Development of an ammonia-based process for CO 2 capture in cement plants: absorption tests, rate-based modelling and process optimization

Transcription

1 Development of an ammonia-based process for CO 2 capture cement plants: absorption tests, rate-based modellg and process optimization José-Francisco Pérez-Calvo, Daniel Sutter, Federico Milella, Matteo Gazzani, Marco Mazzotti, ETH Zurich PCCC4, 5 th 8 th September, 2017

2 Talk outle Introduction and scope of the study Model CAP heuristic optimization for cement plants Pilot plant tests of the CO 2 absorber Regression of ketic parameters Conclusions José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 2

3 CO 2 emissions from cement & the CAP 5% of global anthropogenic CO 2 emissions ~ 0.58 t CO 2 /t cement (BAT) ~ 50 60% process-related emissions CCS required Why the Chilled Ammonia Process (CAP)? Low thermal energy for regeneration required Limited waste heat available cement plants Stable the presence of impurities Technology demonstrated various facilities of different scale José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 3

4 The Chilled Ammonia Process José-Francisco Pérez-Calvo, 06-Sep-17 4

5 The Chilled Ammonia Process José-Francisco Pérez-Calvo, 06-Sep-17 5

6 The Chilled Ammonia Process José-Francisco Pérez-Calvo, 06-Sep-17 6

7 The Chilled Ammonia Process José-Francisco Pérez-Calvo, 06-Sep-17 7

8 From power plants to cement plants NG power plants ~ 4 %vol. CO 2 Coal-fired power plants ~ 14 %vol. CO 2 Cement plants 15 35%vol. CO 2 Model-based optimization Process complexity Heat tegration Thermodynamics Ketics Adaptation of operatg conditions José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 8

9 Cement plant Power plant k app(s 1 ) Introduction Model CAP optimization Pilot tests Model regression Conclusions Scope of the study Equilibrium model (Aspen Plus) Pilot plant tests Reaction ketics Rate-based model (Aspen Plus) r = k appc CO2 T( C) Startg pot Thomsen model [1,2] Murphree effs. for power plants from literature [3,4] Ultimate goal Thomsen model [1,2] Mass transfer resistance Reaction ketics [1] Thomsen and Rasmussen Chem Eng Sci 54 (1999) [2] Darde et al. Ind Eng Chem Res 49 (2010) [3] Sutter et al. Faraday Discuss 192 (2016) [4] Jilvero et al. Ind Eng Chem Res 53 (2014) Model-based optimization José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 9

10 Cement plant Power plant k app(s 1 ) Introduction Model CAP optimization Pilot tests Model regression Conclusions Scope of the study Equilibrium model (Aspen Plus) Pilot plant tests Reaction ketics Rate-based model (Aspen Plus) r = k appc CO2 T( C) Startg pot Thomsen model [1,2] Murphree effs. for power plants from literature [3,4] Ultimate goal Thomsen model [1,2] Mass transfer resistance Reaction ketics [1] Thomsen and Rasmussen Chem Eng Sci 54 (1999) [2] Darde et al. Ind Eng Chem Res 49 (2010) [3] Sutter et al. Faraday Discuss 192 (2016) [4] Jilvero et al. Ind Eng Chem Res 53 (2014) Model-based optimization José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 10

11 Cement plant Power plant k app(s 1 ) Introduction Model CAP optimization Pilot tests Model regression Conclusions Scope of the study Equilibrium model (Aspen Plus) Pilot plant tests Reaction ketics Rate-based model (Aspen Plus) r = k appc CO2 T( C) Startg pot Thomsen model [1,2] Murphree effs. for power plants from literature [3,4] Model validation with Literature research Thomsen model [1,2] CSIRO pilot tests Validated ketic model Ultimate goal Ad-hoc Murphree efficiencies for cement plants Thomsen model [1,2] Defition of test matrix Cement-like flue gas composition Regression of ketic parameters Thomsen model [1,2] Mass transfer resistance Reaction ketics 5 [1] Thomsen and Rasmussen Chem Eng Sci 54 (1999) [2] Darde et al. Ind Eng Chem Res 49 (2010) Model-based optimization [3] Sutter et al. Faraday Discuss 192 (2016) [4] Jilvero et al. Ind Eng Chem Res 53 (2014) Model-based optimization José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 11

