which has been applied to a much smaller number of systems. Space does not permit its detailed description here; see the original papers or the review of Turner et al. [12]. It can be considered to be the molecular counterpart of macroscopic methods that augment the ideal-gas part of the species chemical potential value with a nonideal contribution modelled by an empirical equation or derived from an EOS model. The REMC algorithm treats the total species chemical potentials similarly, by separately incorporating the ideal-gas part of the partition function and accounting for the nonideality by simulating the configurational part of the partition function. Finally, we mention in passing that other methods for molecular-level modeling include the DFT-based ab-initio molecular dynamics [47] and COSMO-RS [48] approaches and the “reactive force field (ReaxFF) approach [49], all of which we deem to be less suitable for studying the thermodynamics of the complex systems considered here, and which are outside the scope of our study. We remark that the implementation of REMC methodology is not straightforward, particularly for dense liquid systems containing complex molecules, and has technical challenges similar to those encountered by other simulation methodology related to chemical potentials (such as the GEMC method for phase equilibrium). The challenges include the requirement for the delicate treatment of the transition probabilities for reaction steps involving flexible molecular FFs [50] and the general need for computational efficiency enhancements for dense complex systems, the latter implemented by means of various “gradual particle insertion methods” [51–55]. In addition, special techniques will be necessary to handle species present in small amounts, in order that system sizes not be too large; we expect that such species can be treated separately from the simulations by means of a Henry–Law type of model. Improved Solvents for CO2 Capture 153 8.4 Molecular-Level Models and Methodology for MEA–H2O–CO2 We will initially consider the monoethanolamine (MEA) solvent and MEA-H2O–CO2 solutions. This system has traditionally been described at the macroscopic level by means of a chemical model, for example in conjunction with a UNIQUAC model [32], the NRTL EOS [31, 56], or a cubic-plus-association (CPA) EOS [37]. The SAFT EOS has also recently been employed to model this system [39]. A first step in modeling the chemical equilibrium composition of a ­system typically entails the construction of a set of chemical reactions modeling the chemical changes. In this regard, it is important to emphasize that the equilibrium composition (assuming the absence of any kinetic or other restrictions (see [9, 57], Sections 2.4 and 9.6) is independent of any particular kinetic mechanism, and is governed only by thermodynamic considerations for the set of species present at equilibrium (the “species list”). The assumption of a particular species list is an important part of the thermodynamic analysis, and may be motivated by a number of considerations. Spectroscopic analysis of the solution may provide some indication of the species present, but this may also indicate association between molecules rather than intramolecular chemical bonds resulting from distinct chemical species. In order to obtain an economical model entailing a small species list, it is important to treat association by means of a FF whenever possible. If one assumes that the 9 species {H2O, H3O+, OH−, CO2, HCO3−, CO32–, RNHCO2−, RNH2, RNH3−} comprise the system, where R = HO – CH2 – CH2−, the rank of the system formula matrix is 4, and a “complete” set of chemical reactions required to describe all possible chemical change consists of 5 linearly independent reactions (a more appropriate term is “chemical equations (see [9], Chapter 2). The choice of the particular set is arbitrary, but the choice affects the computational efficiency of an equilibrium calculation. For example, the following is a possible complete set, based primarily on mechanistic considerations (see [31, 39, 58] for other sets)): 2H2O = H3O+ + OH−(8.2) CO2 + 2H2O = H3O+ + HCO3−(8.3) H2O + HCO3− = H3O+ + CO32−(8.4) CO2 + 2RNH2 = RNHCOO– + RNH3+ (8.5) 154 Carbon Dioxide Capture and Acid Gas Injection H3O+ + RNH2 = H2O + RNH3+ (8.6) Reaction (8.2) represents H2O dissociation, (8.3) bicarbonate formation, (8.4) carbonate formation, (8.5) carbamate formation and (8.6) MEA protonation. The reaction set (8.2)–(8.6) is problematic in a computer simulation because it contains reactions that have net charges on one or both sides of the reactions; as a result, special procedures must be added in order to ensure that the simulation box always contains a zero net charge. Additionally, experience from macroscopic reaction equilibrium problems suggests that it is advantageous to have the most abundant species on the left sides of the reactions. With the additional assumption that the concentrations of OH−and CO32– are small, the following may be a computationally convenient reaction set (the last of which arises from the combination (8.3) + (8.6): CO2 + RNH2 + H2O = RNH3+ + HCO3– (8.7) The system is typically specified by its temperature, T, and pressure, P, the initial solvent composition weight fraction ω, and a solution CO2 loading parameter L. ω and L are defined by n0 RNH2 M NH2 n0 RNH2 MRNH2 L n0 H2O MH2O n0 CO2 n0 RNH2 (8.8) (8.9) where n0 denotes a molar amount and M denotes the molecular weight. These define the initial amounts of CO2 and RNH2 per mole of H2O solvent. The composition of the vapour phase species H2O, CO2 and RNH2 in equilibrium with the calculated composition of the solution phase may be calculated from the equality of their solution phase chemical potentials. This may be performed using an equation of state, or by means of a Grand Canonical Ensemble (GCE) simulation in the vapour phase. To implement the REMC approach for a given system, for each species we require a FF and an ideal-gas standard chemical potential, µ0(T). The latter can be obtained from thermochemical tables (e.g. [59, 60]), or calculated by means of quantum mechanical software such as Gaussian09 [61]. At all stages, we will develop force fields that are transferable, in the sense that the same FF parameters can be used to describe a given interaction Improved Solvents for CO2 Capture 155 site in different molecules and that the FF is applicable for the prediction of different properties (e.g., thermodynamic, structural, or transport) across a wide range of state points (e.g., pressure, temperature, or composition). This approach is similar in philosophy to the Group Contribution approach of macroscopic thermodynamic models (e.g., [62]). As necessary, a small number of pure fluid FF parameters may be fitted to appropriate experimental data. For mixtures, either Lorentz–Berthelot or geometric mean combining rules will be used for the cross-species FF parameters, or we will adjust them to experimental data. We will use the TIP4P [28] FF for H2O, the FF of Simond et al. [63] for the MEA-H2O subsystem, the FF of Vácha et al. [64] for H3O+ and a TraPPE FF for CO2 [65]. OPLS FFs are available for many of the remaining species at http://virtualchemistry.org/ moldb.php. Once all FF parameters have been determined for the constituent binary systems by comparing against binary thermodynamic data, predictions will be made for a range of thermodynamic properties of the ternary MEA–H2O–CO2 system without the need for any additional parameters, and the results will be compared with available experimental data. Chief among the properties of interest will be enthalpies, densities, vapor-liquid equilibria, CO2 solubility and the T and P dependence of these properties. In the course of model development, we will compare our results with those obtained from available thermodynamic models in the literature and with calculations using the ASPEN [40] and/or PROSIM [41] process simulation software packages. 8.4.1 Extensions to Other Alkanolamine Solvents and Their Mixtures Once the strategy has been finalized for the MEA–H2O–CO2 system, we will apply the same approach to other alkanolamine–H2O–CO2 systems of increasing complexity. The first of these involves the family of primary alkanolamines based on the N–C–C–O backbone considered by Simond et al. [19,63, 66], illustrated in Figure 8.2. This group has developed FFs for their binary mixtures with H2O. Simond first developed FFs for the pure alkanols [63] based on the OPLS–AA functional form [67] using Gaussian09 [61] quantum mechanical calculations, based on a training set involving the simplest molecules of the family, MEA and 2A1P, using experimental heat of vaporization and liquid density data to refine the parameters. They were then extended to the test set consisting of the remaining members of the family, and the densities and enthalpy of vaporization predictions were found to be in 156 Carbon Dioxide Capture and Acid Gas Injection N O C C MEA 2A1P ABU AMP MIPA AMP2 1A2B Figure 8.2 Primary alkanolamines based on the structure N–C–C–O. (Reproduced with permission from the PhD thesis of Micka¨el Simond [19]. good agreement with the experimental values. They then used these FFs to study predictions of the excess enthalpy of their binary mixtures with water. The TIP4P [28] FF for water was used, and geometric combining rules were used for all the cross Lennard–Jones parameters between nonbonded sites of the components. They found that this did not produce quantitative agreement with the heats of mixing as a function of concentration. By introducing a Lennard–Jones cross interaction site (involving 2 parameters, E and σ) corresponding to the hydrogen bond between the OH group of the alkanolamine and the oxygen atom of water, they could quantitatively reproduce the experimental excess enthalpies of mixing for the MEA+H2O and ABU+H2O mixtures as a function of concentration. They then found that incorporating the same site for the other mixtures produced similarly quantitative agreement with the heat of mixing data. The next stage of our study will be to extend the MEA system methodology for constructing FFs to the binary mixtures of this family with CO2. As in the case of MEA, we will perform pure predictions of the ternary properties using the FFs determined from the underlying binaries. Finally, as time permits, we will use the same FFs to perform pure predictions for mixtures of solvents with H2O and CO2. We do not anticipate that major adjustments may be required in the case of these quaternary mixtures, Improved Solvents for CO2 Capture 157 although the worst-case scenario is that adjustments may be required for the binary solvent mixtures. Also as time permits, we will extend the foregoing methodology to solvents selected from other (secondary and tertiary) alkanolamine families, and finally to their mixtures. In all cases, we will perform predictions of the CO2 solubilities and of other thermodynamic properties (CO2 solubility, densities, enthalpies and vapor-liquid equilibria, and their T and P dependence), and compare these with experiment. In many cases, experimental results will not be available, and these will be measured by the ANR group. Experience with the MEA–H2O–CO2 ternary system will guide our strategies to determine FF parameters and conduct experiments to measure required properties of these systems. Similarly, as for the simpler systems, we will compare with the predictions of chemical process simulation software such as ASPEN [40] and PROSIM [41] where these are available. A final result of the project will be a set of force fields for molecular segments by means of which the force field for an arbitrary molecule of interest in a large family of primary, secondary and tertiary amine solutions with H2O and CO2 may be constructed, and from which the CO2 solubility and other thermodynamic properties may be calculated by means of molecular simulation. These results can be fitted to macroscopic thermodynamic models for use in chemical process simulators involving such solutions. Acknowledgements The financial support of the Natural Sciences and Engineering Research Council of Canada under Grant STPGP 479466, and computational facilities provided by SHARCNET (Shared Hierarchical Academic Computing Network), a partner organization of Compute Ontario and the Compute/ Calcul Canada national advanced research computing platform, are gratefully acknowledged. References 1. Prospects for CO2 capture and storage, 2004, IEA-International Energy Agency. Head of Publication Service, IEA/OECD, Paris, France. 2. H. Yang et al., J. Env. Sci. 20, 14, 2008. 3. J. D. Figueroa, T. Fout, S. Plasynski, H. McIlvrie, and R. D. Srivastava, Int. J. Greenhouse Gas Control 2, 9, 2008. 4. G. T. Rochelle, Science 325, 1652, 2009. 5. A. Schäffer, K. Brechtel, and G. Scheffknecht, Fuel 101, 148, 2012. 158 Carbon Dioxide Capture and Acid Gas Injection 6. M. Ramdin, T. W. de Loos, and T. J. Vlugt, Ind. Eng. Chem. Res. 51, 8149, 2012. 7. M. R. Abu-Zahra, Z. Abbas, P. Singh, and P. Feron, Carbon dioxide post– combustion capture: Solvent technologies overview, status and future directions, in Materials and processes for energy: Communicating current research and technological developments, edited by A. Méndez-Vilas, pages 923–934, Formatex Research Center, Bardajo Spain, 2013. 8. S. Kumar, J. H. Cho, and I. Moon, Int. J. Greenhouse Gas Control 20, 87, 2014. 9. W. R. Smith and R. W. Missen, Chemical Reaction Equilibrium Analysis: Theory and Algorithms (ISBN 0-89464-584-6, Reprint with corrections, of 1982 Wiley–Interscience publication of same name, (ISBN 0-471-09347-5)), Krieger Publishing, Malabar, Florida, 1990. 10. W. R. Smith and B. Tr´ıska, J. Chem. Phys. 100, 3019, 1994. 11. J. K. Johnson, A. Z. Panagiotopoulos, and K. E. Gubbins, Molec. Phys. 81, 717, 1994. 12. C. H. Turner et al., Molec. Simulation 34, 119, 2007. 13. F. Mouˇcka, I. Nezbeda, and W.R.Smith, J. Chem. Theory Comput. In press, DOI:10.1021/acs.jctc.5b00018, 2015. 14. W. R. Smith, S. Figueroa-Gerstenmaier, and M. Skvorova, J. Chem. Eng. Data 59, 3258, 2014. 15. H. Arcis, K. Ballerat-Busserolles, L. Rodier, and J.-Y. Coxam, J. Chem. Eng. Data 56, 3351, 2011. 16. H. Arcis, L. Rodier, K. Ballerat-Busserolles, and J.-Y. Coxam, J. Chem. Thermodynamics 41, 783, 2009. 17. R. R. Bottoms, US Patent 1783901, 1930, Girdler Corp. 18. F. Daniel, Engineering and economic evaluation of CO2 removal from fossil– fuel–fired power plants (IR–7365), December, 1991, Electric Power Research Institute, Palo Alto, CA. 19. M. R. Simond, E´tude des interactions mol´eculaires dans les solvants d’int´erˆet pour le captage des gaz acides, PhD thesis, Institute of Chemistry, Un. Of Clermont-Ferrand, 2013. 20. N. MacDowell, F. Llovell, C. S. Adjiman, G. Jackson, and A. Galindo, Molec. Phys. 110, 1325, 2012. 21. A. Chremos et al., Chem. Eng. Transactions 35, 427, 2013. 22. J. J. van Laar, Z. Phys. Chem. 72, 723, 1910. 23. J. J. van Laar, Z. Phys. Chem. 83, 4599, 1913. 24. F. Dolazalek, Z. Phys. Chem. 64, 727, 1908. 25. F. Dolazalek, Z. Phys. Chem. 71, 191, 1910. 26. F. Dolazalek, Z. Phys. Chem. 83, 40, 1913. 27. L. M. S. G. A. Applegarth, C. R. Corbeil, D. J. W. Mercer, C. C. Pye, and P. R. Tremaine, J. Chem. Thermodynamics 87, 153, 2014. 28. W. L. Jorgensen, J. Chandrasekhar, J. D. Madura, R. W. Impey, and M. L. Klein, J. Chem. Phys. 79, 926, 1983. 29. H. J. C. Berendsen, J. R. Grigera, and T. P. Straatsma, J. Phys. Chem. 91, 6269, 1987. Improved Solvents for CO2 Capture 159 30. B. Guillot, J. Molec. Liq. 101, 219, 2002. 31. D. M. Austgen, G. T. Rochelle, X. Peng, and C. C. Chen, Ind. Eng. Chem. Res. 28, 1060, 1989. 32. L. Faramarzi, G. M. Kontogeorgis, K. Thomsen, and E. H. Stenby, Fluid Phase Equilib. 282, 121, 2009. 33. K. Thomsen, P. Rasmussen, and R. Gani, Chem. Eng. Sci. 54, 1787, 1999. 34. K. S. Pitzer, J. Am. Chem. Soc. 102, 2902, 1980. 35. W. Furst and H. Renon, Am. Inst. Chem. Eng. J. 39, 335, 1993. 36. A. S. Avlund, G. M. Kontogeorgis, and M. L. Michelsen, Ind. Eng. Chem. Res. 47, 7441, 2008. 37. A. T. Zoghi and F. Feyzi, J. Chem. Thermodynamics 87, 153, 2013. 38. G. N. I. Clark, A. J. Haslam, A. Galindo, and G. Jackson, Molec. Phys. 104, 3561, 2006. 39. N. MacDowell, F. Llovell, C. S. Adjiman, G. Jackson, and A. Galindo, Ind. Eng. Chem. Res. 49, 1883, 2010. 40. Aspen Plus software, 2015, Aspen Technology, Inc., 20 Crosby Drive, Bedford, Massachusetts. 41. ProSim software, http://www.prosim.net/.\ 42. W. R. Smith and I. Nezbeda, J. Chem. Phys. 81, 3694, 1984. 43. M. S. Wertheim, J. Chem. Phys. 88, 1145, 1988. 44. M. P. Allen and D. J. Tildesley, Computer Simulation of Liquids, Oxford University Press, Oxford, 1987. 45. D. Frenkel and B. Smit, Understanding Molecular Simulation: From Algorithms to Applications (2nd Edition), Elsevier, Amsterdam, 2002. 46. A. Z. Panagiotopoulos, Molec. Phys. 61, 813, 1987. 47. D. Marx and J. Hutter, Ab Initio Molecular Dynamics: Basic Theory and Advanced Methods, Cambridge U. Press, Cambridge, UK, 2009. 48. A. Klamt, COSMO–RS: From Quantum Chemistry to Fluid Phase Thermodynamics and Drug Design, Elsevier, Amsterdam, 2005. 49. A. C. T. van Duin, S. Dasgupta, F. Lorant, and W. A. Goddard, J. Phys. Chem. A 105, 9396, 2001. 50. T. W. Rosch and E. J. Maginn, J. Chem. Theory Comput. 7, 269, 2011. 51. N. Hansen, S. Jakobtorweihen, and F. J. Keil, J. Chem. Phys. 122, 164705, 2005. 52. S. Jakobtorweihen, N. Hansen, and F. J. Keil, J. Chem. Phys. 125, 224709, 2006. 53. A. Torres-Knoop, S. P. Balaji, T. J. H. Vlugt, and D. Dubbeldam, J. Chem. Theory Comput. 10, 942, 2014. 54. W. Shi and E. J. Maginn, J. Chem. Theory Comput. 3, 1451, 2007. 55. W. Shi and E. J. Maginn, J. Comput. Chem. 29, 2520, 2008. 56. Y. Zhang and C.-C. Chen, Ind. Eng. Chem. Res. 50, 163, 2011. 57. G. W. Norval, M. J. Phillips, R. W. Missen, and W. R. Smith, Can. J. Chem. Eng. 69, 1184, 1991. 58. P. M. M. Blauwhoff, G. F. Versteeg, and W. P. M. van Swaaij, Chem. Eng. Sci. 39, 207, 1984. 160 Carbon Dioxide Capture and Acid Gas Injection 59. M. W. Chase, Jr., NIST–JANAF Thermochemical Tables, J. Phys. Chem. Ref. Data, Monograph No. 9, Am. Chem. Soc. and Am. Inst. of Physics, Woodbury, New York, 1998. 60. NIST Chemistry Webbook, http://webbook.nist.gov, 2011. 61. M. J. Frisch et al., Gaussian09 Revision D.01, 2009, Gaussian Inc. Wallingford CT 2009. 62. J. Gmehling, Fluid Phase Equilib. 144, 37, 1998. 63. M. R. Simond, K. Ballerat-Busserolles, J.-Y. Coxam, and A. A. Pádua, Chem. Phys. Phys. Chem. 13, 3866, 2012. 64. R. Vácha, V. Buch, A. Milet, J. P. Devlind, and P. Jungwirth, Phys. Chem. Chem. Phys. 9, 4736, 2007. 65. J. J. Potoff and J. I. Siepmann, Am. Inst. Chem. Eng. J. 47, 1676, 2001. 66. M. R. Simond, K. Ballerat-Busserolles, J.-Y. Coxam, and A. A. H. Pádua, J. Chem. Theory Comp. 10, 2471, 2014. 67. R. C. Rizzo and W. L. Jorgensen, J. Am. Chem. Soc. 121, 4827, 1999. 9 Strategies for Minimizing Hydrocarbon Contamination in Amine Acid Gas for Reinjection Mike Sheilan, Ben Spooner and David Engel Amine Experts – division of Sulphur Experts International Inc. Calgary, Alberta, Canada Abstract Amine units are used by gas processing operations around the world to remove acid gases such as H2S and CO2 from gas and liquid streams. Once regenerated from the amine, the H2S and CO2 is then sent to reinjection. There are many problems associated with acid gas compression when hydrocarbons are present in the gas. Excess hydrocarbons will be present in the acid gas stream from the amine unit if the amine unit is not designed or operated properly. Undesirable hydrocarbons exist in the inlet gas of most amine absorbers, in either gas or liquid phase. Either way, the amine solution can and will absorb these hydrocarbons to a certain degree. Various types of amines are somewhat soluble in hydrocarbons depending on the conditions, which explains the absorption in part. Entrainment of free hydrocarbons and emulsification of hydrocarbons in the amine solution are also possible. Hydrocarbons can create numerous operational and performance problems in the amine unit or the reinjection system and must be minimized. This paper discusses how to minimize the level of hydrocarbons in rich amine streams, and thereby protect and enhance the efficiency of the amine units and reinjection equipment. Several strategies that can be implemented by gas processing operations to mitigate hydrocarbon contamination of the rich amine stream will be outlined. These strategies range from reducing hydrocarbons in the gas feed, to operation of the amine absorbers, to using hydrocarbon separation technologies on the rich amine stream feeding the regenerator. Ying Wu, John J. Carroll and Weiyao Zhu (eds.) CO2 Capture and Acid Gas Injection, (161–184) 2017 © Scrivener Publishing LLC 161 162 Carbon Dioxide Capture and Acid Gas Injection The problems immediately resulting from hydrocarbon contamination of the amine acid gas being routed to the acid gas compressor and their effect on the plant operation and efficiency are also identified and discussed. 