Main Heading - University of Auckland

Main Heading - University of Auckland

New Zealand Modelling Dual Reflux Pressure Swing Adsorption for Methane and Nitrogen Separations in LNG Production The University of Auckland Thursday, February 13, 2020 Yechun Zhang1 [email protected] Thomas Saleman2, Eric May2 and Brent Young1 1: The Department of Chemical and Materials Engineering, The University of Auckland, Auckland 1010, New Zealand 2: School of Mechanical and Chemical Engineering, The University of Western Australia, Crawley WA 6009, Australia WA's North West Shelf Thursday, February 13 , 2020 Project Pluto Gorgon Prelude Wheatstone The University of Auckland New Zealand Ichthys Resource 5.0 TCF 47.3TCF 295MMbblc 2.8 TCF 120MMbblc

4.5 TCF 12.8 TCF 527MMbblc Estimated Start-up Proposed Capacity Concept [mmtpa] 2012 4.3 2014 15 2016 3.6 2016 8.9 2018 8.4 LNG Plant at Burrup Peninsula LNG Plant on Barrow Island Floating LNG LNG Plant at Onslow LNG Plant at Darwin Significant offshore gas fields with production since 1989

Estimated $140B investment in next 10 years (most committed) N2 vent streams from plants contain 3% CH4: 435000 CO2e Page 2 tonnes per annum in greenhouse emissions Thursday, February 13 , 2020 Conventional PSA to Capture the CH4 Consider conventional PSA cycles to Lean Product (< 100 ppm CH4) achieve 100 ppm CH4 in N2 vent The University of Auckland New Zealand Challenges - Low CH4 selectivity for current adsorbents - Waste stream barely enriched HP LP Feed Waste (CH4 enriched) - Unviable numbers of beds required Solutions - Require a new adsorbent with high selectivity (6-10) - Process engineering design Page 3

New Zealand Thursday, February 13 , 2020 Dual Reflux PSA Theoretically produces 2 pure products from binary mixture1-3 Experimental demonstrations show: The University of Auckland 1. Enrichment ratios well beyond standard thermodynamic limit 2. Lean products with <100 ppm contaminants possible 1 Diagne, D., Goto, M., Hirosi, T., 1994.. Journal of Chemical Engineering Japan 27 (1), 8589. 2. A. Ebner and J. Ritter AIChE Journal. Volume 50, Issue 10, pages 24182429. 3. J. A. Mc Intyre, C. E. Holland, and J. A. Ritter. Ind. Eng. Chem. Res. 2002, 41, 3499-3504. Page 4 Possible Process Design Solution: Dual Reflux PSA Thursday, February 13 , 2020 Lean Product (N2) The University of Auckland New Zealand Intermediate feed position HP LP

Feed to HP or LP Stripping PSA HP LP Feed Enriching PSA Rich Product (CH4) Page 5 Thursday, February 13 , 2020 Evaluating DR-PSA Viability Question: How to evaluate DR-PSA for N2/CH4 separation required by LNG industry? The University of Auckland New Zealand Strategy: a) Conduct modelling to establish whether DR-PSA solution is economically viable b) If true, then proceed with prototype experiments. What modelling tools available for viability screening purpose? Page 6 Analytical Model of Kearns & Webley for Screening Started with understanding Kearns & Webleys analytical model of DR-PSA Method of characteristics to reduce PDE to ODEs Analytical solution solved using numerical techniques

Thursday, February 13 , 2020 Pros New Zealand - Insight into the process - Fast Solution - Search optimized parameters Cons - Idealized: assumes perfect separation - Can not be tested against experimental data The University of Auckland - Cant be sure optimum found will actually achieve product specifications Need a more realistic model to test optimum solutions ability to achieve product specifications - Transfer function model achieve goals faster and easier - Full numerical simulation of DR-PSA still under development Page 7 Thursday, February 13 , 2020 Transfer Function: Stripping What is transfer function model? Lean Product Feed Stripping PSA The University of Auckland New Zealand