12 Thermodynamic model: CO 2 -NH 3 -H 2 O system Thomsen model to predict the system thermodynamics Vapor CO 2 H 2 O NH 3 SRK for gas fugacities Liquid CO 2 H 2 O NH 3 H 2 O OH - + H + NH 3 + H + NH 4 + HCO 3 - CO H + CO 2 + OH - HCO 3 - CO 2 + 2NH 3 NH 2 COO - + NH 4 + VLE Extended UNIQUAC (Thomsen-Darde [1] ) Solubility data for solids fitted on Jänecke [2] Solid (NH 4 )HCO 3 (NH 4 ) 2 CO 3 2NH 4 HCO 3 (NH 4 ) 2 CO 3 H 2 O NH 2 COONH 4 Ice [1] Darde et al. Ind Eng Chem Res 49 (2010) [2] Jänecke Z Elektrochem 35 (1929) 9: SLE Pure solids, activity = 1 External route Aspen from Thomsen group José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 12

13 Phase diagram: CO 2 -NH 3 -H 2 O system S: Solid L: Liquid V: Vapor 10 C 1 bar Pure solids BC: ammonium bicarbonate (NH 4 )HCO 3 SC: ammonium sesqui-carbonate (NH 4 ) 2 CO 3 2NH 4 HCO 3 CB: ammonium carbonate (NH 4 ) 2 CO 3 H 2 O CM: ammonium carbamate NH 2 COONH 4 Light blue area: Two-phase region where the solid exists its mother liquor Red area: The algorithm does not converge José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 13

14 Phase diagram: CO 2 -NH 3 -H 2 O system [2] [1] [1] Jänecke Z Elektrochem 35 (1929) 9: [2] Sutter et al. Chem Eng Sci 133 (2015) José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 14

15 k app (s 1 ) Introduction Model CAP optimization Pilot tests Model regression Conclusions Rate-based model validation with CSIRO tests x 10 4 CO 2 + 2NH 3 k cm NH 2 COO + NH 4 + r cm = k app C CO2 5%wt NH %wt H 2 O Puxty model [1] Psent et al. Trans Faraday Soc. 52 (1956) [2] Puxty et al. Chem Eng Sci. 65 (2009) Psent model T( C) T lean = 16.4 C T lean = 27.0 C Rate-based model with: T lean = 23.9 C Thomsen model Reaction ketics: Psent model T cooler = 15.3 C T cooler = 23.1 C T cooler = 26.7 C Experiment [3] Psent model Puxty model Puxty model T FG = 16.6 C T FG = 21.8 C T FG = 28.0 C [3] Qi et al. Int J Greenh Gas Con 17 (2013) José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 15

16 Cement plant Power plant k app(s 1 ) Introduction Model CAP optimization Pilot tests Model regression Conclusions From the power plant to the cement plant Equilibrium model (Aspen Plus) Pilot plant tests Reaction ketics Rate-based model (Aspen Plus) r = k appc CO2 T( C) Startg pot Thomsen model Murphree effs. for power plants from literature [3,4] Model validation with CSIRO pilot tests Literature research Thomsen model Psent model Ultimate goal Ad-hoc Murphree efficiencies for cement plants Thomsen model Defition of test matrix Cement-like flue gas composition Regression of ketic parameters Thomsen model Mass transfer resistance Reaction ketics 5 Model-based optimization Model-based optimization José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 16

17 First CAP heuristic optimization for cement plants Total Specific Exergy Needs w = W reboilers + W chillg + W auxiliaries abs m CO2 MJ kg CO 2 captured Operatg conditions varied the CO 2 absorber L/G CO 2 loadg NH 3 concentration Pumparound: Temperature Split fraction Specifications and constrats CO 2 capture > 85 % CO 2 purity > 99 %mole NH 3 slip < 200 ppm No solid formation José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 17