9.1 Introduction There is a global trend of increasingly stringent environmental air quality legislation. For instance, in the United States, the Clean Air Act limits the emissions of Volatile Organic Compounds (VOCs) to 250 tons per year and the total amount of Benzene, Toluene, Ethylbenzene and Xylene (BTEX) emissions to 25 tons per year for a facility. Flaring sour gas is also prohibited by strict SO2 emission guidelines. For these reasons, flaring or venting acid gas streams is falling out of favor and reinjection of lower level H2S acid gas streams is growing in popularity. The process of removing H2S and CO2 from natural gas streams using amine systems is well established; however, on occasion heavier hydrocarbons are also removed. Hydrocarbons in amine acid gas are known to cause process issues in downstream Acid Gas Injection Systems (AGIS). Thus, there is strong operational and legislative motivation to minimize the amount of hydrocarbons in the acid gas and in the amine system. Reducing hydrocarbons in amine systems is advantageous to not only the AGIS, but also within the amine system. Hydrocarbons in amine can result in foaming, fouling, pipe vibration, and destruction of gaskets in plate and frame lean/rich exchangers. Hydrocarbons in amine also represent lost production for the plant. Amines are not meant or designed to remove hydrocarbons, which are the components of natural gas (or Liquefied Petroleum Gas (LPG)), which are used as fuel or sold. 9.2 Amine Sweetening Process Utilizing alkanolamines for acid gas removal is a process that has been in use since 1931. Today, amine treating is used for H2S and CO2 removal worldwide in gas plants, refineries, steel plants and power plants. Simplistically, an amine system absorbs CO2 and H2S contaminants out of a gas or LPG stream in a contactor unit (absorber) at high pressure and low temperature, and binds the contaminants to an amine molecule Strategies for Minimizing Hydrocarbon Contamination 163 through a set of chemical reactions. The binding reaction is then reversed in a stripper unit (regenerator) at low pressure and high temperature. The CO2 and H2S loaded amine (referred to as “rich” amine) is thus regenerated for reuse, and the CO2 and H2S exit the system via the acid gas stream and may be further processed (see Figure 9.1). Other than some specialty applications, the inlet gas to most amine absorbers will contain a variety of hydrocarbons. When this occurs, most gas-phase hydrocarbons will remain as gas and flow harmlessly up though the amine and exit out of the top of the tower. In the event of liquid-­ liquid treating, liquid hydrocarbons in treater will mix with the amine as intended, but then separate if a proper amount of residence time is given to the amine in the bottom of the treater. In a gas-liquid contactor, very little separation time is included in most designs and therefore, all liquid hydrocarbons will travel with the rich amine. If the rich amine entering the regenerator contains hydrocarbons, they will then vaporize and likely travel with the acid gas (or be condensed and recirculated to the regenerator, increasing the foaming risk). If the acid gas is routed to an acid gas injection system, the hydrocarbons can create problems, both operationally and mechanically. In the experience of Sulphur Experts, Amine Acid Gas (AAG) containing more than 2% hydrocarbon is an indication of problems upstream of the regenerator. This paper discusses how to mitigate hydrocarbons in the rich amine feeding the regenerator, thus reducing fouling, foaming, corrosion and many associated operating problems both in the Amine Regeneration Unit (ARU) and the AGIS. Amine gas-treating plant Sweet gas Acid gas Amine storage tank Regenerator Absorber Feed gas Rich solution Figure 9.1 Generic Amine Plant. Lean solution 164 Carbon Dioxide Capture and Acid Gas Injection 9.3 Hydrocarbons in Amine Hydrocarbons first enter the amine system in the contactor. Any hydrocarbons that end up in the rich amine will arrive there either by absorption (solubility), entrainment, condensation or emulsification. Amines are organized into three categories: primary, secondary and tertiary. A paper published by Critchfield et al. [4] explained how hydrocarbon solubility in amines relates to the molecular weight of the amine. The order of hydrocarbon solubility, in order of lowest to highest is as follows: Monoethanolamine Diglycolamine Diethanolamine Methyldiethanoamine Diisopropanolamine There are two main types of hydrocarbons that can enter an amine plant: polar and non-polar. Non-polar hydrocarbons are generally free hydrocarbons, which do not blend well with amines and can be relatively easily separated and removed. Polar hydrocarbons have a unique chemical characteristic, whereby part of the molecule is hydrophobic and the other part is hydrophilic. This basically results in an emulsion of amine and hydrocarbon. If the level of contamination is high enough in the amine, significant amounts of hydrocarbons (being pulled along by the surfactant) are carried with the solution to the flash tank. Although slightly different in chemistry, aromatics (i.e., BTEX) exhibit characteristics similar to polar hydrocarbons, and are also harmful to the AGIS. The thesis of Borda [1] provides a good review of the available data on BTEX and VOC solubility in amine solutions. Bullin & Brown [2] proposed theoretical conditions for minimizing the pick-up of hydrocarbons and BTEX. This was based on simulation. The accuracy of these simulations has been called into question by Borda [1]. Borda [1] presents new BTEX data fitted to a different correlation and compares the results from this, and results from the same simulator used by Bullins and Brown [2], to plant data. The simulator significantly overpredicted the actual hydrocarbon solubility. Overprediction of hydrocarbons in rich amine by a simulator is hard to rationalize. It is expected that real plant results would always have more hydrocarbons than predicted by a simulator. It is important to note that most data for hydrocarbon solubility in amine solutions is based on lean amine solutions that are not loaded with aqueous CO2 and H2S. Hatcher et al. [5] explain the overprediction of hydrocarbon solubility in rich Strategies for Minimizing Hydrocarbon Contamination 165 Figure 9.2 Amines Contaminated with Varying Degrees of Hydrocarbon. amine through a “salting-out” phenomenon, as described further in the “operational parameters” section of this report. Hatcher et al. [5] use an ­alternative simulator to Borda [1] and it is based on mass transfer rather than equilibrium models. Figure 9.2 is an example of different levels of hydrocarbon contamination that can occur within various amine systems. Hydrocarbons in rich amine streams can be present essentially in the following three forms, as previously detailed by Spooner and Engel [9]: Free Hydrocarbons. These non-polar hydrocarbons will float on top of the amine solution within a few minutes if given the chance (see Figure 9.3). This typically occurs in the flash tank. Soluble Hydrocarbons. All hydrocarbons will have certain solubility in amine solutions (see Figure 9.4). The extent of the solubility will depend on the following: Type and concentration of amine (common hydrocarbons such as C1/C2/C3 are two to three times more soluble in amine compared to pure water [3]), pH of the amine, Amine contactor pressure and temperature, and Type of hydrocarbon and polar functional groups such as carboxylic acids and alcohols. Aromatics are included in this group, as is explained further in this paper. Emulsified Hydrocarbons. When surfactants are present, hydrocarbon contaminants can form very small droplets in the amine solutions (see Figure 9.5). These droplets are stabilized by molecular surfactants (similar to soaps or detergents) and also by small-size suspended solids. Emulsion droplet sizes can range from a few microns to about 500 microns. Micro-emulsions, which are the most stable emulsions available (and can take weeks to separate) are typically found when droplet sizes are <10 microns. 166 Carbon Dioxide Capture and Acid Gas Injection Figure 9.3 Free Hydrocarbons in Amine. 9.4 Effect of Hydrocarbons on the Acid Gas Reinjection System Traditionally, acid gas streams were vented, flared, or sent to a sulphur plant. The modern approach to low-level H2S acid gas management is to reinject the gas stream leaving the amine system back downhole. It is advised to minimize hydrocarbons in reinjection systems for several reasons: Light hydrocarbon reduce the density of the acid gas which increases the required injection pressure. This in turn means more compression power. Non-acid gas components in the mixture means more volume to compress. The power needed by the compressor is almost directly proportional to the volume. Strategies for Minimizing Hydrocarbon Contamination 167 Figure 9.4 Soluble Hydrocarbons Concentrated in Reflux Water. Heavier hydrocarbons (C4+) increase the dew point of the acid gas mixture and may lead to unwanted condensation of the acid gas (especially on the interstage). The presence of too much methane in the acid gas may result in a gas bubble forming in the well which impairs and may even prevent injection. 9.5 Effect of Hydrocarbons on the Amine Plant Hydrocarbons do not chemically bond to the amine, so there is no direct harmful effect on the actual amine solution. Given the right amount of time, or use of technology, any hydrocarbons mixed into the amine can be removed and the amine can then be reused. In the meantime, however, while a mixture of hydrocarbon and amine is circulating through the plant, many negative consequences can result, the most common being as follows: Foaming. Probably the first and foremost concern when amines are contaminated with hydrocarbons is foaming. Hydrocarbons have a lower surface tension than amine, 168 Carbon Dioxide Capture and Acid Gas Injection Figure 9.5 Emulsified Hydrocarbons in Amine. which allows the surface of the liquid to expand quite easily. When gas or steam is bubbled through the amine, the bubble reaches the surface of the liquid, but does not “pop”. The bubble remains, and more and more bubbles build up on top of it until the entire vapor space is filled with this foam, as illustrated in Figure 9.6. Foaming can be a concern in both the contactor and the regenerator. When amines are in a foam state, they do not remove H2S in the contactor and cannot be regenerated in the regenerator. Fouling. Hydrocarbons contribute to the “black shoe polish” that commonly fouls amine filters, lean/rich exchangers and packed towers (see Figure 9.7). A carbon bed is used to remove hydrocarbons, but because the carbon is generally on the lean side, the amine has to flow through many pieces of equipment beforehand, where hydrocarbons can form a Strategies for Minimizing Hydrocarbon Contamination 169 Figure 9.6 Foaming Tendency of Hydrocarbon Contaminated Amine. Figure 9.7 Black shoe polish on rich amine filters. 170 Carbon Dioxide Capture and Acid Gas Injection matrix along with iron sulphides, salts, degraded amine, and antifoam. This is what often fouls amine systems. Gasket Destruction of Plate and Frame Exchangers. There are several different types of gaskets available for separating the plates in plate-and-frame exchangers. None of these gaskets are immune to the harmful effects of hydrocarbons flashing across the exchanger. Liquid hydrocarbons can cause polymerization of the gasket material and flashing hydro­carbons can erode the gaskets. Operators must rely on the flash tank to minimize the hydrocarbon content of the rich amine entering the exchanger. Otherwise the risk of gasket failure and leaking substantially increases, as shown in Figure 9.8. Figure 9.8 Leaking exchanger gaskets. Strategies for Minimizing Hydrocarbon Contamination 171 Loss of Treated Product. Hydrocarbons in amine represent a loss of hydrocarbons in the treated gas (i.e., result in lower volumes of sales gas or LPG). These losses directly impact the profitability of the gas processing system, and are obviously undesirable. 9.6 Minimizing Hydrocarbon Content in Amine Acid Gas The best way to minimize hydrocarbons in amine acid gas is to ensure that excess hydrocarbon does not enter in contact with the amine solution in the first place. This requires a comprehensive and thorough evaluation of the inlet gas stream to each amine contactor in the system. Typical amine plant designs will include at least an inlet separator before the contactor to knock out free liquids. However, many of these inlet separators are inadequate for complete hydrocarbon liquid removal. Other options for minimizing the hydrocarbon content of the amine acid gas include the following: Option 1. Optimizing the operation of the actual amine plant. Option 2. Optimizing the amine flash tank operation. Sometimes design changes to the tank interior may be necessary to ensure minimal hydrocarbon breakthrough. Option 3. Using filtration and/or coalescing technology on the rich amine to remove hydrocarbons. Option 4. Using potential skimming capabilities on contactor, flash tank and reflux water. Option 5. Technological solutions. 9.6.1 Option 1. Optimization of the Amine Plant Operation Once the plant is designed, built and operating, there are two strategies operators can employ to minimize the level of hydrocarbon pick up by the amine: (i) preventing hydrocarbon entering the contactor with the sour gas in the first place, and (ii) operating the plant at conditions that minimize hydrocarbon solubility in amine, noting that examples include choice of amine type and strength, circulation rate, rich loading, differential temperature between lean amine and inlet gas, and reflux operations. 172 Carbon Dioxide Capture and Acid Gas Injection Preventing Hydrocarbon Ingress. Assessing the amount of hydrocarbon entrainment in gas can be directly analyzed. Several companies perform entrainment testing at moderate costs. When done correctly, the associated optimizations that are possible as a result of the new data rapidly pays back the cost of the testing. The resulting changes that are made to process conditions can ensure reliability, integrity, capacity and energy/chemical utilization, thereby resulting in valuable cost savings to the plant. Bulk liquid hydrocarbons are meant to be removed from inlet gas streams by the inlet separator. More precision removal can be done using a cyclone separator and/or a coalescing filter. The inlet separator is the most important piece of equipment as far as hydrocarbons in amine acid gas are concerned—if the inlet separator fails, there will be serious consequences for the sulphur plant. Cyclones, centrifuges and coalescers are primarily used to prevent foam promoting contaminants from entering the amine contactor, which is of course important, but will not necessarily make much of an impact on the level of hydrocarbons in the AAG. Inlet separators rely on four basic parameters which ­determine the effectiveness of liquid separation from gas: Density Difference between the Liquid and Gas. Higher density liquids will be removed easier from gas than lighter ones because of the lack of gas solubility in the hydrocarbon. The two phases want to separate. Flow Directional Change. Flow direction changes are possible for the gas, but not so much for the droplets of liquid. Forcing the gas around an impingement plate followed by a demister pad in the top of the vessel creates something similar to an obstacle course. The gas can go through it, but liquid droplets impinge on the surface of the obstruction and eventually build in size until the droplets fall to the bottom of the separator. Figure 9.9 is a typical inlet gas separator, with a diverter plate and demister pad for flow directional change. Velocity. Velocity has a large effect on the volume of liquid hydrocarbon in a gas stream. The velocity of the gas stream imparts drag force on each liquid droplet, pulling the droplet along the pipe. The only opposing force Strategies for Minimizing Hydrocarbon Contamination 173 Gas outlet Mist eliminator Diverter plate Inlet gas Liquid Liquid outlet Figure 9.9 Typical Inlet Gas Separator. to counter this is gravity. It is important that gas velocities not be so high that they overcome gravity. Gas flow, as well as gas pressure, determines the velocity through the piping. This is why a separator has to be carefully designed to minimize pressure drop across the vessel, since a drop in pressure results in the gas expanding and a corresponding increase in velocity and drag force. Time. Time is the final separation parameter. It takes time for gravity to pull droplets out of a gas stream. Therefore, a larger separator tends to remove more liquids than a smaller vessel, assuming proper design of each. Proper operation of an inlet separator involves ensuring the liquid level is kept low at all times. The frequency of the level control valve opening should be noted, since frequent dumping of the vessel could mean an excessive amount of liquids having to be removed from the gas stream. This could indicate that that there is a possible problem upstream that requires investigation. Separators should always operate with a low and consistent pressure drop. Low pressure is required to prevent excessive drag forces on liquid 174 Carbon Dioxide Capture and Acid Gas Injection droplets and consistent pressure is to ensure that there is no fouling or plugging of the demister pad within the vessel. Plant Operating Conditions. There are several operational parameters that directly affect the amount of hydrocarbon in amine acid gas: Amine Circulation Rate. Because of the inevitable solubility of hydrocarbons in amine (described earlier in this paper), a higher amine circulation rate of amine will carry more hydrocarbons into the circulating rich solution. Furthermore, higher amine circulation rates decrease the flash tank residence time, lowering the hydrocarbon removal efficiency. Amine Rich Loading. By increasing the loading (aqueous CO2 and H2S amine salt concentration) of the amine solution, less amine and water is available for interaction with the charged part of the hydrocarbon. This renders the hydrocarbon-hydrocarbon interactions stronger than the amine-hydrocarbon interactions, causing the hydrocarbon molecules to coagulate through hydrophobic interactions with one another. The fact that higher rich loadings reduce the amine-hydrocarbon solubility levels is even further reason to lower the amine circulation rate if possible. Differential Temperature between Lean Amine and Inlet Gas. Before gas streams enter the amine contactor, the gas streams pass through an inlet gas separator, which allows for liquid hydrocarbons to be separated. Therefore, the gas leaving the separator should be at the hydrocarbon dewpoint, meaning if the gas pressure were to increase or the temperature to decrease, hydrocarbons would condense out and form droplets in the gas line. It is important to not allow this to occur as it will result in liquid hydrocarbons in the rich amine. For this reason, it is recommended the inlet gas separator be located within 10 m of the amine contactor, which will minimize the risk of condensation of hydrocarbons along the pipeline. It is also recommended that this line be insulated. It is also possible for hydrocarbons to condense inside the actual amine absorber, which can happen if the gas is cooled while travelling up the contactor. This will hap- Strategies for Minimizing Hydrocarbon Contamination 175 pen if the lean amine being injected into the contactor is cooler than the inlet gas stream. In fact, because the hydrocarbon dewpoint of the gas changes as acid gases are removed (because the removal of acid gases, the hydrocarbon dewpoint temperature will be higher at the top of contactor as compared to the bottom), it is recommended that operators maintain a minimum five Celsius or Fahrenheit degree temperature differential between the lean amine and the inlet gas. On especially rich gas streams, a greater differential temperature may be required. Reflux Operation. Despite optimizing amine circulation rates, rich loadings and flash tank operations, there will still likely be hydrocarbons entering the regenerator. The last area where hydrocarbons can be removed before leaving with the acid gas is in the reflux system. Proper condensing of the regenerator overhead stream will minimize the level of hydrocarbon vapor and the hydrocarbons will instead circulate with the reflux. The recommended reflux temperature is between 35 and 45 °C, or 95 to 115°F. Within the amine industry, there is some variation of which end of this temperature range is recommended. Refiners tend to run on the higher end of this range to minimize the risk of ammonia salt precipitation. Gas plants, or systems with no ammonia ingress (these are the systems who would be utilizing an AGIS), should target lower reflux temperatures. Since hydrocarbons (including methanol) are condensed in the reflux, the concentration will increase if there is no reflux purge. An increase in hydrocarbon content in the reflux is undesirable, as this increases the risk of hydrocarbon carryover with the acid gas, and can also cause foaming and other operational problems in the regenerator. Hydrocarbon contaminated reflux streams should be either continuously purged to a sour water stripper or disposal tank. If purging is not available, completely emptying the reflux tank to disposal is acceptable. This should be done based on visual or laboratory analysis of reflux water (or better yet reflux skimming) for hydrocarbons. It is important to recognize the implications of operating outside the recommended reflux temperature range. 