(Transfer function) Waste Aspen Adsorption Page 8 The University of Auckland New Zealand Thursday, February 13 , 2020 Transfer Function: Enriching Rich Product Feed Enriching PSA (Transfer function) Waste Aspen Adsorption Page 9 Transfer Function Model The University of Auckland New Zealand Thursday, February 13 , 2020 Steady-state results from stripping & enriching PSA cycle simulations give half-cycle transfer functions (TFs) Combine TFs & iteratively solve

material balance Test against McIntyre et al.s experimental results1 Test optimized parameters from analytical model 1 J. A. Mc Intyre, C. E. Holland, and J. A. Ritter. Ind. Eng. Chem. Res. 2002, 41, 3499-3504. 2 Lean Product Stripping Reflux PSA (Transfer function) 1 3 6 4 Feed 7 Enriching Reflux PSA (Transfer function) 5 Rich Product Page 10 2010 Scenario1 Transfer (RS=0.382, Function 1 RE=72.3) Model Results Item Inputs Feed Rate (sccm)

Results The University of Auckland New Zealand Thursday, February 13 , 2020 Comparison with C2H6/N2 result of McIntyre et al. Lean Product Flow (sccm) Rich Product Flow (sccm) Total Cycle Flow (sccm) yFeed yLean Product (ppm) yEnriched Product N2 Recovery C2H6 Recovery 583.6 573.0 10.63 733 1.35% 2900 62.6% 99.2% 80.8% 583.3 573.1 10.30 723 1.35% 2000 63.9% 99.4% 83.2% 1 J. A. McIntyre, Armin D. Ebner, and J. A. Ritter. Ind. Eng. Chem. Res. 49, 2010, 18481858. Page 11

The University of Auckland New Zealand Thursday, February 13 , 2020 Comparison with C2H6/N2 result of McIntyre et al. Mean Deviations Item 2002 2002 2010 2010 Mean Scenario 1 Scenario 3 Scenario 1 Scenario 6 Deviation 0.05% 0.04% -0.09% 0.06% 0.06% -2.90% -4.60% 1.27% -5.10% 3.47% RecLP -1.70% -0.40% 0.14%

-0.10% 0.59% RecHP -5.70% -4.70% 2.42% -5.10% 4.48% yL (full %) yE (full %) Page 12 Item Inputs Pressure Ratio Results The University of Auckland New Zealand Thursday, February 13 , 2020 Apply TF model to Analytical Optimum Transfer Function Transfer Function Analytical Model Model using Model using

with Optimized Analytically Alternate Parameters Optimized Parameters Parameters 1.9 1.9 3 5.0MMscfd 5.0MMscfd 0.85MMscfd 0.445 0.445 0.5 Stripping Reflux Ratio 2.7 2.7 10.1 Enriching Reflux Ratio 199 199 8.1 Lean Product 0%

2.91% 0.085% Rich Product 100% 10.04% 5.1% Methane Recovery Rate 100% 4.42% 98.8% Throughput Feed position Page 13 The University of Auckland New Zealand Thursday, February 13 , 2020 Conclusions and Future Work Conclusions - Transfer Function Model were proposed - Showed close match with experimental data

Future work - Full DR-PSA model is available and under test - Proposed the first DR-PSA sizing tool and being extended - Conduct optimum search with transfer function model - Laboratory DR-PSA apparatus to test real CH4/N2 separation under construction Page 14 Thursday, February 13 , 2020 Full Numerical Model of DR-PSA Four column configuration with intermediate feed Available for Feed to High Pressure New Zealand column Unstable pressure-flow network observed The University of Auckland Backflow at feed

Switch to ideal compressor Ideal compressor Preliminary tests done (CH4/N2) T, CH4% in feed, axial feed position, Real compressor cycle time Page 15 Preliminary Sizing Tool of DRPSA Thursday, February 13 , 2020 Preliminary Sizing Algorithm for DR-PSA First-order DR-PSA sizing tool Enables rough estimation of DR-PSA New Zealand Equilibrium capacity Qads = Qi(T,P) 8a. Calculate required adsorbent mass for stripping part Integrate stripping PSA with enriching PSA The University of Auckland