18 CO 2 absorber profiles for optimum operatg conditions 22% 18% Cement Plant 14% vol. CO 2 Power Plant Adaptation of operatg conditions Higher L/G Lower NH 3 content Constant CO 2 loadg José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 18

19 Defition of the test matrix for new absorber pilot tests CSIRO [1] %vol CO 2 the let fluegas CSIRO pilot tests [1] This work Pilot tests This work [1] Yu et al. Chem Eng Res Des 89 (2011) José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 19

20 Test rig Inlet liquid conditions l CO2 = mol CO2 /mol NH3 = mol NH3 /kg H2 O m NH3 T L = 5 45 C L G = (kg kg) Inlet gas conditions Q G y CO2 = Am 3 /h 0.35 Boundary box for experimental results José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 20

21 Systematic treatment of pilot raw data Raw data PG out, kpa QG out, Am 3 /h y, vol% MCO2, mol/l M NH3, mol/l T, C Q L, L/h Introduction Model CAP optimization Pilot tests Model regression Conclusions T L out T L T G out y CO2 out y NH3 out y CO2 out M CO2 M CO2 out M NH3 M NH3 [1] Martínez-Maradiaga et al. Appl Therm Eng. 51 (2013) t, m José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 21

22 σ 180s T G out, C Systematic treatment of pilot raw data Raw data Steady-state detection QG out, Am 3 /h PG out, kpa M NH3, mol/l y, vol% T, C Q L, L/h Introduction Model CAP optimization Pilot tests Model regression Conclusions τ(t G out ) SS detected Average values & st.devs No SS T L out Methodology based on a movg wdow [1] T L T G out y CO2 out y NH3 out y CO2 MCO2, mol/l out M CO2 M CO2 out M NH3 M NH3 [1] Kim et al. Int J Refrig. 31 (2008) t, m José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 22

23 H G out HL out, kw Introduction Model CAP optimization Pilot tests Model regression Conclusions Systematic treatment of pilot raw data CO 2 mass balance Raw data +10% NH 3 mass balance +10% Steady-state detection Average values & st.devs out out Gas CO Liq 2 CO 2, kmol/h out out Gas NH Liq 3 NH, kmol/h 3-10% -10% Gas CO2 Liq CO2, kmol/h Total mass balance Gas NH3 Liq NH3, kmol/h Energy balance +10% out Gas tot out Liqtot, kmol/h -10% -10% +10% Gas tot Liq tot, kmol/h H G H L, kw José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 23

24 H L HL out, kw Introduction Model CAP optimization Pilot tests Model regression Conclusions Systematic treatment of pilot raw data CO 2 mass balance Raw data +20% NH 3 mass balance +20% Steady-state detection Average values & st.devs out Liq CO Liq 2 CO, kmol/h 2 out Liq NH Liq 3 NH, kmol/h 3-20% -20% Malfunctiong struments? Gas CO2 Gas out CO2, kmol/h Total mass balance Gas NH3 Gas out NH3, kmol/h Energy balance Deceptive measurements? Unreliable experiments? Which experiments and measurements can we trust? out Liq tot Liqtot, kmol/h +20% -20% -20% +20% Gas tot Gas out tot, kmol/h H G out H G, kw José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 24

25 Systematic treatment of pilot raw data Thomsen model MB s & EB Raw data Steady-state detection Average values & st.devs DoF analysis Data reconciliation Data reconciliation [1] F = m subject to: f u DR, z = 0 i Reconciled value u i DR u i DM σ i 2 Measured value CO 2 mass balance NH 3 mass balance Air mass balance Total mass balance Energy balance Gross error detection GED [1] if e i = u i DR u i DM σ i e c GED? Yes No Pilot test results [1] Martínez-Maradiaga et al. Appl Therm Eng. 51 (2013) José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 25