176 Carbon Dioxide Capture and Acid Gas Injection Specifically operating at less than 35 °C/95 °F will not only have limited effect on the amount of water and hydrocarbon in the acid gas, but will also increase the necessary reboiler duty since the cold reflux will be condensing an inordinate amount of steam traffic in the upper section of the regenerator. Operating at higher than the recommended reflux temperature will allow excessive water and hydrocarbon to escape with the acid gas, negatively affecting acid gas compressor operations as well as increasing the make-up water demand on the amine plant. 9.6.2 Option 2. Amine Flash Tanks Separation technologies based on pressure drop, velocity changes and residence time are among the most common separation systems used in oil and gas operations. All these technologies have the common theme of using simple concepts to attempt to solve a separation problem. One such example is the amine flash tank. This device removes off-gases by reducing the rich amine pressure downstream of an amine contactor. If designed correctly, these systems also provide limited liquid-liquid separation capabilities for free hydrocarbon removal since these hydrocarbon liquids will float to the top of the amine solution within 30 minutes and can then be skimmed. No emulsified or dissolved contaminant is affected. For a flash tank to effectively separate hydrocarbons from amine, there must be sufficient residence time as well as sufficient pressure reduction. Entrained hydrocarbon gases will flash off within three to five minutes, noting that the lower the flash tank pressure, the faster and more efficient hydrocarbon gases will flash. Flash tank pressure is determined by the necessary flash gas and rich amine pressures. Common flash gas destinations include the following: Low pressure fuel gas absorber inlets Flare Incineration Acid gas The rich amine must leave the flash tank and flow through the following: Rich amine filters (if they exist) Lean/rich exchanger The vertical piping up to almost the top of the regenerator Strategies for Minimizing Hydrocarbon Contamination 177 Generally, 70 psig/475 kPag is enough pressure to push both the gas and amine to their respective destinations. It is important to not pressure the flash tank any higher than necessary since this will reduce the amount of hydrocarbon flashing. If the flash gas is routed to an extremely low pressure flare or incinerator, some plants will operate the flash tank at virtually atmospheric pressure and will install rich amine pumps immediately after the flash tank in order to push the amine through the exchanger and into the regenerator. Flash tank residence time is a function of the size of the vessel, the level at which the amine is maintained inside the vessel, and the circulation rate of the amine. If the tank is designed only for two-phase separation, the tank will simply be an open tank with a gas outlet in the top and liquid drain at the bottom. Operators should set the level at 50 to 60% full. This maximizes residence time, while still allowing for vapor disengaging space, noting that when hydrocarbons flash from liquid to gas, the hydrocarbons expand in size by up to 300% and can therefore carry gas upwards as a result of high velocities. Having at least 40% of the tank as vapor space will allow for gravity to pull the droplets of that amine back out of the flash gas and into solution. For designers, sizing a 30-minute residence time is often achieved by assuming some variables in the Stokes Law calculations, such as hydrocarbon density and hydrocarbon droplet size. The typical values for these tend to be a specific gravity of 0.6 to 0.7 and droplet sizes of 150 microns and larger. It is important to note that these values do not account for the possibility of heavier hydrocarbons and micro-emulsions that exist in rich amine streams. The Stokes Law calculations also do not account for any possible surfactant contaminants, which aid in stabilizing micro-emulsions. Granted, designers must balance the size and cost of the flash tank with expected performance. Designing a flash tank for 10 micron droplet removal would make the flash tank so large that it would not be economical to build or fit within the battery limits of the unit. Assuming the residence time is adequate, free hydrocarbons floating on the amine surface can and should be skimmed. Depending on the flash tank design, these hydrocarbons may flow over a weir or into an internal bucket that is attached to the inside of the vessel walls. When skimming, it is preferential to be able to sample the skimmed liquid to ensure it is indeed hydrocarbon and not amine. Some skimming sections have site glasses where the interface between amine and hydrocarbon can be seen. These make it very simple for operators to prevent the skimming of amine to disposal. 178 Carbon Dioxide Capture and Acid Gas Injection A number of different flash tank designs are available, some certainly more effective than others. A growing trend is to incorporate metal mesh internals to promote coalescence, which can compensate for low residence times. These promising systems so far have provided marginal results because of poor understanding of highly fouling rich amine streams. Most flash tanks, whether two- or three-phase separators, are often sized based on correct parameters, but can lack the understanding of liquid and solid contaminant loading. To use coalescing mesh-pads correctly, these have to be designed not only according to the gas velocity across the pad (using the modified Souders-Brown equation), but should also consider liquid and solid properties in addition to internal flow geometry. Any disregard of these aspects will invariably lead to element flooding and liquid carry-over or fouling with an increase in differential pressure. Case Study In one case study used to verify the strategies presented in this paper, several operational changes were made to a gas plant in Wyoming. The effects of the hydrocarbon content of the acid gas were measured. This plant had an inlet gas composition involving: 614 kmol/h hydrocarbons 11.8 kmol/h BTEX 5% H2S 8% CO2 System pressure of 5740 kPa(g) As is shown in Table 9.1, as the amine circulation rate, flash tank pressure and reflux temperature were dropped, a significant decrease in hydrocarbon and BTEX content of acid gas was seen. 9.6.3 Option 3. Rich Amine Liquid Coalescers Presently, many amine units only separate hydrocarbons in the flash tank. It is commonly believed that flash tanks will be able to separate any hydrocarbon in the rich amine streams. This is correct to some extent. However, the reality is that emulsions in rich amine streams are very stable, with droplet sizes near or less than the micro-emulsion range (10 microns and smaller). If Stokes Law is used to calculate the required residence time for the separation of a 15 micron emulsion, the result will indicate approximately two days. If the Strategies for Minimizing Hydrocarbon Contamination 179 Table 9.1 Hydrocarbon reduction case study. Optimization strategies (in order of implementation) Acid gas hydrocarbon content Btex content Treated gas Original operational conditions: 50% MDEA 160 m3/h Lean amine temperature 50 °C Inlet gas temperature 40 °C Rich loading 0.14 mol/mol Flash tank pressure 758 kPa Reflux temperature 60 °C 1.36 kmol/h 12.24 kmol/h 2.4 ppm H2S 0.174% CO2 Decrease Amine to 70 m3/h (0.314 mol/mol rich loading) 0.44 kmol/h 5.19 kmol/h <1 ppm H2S 0.479% CO2 Flash tank pressure dropped to 480 kPag 0.275 5.10 <1 ppm 0.48 % CO2 Flash tank pressure dropped to 35 kPag, rich pump installed 0.017 3.16 <1 ppm 0.48 % CO2 Decrease reflux to 35°C 0.016 3.14 <1 ppm 0.48 % CO2 particle size is slightly smaller, then the separation time can be in the order of weeks or even months. Based on the low efficacy of residence tanks to properly separate these emulsified contaminants to the level required for feed into the regenerator, it is necessary to use secondary systems, such as coalescers. Coalescence is the recombination of two or more small liquid droplets to produce a single droplet that is larger in size. This phenomenon also takes advantage of Stokes Law, which relates the velocity of separation of a particle or droplet in a medium to the diameter of the contaminant, densities, viscosity and gravitational pull. As coalescence takes place, small micronsize contaminant droplets coalesce into fairly large droplets, resulting in an almost immediate separation from the continuous phase. Mechanical coalescing systems are basically comprised of the following three technologies: Inclined plates Metal mesh Microfiber based Because of the particle size and the high fouling properties of the solids and emulsified hydrocarbons in rich amine steams, only disposable filters 180 Carbon Dioxide Capture and Acid Gas Injection and microfiber-based coalescers are able to provide the proper particle removal and emulsion separation. Other systems, such as back-flushable metal-based filters, do not work because the adhesion of solids to the surface is too strong, and prevents a backwash from being effective. Suspended solids removal upstream of the coalescer is mandatory, as this will protect the coalescer elements and will also help destabilize the emulsion. If solids are introduced into a coalescing filter, it will plug off almost immediately and become ineffective. Liquid-liquid coalescers are devices designed to separate small emulsified liquid contaminants in a liquid stream. These devices are segregated into the following two general categories: Low efficiency systems with metal-based internals Coalescers with microfiber internals The two types of coalescers perform rather differently and should be used for different objectives and specific cases. While metal-based internal coalescers are effective for separating free liquids and macro-emulsions (~100 microns and larger), coalescers with microfibers are more suited for separating micro-emulsions (100 microns and smaller). 9.6.4 Option 4. Use of Skimming Devices Hydrocarbons which float to the surface of amine or reflux water may be skimmed, assuming the design of the vessel allows for skimming. Most flash tanks have skim connections or hydrocarbon carryover weirs included in the design. Some contactors, reflux accumulators and regenerators also have skim connections. Whenever possible, hydrocarbons should be skimmed out of the system in order to prevent any chance of evaporation. Areas with higher residence times, such as flash tanks and regenerator bottoms have the potential for particularly high hydrocarbon volumes, as shown in Figure 9.10, which presents a side-by-side comparison of regenerator skimmings and regenerator bottoms, taken less than one minute apart from the same regenerator. Although not commonly thought of being a “filter”, the regenerator absolutely will drive hydrocarbons out of the amine, in effect filtering the solution. The vaporized hydrocarbons enter the reflux system where the hydrocarbons are possibly condensed in the reflux and where if not removed, will eventually be recycled back to the regenerator with the reflux water. Reflux water that is contaminated should therefore be partially purged, or better yet skimmed. Figure 9.11 shows hydrocarbons floating on reflux water, a common sight in hydrocarbon-contaminated amine systems. Strategies for Minimizing Hydrocarbon Contamination 181 Figure 9.10 Regen skimmings (left), Regen bottoms (right). Figure 9.11 Reflux water with skimmable hydrocarbon. 182 Carbon Dioxide Capture and Acid Gas Injection 9.6.5 Option 5. Technological Solutions More stringent legislation has led to additional treatments being proposed in order to meet current and future environmental specifications. Morrow [7] proposed the addition of a stripping column for hydrocarbons on the rich amine line (see Figure 9.12). This design uses a portion of the sweet gas to strip VOC and BTEX from the amine. The design has been patented and is described by Morrow [7], Morrow & Wallender [10], Morrow & Lunsford [6] and McIntyre et al. [8]. A VOC and BTEX removal of greater than 75% is apparently possible, Morrow & Wallender [10]. Bullin & Brown [2] modelled the performance of a hypothetical hydrocarbon stripping column on an Methyldiethanolamine (MDEA) plant and found it to remove 70% of the benzene (although 10% of the CO2 in the stream was also liberated, which could certainly be a problem). Bullin & Brown [2] also model a hot flash vessel (see Figure 9.13) that is located downstream of the lean-rich heat exchanger on the rich amine stream. Essentially, a higher temperature flash leads to more hydrocarbons being removed. The simulations show that this is more effective at removing VOCs and does not have a significant impact on BTEX. Unfortunately, substantial amounts of CO2 and H2S are liberated at these temperatures with the VOCs requiring an additional amine contactor for these gases. Sweet gas Acid gas to SRU Hydrocarbons Absorber Regenerator Lean/rich exchanger Flash gas Feed gas Rich amine flash Hydrocarbon stripper Lean amine pump Figure 9.12 Amine plant with hydrocarbon stripping column. Strategies for Minimizing Hydrocarbon Contamination 183 Sweet gas Acid gas to SRU Hot flash gas Absorber Lean/rich exchanger Flash gas Regenerator Hot flash & amine treating Feed gas Rich amine flash Lean amine pump Figure 9.13 Amine system with hot flash vessel. References 1. Borda, R.T. 2011. Experimental Measurement of Multi-component BTEX Solubility in Amine Solutions. M.Sc. Thesis, University of Oklahoma, 2011. 2. Bullin, J.A., and W.G. Brown. 2004. Hydrocarbon and BTEX Pickup and Control from Amine Systems. Proceedings of the 83rd Gas Processors Association Convention, 2004. 3. Carroll, J.J., J. Maddocks, and A.E. Mather. 1998. The Solubility of Hydrocarbons in Amine Solutions. Laurence Reid Gas Conditioning Conference, March 1998. 4. Critchfield, J., P. Holub, H. Ng, A. E. Mather, F. Jou, and T. Bacon, “Solubility of Hydrocarbons in Aqueous Solutions of Gas Treating Amines”, Proceedings of the 2001 Laurance Reid Gas Conditioning Conference. 5. Hatcher, N.A., C.E. Jones, and R.H. Weiland. 2013. Hydrocarbon Solubility in Amine Treating Solvents: A Generalized Model. Laurence Reid Gas Conditioning Conference, February 2013. 6. Morrow, D.C., and K.M. Lunsford. 1997. Removal and Disposal of BTEX Components from Amine Plant Acid Gas Streams. Pages 171 to 173 of the Proceedings of the 76th Annual GPA Convention. San Antonio, Texas, 1997. 7. Morrow, D.C. 1996. Removal/Disposal of BTEX Components in Amine Systems. Permian Basin Regional GPA Meeting, 1996. 184 Carbon Dioxide Capture and Acid Gas Injection 8. McIntyre, G.D., V.N. Hernandez-Valencia, and K.M. Lunsford. 2001. Recent GPA Data Improves BTEX Predictions for Amine Sweetening Facilities. Proceedings of the 80th Gas Processors Association Convention, 2001. 9. Spooner, B.H., and D.L. Engel. 2012. Reducing Hydrocarbons in Sour Water Stripper Acid Gas. Sulphur 2012 Conference, November 2012. 10. Wallender, J.W., and D.C. Morrow. 1999. Reduction of BTEX Emissions From Amine Plant Acid Gas Streams. GRI Gas Industry Air Toxics Conference, San Antonio, Texas, May 24–26, 1999. 10 Modeling of Transient Pressure Response for CO2 Flooding Process by Incorporating Convection and Diffusion Driven Mass Transfer Jianli Li and Gang Zhao University of Regina, Regina, Saskatchewan, Canada Abstract Traditionally well testing models simulate the CO2 flooding process as tworegion or three-region composite models, which normally neglect the dynamic mass transfer process and over-simplify the transient viscosity in transition zone. This leads to insufficient technical capacity to deal with more complicated field ­situations where reservoir heterogeneity largely affects CO2 injection process. Aiming at eliminating this restriction/limitation, this study proposes a comprehensive transient pressure model for CO2 flooding. It consists of two submodels for the pressure and mass transport processes, respectively. The proposed reservoir physical system is actually an enhanced three-region composite model which includes CO2 bank, transition zone, and oil zone. Pressure change in each region are ­properly modeled and CO2 concentration change in the transition zone is also reasonably described. The entire model is solved in a semi-analytical manner and a trapezoidal approximation scheme is used for variable flow velocity and oil ­viscosity. Type curves of the proposed comprehensive transient pressure model are plotted. Four flow regimes are identified: the early radial flow, the transition flow, the pseudo-radial flow, and the boundary-dominated flow regimes. The investigation of the impact of mass transfer process in the transition zone concludes that it mainly affects the transition flow regime with much slower slop change and the pseudo-radial flow regime with lower straight line compared with the case without considering the mass transfer process in the transition zone. This is caused by a gradual change in properties, such as total compressibility and viscosity, from a CO2 bank to a transition zone rather than an abrupt change for properties from a CO2 bank to an untouched oil zone. Sensitivity analysis shows that an injection Ying Wu, John J. Carroll and Weiyao Zhu (eds.) CO2 Capture and Acid Gas Injection, (185–198) 2017 © Scrivener Publishing LLC 185 186 Carbon Dioxide Capture and Acid Gas Injection scheme with smaller injection rate and longer injection period is better for viscosity reduction due to time dependent dynamic process than that of a larger injection rate and a shorter injection period in terms of viscosity reduction with the same amount of injected CO2. 