3. Specify adsorption and sorption properties i, bulk, Qi column size & scale using idea of Transfer Function Model 2. Use TFM to calculate total material balance and obtain local flow rate and concentration for stripping/enriching PSA part according to configuration Stripping: FFS, yFS, yL, yLR, RS Enriching: FFE, yFE, yER, yE, RE 1. Define Objectives FF, yF, yE, yL and configuration (feed to HP/LP) Estimate size of stripping part Estimate size of enriching part Adding up and check output Compromise in calculating gas residence time m adsS t ads F FS ( y FS y )/Q L 4. Specify adsorption

time tads 8a. Calculate required adsorbent mass for enriching part m ads ,eff adsE t ads F FE ( y E y FE )/Q ads ,eff 5. Specify temperature and pressure PH, PL, T 9a. Calculate volume of adsorbent V m adsS / adsS bulk 9b. Calculate volume of adsorbent V 6. Calculate gas properties g, , MW 10a. Calculate h satE and number of beds required Nads

N g MW 2 bed MTZS bedS v ); v int S v gS / i 12a. Calculate height MTZ and bed, tres for stripping part h h ads v int S / MTC h satS h MTZS V adsS /( 2 2 ) /( ) 4 h satS D bed V adsE 4 h satE D bed 11b. Calculate gas velocities: superficial (vg) and interstitial (vins) for enriching part

11a. Calculate gas velocities: superficial (vgS) and interstitial (vinsS) for stripping part D bulk 10b. Calculate h satS and number of beds required Nads 7.Specify Dbed and hsatS (or hsatE) 2 2 N ads V adsS / ( 4 h satS D bed ) V adsE / ( 4 h satE D bed ) v gS F FS / ( 4 N ads m adsE / adsE 13. Combination of stripping part and enriching part h bed gE F FE / ( N 4 D g

MW 2 bed ); v int E v gE / i 12b. Calculate height MTZ and bed, tres for enriching part h bedS h bedE h h tres=hbed/vintS (Feed to HP) tres=hbed/vintE (Feed to LP) MTZE bedE v int E / MTC h satE h MTZE a,b,c 14. Output of algorithm result Nads, Dbed, hbed, tads, tres ads e 15. Check algorithm result a. Pressure drop: Is P/L<7.5kPa/m AND P/L*hbed<55kPa? Use Ergun equation and vg=vgS(feed to HP) or vg=vgE (feed to LP) b. Sufficient time for adsorption: Is tres>tads ? c. Practical bed dimensions: Is 2>1, Qi,eff=Qi,eqbm (ii) (tres*MTC)<<1, Qi,eff=0 (iii) (tres*MTC)~1, Qi,eff=0.63Qi,eqbm

e. Is Nads too large? Page 16 The University of Auckland New Zealand Thursday, February 13 , 2020 Thank you Page 17 High pressure -> intermediate pressure Extract (N2 enriched) 1 2 Pressure Equalization Low pressure desorption High pressure adsorption New Zealand Thursday, February 13 , 2020 Conventional PSA Cycles The University of Auckland 1 Feed Raffinate (C1 enriched)

Adsorption/Desorption Low pressure -> intermediate pressure 1 2 Depressurization/ Pressurization 2 1 Intermediate pressure > low pressure 2 Intermediate pressure -> high pressure Page 18 Transfer Function Model New Zealand Thursday, February 13 , 2020 Pros and Cons Pros - Flexible source of TF (experiment or simulation) - Faster & more stable than full numerical model of DR- The University of Auckland PRSA Cons - CCS only - Operates as separate columns - Averaged flow between stripping & enriching PSA

- Reasonable match to - Slower than analytic available experimental data solution Page 19 Thursday, February 13 , 2020 Research Motivation Greater Gorgon LNG production: 1000 MMSCFD N2 contamination in raw gas New Zealand Methane concentration in N2 vent: ~3% Substantial GHG emission and carbon tax! The University of Auckland (AU$23/ton) 435,000t CO2e -> 10M AU$ for Gorgon Objective Source: Finn et al. 2007 3%->100 ppm Page 20 The University of Auckland New Zealand Thursday, February 13 , 2020 DR-PSA Model Results

Page 21 The University of Auckland New Zealand Thursday, February 13 , 2020 DR-PSA Model Results Page 22 The University of Auckland New Zealand Thursday, February 13 , 2020 DR-PSA Model Results Page 23 The University of Auckland New Zealand Thursday, February 13 , 2020 DR-PSA Model Results Page 24

Recently Viewed Presentations