26 Systematic treatment of pilot raw data Thomsen model MB s & EB Raw data Steady-state detection Average values & st.devs DoF = 2 Data reconciliation Malfunctiong struments? Gross error detection Inlet gas flowrate Deceptive measurements? Outlet liquid flowrate Reconciled Measured 82 experimental pots GED? No Pilot test results Yes Outlet liquid temperature Unreliable experiments? 7 experimental pots discarded José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 26

27 N CO 2, kg/h Introduction Model CAP optimization Pilot tests Model regression Conclusions Analysis of pilot test results CO 2 absorption rate P CO2 = 20 kpa P CO2 = 16 kpa P CO2 = 8 kpa P CO2 = 20 kpa P CO2 = 30 kpa, mol CO2 /mol NH3 l CO2, mol CO2 /mol NH3 l CO2 m NH3 = 5.7 mol NH3 /kg H2 O T L = 16 C L = 8.7 kg/kg = 1.0 m/s G v s,g m NH3 = 6.0 mol NH3 /kg H2 O T L = 40 C L = 11.3 kg/kg = 0.7 m/s G v s,g José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 27

28 Model regression Carbamate ion reaction ketics Pilot plant CO 2 capture tests for cement plant Experimental value from pilot tests Carbamate ion reaction ketics? N CO2 = A t K G,CO2 (P CO2 P CO2 ) Rate-based modellg Aspen Plus A t = f hydrodynamics transport properties (P CO2 P CO2 ) = f thermodynamics Rochelle model [1] to compute the effective G-L terfacial area Thomsen thermodynamic model to compute the drivg force 1 K G,CO 2 = RT k g,co 2 + H CO2 0 Ek l,co 2 k g,co2 0 = f k l,co2 hydrodynamics transport properties Rochelle model [1] to compute the G-film and L-film mass-transfer coefficients H CO2 = f thermodynamics Partition coefficient computed by Thomsen model [1] Wang et al. Ind Eng Chem Res 55 (2016) E = f L phase reaction ketics José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 28

29 K G,CO2 model, mmol/(s m2 kpa) Introduction Model CAP optimization Pilot tests Model regression Conclusions Model regression Carbamate ion reaction ketics Tests with l CO2 < 0.5 mol CO2 /mol NH3 BC formation negligible due to low C OH Sgle irreversible reaction [1,2] E = f Ha [3] CO 2 + 2NH 3 k cm NH 2 COO + NH 4 + Ha = n k cm C NH3 D CO2,L 0 k l,co2 n r cm = k cm C NH3 C CO2 +20% +10% k cm = k 0cm,Tref exp E a,cm R 1 T 1 T ref Psent model -10% -20% k cm (T = T ref = 298 K) E a,cm n m 3 kmol s kj kmol This work This work Psent [1] Jilvero et al. Ind Eng Chem Res. 53 (2014) [2] Ahn et al. Int J Greenh Gas Control. 5 (2011) [3] van Swaaij and Versteeg. Chem Eng Sci. 47 (1992) K G,CO2 exp, mmol/(s m 2 kpa) José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 29

30 Conclusions Introduction CAP process Scope Model CAP optimization Conclusions The Chilled Ammonia Process can be applied for CO 2 capture to cement plants The adaptation of the operatg conditions the CO 2 absorber is required CO 2 absorption pilot tests have been performed successfully, usg: Synthetic flue gases mimickg cement-like flue gas compositions Liquid compositions and operatg conditions derived from the CAP heuristic optimization, which has been done based on: Assessment of the energy requirements Equilibrium model with ad-hoc Murphree efficiencies for cement plant flue gas compositions A ketic model for the formation of the carbamate ion has been regressed order to be used with engeerg purposes José-Francisco Pérez-Calvo, francisco.perezcalvo@ipe.mavt.ethz.ch 06-Sep-17 30

31 Acknowledgements This project has received fundg from the European Union's Horizon 2020 research and novation programme under grant agreement no This work was supported by the Swiss State Secretariat for Education, Research and Innovation (SERI) under contract number José-Francisco Pérez-Calvo, 06-Sep-17 31