10.1 Introduction As one of the most plentiful compounds found on this planet, the idea of using CO2 to remove oil from underground reservoirs originated as early as 1952 when a patent for an oil recovery method with CO2 was proposed. Through six decades’ development, CO2 flooding has been widely applied to enhance oil recovery all over the world. In United States, all the largest CO2 flooding projects were miscible displacements of medium-to-heavy oils with a number of injection and producing wells, except for an immiscible Yates Sand flood in West Texas. There have been 120 CO2 flooding projects by 2012. About three billion cubic feet of CO2 is supplied every day for the flooding projects, leading to an incremental oil production of about 350,000 bbl/d. It is estimated that 42 billion barrels of recoverable reserves has been added [Winslow, 2012]. CO2 flooding is carried out by injecting a large amount of CO2 into a reservoir. This process can be either an immiscible or a miscible process depending on whether the operating pressure achieves the minimum miscibility pressure (MMP) or not. Immiscible processes were studied in the 1950s and 1960s, and the recovery mechanisms were identified to be reduction in oil viscosity, oil swelling, and dissolved-gas drive. Miscible displacement can be achieved when the operating pressure is greater than MMP. It is much more effective than an immiscible displacement in terms of oil recovery. Production mechanisms in a miscible process involve: CO2–hydrocarbon miscible drive, hydrocarbon vaporization, direct ­miscible CO2 drive, and multiple-contact dynamic miscible drive [Holm and Josendal, 1974; 1987]. Traditionally, well testing models simulate the CO2 flooding process as two-region or three-region composite models, which normally neglect the dynamic mass transfer process and oversimplify the transient viscosity in the transition zone. This leads to insufficient technical capacity to deal with more complicated field situations where reservoir heterogeneity largely affects the CO2 injection process. This study proposes and develops a comprehensive well testing model for CO2 flooding process by incorporating convection–diffusion driven mass transfer process in modeling transient pressure response. The pressure model Modeling of Transient Pressure Response for CO2 Flooding 187 covers three regions (gas bank, transition zone, and oil zone) while the mass transfer model describes fluid dynamics in a transition zone. The two models are reasonably coupled together. Bottom-hole pressure is calculated and type curves are plotted. Effects of reservoir and fluid properties and operating parameters on dimensionless pressure and its derivative are analyzed. 10.2 Model Development 10.2.1 Pressure Diffusion A reservoir undergoing a CO2 miscible flooding process consists of three zones: a CO2 bank, a miscible zone, and a crude oil bank (Figure 10.1). CO2 and crude oil banks are assumed to be filled purely with carbon dioxide and crude oil, respectively. The mixing of crude oil and CO2 happens only in the miscible zone. Also, water phase is eliminated in this study and the reservoir has a radial shape with uniform porosity and permeability. The entire model contains two submodels: pressure diffusion over all the three regions and mass transfer in the transition zone. CO2 bank is considered to be filled with gas and residual oil. In the transition zone, the mixture of crude oil and CO2 is assumed to be in liquid state. The untouched oil zone is treated as an extension of the miscible zone in which concentration equals zero. The pressure distribution over the entire model is described by a diffusivity equation [Lee, 1996]: 1 P P r r r z r P Ct P ,r z k t w Well rw rBK CO2 bank rMZ Miscible zone Oil zone Figure 10.1 Schematics of a reservoir undergoing CO2 flooding r re r1 (10.1) 188 Carbon Dioxide Capture and Acid Gas Injection Ct P ,r k t 1 1 P r r r r r re (10.2) where r is the radial distance from an injector, m; P is the pressure, Pa; µ is the viscosity, Pa.s; z is the compressibility factor, dimensionless; is the porosity, dimensionless; k is the permeability, m2; Ct is the total isothermal compressibility, Pa–1; r1 is the location of a preset boundary between a CO2 bank and a miscible zone, m; and t is the time variable, s. The viscosity, compressibility factor, and total isothermal compressibility are all functions of pressure. At the inner boundary (injector borehole radius), CO2 is injected at a constant rate. The outer boundary is considered to have three types: a closed boundary, a constant pressure boundary, and an infinite boundary. Initially, the gas reservoir has a uniform pressure distribution. r P r P re , t q Bg P r (10.3) 2 kh rw 0 (Closed boundary) r re (10.4) Pi (Constant pressure boundary)(10.5) P r ,t P r ,0 Pi (Infinite reservoir)(10.6) Pi j 1, 2, ,N (10.7) where rw is the wellbore radius, m; T is the temperature, K; h is the reservoir thickness, m; and subscript “sc” denotes the standard condition. 10.2.2 Mass Transfer The mixing of CO2 and crude oil is caused by not only molecular diffusion but also convective dispersion across the CO2–oil miscible zone. The ­governing equation of a mass transfer process in a radial system: c t D 2 c r 2 1 c r r V c V c ,r r r 1 r re (10.8) Modeling of Transient Pressure Response for CO2 Flooding 189 where c is the concentration, dimensionless; D is the diffusivity, m2/s; and V is convection velocity, m/s. The convection velocity is calculated by using Darcy’s law: k V P (10.9) The inner boundary of the miscible zone is assumed as completely ­saturated with CO2 all the time, which indicates a Dirichlet BC. The outer boundary of the miscible zone is regarded as a no-flow boundary and a Neumann BC is applied. Initially, the model is free of CO2. The BCs and IC are described: c* cr 1 cr 0 e ct (10.10) (10.11) (10.12) 0 0 Viscosity r i 1 P r ra1 P1 P r N ri Pi P r ran Pn Figure 10.2 Schematics of a trapezoidal approximation of the viscosity profile in the miscible zone 190 Carbon Dioxide Capture and Acid Gas Injection where c* is the saturation concentration under certain operating ­conditions, dimensionless. 10.2.3 Solutions A pseudo-pressure and pseudo-time method is applied to accurately solve nonlinear governing equation, Eq. (10.1), in which the pseudo-terms are defined as m(P ) 2 t ap (t ) P P0 t 0 P dP z (10.13) 1 dt Ct (10.14) For the miscible zone, due to the complex relationships among v­ iscosity, pressure, and concentration, it is difficult to solve Eqs. (10.2) and (10.8) simultaneously. Thereby, this study compute the pressure and ­concentration stepwise. At a step, the true curved viscosity profile is simplified as a trapezoidal profile. A piecewise linearization scheme and Laplace transformation are applied to obtain the semi-analytical solutions to each segment of the transition zone. Then the solutions are coupled together through reasonable two conditions to ensure the ­continuity and smoothness on the pressure profile at the interface of neighbouring segments: Pj (rj , t ) Pj 1 (rj , t ) k Pj r rj k Pj 1 r (10.15) (10.16) rj here the subscript of pressure, “i” (i = 1, 3 … n), denotes the pressure of the ith segments of the miscible-zone. At last, a matrix form of equations can be formed which is solved by using the Guess Elimination algorithm [Muller, 2001]. Detailed solution theory can be found in a previous study [Jia et al., 2013]. The mass transfer model is solved in a similar manner as that for the pressure model. Two coupling conditions are applied at the interface in between two neighboring segments to ensure the continuity and smoothness on the concentration profile: Modeling of Transient Pressure Response for CO2 Flooding 191 Input D, Ct, k, , h, Bg, q Pressure transfer model G.E., BCs, IC Pressure, pressure gradient P, P Convection velocity V Mass transfer model G.E., BCs, IC Viscosity Concentration c No Termination condition Yes Finish Figure 10.3 Flowchart for calculating the solutions to the pressure diffusion and mass transfer models. c j (rj , t ) c j 1 (rj , t ) cj cj r r rj 1 (10.17) (10.18) rj Figure 10.3 shows a flowchart for calculating pressure and concen­tration profiles in the above-developed CO2 miscible model. 10.3 Results and Discussion 10.3.1 Flow Regimes The dimensionless pressure and its derivative for the transient pressure model of CO2 flooding considering mass transfer process are ­calculated. The main flow regimes are identified according to the type curves. Figure 10.4 shows the pressure characteristics of a CO2 miscible flooding with the consideration of mass transfer process. It can be seen that 192 Carbon Dioxide Capture and Acid Gas Injection 1000 4 3 pD, dpD/dln(tD) 100 2 10 1 1 The first radial flow regime 1 2 Transition flow regime 3 Pseudo radial flow regime 4 Boundary dominated flow regime 0.1 1.0E + 02 1.0E + 03 1.0E + 04 1.0E + 05 1.0E + 06 1.0E + 07 1.0E + 08 1.0E + 09 tD Figure 10.4 Type curves of the transient pressure model for CO2 miscible flooding considering mass transfer process (Ra = 10 m, Re = 150 m, Q = 0.01 m3/s, Ct = 1e-8, k = 10 mD, µ = 4.7 cP). there are mainly four flow regimes on the type curves: the first radial flow regime, transition flow regime, the pseudo-radial flow regime, and ­boundary-dominated flow regime. Figure 10.5a shows the relationship between CO2 concentration and distance as time goes. It can be seen that the CO2 concentration in the miscible zone increases and expands outwards with the continuous injection of CO2. Figure 10.5b shows the visco­ sity of CO2–oil mixture versus distance as time goes. It is an obvious trend that the larger the concentration of CO2 in the mixture, the greater the viscosity will be reduced. 10.3.2 Effect of Mass Transfer Viscosity reduction through CO2–oil mass transfer is a key production mechanism of CO2 miscible flooding processes. The varying visco­ sity in the miscible zone, in return, influences the pressure propagation. Figure 10.6 compares the transient pressure responses with and without considering the mass transfer process in the miscible zone. The differences in dimensionless pressure and pressure derivative caused by the viscosity profile in the miscible zone are quite obvious. This means that the mass transfer process should not be ignored for a reliable bottom-hole pressure analysis. The first flat section of pressure and pressure derivative curves with solid lines is due to a radial flow at the early time period. The following Modeling of Transient Pressure Response for CO2 Flooding 193 1 0.9 t = 1e3 seconds t = 1e4 seconds t = 1e5 seconds t = 1e6 seconds Concentration, Vol. % 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 10 15 20 25 30 35 Distance, m (a) 0.005 0.0045 0.004 Viscosity, Pa.s 0.0035 0.003 0.0025 0.002 0.0015 t = 1e3 seconds t = 1e4 seconds t = 1e5 seconds t = 1e6 seconds 0.001 0.0005 0 10 (b) 15 20 25 Distance, m 30 35 Figure 10.5 (a) CO2 Concentration and (b) Oil Viscosity distributions across the miscible zone (Ra = 10 m, Re = 150 m, Q = 0.01 m3/s, Ct = 10–8 Pa–1, k = 10 mD, µ = 4.7 cP). sharp increase of pressure and pressure derivative curves is caused by a sudden change in fluid properties, such as total compressibility and viscosity, from a CO2 bank to a miscible zone, which is defined as a transition flow period. With the decrease of the viscosity in miscible zone, the pressure and pressure derivative curves go down slightly. Then the flow arrives at the pseudo-radial flow period followed by boundary-­ dominated flow indicated by the unit slop in the pressure derivative curve. 194 Carbon Dioxide Capture and Acid Gas Injection 1000 pD, dpD/dln(tD) 100 10 1 With mass transfer With mass transfer Without mass transfer Without mass transfer 0.1 1.0E + 02 1.0E + 03 1.0E + 04 1.0E + 05 1.0E + 06 1.0E + 07 1.0E + 08 1.0E + 09 tD Figure 10.6 The effect of mass transfer process on dimensionless pressure and dimensionless pressure derivative. Table 10.1 Physical properties and operating conditions of the base case. Properties Symbol Value Unit Total Compressibility Ct 1 × 10-8 Pa Initial reservoir pressure Pi 15´106 Pa Well radius Rw 0.091 m CO2 bank radius Ra 10 m Reservoir outer radius Re 150 m Thickness h 1 m Permeability k 1 × 10–14 m2 Porosity f 0.26 1 Injection rate Q 0.01 m3/s Temperature T 333.15 K Pressure at standard condition Psc 1 × 105 Pa Temperature at standard condition Tsc 293 K Critical Pressure Pc 7.377 × 106 Pa Critical Temperature Tc 304.13 K Modeling of Transient Pressure Response for CO2 Flooding 195 10.3.3 Sensitivity Analysis 10.3.3.1 CO2 Bank This section investigates the sensitiveness of four reservoir and fluid ­properties and one operating condition to the pressure responses during a CO2 miscible flooding process. In a CO2 miscible flooding reservoir, the radius of each region indicates the drainage area and has a direct effect on transient pressure responses. Figure 10.7 shows the effect of CO2 bank size on dimensionless pressure and pressure derivative responses. It is found that when the radius of CO2 bank equals to 5 and 10 m, there are two radial flow regimes which are characterized by a 0.5 slope-line at an early time period and a horizontal straight line before the occurrence of boundary effect, respectively. A larger CO2 bank radius leads to a longer first radial flow regime at the early time period and a shorter pseudo-radial flow regime at a later time period. For the case of Ra = 1 m, the first radial flow at the early time period is missing for two reasons. One reason is the big difference in the properties of the CO2 bank and the miscible zone. The viscosity and total c­ ompressibility of the CO2 bank are much lower and larger compared with those in the miscible zone, respectively. The other reason is the small radius (Ra = 1 m) of the CO2 bank, because of which the flow reaches to the radial flow period very quickly, as shown on the green curves of the dimensionless pressure and pressure derivative. In this case, although the first radial flow regime does not exist at the early time, it has the longest pseudo- radial flow regime compared with other cases (Ra = 5, 10, 20 m). For the case of 1000 pD, dpD/dln(tD) 100 10 Ra = 1 m Ra = 5 m Ra = 10 m Ra = 20 m 1 Ra = 1 m Ra = 5 m Ra = 10 m Ra = 20 m 0.1 1.0E + 02 1.0E + 03 1.0E + 04 1.0E + 05 1.0E + 06 1.0E + 07 1.0E + 08 1.0E + 09 tD Figure 10.7 The effect of CO2 bank size on dimensionless pressure and pressure derivative. 196 Carbon Dioxide Capture and Acid Gas Injection 10000 pD, dpD/dln(tD) 1000 Re = 100 m Re = 150 m Re = 200 m Re = 100 m Re = 150 m Re = 200 m 100 10 1 0.1 1.0E + 02 1.0E + 03 1.0E + 04 1.0E + 05 1.0E + 06 1.0E + 07 1.0E + 08 1.0E + 09 tD Figure 10.8 The effect of reservoir outer boundary size on the dimensionless pressure and pressure derivative curves. Ra = 20 m, the first radial flow regime is the longest. But the closed outer boundary effect occurs soon after the appearance of the pseudo-radial flow. 10.3.3.2 Reservoir Outer Boundary Three outer boundary radii of Re = 100, 150 and 200 m are tested to ­analyze the effect of reservoir outer boundary sizes on the transient pressure responses. It can be seen from Figure 10.8 that the longer the outer boundary radius, the longer the pseudo-radial flow regime and the later the appearance of the boundary effect. It is obvious that the bigger the reservoir, a longer time is needed for pressure to reach the outer boundary and thereby leads to a longer pseudo-radial flow regime. It is also found that before the appearance of the pseudo-radial flow regime, the dimensionless pressure and pressure derivatives of the three cases are exactly the same. Therefore, the size of the reservoir can be clearly reflected on the pressure responses. 10.4 Conclusions This thesis developed a comprehensive transient pressure model for a CO2 flooding miscible process. Through this study, some conclusions are made as following: 1. A convection–diffusion mass transfer model is developed to evaluate the dynamic mixing process between oil and CO2 in Modeling of Transient Pressure Response for CO2 Flooding 197 the miscible zone. It is then incorporated into the pressure model through the viscosity profile in the miscible zone. 2. Four flow regimes can be identified in the dimensionless pressure and pressure derivative profiles: the first radial flow regime, transition flow regime, the pseudo-radial flow regime, and boundary-dominated flow regime. 3. A comparison of two pressure responses with or without considering CO2–oil mass transfer in the miscible zone shows that mass transfer process has a significant effect on the transition and the pseudo-radial flow regimes. Thus, it is imperative to take the mass transfer process into account for a reliable modeling CO2 flooding process. 4. The size of the CO2 bank strongly influences the length of the first and the pseudo-radial flow regimes. A larger radius of the CO2 bank leads to a longer first radial flow regime and a shorter pseudo-radial flow regime. A trivial radius of CO2 bank can cause the missing of the first radial flow regime but a much longer the pseudo-radial flow period. Similarly, the size of the reservoir outer boundary can also be identified in the pressure responses. A bigger radius of the reservoir outer boundary results in a longer pseudo- radial flow period. Acknowledgments Funding support from the NSERC Discovery grant is highly appreciated. Thanks also go to KAPPA engineering for offering software licenses for academic use. References 1. Annual Production Reports, Oil and Gas J. 5 April, 1982. 2. Baviere, M. Basic Concepts in Enhanced Oil Recovery Processes. Published for SCI by Elsevier Applied Science, London and New York, 1991. 3. Holm, L. W. and Josendal, V. A. Mechanisms of Oil Displacement by Carbon Dioxide. Journal of Petroleum Technology, 26(12), 1427–1438, December 1974. SPE-4736-PA. 4. Holm, L.W. CO2 Flooding: Its Time Has Come. Journal of Petroleum Technology, 34(12), 2739–2745, December 1982. SPE-11592-PA. 5. Holm, W. L. Evolution of the Carbon Dioxide Flooding Processes. Journal of Petroleum Technology, 39(11), 1337–1342, November 1987. SPE-17134-PA. 198 Carbon Dioxide Capture and Acid Gas Injection 6. Jia, X., Zeng, F., and Gu, Y. Semi-Analytical Solutions to a One-Dimensional Advection, 2013. 7. Diffusion Equation with Variable Diffusion Coefficient and Variable Flow Velocity. Appl. Mathe. & Comput., 221(2), 268−281 8. Lee, J. and Wattenbarger, R. A. Gas Reservoir Engineering. Society of Petroleum Engineers. Richardson, Texas, 1996. 9. Martin, D. F. and Taber, J. J. Carbon Dioxide Flooding. Journal of Petroleum Technology, 44(4), 396–400, 1992. SPE-23564-PA. 10. Moritis, G. CO2 and HC injection Lead EOR Production Increase. Oil & Gas Journal, 49–82, 23 April, 1990. 11. Muller, K.E. Computing the confluent hypergeometric function, M(a, b, x), Numer. Math. 90(1), 179–196, 2001. 12. Mungan, N. Carbon Dioxide Flooding as an Enhanced Oil Recovery Process. Journal of Canadian Petroleum Technology, 31(09), 13–15, November 1992. PETSOC-92-09-01. 13. Stalkup, F. I. Carbon Dioxide Miscible Flooding: Past, Present, and Outlook for the Future. Journal of Petroleum Technology, 30(8), 1102–1112, August 1978. SPE-7042-PA. 14. Winslow, D. Industry Experience with CO2 for Enhanced Oil Recovery. Workshop on California Opportunities for CCUS/EOR, 27 June, 2012. 11 Well Modeling Aspects of CO2 Sequestration Liaqat Ali and Russell E. Bentley WSP | PARSONS BRINCKERHOFF, 16200 Park Row, Suite 200, Houston, TX 77087, USA Abstract The study presents the results of well modeling for CO2 sequestration wells completed into a deep porous brine aquifer reservoir. In CO2 injection projects, it is essential to design an interface between the reservoir and the pipeline. Inflow performance relationship (IPR) and vertical flow performance (VFP) analyses help design this interface. We present the results of well modeling of a vertical well and an extended reach directional (ERD) well in this study for a typical CO2 sequestration field. We also briefly discuss the effect of impurities in CO2 stream. The results showed that the temperature change at lower wellhead pressures along the wellbore was greater than that at higher wellhead pressures. The results also showed that injection rate requirements dictate the number of wells required. 11.1 Introduction In this study we describe CO2 flow through a vertical wellbore into a deep porous brine aquifer reservoir. Like any other oil and gas production and injection field, CO2 sequestration projects also require designing an interface between the pipeline and the reservoir. Therefore, the first and last nodes are at the wellhead and at the well total depth (TD) respectively. The IPR and VFP help design this interface. The IPR for a well is the relationship between the flow rate of the well and the flowing pressure of the well. The VFP relates to estimating the pressurerate ­relationship in the wellbore as the fluids move through the tubing. The CO2 flow through a vertical wellbore could be single phase or two Ying Wu, John J. Carroll and Weiyao Zhu (eds.) CO2 Capture and Acid Gas Injection, (199–220) 2017 © Scrivener Publishing LLC 199 200 Carbon Dioxide Capture and Acid Gas Injection phase depending upon the temperature and pressure condition in the wellbore. The phase change in the wellbore directly affects the well injection pressure. It is also important to know whether for the given tubing size, wellhead temperature, and pressure the required rates can be achieved. The IPR and VFP curves provide great insights into designing an optimal injection program. We have used PROSPER software to develop IPR and VFP curves and investigate the efficacy of the project’s requirements. The main objectives of the study were (1) to verify if the minimum wellhead pressure (WHP) of 110 Bara and a wellhead temperature (WHT) of 35 degrees C would be sufficient for a steady state injection rate of 3 million tons per annum (Mtpa) [4.5 million S m3/d (MSm3/d)] and (2) to quantify the injection capacity of vertical well as well as the ERD well under this scenario for pure CO2. 11.2 Delivery Conditions Due to variations in rates and temperatures and corresponding changes in wellhead pressures, various cases have been performed in this study. The pressure and temperature conditions, along with the pipeline delivery conditions for the VFP analyses, are shown on the phase diagram for pure CO2 (Figure 11.1). Phase envelope 160 Analyses performed: Pressure: 70–160 bara Temperature: 11–35 °C 120 Pressure (bara) Supercritical region Dense phase region 140 100 Delivery conditions (110 bara & 35 °C) 80 Liquid region 60 Gas Region 40 0 Critical temperature: 30.94 °C Critical pressure: 73.98 bara Vapor region 20 0 5 10 15 20 25 30 Temperature (°C) 35 40 45 50 Figure 11.1 Phase envelope of pure CO2 with delivery conditions and the conditions on which the sensitivities were performed. Well Modeling Aspects of CO2 Sequestration 201 11.3 Reservoir and Completion Data Reservoir and completion data are given in Table 11.1. The reservoir pressure and temperature have been calculated using a normal pressure gradient of 0.433 psi/ft at an average injection depth. Please note that the ERD well has much greater completion length compared to the vertical well. 11.4 Inflow Performance Relationship (IPR) and Injectivity Index The IPR analysis was conducted in PROSPER software to establish the required flowing bottom hole pressure to inject at a given rate for any given reservoir pressure. The multi-rate Forchheimer IPR model with pseudo-pressure was used to fit a set of rates and pressures obtained from reservoir simulations for vertical as well as deviated wells (Figure 2). The Forchheimer equation accounts for non-Darcy pressure losses in high flow rates situations as given below. (Pr2–Pwf2) = aQ2 + bQ(11.1) where: Pr = Reservoir pressure Pwf = Flowing bottom hole pressure Q = Flow rate Table 11.1 Reservoir and completion data. Well Vertical well ERD well TVD, mss 1500 1330 Completion length, m 140 960 7(0.1778) 7(0.1778) Average injection depth, m 1400 1300 Turbing size, in (m) BHP, Bara 138 126 BHT °C 64 60 Reservoir permeability, md 650 550 Reservoir porosity 0.17 0.17 Reservoir thickness, m 200 200 Wellbore thickness, m 200 200 Wellbore radius, in (m) 9 5/8 (0.22425) 9 (5/8) 0.22425 202 Carbon Dioxide Capture and Acid Gas Injection 160 Injectivty index: Vertical well: 0.5 (MSm3/d)/Bara ERD well: 0.3 (MSm3/d)/Bara Flowing BHP pressure (Bara) 155 150 145 140 135 Simulation data-vertical well IPR-vertical well Simulation data-ERD well IPR-ERD well 130 125 120 0 2 4 6 Gas rate (MSm3/day) 8 10 Figure 11.2 Comparison of the reservoir simulation data and the calculated IPR using Prosper for the vertical well and ERD well. a = Darcy pressure loss coefficient b = Non-Darcy pressure loss coefficient Figure 11.2 also presents the corresponding injectivity index for each well. The injectivity index is defined as the gas rate divided by the difference of flowing BHP and the reservoir pressure. The reservoir pressures of 138 Bara and 128 Bara for the vertical well and the ERD well are used respectively. The IPR relationships were obtained by matching the simulation data with the Forchheimer IPR model of PROSPER software. The IPR curves calculated using PROSPER are shown by the grey and black dotted curves for the vertical well and ERD well respectively. The simulation data is also presented as open circles and squares to show the fitness of the curves. 11.5 Equation of State (EOS) Two equations of state were available in PROSPER software: SoaveRedlich-Kwong (SRK) and Peng-Robinson (PR). Recently, Du et al. (2014) compared different equation of states including SRK and PR. They found that PR equation is the most precise EOS to compute density and predict CO2 phase behavior. Furthermore, Boyle and Carroll (2002) performed a study on predicting acid-gas (pure H2S, pure CO2 and mixture of the two) densities for the pressure and temperature ranges specific to acid-gas injection applications. They found that the SRK equation was unsatisfactory Well Modeling Aspects of CO2 Sequestration 203 for predicting densities over the entire range of pressure and temperature conditions used. Mazzoccoli et al. (2013) also reached a similar conclusion that the SRK equation showed higher errors in calculating densities for pure CO2 as well as a binary mixture of CO2 with Ar, O2, N2 and CH4, respectively. Based on this review, the PR equation was used for this study. Consistency checks were made to ensure that the PROSPER PR equation reproduces the experimental results in the pressure and temperature conditions of the well modeling for VFP analyses. These checks included comparing the PROSPER PR calculations of phase diagrams and d ­ ensity for pure CO2 and CO2 mixtures. The checks also included comparing viscosity calculated using Lohrenz-Bay-Clark with that calculated using Freng-Wakeham correlation. For pure CO2, the Span & Wagner (SW) EOS (1996) is based on extensive experimental data and is considered to be the most accurate EOS. Comparison of phase diagrams of PR and SW equations showed that the difference in the critical point is less than 1% (Figure 11.3). Comparison of PR and SW equations showed that the error in densities is less at lower temperatures and higher at higher temperatures at a pressure of 110 Bara for both PR and PR with volume shift (Figure 11.4). However, PR yields better results than that of PR with volume shift for operating conditions of this study. The error for PR ranges from 0.4 at 11 degrees C (dense phase region) to as high as 7.7% at 35 degrees C (supercritical region) at a pressure of 110 Bara. Phase envelope 80 Pressure (Bara) 70 60 50 EOS Temperature °C Pressure bara Peng-Robinson 30.94 73.9777 Span & Wagner 31.1282 73.773 Difference, % 0.60 0.28 40 30 20 Span & Wagner Peng-Robinson 0 5 10 15 20 25 30 Temperature (°C) 35 40 45 Figure 11.3 Comparison of phase diagrams of Peng-Robinson and Span & Wagner equations. 50 204 Carbon Dioxide Capture and Acid Gas Injection 1000 Span & Wagner PR PR with volume shift 900 800 12.0 600 Absolute error (%) Density (Kg/m3) 700 500 400 300 200 10 8.2 7.4 6.0 4.8 7.7 4.2 4.0 2.0 11.3 PR PR with volume shift 8.0 0.0 100 0 10.0 2.7 0.4 11 20 30 Temperature (°C) 15 20 35 25 Temperature (°C) 30 35 40 Figure 11.4 Comparison of errors and densities calculated from PR, PR with volume shift and Span & Wagner equations for WHP pressure of 110 Bara for pure CO2. 0.12 Fenghour Lohrenz (PROSPER) 0.08 0.06 Absolute error (%) Viscosity (mPa.s) 0.10 0.04 0.02 0.00 0 Errors in viscosity of pure CO2 20 18 16 14 12 10.8 10.110.2 10.5 10.6 10.8 10.5 10.4 9.3 10 8.0 8.7 8.0 8 6 4 2 0 0 4 8 12 16 19 23 27 31 35 Min Max Temperature (ºC) 10 20 Temperature (°C) 30 40 Figure 11.5 Comparison of viscosities and errors associated for pure CO2 for WHP pressure of 110 Bara. Viscosity was calculated using the Lohrenz-Bray-Clark correlation which was chosen among several correlations available in PROSPER software. A comparison of viscosities calculated using this correlation and the FenghourWakeham correlation (1998) was made for pure CO2. The Fenghour-Wakeham correlation was developed based on extensive experimental data with pure CO2 and is considered to be more accurate for pure CO2. Lohrenz-Bray-Clark correlation is based on reservoir fluids containing methane through heptanes plus, H2S, N2 and CO2. The results are presented in Figure 11.5. The error for Lorenz viscosities is in the range of 8 to 10.8% for viscosities. Well Modeling Aspects of CO2 Sequestration 205 1400 225 K 245 K 265 K 285 K 300 K 320 K 350 K 400 K 450 K Density (Kg/m3) 1200 1000 800 600 400 Operating conditions (table 1) Temperature: 284–308 K Pressure: 7–16 Mpa 200 0 0 10 20 30 40 Pressure (MPa) 50 60 70 Figure 11.6 Density of a mixture: CO2 (91%) and N2 (9%) showing experimental data (Carroll and Roberts, 2013) and predictions from PR EOS of PROSPER. The continuous lines are the calculated values. Experimental data from Roberts and Carroll (2013) was used to compare the densities calculated by PROSPER using the same mixture: CO2 (91%) and N2 (9%) to further check robustness of PR of PROSPER software. Figure 11.6 presents the results that confirm the conclusions arrived at by Roberts and Carroll (2013) in their paper that the greatest errors occur at lower temperatures. Low temperatures are not typically of interest in carbon sequestration applications. Figure 11.6 shows that the match between experimental and PROSPER PR equation density calculations is very good in the range of the operating conditions being considered for this study. 11.6 Vertical Flow Performance (VFP) Curves Several sets of VFP curves were developed for the vertical and ERD wells. These curves cover a wide range of flowing bottom hole (BHP) and WHP for wellbore geometries, wellhead conditions and reservoir pressures. We present the results of IPR-VFP analyses for reservoir pressures of 138 Bara and 128 Bara for the vertical well and ERD well respectively. A WHP of 110 Bara and a WHT of 35 degrees C were used to develop the VFP curves. The VFP curves were also developed for WHT of 11 degrees C, 20 degrees C (only for the vertical well), 30 degrees C and 35 degrees C for each well. These curves are not shown here; however, a summary of their results are presented. 206 Carbon Dioxide Capture and Acid Gas Injection Table 11.2 Critical points of the cases considered. Critical ­temperature °C Case Critical ­temperature Cricondentherm Cricondenbar Bara °C bara Pure CO2 30.9400 73.9777 30.9400 73.9777 CO2 with 5% contaminants 30.0089 77.7632 30.0347 72.3256 Phase envelope 80 75 Pressure (bara) 70 65 60 55 50 45 40 Pure CO2 CO2 with 5% contaminants 35 30 0 5 10 15 20 Temperature (°C) 25 30 35 Figure 11.7 Phase diagram of the compositions used in well engineering analysis for pure of CO2 and CO2 with 5% contaminants. In addition to pure CO2, CO2 mixtures (5% contaminants) were also used to perform VFP analyses. The graphical results of VFP analyses are not presented here; however, they are briefly discussed. The contaminants mainly consisted of H2O, N2, H2, Ar, O2, CH4, CO, H2S, C2H6, SO2 and NOx. For the worse case of NOx, only NO2 was used in the analysis. Table 11.2 presents the critical temperature and critical pressure for each case and Figure 11.7 presents the phase diagram. Addition of impurities and their molar concentration increase the two-phase region of phase diagram as shown in Figure 11.7. Figure 11.8 presents VFP curves for the vertical well for the delivery conditions: WHP of 110 Bara, 35 degrees C and an initial reservoir pressure of 138 Bara. Figure 11.8 shows that this well can handle gas rates up to 6.5 MSm3/d (4.3 Mtpa) at the required operating conditions which is well above the designed gas rate of 4.5 MSm3/d (3 Mtpa) for this study. WHT = 35 °C At current reservoir pressure, this well can make 6.5 MSm3/d (4.3 Mtpa) Reservoir pressure 150 bara 138 bara Pc rit 60 1,000 ba 80 ra ra ra 3,000 2,000 8b ara 10 0b ba ba 3.9 120 bara 110 bara 14 0b ar a 5 Mtpa 70 :7 3 Mtpa 260 250 240 230 220 210 200 190 180 170 160 150 140 130 120 110 100 90 80 70 60 50 40 30 20 1 Mtpa Pressure (bara) Well Modeling Aspects of CO2 Sequestration 207 4,000 5,000 6,000 Gas rate (1000 Sm3/d) 7,000 11 0b 1 ar 20 ba a ra ar a 8,000 9,000 10,000 At current reservoir pressure, this well can make 5.45 MSm3/d (3.6 Mtpa) WHT = 35 °C Reservoir pressure 150 Bara 140 Bara 128 Bara 110 Bara ra 2,000 3,000 ra 11 0B 4,000 5,000 6,000 Gas rate (MS3/d) 16 0B ar a ara Ba ar 8B 90 0B 3.9 Ba 1,000 :7 10 rit a 3 Mtpa Pc Bar ar a 5 Mtpa 80 70 270 260 250 240 230 220 210 200 190 180 170 160 150 140 130 120 110 100 90 80 70 60 50 40 1 Mtpa Pressure (Bara) Figure 11.8 The VFP curves for the vertical well showing the gas rate of 6.5 MSm3/d (4.3 Mtpa) at WHP of 110 Bara, WHT of 35 degrees C and reservoir pressure of 138 Bara. 7,000 12 0 a 14 Ba 0B ra 8,000 ar a 9,000 Figure 11.9 The VFP curves for ERD well showing the designed rate of 3 Mpta (4.5 MSm3/d) and full potential of 3.37 Mtpa (5.45 MSm3/d) at WHP of 110 Bara, WHT of 35 degrees C and reservoir pressure of 128 Bara. Figure 11.9 presents the results of the VFP analysis for the ERD well. This well can make 5.47 MSm3/d (3.62 Mtpa) which is also above the designed gas rate of 4.5 MSm3/d (3 Mtp) at the required operating conditions. The overall results of the VFP analysis at the WHT of 11 degrees C, 20 degrees C, 30 degrees C, and 35 degrees C temperatures are presented in Figure 11.10. For the vertical well, gas rate ranges from 4.3 to 6.0 Mtpa (6.5 to 9.1 MSm3/d) whereas it ranges from 3.6 to 4.8 Mtpa (5.4 to 7.2 MSm3/d) for the ERD well. The results also indicate that the higher the injection temperature, the lower the gas rates and vice versa. However, the vertical well has higher rates than that of the ERD well at each temperature. 208 Carbon Dioxide Capture and Acid Gas Injection 7 6.0 5 4 5.2 5.6 4.8 4.3 3.6 4.1 0 Vertical well 11 °C 30 °C 35 °C 20 °C 1 11 °C 2 30 °C 3 35 °C Gas rate (Mtpa) 6 ERD well Pure CO2 Figure 11.10 Comparison of gas rate for the vertical and ERD wells for WHT temperature of 11 degrees C, 30 degrees C and 35 degrees C and for WHP of 110 Bara. 11.7 Impact of the Well Deviation on CO2 Injection The impact of the well trajectory and well deviation on CO2 injection was investigated by studying the change in properties such as pressure, temperature and density profiles along the wellbore at three different temperatures (11 degrees C, 30 degrees C and 35 degrees C) for the vertical and ERD wells for pure CO2. Change in properties along the wellbore is a function of pressure loss, gas rate, temperature and phase behavior of the injected fluid. Figure 11.11 is an illustration of the pressure losses in the wellbore for the vertical and ERD wells at an injection temperature of 11 degrees C. In this figure, the 1st curve from the left is the friction pressure, the 2nd curve is the hydrostatic pressure, the 3rd curve is the net loss/gain and the 4th curve is the pressure along the wellbore. As can be seen, the pressure curve is the mirror image of the net loss/gain curve. Inflections in the pressure curves are caused by the pressure losses as seen clearly just below the tubing/packer depth in each well. Temperature and pressure conditions also change at this location causing a pronounced effect on the pressure losses. Figure 11.12 presents the change in properties for the vertical and ERD wells at an injection temperature of 35 degrees C and WHP of 110 Bara. This figure shows that most pronounced change in properties occurs at the tubing/packer depth for each well and at the heel of the ERD well. At the tubing/packer depth, the configuration of the wellbore changes from 7"(0.1778 m) tubing to 9 5/8" (0.22425 m) casing. Figure 11.13 shows the change in pressure and temperature along the wellbore at three different injection temperatures with respect to phase envelope. Well Modeling Aspects of CO2 Sequestration 209 0 0 200 1000 Tubing/packer depth: 1100 m 400 2000 Measured depth (m) Measured depth (m) 600 800 1000 3000 4000 Heel: 4810 m 1200 5000 1400 Tubing/packer depth: 1330 m Vertical well 1600 Net loss/gain Res. pressure = 138 Bara Hydrostatic pressure WHP = 110 Bara Friction pressure Gas rate = 9.1 (MSm3/d) Wellbore pressure 1800 –150 –100 –50 0 50 100 150 200 Pressure (Bara) 6000 ERD well Res. pressure = 128 Bara WHP = 110 Bara Gas rate = 7.31 (MSm3/d) 7000 –100 Friction pressure Net loss/gain Hydrostatic pressures Wellbore pressure 0 100 Pressure (Bara) 200 Figure 11.11 Pressure losses in the vertical well and ERD well at an injection temperature of 11 degrees C. Figure 11.13 shows that the wells with wellhead temperatures of 11 degrees C and 35 degrees C experienced greater change in temperature (13 degrees C for 11 degrees C curve and 10 degrees C for 35 degrees C curve) along the wellbore compared to the same wells at 30 degrees C (only 5 degrees C). Figure 11.14 shows the effect of well deviation on VFP curves at an injection temperature of 35 degrees C and WHP of 110 Bara for pure CO2. This figure shows that at an injection temperature of 35 degrees C, the ERD well has reduced injection capacity of 3.6 Mtpa (5.4 MSm3/d) which is 24% less than that of the vertical well that had an injection capacity of 4.7 Mtpa (7.1 MSm3/d). 11.8 Implication of Bottom Hole Temperature (BHT) on Reservoir Figure 11.15 presents BHT as a function of gas rate for different injection temperatures for the vertical and ERD wells. This figure shows that for the vertical well, from 1 to 5 Mtpa (1.5 to 7.6 MSm3/d), the decrease in BHT 210 Carbon Dioxide Capture and Acid Gas Injection 0 0 Tubing/packer depth: 1330 m 1000 Tubing/packer depth: 1100 m 1000 3000 4000 Heel 5000 1000 2000 Tubing/packer depth: 1100 m Measured depth (m) 2000 Tubing/packer depth: 1100 m Measured depth (m) Measured depth (m) 2000 0 Tubing/packer depth: 1330 m 3000 4000 Heel 5000 6000 7000 100 120 140 Pressure (Bara) 160 3000 4000 Heel 5000 6000 Vertical well, gas rate = 6.5 mSm3/d ERD well, gas rate = 5.4 mSm3/d Tubing/packer depth: 1330 m 6000 Vertical well,gas rate = 6.5 mSm3/d ERD well, gas rate = 5.4 mSm3/d 7000 30 35 40 45 Temperature (°C) 50 Vertical well, gas rate = 6.5 mSm3/d ERD well, gas rate = 5.4 mSm3/d 7000 680 690 700 710 720 Density (Kg/m3) 730 Figure 11.12 Change in properties along the wellbore in vertical and ERD wells at an injection temperature of 35 degrees C and WHP of 110 Bara. 180 Vertical well ERD well Phase envelope 160 WHT = 30 °C WHT = 35 °C Pressure (Bara) 140 WHT = 11 °C 120 Supercritical region 100 80 60 Critical point 40 20 0 0 5 10 15 20 25 30 Temperature (°C) 35 40 45 50 Figure 11.13 Change in pressure and temperature along the wellbore with reference to phase envelope. ranges from 5 degrees C to 10 degrees C. For the ERD well, the BHT reduction ranges from 15 degrees C to 18 degrees C depending upon the surface temperature. Figure 11.15 also shows that from 1 to 3 Mtpa (1.5 to 4.5 MSm3/d), the decrease in temperature is only 1 degrees C for vertical well and 6 degrees C Well Modeling Aspects of CO2 Sequestration 211 250 IPR-vertical well VFP-vertical well IPR-ERD well VFP-ERD well 150 0 1 Mtpa 50 0 2000 5 Mtpa 100 3 Mtpa Pressure (Bara) 200 4000 6000 8000 Gas rate (1000 Sm3/d) WHT = 35 °C 10000 12000 Figure 11.14 Effect of well deviation on VFP curves for an injection temperature of 35 degrees C and WHP of 110 Bara. for the ERD well for WHT of 35 degrees C. For both wells, the temperature change in the fluid’s BHT is about the same (5 to 6 degrees C) for a WHT of 11 degrees C. There is no apparent change in the fluid’s BHT for a WHT of 20 degrees C for the vertical well. For a WHT of 30 degrees C at the vertical well, the fluid’s BHT temperature fluctuates at injection rates between 1 Mtpa (1.5 MSm3/d) and 3 Mtpa (4.5 MSm3/d), however, 1 Mtpa (1.5 MSm3/d) and 3 Mtpa (4.5 MSm3/d), the fluid’s BHT is the about the same. For the ERD well, the change in the fluid’s BHT is about the same (5 to 6 degrees C) at a WHT of 30 degrees C. Since the maximum designed injection rate per well was 3 Mtpa (4.5 MSm3/d), the maximum difference between the injected fluid’s BHT and the initial reservoir temperature is 18 degrees C (Initial reservoir temperature (Tr) = 67 degrees C) for the vertical well. For the ERD well, it is 15 degrees C (Initial Tr = 63 degrees C) for a WHT of 35 degrees C. Impact of cold water in waterflood reservoirs is well documented (Gadde and Sharma, 2001; Mitchell et al., 2013). Due to the cold fluid ­temperature at the bottom of the hole, CO2 has been suggested as a stimulation fluid in hydraulic fracturing of the tight shale reservoirs (Amro et al., 2011; Enayatpour and Patzek, 2013). Goodarzi et al. (2010) presented their work on thermal aspects of geomechanics and induced fracturing in CO2 sequestration applications for a vertical well. They concluded by comparing the results of the thermal model with those of the iso-thermal case 212 Carbon Dioxide Capture and Acid Gas Injection 70 Vertical well WHP = 110 Basra Reservoir pressure = 138 Bara 60 BHT (C) 50 40 35 30 °C 30 °C 20 °C 20 0 0 2000 5 Mtpa 1 Mtpa 10 3 Mtpa 11 °C 4000 6000 8000 Gas rate (1000 Sm3/d) 10000 12000 70 ERD well WHP = 110 Basra Reservoir pressure = 128 Bara 60 BHT (C) 50 40 35 °C 30 °C 30 20 0 3 Mtpa 0 1 Mtpa 10 2000 4000 5 Mtpa 11 °C 6000 8000 Gas rate (1000 Sm3/d) 10000 12000 Figure 11.15 BHT as function of gas rates for different temperatures for vertical and ERD wells. that the total minimum stress at the wellbore decreases with time and falls below the injection pressure quite early during the injection. Also, thermal effects could increase the speed of fracture propagation in the storage layer depending on the injection rate. The study cited was conducted for Well Modeling Aspects of CO2 Sequestration 213 lower injection rates than this study; however, the temperature differences between the BHT and reservoir temperature were very similar. Based on the injection rates used in their study they concluded that injection of CO2 at a temperature lower than reservoir temperature reduces the fracture pressure, which in turn reduces the injectivity. The results also pointed out that it may not be practical to avoid induced fracturing in CO2 projects. Therefore, they recommended optimizing the injection rate and temperature by maximizing the injection capacity, while maintaining the safety of the storage reservoir by limiting the fracture length. Based on the insights from the above-mentioned study, we recommend that a geomechanical study be performed to study the impact of injection rate and temperature on injectivity, fracturing and fracture length. 11.9 Impact of CO2 Phase Change Transient flow occurs during events such as start-up, shutdown, planned or uncontrolled depressurization of pipeline. Aursand et al. (2013) presented a review of current research challenges for modeling of transient flow of CO2 mixtures in pipelines. Liljemark et al. (2011) concluded that quick shutdown and load change caused two-phase flow in the vertical section of the pipeline (injection well). Single-phase flow exists within the borehole above the critical point (73.98 Bara and 30.94 degrees C) for pure CO2. At low injection rates (e.g., during shutdown/start-up of injection operations), the pressure could be significantly lower than the critical pressure and two-phase flow will most likely occur near the surface down to a depth at which the injectate goes back into a dense state again with increasing hydrostatic pressure. There is no particular concern for any of the subsurface components while operating in this state and it is not unusual to operate wells with two-phase flowing conditions. We have recent experience with an acid gas injection well (15% H2S / 85% CO2) which, when going from a transient to steady state, needed a somewhat higher wellhead pressure (a pressure “bump”) to overcome the two phase phenomena and return the injectate to a single phase again. The lower the injection rate, the longer the transient period and the lower the pressure bump. This would imply that should such phenomena appear, a certain time period would need to be accounted for when going from start-up to steady state. (A longer “ramp up” time would be needed for a smaller pressure bump.) The length of the transient period and the additional pressure necessary to resume steady state flow is apparently velocity dependent. At faster 214 Carbon Dioxide Capture and Acid Gas Injection velocities the phenomena may be less pronounced or even non-existent. A full transient study would need to be undertaken to properly gauge whether or not pressure anomalies appear, how large they might be, and how long they might last. 11.10 Injection Rates, Facility Design Constraints and Number of Wells Required Figures 11.8 and 11.9 present the VFP sensitivities for the vertical and ERD wells, and show that designed injection rates of 3 Mtpa (4.5 MSm3/d) can be achieved. To access the full capacity rate of 5Mtpa (7.6 MSm3/d), either two wells are required or the injection pressure has to be increased from the designed 110 Bara to 120 Bara for the vertical well or to 130 Bara for the ERD well. For a steady state injection of 1 Mtpa (1.5 MSm3/d), one vertical or ERD well is sufficient. However, a second well will allow continued injection in the event the first well needs to be shut down for intervention operations. For steady state injection of 3 Mtpa (4.5 MSm3/d), two ERD wells are recommended. For contingencies such as well interventions, one additional well will be required. For steady state injection of 5 Mtpa (7.6 MSm3/d), two vertical wells or two ERD wells are recommended. For contingencies such as well interventions, a third well will be required. 11.11 Wellhead Temperature Effect on VFP Curves The effect of WHT on gas rates has been discussed briefly in the previous sections. In this section, the effect of WHT on gas rates for the vertical and ERD wells for temperatures of 11 degrees C, 30 degrees C and 35 degrees C for pure CO2 will be discussed. Figure 11.16 presents IPR-VFP curves for the vertical and ERD wells. Figure 11.16 clearly shows that the gas rate is significantly affected as the WHT is increased. For vertical well, at 11 degrees C, the well’s gas rate of 5.98 Mtpa (9.1 MSm3/d) is 16% more than 5 Mtpa (7.6 MSm3/d). At 30 degrees C, the gas rate is 5.18 Mtpa (7.8 MSm3/d) which is a 13% decrease. With increase of 5 degrees C in WHT (at 35 degree C), the gas rate further decreases to 4.2 Mtpa (6.5 MSm3/d) which is a 9% decrease and overall decrease of 21% from its full injection capacity at 11 degrees C. Well Modeling Aspects of CO2 Sequestration 215 250 Vertical well (WHP = 110 Bara) 138 Bara 150 11 C 100 C 35 Pressure (Bara) 200 30 0 2000 5 Mtpa 1 Mtpa 0 3 Mtpa C 50 4000 6000 8000 Gas rate (1000 Sm3/d) 10000 12000 250 ERD well (WHP = 110 Bara) 150 126 Bara 35 100 C 0 2000 5 Mtpa 3 Mtpa 1 Mtpa 0 C C 50 11 30 Pressure (Bara) 200 4000 6000 8000 Gas rate (1000 Sm3/d) 10000 12000 Figure 11.16 Effect of injection temperature on VFP curves for the Vertical and ERD wells. At an injection temperature of 11 degrees C, the ERD well has a full gas injection capacity of 4.83 Mtpa ((7.31 MSm3/d) which is 38% more than 3 Mtpa (4.5 MSm3/d). At a WHT of 30 degrees C, this gas rate reduces to 4.14 Mtpa (6.26 MSm3/d) which is a 14% decrease in gas injection capacity. At an increase of 5 degrees C (at 35 degrees C), the gas rate further reduces to 3.62 Mtpa (5.48 MSm3/d) which is a further decrease of 13% in gas rate and an overall decrease of 25% from the full gas injection capacity. A gas rate of 3.6 Mtpa (5.4 MSm3/d) is still 17% higher than 3 Mtpa (4.5 MSm3/d). 216 Carbon Dioxide Capture and Acid Gas Injection 11.12 Effect of Impurities in CO2 on VFP Curves Figure 11.17 shows the effect of impurities on wellhead pressures and injection rates. Only cases for pure CO2 and CO2 with 5% contaminants are plotted in this figure for vertical as well as ERD wells to show the maximum effect. This figure shows that at a given injection rate the WHP for CO2 with 5% contaminants will be higher than that of pure CO2 for both the wells. The vertical well shows higher wellhead pressures for CO2 with 5% contaminants than the ERD well. The results of VFP curves for CO2 with 5% contaminants (not presented in the paper) showed that the addition of impurities reduces the gas rates from 5 to 10% at all temperatures. The higher the temperature, the higher is the reduction in gas rates. Vertical well (Reservoir pressure = 138 Bara, WHT = 30 C) 180 Pure CO2 CO2 with 5% contaminants WHP, Bara 160 140 120 100 80 60 0.0 2.0 3.0 4.0 Injection rate (MSm3/d) 5.0 6.0 7.0 ERD well (Reservoir pressure = 128 Bara, WHT = 30 C) 200 Pure CO2 CO2 with 5% contaminants 180 WHP, Bara 1.0 160 140 120 100 80 60 0.0 1.0 2.0 3.0 4.0 Injection rate (MSm3/d) 5.0 6.0 Figure 11.17 Effect of impurities on wellhead pressures and injection rates. 7.0 Well Modeling Aspects of CO2 Sequestration 217 11.13 Concluding Remarks The objective of this study was to design an interface between the reservoir and the pipeline. Therefore, the first and last nodes were wellhead and well bottom hole respectively. IPR and VFP help design this interface. The Forchheimer IPR Model in PROSPER software was chosen and fitted with the simulation data to obtain the representative IPR. The IPR analyses were performed for reservoir pressures of 138 Bara for the vertical well and 128 Bara for the ERD well. In addition, IPR curves were also generated for pressures other than the reservoir pressure such as 110 Bara, 120 Bara and 150 Bara for the vertical well; 110 Bara, 120 Bara, and 150 Bara for the ERD wells. VFP analyses were performed for a wide range of wellhead pressures. Consistency checks were made and a PR EOS equation was selected for the work performed. These checks indicated that for pure CO2, the PR equation can have errors in density ranging from 0.4% at 11 degrees C (dense phase region) to as high as 7.7% at 35 degrees C (supercritical region) at a pressure of 110 Bara. Experimental data (Roberts and Carroll, 2013) was used to compare the computed densities using the same mixture: CO2 (91%) and N2 (9%) as the experimental data to further check robustness of PR of PROSPER software. Results confirm the conclusions arrived by Roberts and Carroll (2013) in their paper that the greatest errors occur at lower temperatures. Low temperatures are not typically of interest in carbon sequestration applications. VFP analyses indicated that all pure CO2 cases showed higher gas rates for the corresponding wells compared to CO2 with 5% contaminants. Furthermore, lower temperatures showed higher gas injection capacities for all the wells. For pure CO2 at a WHT of 11 degrees C, maximum gas rates of 6 Mtpa (9.1 MSm3/d) and 4.8 Mtpa (7.2 MSm3/d) were observed for the vertical well and ERD well respectively. Overall, the ERD well had lower gas injection capacity than the vertical well at all temperatures. Moreover, at a given injection rate the WHP for CO2 with 5% contaminants will be higher than that of pure CO2 for both the wells. The vertical well shows higher wellhead pressures for CO2 with 5% contaminants than the ERD well. Addition of impurities and their molar concentration increase the two-phase region of phase diagram. Also, the addition of impurities reduces the gas rates from 5 to 10% at all temperatures. The higher the temperature, the higher is the reduction in gas rates. Pressure losses are more pronounced at the tubing/packer depth for all the wells and at heel for the ERD well. The pressure along the wellbore mirrors the net pressure losses experienced in the wellbore. The change in 218 Carbon Dioxide Capture and Acid Gas Injection density correspondingly is more pronounced at the tubing/packer depth of the ERD well. Conversion Factors tpa × 1.5 E+00 bar × 14.503773 E+00 m × 3.28 E+00 m × 39.37 E+00 m3 × 35.31 E+00 kg/m3 × 6.242796 E-02 Mpa × 145.038 = m3/d = psi = ft = in = ft3 = lb/ft3 = psi References 1. Amro, M., Haefner, F. and Mueller, M., Synergetic CO2 Storage and Gas/Oil Production from Tight Reservoirs. The 6th Jordanian International Mining Conference, Amman, Jordan, Nov. 1–3, 2011. 2. Aursand, P., Hammer, M., Munkejord, S.T. and Wilhelmsen, O., Pipeline Transport of CO2 Mixtures: Models for Transient Simulation. International Journal of Greenhouse Gas Control 15, p. 174–185, 2013. 3. Boyle, T.B. and Carroll, J.J., Study Determines Best Methods for Calculating Acid-Gas Density. Drilling & Production, Oil & Gas Journal, Jan. 14, 2002. 4. Du, B., Cheng, L. and Ding, J., Simulation Studies of Temperature and Pressure Distribution in Carbon Dioxide Injection Well. EJGE Vol. 19, Bund. F. p. 1467–1476, 2014. 5. Enayatpour, S. and Patzek, T., Thermal Shock in Reservoir Rock Enhances the Hydraulic Fracturing of Gas Shales. SPE Unconventional Resources Technology Conference. Denver, Colorado, August 12–14, 2013. 6. Fenghour, A. and Wakeham, W.A., The Viscosity of Carbon Dioxide. J. Phys. Chem. Ref. Data, Vol. 27, No. 1, 1998. 7. Gadde, P.B. and Sharma, M.M., Growing Injection Well Fractures and Their Impact on Waterflood Performance. SPE Annual Technical Conference and Exhibition. New Orleans, Louisiana, Sept. 30-Oct. 3, 2001. 8. Goodarzi, S., Settari, A., Zoback, M. and Keith, D.W., Thermal Aspects of Geomechanics and Induced Fracturing in CO2 Injection with Application to CO2 Sequestration in Ohio River Valley. SPE 139706. New Orleans, Louisiana, Nov. 10–12, 2010. 9. Liljemark, S., Arvidsson, K., McCann, M.T.P., Tummescheit, H. and Velut, S., Dynamic Simulation of a Carbon Dioxide Transfer Pipeline for Analysis of Normal Operation and Failure Modes. Energy Procedia 4, p. 3040–3047, 2011. Well Modeling Aspects of CO2 Sequestration 219 10. Mazzoccoli, M., Guido, G.D., Bosio, B., Arato, E. and Pellegrini, L.A., CO2mixture Properties for Pipeline Transportation in the CCS Process. Chem. Eng. Trans., Vol. 32. p. 1861–1866, 2013. DOI: 10.3303/CET1332311 11. Mitchell, P., Smith, K. and Podgorney, R.K., Cold Water Injection Effects in Fractured Reservoirs. The Geological Society of America, Denver, Colorado, Oct. 27–30, 2013. 12. Roberts, E.L. and Carroll, J.J., Densities of Carbon Dioxide-Rich Mixtures: Part 2: Comparison with Thermodynamic Models. 4th International Acid gas Injection Symposium. Calgary, Canada, September 24–27, 2013. 13. Span, A. and Wagner, W., A New Equation of State for Carbon Dioxide Covering the Fluid Region from the Triple-Point Temperature to 1100 K at Pressure up to 800 Mpa. J. Phys. Chem. Ref. Data, Vol. 25, No. 6, 1996. 12 Effects of Acid Gas Reinjection on Enhanced Natural Gas Recovery and Carbon Dioxide Geological Storage: Investigation of the Right Bank of the Amu Darya River Qi Li, Xiaying Li, Zhiyong Niu, Dongqin Kuang, Jianli Ma, Xuehao Liu, Yankun Sun and Xiaochun Li State Key Laboratory of Geomechanics and Geotechnical Engineering, Institute of Rock and Soil Mechanics, Chinese Academy of Sciences, Wuhan, China Abstract Due to the environmental pollution of H2S, greenhouse effect of CO2 and economic problems of sulfur recovery, the reinjection of acid gas (H2S + CO2) into subsurface formations, which may dissolve some of the rock and a great deal of the cementing material, is an effective option to treat acid gas. The Amu Darya right bank, Turkmenistan, is one of the most important sources of China’s natural gas supply in Central Asia. However, most of the gas reservoirs contain CO2 and H2S, which increases both the difficulties of development and the financial burdens of the developers. The paper simulates the reinjection of the produced acid gas into the reservoirs using the ECLIPSE simulator and analyzes the effect of the reservoir pressure, gas sequestration, production and recovery ratio on the reservoir formations. The numerical results show not only that the reinjected acid gas is sequestered by gas-gas displacement but also that there is residual storage space. When the acid gas injection rate is set to 8 × 104 m3/day and the maximum daily production rate is set to 1 × 105 m3/day, more acid gas may be stored, thereby enhancing the natural gas recovery. The injected acid gas breaks through the production well in the seventh year, and the reservoir pressure maintains the stability of the reservoir. The production time is overly shortened and the reservoir energy rapidly declines when the daily production is oversized, which is adverse to long-term stability during gas reservoir production. Therefore, in this Ying Wu, John J. Carroll and Weiyao Zhu (eds.) CO2 Capture and Acid Gas Injection, (221–244) 2017 © Scrivener Publishing LLC 221 222 Carbon Dioxide Capture and Acid Gas Injection study, the reinjection of all acid gas produced in the research area is the reservoir exploitation countermeasure, considering safety, the environment, and economic efficiency. This can efficiently enhance the gas recovery and lead to acid gas storage to solve the problems of acid gas reservoir exploration. 12.1 Introduction Compared with conventional gas reservoirs, the difficulties of exploiting acid gas reservoirs include economic, environmental, safety and market issues [1, 2]. These issues are embodied in the capacity of the evaluation techniques, fluid phase safety testing, and corrosion studies, among others [3, 4]. The acid gas reservoir development strategy is related to the protection and storage of sulfur, acid gas reinjection, electricity generation by acid gas and safety issues [1, 5]. Taking the Puguang gas reservoir as an example, the difficulties of exploiting acid gas reservoirs include that there is no standard for reservoir development and that acid corrosion, sulfur deposit, and safe mining occur [6, 7]. For acid gas reservoirs, during the processes of gas development and exploitation, solid sulfur precipitation may result in the plugging of pore throats when the reservoir temperature and pressure decline. Injected acid gas may compensate for the energy loss, relieve the temperature and pressure change, and dissolve part of the rocks [8, 9]. Several scholars have performed research related to Acid Gas Reinjection [10]. In 2007, at the Z3Z oil reservoir in the main rivers in Zama, Trivedi et al. (2007) studied the optimal injection strategies and importance of developing operational parameters for acid gas sequestration during tertiary oil recovery using the ECLIPSE simulator [11]. In 2008, Pooladi-Darvish et al. studied CO2 injection for the enhanced gas recovery and geological storage of CO2 in the Long Coulee Glauconite F Pool, Alberta, Canada using the GEM simulator [12]. In 2012, Khan et al. studied the effects of CO2 and acid gas injection on EGR and storage using the TEMPEST simulator [13]. Lubas et al. (2012) investigated the effects of acid gas reinjection on the CO2 concentration in natural gas produced in the Borezcin reservoir [14]. Hou et al. (2012) thermo-hydromechanically modeled the CO2 injection for EGR using the TOUGH-FLAC simulator [15]. In 2014, Li and Elsworth studied the geochemistry associated with CO2-enhanced shale gas recovery [16]. In 2015, Wang et al. studied the effect of impurities on the CO2 storage capacity in geological formations [17]. This paper investigated the impact of the acid gas storage capacity and EGR on the parameters governing acid gas reinjection using the ECLIPSE simulator. Investigation of the Right Bank of the Amu Darya River 223 The Amu Darya right bank gas project in Turkmenistan is a major overseas project of the China National Petroleum Company. Most of the reservoirs are acid gas reservoirs, which causes many difficulties for development and exploitation [18]. Thus, based on the combination of security development strategies for acid gas reservoirs and the rational use of resources, this paper suggests that acid gas reinjection can be considered a development strategy for acid gas reservoirs. 12.2 The Amu Darya Right Bank Gas Reservoirs in Turkmenistan Most of the gas reservoirs are carbonate gas reservoirs containing H2S and CO2. Some gas reservoirs in the study area, such as the BP gas reservoir, are in the reservoir geological study stage [19]. Other gas reservoirs are in the drilling stage, which uses the directional drilling technique [20]. Some gas reservoirs are exploited. There are difficulties associated with old wells that need to be repaired and the recovering capacity. For example, 33 old wells were repaired in the Sa gas reservoir [21]. The highlights of this research are the variation of the gas amount, the stability of the reservoir evaluation and changes in production that occur when reinjecting acid gas from production wells into formations at a certain proportion of H2S and CO2 for environmental protection and economic maximization. It is well known from previous studies that acid gas components contain methane, ethane, propane, n/iso-butane, n/iso-pentane, C5 above, H2S, CO2, N2 and He [19–23]. According to the data from the Sam53-1 well, the burial depth in the Callovian-Oxfordian formation is 3500~3600 m, corresponding to a rock-forming temperature of 90~100 °C and a main temperature range of 90~110 °C [24]. The pressure factor in western Callovian-Oxfordian formation in the study was 0.85~1.08. The pressure factor in eastern and southeastern Callovian-Oxfordian formation was 1.65~1.90 [25]. 12.3 Model Development At present, it is of economic and environmental importance to address the acid gas from production wells by reinjecting acid gas into the reservoirs to enhance the gas recovery. This paper uses numerical simulation using the compositional simulator ECLIPSE [26]. 224 Carbon Dioxide Capture and Acid Gas Injection 12.3.1 State equation The model includes only two phases: the gas phase and the water phase. The equation of state (EOS) used is the modified Peng-Robinson (PR) EOS. 12.3.1.1 Introduction of Traditional PR State Equation The A coefficient in the PR state equation is: A jk X j X k A jk In the formula, x is the component mole fraction, A jk 1 k jk A j Ak and kjk are binary interaction coefficients. Ajk is defined as: A j 1/2 j Prj Trj2 , . The reduced critical temperatures and pressures are Tr = T/Tc and Pr = P/Pc, where Tc is the critical temperature and Pc is the critical pressure. Arj P The simplified coefficient A is defined by T and P: A j 2 T2 ojTcj j 1/2 1/2 r In the formula, A j ; j is defined by j 1 f w j 1 Trj1/2 ; Pcj is a constant; wj is the acentric factor; f(wj) is the polynomial of the oj acentric factor; and Trj is the simplified temperature. There are three coefficients in the compositional simulator (ECLIPSE): A1jk A2jk A3jk 2 ojTck 1 k jk 1 k jk 1 k jk 1/2 pcj 2 ojTck 2 okTck 1/2 p j pk pck 1/2 2 okTck pck pcj 2 ojTck 1/2 pcj 1/2 p j qk 2 okTck pck pk q j 1/2 q j qk p . T2 In the formula, P and T are the pressure and temperature in each grid. In every grid, Ajk is simplified as Ajk A1jk A2jkT 1/2 A3jkT 12.3.1.2 Modifications for the Vapor-Aqueous System Two parts of the traditional PR EOS were modified according to Soreide and Whitson’s proposal. Investigation of the Right Bank of the Amu Darya River 225 For the water component, α1/2 was modified and replaced by an expre­ ssion with a more complex temperature dependence: 1/2 1 aq1 1 aq2Tr aq3 Tr 3 1 . In the formula, aq1 = 0.4530, aq2 = 1–0.0103cs1.1, aq3 = 0.0034, cs is the binary concentration expressed as a molality. The remainders take the usual form. A general form for all components is 1/2 = p + qTr1/2 + rTr + sTr–3 Different binary interaction coefficients are used for the gas phase and the aqueous phases. This may cause problems close to the critical point, but the critical temperature of a water-dominated system is in the region of Tcw = 647 K, which should be well above the area of interest. Soreide and Whitson proposed adding a temperature dependence to the aqueous phase binary interaction coefficients: k ajw bq1 bq2Trj bq3Trj2 In the formula, bq1 – A0(1+S0cs), bq2 = A1(1 + S1cs), bq3 = A2(1 + S2cs) A0 = 1.112 – 1.7369wj–0.1, A1 = 1.1001 + 0.836wj, A2 = –0.15742 – 1.0988wj and S0 = 0.017407, S1 = 0.033516, S2 = 0.011478. Special values are used for the binary interaction coefficients of the aqueous phase and N2 and those of CO2 and H2S: kNa 2w a kCO 2w 1.70235(1 0.025587cS3/4 ) 0.44338(1 0.08126cs3/4 )TrN 2 0.31092(1 0.15587cS0.7505 ) 0.23580(1 0.17837cs0.979 )TrCO2 21.2566 exp( 6.7222TrCO2 kHa 2Sw cs ) 0.20441 0.23426TrH2S These fit into the general quadratic form. The modifications of α1/2 and the binary interaction coefficients introduce a coefficient of the PR state equation. The model uses the modified PR state equation. 12.3.2 Salinity The salinity in the model is defined by the molality: cs = 1000·ns/mw, where ns is the number of moles of salt and mw is the mass of the water. 226 Carbon Dioxide Capture and Acid Gas Injection 12.3.3 Diffusion 12.3.3.1 Diffusion Coefficients There are two diffusion models that can be used in the simulator ECLIPSE 300. The first diffusion model is driven by the concentration gradient: x Ji cDi i . The second diffusion model is driven by the gradient of the d chemical potential: Ji cDia xi 1 RT d i MiG h h0 Mi DiT In T , where Ji is the molar flux of the component i per unit area; C is the total molar concentration given by c = 1/vm; vm is the molar volume of the mixture; Di is the diffusion coefficient of the component i; Dia is the activity-­corrected diffusion coefficient of the component i; DiT is the thermal diffusion coefficient of the component; xi is the mole fraction of the is the gradient in the direction of flow; Mi is the molecucomponent i; d lar weight of the component i; G is the acceleration due to gravity; h is the height; h0 is the reference height; T is the temperature; R is the gas constant; and μi is the chemical potential of the component i. 12.3.3.2 The Cross-Phase Diffusion Coefficients There are two models in cross-phase diffusion in ECLIPSE. One model is the activity-driven diffusion. The diffusion between the cells is driven by the chemical potential, resulting in the diffusion over phase boundaries. The diffusion coefficients for the cross-phase diffusion (oil phase and gas phase) are set by combining the diffusion coefficients given by DIFFAOIL Doia Dg ia . and DIFFAGAS, Dog ia The other model is molecular-driven diffusion. The fluids are mixed on the border of each cell, and the resulting composition is used to set up the potential for molecular diffusion of each component. The water-gas and gas-water diffusion coefficients of the GASWAT model are set by the keywords DIFFCWG and DIFFCGW. The molecular diffusion coefficient constructs a border composition by mixing the composition of neighboring cells. As an example, we consider a cell I containing oil and a cell J containing gas. An artificial border composition is constructed by m̃i = (bo·xi)I + (bg·yi)J, where bo is the oil molar density, xi is the oil mole fraction of component i, bg is the gas molar density, and yi is the gas mole fraction of the component i. The border pressure and temperature are set to Investigation of the Right Bank of the Amu Darya River 227 P PI PJ and T TI TJ . The cross-phase diffusive flow between the 2 2 oil in cell I and the gas in cell J for component i is given by Fcross = FXog + FXgo + FYog + FYgo. FXog Diff Dog i min SoI , Sg J bo max xi xi ,0 FXgo Diff Dgoi min SoI , Sg J bo max xi xi ,0 FYog Diff Dog i min SoI , Sg J bg max yi yi ,0 FYgo Diff Dgoi min SoI , Sg J bg max yi yi ,0 In the formula, SoI and SgI are the oil and gas saturation in cell I and J, Diff is the diffusivity, DogI and DgoI and are the cross-phase oil-gas and gas-oil diffusion coefficients. 12.4 Simulation Model 12.4.1 Model Parameters Based on previous studies [27] ½C, this paper constructs the stratum data of the Sam53-1 well to determine the horizons and thicknesses of the layers (Figure 12.1) [28, 29]. Combined with the simulator, the components of the model include C1, C2, CO2, H2S and H2O. The initial porosity, initial permeability and relative permeability are shown in Table 12.1 and Figure 12.2 [30]. The initial temperature is 101 °C [31]. The original rock compressibility is 1.78 × 10–5 [32]. 12.4.2 Grid-Sensitive Research of the Model The size of the model is 2000 m × 2000 m. Based on experience, we choose the gridding schemes 5×5, 10×10 and 20×20 for grid-sensitive research to determine the most suitable gridding scheme. The paper analyzes the reservoir pressure and saturation of the gridding schemes by keeping the other conditions the same (Figure 12.3). At the early stage, the reservoir pressure of the 5×5 model is slightly low, and those of the 10×10 and 20×20 models are basically the same. In the middle and later stages, the reservoir pressure of the 5×5 model is moderate, and those of the 10×10 and 20×20 models are the highest and lowest, respectively. The saturation of 228 Carbon Dioxide Capture and Acid Gas Injection Stratum system System Series Stage Formation Upper jurassic Kimmeridgian HA Depth (m) Simplified lithologic section 2300 2310 2320 2330 2340 XVac 2350 2360 2370 2380 XVp 2390 2400 2410 2420 2430 2440 2450 Jurrasic Medium-upper jurassic series Callovianoxfordian XVm 2460 2470 2480 2490 2500 2510 2520 XVhp 2530 2540 2550 XVal 2560 2570 Z 2580 2590 2600 2610 XVa2 2620 2630 2640 2650 2660 XV1 Medium-lower jurassic series 2670 2680 2690 2700 Figure 12.1 Synthetic histogram of the Callovian-Oxfordian of the Sam53-1 well in this study [28, 29]. Investigation of the Right Bank of the Amu Darya River 229 Table 12.1 Physical reservoir properties. Parameter Caprock Reservoir Basement Depth (m) 2270~2310 2310~2690 2690~2730 Porosity 0.00001 0.098 0.00001 Permeability (mD) 0.00001 54 0.00001 0.5 Krg Krw krg / Krw 0.4 0.3 0.2 0.1 0 0.2 0.4 0.6 Sw 1 0.8 Figure 12.2 The relative permeability relationship diagram for the Callovian-Oxfordian in the study area [21]. 3.10E + 02 20-FGSAT 5-FGSAT 2.50E – 02 2.70E + 02 5-FPR 2.30E + 02 1.50E – 02 10-FPR 2.10E + 02 Saturation 2.00E – 02 10-FGSAT 2.50E + 02 1.00E – 02 1.90E + 02 5.00E – 03 0.00E + 00 31/1/1 29/1/1 23/1/1 21/1/1 19/1/1 17/1/1 15/1/1 13/1/1 11/1/1 09/1/1 07/1/1 05/1/1 03/1/1 01/1/1 1.50E + 02 27/1/1 20-FPR 1.70E + 02 25/1/1 Reservoir pressure (bar) 2.90E + 02 3.00E – 02 Time(Y/M/D) Figure 12.3 Time-dependent diagram of reservoir pressure and gas saturation. Note: FPR is the reservoir pressure. FGSAT is the reservoir gas saturation. The time period is January of the first year to December of the 30th year. 230 Carbon Dioxide Capture and Acid Gas Injection the 5×5 model is somewhat high at the early stage, and those of the 10×10 and 20×20 models is basically the same. In the middle and later stages, the saturation of the 5×5 model slowly decreases and stays in the middle, that of the 10×10 model is the lowest, and that of the 20×20 model is the highest. Theoretically, the 5×5 gridding scheme is better, but it can lead to more grids, and additional calculation time is needed. This paper analyzes the relative errors of the 5×5 and 20×20 grids compared to the 10×10 grid to determine the optimal gridding scheme. Comparing the 5×5 grid to the 10×10 grid, the maximum relative reservoir pressure error is 0.000021, the minimum is –0.008466 and the average value is 0.006404. The maximum relative saturation error is 0.009934, the minimum error is 0.000256 and the average value is 0.006505. Comparing the 20×20 grid to the 10×10 grid, the maximum relative reservoir pressure error is 0.003596, the minimum value is –0.003596 and the average value is –0.008175. The maximum relative saturation error is 0.015857, the minimum error is –0.004544 and the average value is 0.008755. The relative error of the 5×5 grid and 10×10 grids is therefore very small. With respect to the whole, the small error may be negligible. In general, based on experience, the operational efficiency and acid gas reinjection, the grid scheme of 10×10 is optimal. 12.4.3 The Development and Exploitation Mode According to the 10×10 grid scheme and the above-mentioned parameters, the paper establishes a geological model (Figure 12.4). The acid gas that is reinjected into the reservoir is produced for the purposes of this research. Nine development and exploitation modes are designed in nine different models (Table 12.2). According to the produced acid gas, the gas component is the total CO2, and the component ratios are 8:2 and 7:3. 12.5 Results and Discussion 12.5.1 Reservoir Pressure The changes of the reservoir pressure can reflect reservoir energy losses and are an important indicator for measuring the production mode. The changes of the reservoir pressure are basically identical in models 1, 2, 3, 4, and 5, (i.e., the pressure decreases slowly, and the pressure lapse rate decreases gradually). In model 6, during the flow period, the reservoir pressure decreases to 197 bar, and after 35 years, the reservoir pressure Investigation of the Right Bank of the Amu Darya River 231 2000 m 460 m 2000 m (a) –2270 m –2310 m Caprock Reservoir –2690 m Baserock –2730 m (b) Figure 12.4 The geological model in the study area (a: 3D view; b: sectional view). Table 12.2 The differences of the nine development models. Mode Maximum of CO2/H2S Injection rate daily production gas ratio (m3/day) (m3/day) Productive life without injection + injection (year) Model 1 7:3 187 10000 10+20 Model 2 7:3 187 10000 15+20 Model 3 7:3 187 10000 20+20 Model 4 8:2 281 10000 15+20 Model 5 10:0 261 10000 15+20 Model 6 7:3 80000 10000 15+20 Model 7 7:3 80000 1×105 15+20 Model 8 7:3 50000 1×106 20+20 Model 9 7:3 60000 1×107 15+20 232 Carbon Dioxide Capture and Acid Gas Injection 3.00E + 02 Reservoir pressure (bar) 2.50E + 02 2.00E + 02 1.50E + 02 1.00E + 02 5.00E + 01 Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 0.00E + 00 01/1/1 06/1/1 11/1/1 16/1/1 21/1/1 26/1/1 31/1/1 36/1/1 Time (Y/M/D) Figure 12.5 Time-dependent reservoir pressure diagram. increases to 227 bar. In model 7, during the flow period, the reservoir pressure drops to 27 bar, and during the gas injection period, the reservoir pressure is approximately 27 bar. In model 8, during the flow period, the reservoir pressure begins to drop quickly, with a change in reservoir pressure of approximately 25 bar after 7 years, which is small. After the acid gas is injected, the change of the reservoir pressure is also small, approximately 26 bar. In model 9, the reservoir pressure first declines quickly and then declines slowly; 5 years later, the reservoir pressure is basically stable. After injecting the gas, the pressure increases slightly, and the increment is very small (Figure 12.5). Due to the rapid rates of pressure losses, which almost reach 1 in models 8 and 9, production models such as these are not accepted because they counteract the stability of the reservoirs. 12.5.2 Gas Sequestration Gas sequestration is an indicator for optimizing the production models. Figures 12.6 and 12.7 show that the gas sequestration decreases linearly in models 1, 2, 3, 4, and 5, and the injected acid gas amount does not affect the acid gas amount in the reservoir because the injection amount is very small compared with the original amount of gas (6.56 × 108 m3 on the ground surface). The injection amount of these models should be stored in the Investigation of the Right Bank of the Amu Darya River 233 1.20E + 09 Gas in place (m3) 1.00E + 09 Model 1 Model 2 8.00E + 08 Model 3 Model 4 6.00E + 08 Model 5 4.00E + 08 Model 6 Model 7 2.00E + 08 Model 8 0.00E + 00 01/1/1 06/1/1 11/1/1 16/1/1 21/1/1 26/1/1 31/1/1 36/1/1 Model 9 Time (Y/M/D) Figure 12.6 Time-dependent diagram of the total amount of gas sequestration. reservoir, although it is of no use for enhancing gas recovery. In models 6, 7, 8 and 9, during the flow period, the gas in place decreases, and the lowering rates differ due to different maximum daily production amounts. The rate of model 9 is the highest, followed by the rates of models 8 and 7. The gas in place increases during the gas injection period. The rising rate is also different because of the different acid gas injection amounts. The rate of model 6 is the highest, followed by that of model 7. Based on a detailed analysis of model 6, the CO2 sequestration in the reservoir slowly increases after 21 years, and the H2S sequestration in the reservoir slowly increases after 25 years. In model 7, CO2 and H2S sequestration drop slowly ­during the flow period. The CO2 and H2S sequestration increases first at a fast speed and later at a slower speed after the acid gas is injected. At the same time, the CO2 sequestration is higher than the H2S sequestration. In model 8, the CO2 and H2S amounts of the reservoir decrease slowly over a 20-year flow period. The CO2 amount of the reservoir begins to quickly increase in June of the 22nd year, and the H2S amount of the reservoir begins to quickly increase in February of the 23rd year. In model 9, after acid gas is injected, the gas in place exhibits a small increase, the CO2 and H2S sequestration increase, and the rising rates gradually decrease. Based on this phenomenon and the combination with the differences of the models, the rising rate of the gas in place is the maximum during the gas i­njection period of model 6, although the final total amount is greater than that of the original gas in place, which counteracts the reservoir stability. The CO2 and H2S sequestration dramatically increase, which may result in reservoir “burdens” such as sulfur and hydrate blockage. Considering the gas 234 Carbon Dioxide Capture and Acid Gas Injection Model 1 CO2 8.00E + 08 Model 2 CO2 Model 3 CO2 7.00E + 08 Model 4 CO2 Model 5 CO2 CO2, H2S sequestration 6.00E + 08 Model 6 CO2 5.00E + 08 Model 7 CO2 4.00E + 08 Model 9 CO2 3.00E + 08 Model 2 H2S Model 8 CO2 Model 1 H2S Model 3 H2S Model 4 H2S 2.00E + 08 Model 5 H2S 1.00E + 08 Model 6 H2S 0.00E + 00 01/1/1 06/1/1 11/1/1 16/1/1 21/1/1 26/1/1 31/1/1 36/1/1 Time (Y/M/D) (a) Model 2 CO2 1.80E + 08 CO2, H2S sequestration Model 8 H2S Model 1 CO2 2.00E + 08 Model 3 CO2 1.60E + 08 Model 4 CO2 1.40E + 08 Model 6 CO2 Model 5 CO2 1.20E + 08 Model 7 CO2 1.00E + 08 Model 9 CO2 Model 8 CO2 Model 1 H2S 8.00E + 07 Model 2 H2S Model 3 H2S 6.00E + 07 Model 4 H2S 4.00E + 07 (b) Model 7 H2S Model 5 H2S 2.00E + 07 Model 6 H2S 0.00E + 00 01/1/1 06/1/1 11/1/1 16/1/1 21/1/1 26/1/1 31/1/1 36/1/1 Model 8 H2S Model 7 H2S Time (Y/M/D) Figure 12.7 (a) Time-dependent diagram of the amounts of CO2 sequestration and H2S sequestration, (b) zooming into (a) by shortening the Y-axis. Investigation of the Right Bank of the Amu Darya River 235 sequestration during the flow period, the gas sequestration of model 7 decreases steadily, and those of models 8 and 9 decrease quickly. The production modes of models 8 and 9 belong to the depletion development. The storage amounts increase after acid gas injection, the CO2 rising rate is greater and the amount of CO2 increases more rapidly because there is more CO2 than H2S in the reservoir and because there is more CO2 than H2S in the injected acid gas. 12.5.3 Production In models 1, 2, 3, 4, 5 and 6, the daily production is always the maximum value, which reveals that there is a possibility of daily production increase. Therefore, changes in daily production are observed after it is enhanced by one order of magnitude in models 7, 8 and 9. In model 7, the daily production reaches to 1×105 m3/d in the first 12 years and decreases until the acid gas is injected. It then increases up to year 30, while the increasing speed gradually decreases and inclines to stability, approximately 1×105 m3/d. In model 8, daily production decreases from 1×106 m3/d to 411 m3/d rapidly at first, but it later gradually slows down. When the acid gas is injected, it begins to increase step by step within 2 years until it remains constant after 8 years. In model 9, the daily production declines rapidly in the first few months and then slowly reduces until it equals zero in year 7. Following the injection of acid gas, it again increases with a small value, as shown in Figure 12.8. The daily production in model 6 remains constant for 35 years. The natural gas in the reservoirs is much richer, which suggests that the maximum daily production will be enhanced. The model 7 maintains the maximum daily production for 12 years and increases slightly 15 years after the acid gas injection, which indicates stable and decreasing production stages, which is helpful for understanding the reservoirs and verifies the enhanced function of the acid gas injection. Models 8 and 9 belong to the depletion development with a much lower production speed. Therefore, the enhancement of the production in the short term can use the depletion development, although this method is not advisable for the enhanced acid gas injection. The breakthrough of the acid gas should be noted during the acid gas injection period. In model 7, the CO2 amount begins to increase on January 1 of the 17th year, and the H2S amount increases on June 1 of the 17th year (Figure 12.9). When the acid gas is driven to the production well, the injected acid gas will break through, as shown in Figures 12.10 236 Carbon Dioxide Capture and Acid Gas Injection 1.40E + 06 Daily production (m3/d) 1.20E + 06 1.00E + 06 Model 3 Model 4 8.00E + 05 Model 5 6.00E + 05 Model 6 Model 7 4.00E + 05 Model 8 Model 9 2.00E + 05 0.00E + 00 01/1/1 06/1/1 11/1/1 16/1/1 21/1/1 26/1/1 31/1/1 36/1/1 Time (Y/M/D) (a) Daily production (m3/d) 2.00E + 05 1.50E + 05 Model 3 Model 4 Model 5 1.00E + 05 Model 6 Model 7 Model 8 5.00E + 04 Model 9 0.00E + 00 01/1/1 06/1/1 11/1/1 16/1/1 21/1/1 26/1/1 31/1/1 36/1/1 (b) Time (Y/M/D) Figure 12.8 (a) Time-dependent diagram for the daily production, (b) zooming into (a) by shortening the Y-axis. and 12.11. Near the ground of the production well, there are CO2 and H2S plumes, and the CO2 plume is larger than the H2S plume. In model 9, the CO2 amount begins to increase on March 1 of the 17th year, and the H2S amount increases on October 1 of the 17th year (Figure 12.12). When the acid gas is driven to the production well, the injected acid gas will break through, as shown in Figures 12.13 and 12.14. Near the ground of the production well, there are CO2 and H2S plumes, and the CO2 plume is larger Investigation of the Right Bank of the Amu Darya River 237 Daily production(m3/d) 2.50E + 04 2.00E + 04 CO2 1.50E + 04 1.00E + 04 H2 S 5.00E + 03 15/1/1 15/2/1 15/3/1 15/4/1 15/5/1 15/6/1 15/7/1 15/8/1 15/9/1 15/10/1 15/11/1 15/12/1 16/1/1 16/2/1 16/3/1 16/4/1 16/5/1 16/6/1 16/7/1 16/8/1 16/9/1 16/10/1 16/11/1 16/12/1 17/1/1 17/2/1 17/3/1 17/4/1 17/5/1 17/6/1 17/7/1 17/8/1 17/9/1 17/10/1 17/11/1 17/12/1 18/1/1 18/2/1 18/3/1 18/4/1 18/5/1 18/6/1 18/7/1 18/8/1 18/9/1 18/10/1 18/11/1 18/12/1 0.00E + 00 Time (Y/M/D) Figure 12.9 Time-dependent diagram of the acid gas production of the production well over 15~18 years in model 7. Note: WCO2 is the CO2 concentration of the production well. WH2S is the H2S content of the production well. 0 1000 Production well –2300 –2400 –2500 –2600 –2700 m 0.000 0.241 0.482 0.723 2000 m 0.964 Figure 12.10 The CO2 plume in the 18th year of model 7. 0 1000 Production well –2300 –2400 –2500 –2600 –2700 m 0.000 0.078 0.157 0.235 2000 m 0.313 Figure 12.11 The H2S plume in the 18th year of model 7. than the H2S plume. The acid gas amount dramatically increases during the production stage, which counteracts the acid gas storage because injected acid gas breaks through the production well. The H2S rising rate in the production well is smaller than the CO2 rising rate because the H2S content of the injected acid gas is very small. 1.40E + 04 1.20E + 04 1.00E + 04 CO2 8.00E + 03 6.00E + 03 4.00E + 03 H2S 2.00E + 03 0.00E + 00 15/1/1 15/2/1 15/3/1 15/4/1 15/5/1 15/6/1 15/7/1 15/8/1 15/9/1 15/10/1 15/11/1 15/12/1 16/1/1 16/2/1 16/3/1 16/4/1 16/5/1 16/6/1 16/7/1 16/8/1 16/9/1 16/10/1 16/11/1 16/12/1 17/1/1 17/2/1 17/3/1 17/4/1 17/5/1 17/6/1 17/7/1 17/8/1 17/9/1 17/10/1 17/11/1 17/12/1 18/1/1 18/2/1 18/3/1 18/4/1 18/5/1 18/6/1 18/7/1 18/8/1 18/9/1 18/10/1 18/11/1 18/12/1 Darily production(m3/d) 238 Carbon Dioxide Capture and Acid Gas Injection Time (Y/M/D) Figure 12.12 Time-dependent diagram of the acid gas production of the production well over 15~18 years in model 9. 0 1000 Production well –2300 –2400 –2500 –2600 –2700 m 0.000 0.238 0.475 0.713 2000 m 0.950 Figure 12.13 The CO2 plume in the 18th year of model 9. 0 1000 Production well –2300 –2400 –2500 –2600 –2700 m 0.000 0.074 0.148 0.222 2000 m 0.295 Figure 12.14 The H2S plume in the 18th year of model 9. 12.5.4 Recovery Ratio and Recovery Percentage In model 7, the recovery ratio increases linearly during the flow period and decreases after 12 years. The recovery ratio first increases slowly during gas injection and later increases linearly. The recovery ratio first increases exponentially in model 9; later, the value is invariant. During the gas Investigation of the Right Bank of the Amu Darya River 239 160 Recovery ratio (%) 140 120 100 80 Model 7 Model 9 60 40 20 0 01/1/1 06/1/1 11/1/1 16/1/1 21/1/1 26/1/1 31/1/1 36/1/1 Time (Y/M/D) Figure 12.15 Time-varying diagram of recovery ratio of models 7 and 9. injection period, the recovery ratio first increases slowly and later increases linearly, but the rising rate is smaller than that in model 7 (Figure 12.15). In model 7, the change trend of daily recovery percentage is the same as that of the daily production. The daily recovery percentage is 0.015% during the flow period. The value decreases until the 13th year, and the minimum is 0.0035%. The value increases to 0.011% during the gas injection. The daily recovery percentage decreases first in model 9. The daily recovery percentage increases until the gas is injected. However, the value after 35 years is smaller than that in model 7 (Figure 12.16). 12.6 Conclusions 1. The natural gas production mode is usually the flowing production. On the right bank of the Amu Darya River, we put forward a win-win approach, which is to inject acid gas from the production wells of tertiary recovery. 2. Acid gas reinjection is feasible for the development of Amu Darya blocks. 3. Using the optimal strategy can enhance the daily production and gas recovery with a suitable injection rate. Based on the long-term and sustainable development in the study area, the best development strategy includes an acid gas ratio of 7:3, an injection rate of 8×104 m3/day, a maximum daily production of 1 × 105 m3/day, a flowing 15-year period and the injection of 20-year acid gas. 240 Carbon Dioxide Capture and Acid Gas Injection 0.2 Recovery percent/day (%) 0.18 0.16 0.14 0.12 0.1 Model 7 0.08 Model 9 0.06 0.04 0.02 0 01/1/1 06/1/1 11/1/1 16/1/1 21/1/1 26/1/1 31/1/1 36/1/1 Time (Y/M/D) (a) 0.04 Recovery percent/day (%) 0.035 0.03 0.025 0.02 Model 7 Model 9 0.015 0.01 0.005 0 01/1/1 (b) 06/1/1 11/1/1 16/1/1 21/1/1 26/1/1 31/1/1 36/1/1 Time (Y/M/D) Figure 12.16 (a) Time-dependent diagram of the recovery percentage of models 7 and 9, (b) zooming into (a) by shortening the Y-axis. 12.7 Acknowledgments We acknowledge financial support from the National Natural Science Foundation of China (Grant No. 41274111) and the Hundred Talent Program of the Chinese Academy of Sciences. Investigation of the Right Bank of the Amu Darya River 241 References 1. H. Ye, J. Wang. Challenges and Solutions to Acid Gas Reservoir Development. Oil Forum. vol. 28, no. 4, pp. 63–65, DOI: 10.3969/j.issn.1002-302x.2009.04.018. 2009. 2. X. Liu, Q. Li, L. Du, X. Li. Economic Comparison of Both Acid-gas Reinjection and Sulfur Recovery in High-sour Gasfields. Natural Gas Technology and Economy. vol. 6, no. 4, pp. 55–59, DOI: 10.3969/j.issn.2095-1132.2012.04.015. 2012. 3. Z. Du. Oversea Experiences of Acid Gas Reservoir Development and their Enlightenments. Natural Gas Industry. vol. 26, no. 12, pp. 35–37, 2006. 4. Q. Li, G. Liu, X. Liu. “Development of Management Information System of Global Acid Gas Injection Projects”. In: Y. Wu, J.C. Carroll, Q. Li, Eds., Gas Injection for Disposal and Enhanced Recovery. Wiley. pp. 243–254, 2014. 5. Q. Li, X. Li, L. Du, G. Liu, X. Liu, N. Wei. “Potential Sites and Early Opportunities of Acid Gas Re-injection in China”. In: Y. Wu, J.J. Carroll, W. Zhu, Eds., Sour Gas and Related Technologies. Wiley Scrivener. pp. 131–140, 2012. 6. X. Wu, S. Zeng. On the difficulties and countermeasures of security development in Puguang high sour gas field. Chemical Enterprise Management 34, pp. 147, 2015. 7. Q. Li, X. Liu, L. Du, B. Bai, Z. Fang, M. Jing, X. Li. Economics of acid gas reinjection with comparison to sulfur recovery in China. Energy Procedia. vol. 37, pp. 2505–2510, DOI: 10.1016/j.egypro.2013.06.132. 2013. 8. Q. Li, X. Liu, L. Du, X. Li. “Review of Mechanical Properties Related Problems for Acid Gas Injection”. In: Y. Wu, J.C. Carroll, Q. Li, Eds., Gas Injection for Disposal and Enhanced Recovery. Wiley. pp. 275–292, 2014. 9. B. Yu, Y. Z. Liu, X. Q. Liang, L. Wang. Interfacial Evolution and Migration Characteristics of Acid Gas Injected into a Saline Aquifer. Chinese Journal of Chemical Engineering. vol. 21, no. 4, pp. 434–441, DOI: 10.1016/s10049541(13)60481-4. 2013. 10. Y. Wu, J. C. Carroll, Q. Li, eds., Gas Injection for Disposal and Enhanced Recovery. Hardcover ed. Advances in Natural Gas Engineering. eds. Y. Wu, J.C. Carroll, Wiley-Scrivener: New York. p. 400, 2014. 11. J. J. Trivedi, T. Babadagli, R. G. Lavoie, D. Nimchuk. Acid Gas Sequestration During Tertiary Oil Recovery: Optimal Injection Strategies and Importance of Operational Parameters. Journal of Canadian Petroleum Technology. vol. 46, no. 3, pp. 60–68, 2007. 12. M. Pooladi-Darvish, H. Hong, S. Theys, R. Stocker, S. Bachu, S. Dashtgard. CO2 Injection for Enhanced Gas Recovery and Geological Storage of CO2 in the Long Coulee Glauconite F Pool, Alberta. In: 2008 SPE Annual Technical Conference and ExhibitionSPE: Denver, Colorado, USA. pp. SPE 115789, 2008. 242 Carbon Dioxide Capture and Acid Gas Injection 13. C. Khan, R. Amin, G. Madden. Effects of CO2 and acid gas injection on enhanced gas recovery and storage. Journal of Petroleum Exploration and Production Technology. vol. 3, no. 1, pp. 55–60, DOI: 10.1007/s13202-0120044-8. 2012. 14. J. Lubas, W. Szott, P. Jakubowicz. Effects of Acid Gas Reinjetcion on CO2 Concentration in Natural Gas Produced from Borzecin Reservoir. NAFTAGAZ. vol. ROK LXVIII, no. 7, pp. 405–410, 2012. 15. Z. Hou, Y. Gou, J. Taron, U. J. Gorke, O. Kolditz. Thermo-hydro-mechanical Modeling of Carbon Dioxide Injection for Enhanced Gas-recovery (CO2EGR): A Benchmarking Study for Code Comparison. Environmental Earth Sciences vol. 67, no. 2, pp. 549–561, DOI: 10.1007/s12665-012-1703-2. 2012. 16. X. Li, D. Elsworth. Geomechanics of CO2 Enhanced Shale Gas Recovery. Journal of Natural Gas Science and Engineering vol. 26 no. pp. 1607–1619, DOI: 10.1016/j.jngse.2014.08.010. 2014. 17. J. Wang, Z. Wang, D. Ryan, C. Lan. A Study of the Effect of Impurities on CO2 Storage Capacity in Geological Formations. International Journal of Greenhouse Gas Control. vol. 42, no. pp. 132–137, DOI: 10.1016/j.ijggc.2015.08.002. 2015. 18. Q. Li, D. Kuang, G. Liu, X. Liu. Acid Gas Injection: A Suitability Evaluation for Sequestration Site in Amu Darya Basin, Turkmenistan. Geological Review. vol. 60, no. 5, pp. 1133–1146, 2014. 19. M. Lie, S. Xu, L. Wu, J. He, H. Ming, R. Zhang. Features of Gas Reservoirs in Sub Salit Faulted Reef Complex in the Right Bank of the Amu Darya River, Turkmenistan, and their significance in Exploration: a Case Study of the BP Gas Reservoir. Natural Gas Industry. vol. 35, no. 6, pp. 24–29, 2015. 20. S. Guo, K. Xu, X. Xu, G. Zhang, S. Xiong. Directional Drilling Technology in the Right Bank of the Amu Darya River, Turkmenistan. Drilling & Production Technology. vol. 38, no. 4, pp. 122–124, 2015. 21. H. Zou, X. Wang, J. Kang, H. Zhu, Q. Wang. Key Technologies of Restoring the Productivity of Sour Gas Fields in Turkmenistan. Natural Gas Industry. vol. 29, no. 9, pp. 94–96, 2009. 22. Q. Wang, X. Cheng, H. Fei, L. Wu, R. Chen. Reservoir-Forming Process of a Block in the Right Bank of the Amu Darya River. Natural Gas Exploration & Development. vol. 35, no. 2, pp. 1–4, 2012. 23. B. Liu, X. Wang, W. Li, S. Zhang, J. Dong, Y. Yang. The Problems and Solutions in The Drilling at The Right Bank of Amu Darya in Turkmenistan. In: IADC/ SPE Asia Pacific Drilling Technology Conference and Exhibition: Tianjin, China. pp. SPE 155825, 2012. 24. H. Wen, B. Gong, R. Zheng, H. Liu, L. Wu, R. Chen, S. Li, S. Chen. Deposition and Diagenetic System of Carbonate in Callovian-Oxfordian of Samandepe Gasfield, Turkmenistan. Journal of Jilin University (Earth Science Edition). vol. 42, no. 4, pp. 991–1002, 2012. 25. W. Yan, H. Jiang, X. Wu, G. Xu, R. Chen. Geological Characteristics and Drilling Design of the Amu Darya Right Bank Block, Turkmenistan. Natural Gas Industry. vol. 30, no. 5, pp. 73–76, 2010. Investigation of the Right Bank of the Amu Darya River 243 26. Schlumberger. ECLIPSE reservoir simulation software-Technical Description. 2012. 27. Q. Li, D. Kuang, G. Liu, X. Liu. Acid Gas Injection: A Suitability Evaluation for the Sequestration Site in Amu Darya Basin, Turkmenistan. Geological Review. vol. 60, no. 5, pp. 1133–1146, 2014. 28. Q. Wang, X. Yan, W. Xu, R. Zheng, F. Li, X. Wang, L. Wu. SequencePaleogeographic Characteristics and Evolution of Callovian-Oxfordian in Amu Darya Basin, Turkmenistan. Geology and Exploration. vol. 50, no. 4, pp. 795–804, 2014. 29. R. Zheng, H. Chen, Q. Wang, C. Cui, H. Fei, W. Xu. The Reservoir Characteristics and their Controlling Factors of Callovian-Oxfordian Carbonates in Amu Darya Basin. Acta Petrologica Sinica. vol. 30, no. 3, pp. 779–788, 2014. 30. T. Wang. The Research on Evaluating Carbonate Fractured Reservior of Amu Darya Right Bank Block. Xi’an Shiyou University: Xi’an. 2011. 31. C. Liu, S. Zhou, L. Yan, S. Deng. Drilling Engineering Difficulties in HighSulfur Gas Fields in Turkmenistan and Countermeasures. Natural Gas Industry. vol. 33, no. 1, pp. 79–84, 2013. 32. C. Li. The Relationship between Rock Compressibility and Porosity. China Offshore Oil and Gas (Geology). vol. 17, no. 5, pp. 355–358, 2003. Also of Interest Check out these forthcoming related titles coming soon from Scrivener Publishing Also Available in the Advances in Natural Gas Engineering Series: Acid Gas Extraction for Disposal and Related Topics, edited by Ying Wu, John Carroll, and Weiyao Zhu, ISBN 9781118938614. This fifth volume in the series, Advances in Natural Gas Engineering, offers the most in-depth and up-to-date treatment of acid gas extraction for disposal, an important innovation in natural gas engineering. NOW AVAILABLE! Gas Injection for Disposal and Enhanced Recovery, edited by Ying Wu, John Carroll, and Qi Li, ISBN 9781118938560. This fourth volume in the series, Advances in Natural Gas Engineering, offers the most in-depth and up-todate treatment of the disposal and enhanced recovery of natural gas. NOW AVAILABLE! Sour Gas and Related Technologies, edited by Ying Wu, John J. Carroll, and Weiyao Zhu, ISBN 9780470948149. Written by a group of the most wellknown and knowledgeable authors on the subject in the world, volume three focuses on one of the hottest topics in natural gas today, sour gas. This is a must for any engineer working in natural gas, the energy field, or process engineering. NOW AVAILABLE! Acid Gas Injection and Related Technologies, edited by Ying Wu and John J. Carroll, ISBN 9781118016640. Focusing on the engineering of natural gas and its advancement as an increasingly important energy resource, this volume is a must-have for any engineer working in this field. NOW AVAILABLE! Carbon Dioxide Sequestration and Related Technologies, edited by Ying Wu and John J. Carroll, ISBN 9780470938768. volume two focuses on one of the hottest topics in any field of engineering, carbon dioxide sequestration. NOW AVAILABLE! Ying Wu, John J. Carroll and Weiyao Zhu (eds.) CO2 Capture and Acid Gas Injection, (247–248) 2017 © Scrivener Publishing LLC Other Titles from the Same Authors: Acid Gas Injection and Carbon Dioxide Sequestration, by John J. Carroll, ISBN 9780470625934. Provides a complete overview and guide on the hot topics of acid gas injection and CO2 sequestration. NOW AVAILABLE! Carbon Dioxide Thermodynamic Properties Handbook 2nd Edition, by Sara Anwar and John J. Carroll, ISBN 9781119083580. This second edition of a contemporary classic is the most comprehensive collection of carbon dioxide (CO2) data ever compiled, with new information added to the revision, a “one stop shop” for any engineer or scientist who works with carbon dioxide in any industry. NOW AVAILABLE! Index Amine, 63, 64, 72, 91, 92, 94–99, 105, 109–113, 127–143, 147–157, 161–183 Benedict-Webb-Rubin (see equation of state) Bubble point, 56, 59 Capital cost, 92, 114 Caprock, 229, 231 Casing, 208 Chemical reaction, 85, 91–93, 95, 100, 103, 114, 134, 143, 148–154, 163 Compressibility factor, 185, 188, 193, 195 Compressor, 2, 37, 39, 95, 103, 162, 166, 176 Corrosion, 130, 163, 222 Crude oil, 187, 188 DEA (see amine) DEG (see glycol) Density, 55–60, 63–68, 73, 77–78, 81, 137–138, 143, 155, 166, 172, 177, 202–205, 208, 210, 217, 226 Dew point, 56, 59 Deviation factor, (see compressibility factor) Diethylene glycol, (see glycol) DGA, (see amine) EG, (see glycol) Enhanced oil recovery, 55, 64, 148 Enthalpy, 1–36, 39–52, 85, 140–142, 155–156 EOR (see enhanced oil recovery) Equation of state, 2, 40, 56, 59, 60, 71, 73, 76–77, 133, 149, 202, 224 Ethane, 117, 118, 223 Ethylene glycol, (see glycol) Glycol, 127, 130–131, 135, 137, 139 Hydrostatic gradient, 208–209, 213 Injection pressure, 166, 200, 212, 214 Interfacial area, 101–102 Lee-Kesler (see equation of state) Mass transfer, 93, 95, 96, 100– 103, 130, 140, 165, 185–188, 190–192, 194, 196, 197 MDEA (see amine) MEA (see amine) Mercaptan, 63, 130 Methanol, 175 Operating cost, 91–93, 99, 110, 114 Peng-Robinson (see equation of state) Permeability, 187–188, 194, 201, 227, 229 Ying Wu, John J. Carroll and Weiyao Zhu (eds.) CO2 Capture and Acid Gas Injection, (245–246) 2017 © Scrivener Publishing LLC 245 246 Index Pipeline, 64, 95, 96, 102, 105, 111, 174, 199, 200, 213, 217 Piperidine, 127, 130, 131, 134–135 Porosity, 187, 188, 194, 201, 227, 229 Pump, 57–58, 66, 75, 95–97, 103, 133, 141, 177, 179 Reaction (see chemical reaction) Redlich-kwong (see equation of state) Safety, 64, 92, 213, 222 Soave (see equation of state) TEG (see glycol) Thiol (see mercaptan) Triethylene glycol (see glycol) Tubing, 199–200, 208, 217 Viscosity, 72, 179, 185–186, 188–193, 195, 197, 203–204 z-factor (see compressibility factor)