Surrogate Model based Op1mal Synthesis of Carbon Capture Process

Size: px

Start display at page:

Download "Surrogate Model based Op1mal Synthesis of Carbon Capture Process"

Dulcie Tucker
5 years ago
Views:

Surrogate Model based Op1mal Synthesis of Carbon Capture Process Zhihong Yuan 1, 2 Alison Cozad

Chemical Engineering, Carnegie Mellon University, PiDsburgh, PA David Miller 1 This report was

Neither the United States Government nor any agency thereof, nor any of their employees, makes

accuracy, completeness, or usefulness of any information, apparatus, product, or process

Reference herein to any specific commercial product, process, or service by trade name,

recommendation, or favoring by the United States Government or any agency thereof.

1 Surrogate Model based Op1mal Synthesis of Carbon Capture Process Zhihong Yuan 1, 2 Alison Cozad 1, 2 Nick Sahinidis 1, 2 1 Na$onal Energy Technology Laboratory, Morgantown, WV 2 Department of Chemical Engineering, Carnegie Mellon University, PiDsburgh, PA David Miller 1 This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

2 MOTIVATION Backbone of the energy supply: Fossil fuels Petroleum, coal, natural gas One- third of U.S. CO2 emissions come from power plant Global warming issues Ice mel6ng at poles Rising of ocean levels Greenhouse gas(ghg) emissions CO2 capture, storage & u6liza6on Carnegie Mellon University 2

3 CO2 CAPTURE TECHNOLOGIES Summary of the current CO2 carbon capture technologies (Yuan and Chen, AIChE J, 2012) Most widely inves1gated CO2 capture technology: Post combus1on (Rubin, Prog Energ Combust, 2012) Carnegie Mellon University 3

4 CO2 CAPTURE TECHNOLOGIES MEA solvent based post- combus1on capture technology Highly energy intensive Large solvent makeup due to their thermal and oxida1ve degrada1on High cost (Samanta, I&EC Research, 2012) Cost- effec1ve capture technology is extremely needed to reduce the CO2 emission Growing interest: solid sorbent based adsorp1on process Reduced energy for regenera1on Greater capacity and selec1vity DOE: Carbon Capture Simula1on Ini1a1ve (CCSI) Solid sorbent technology Accelera1ng the commercializa1on of the cost- effec1ve technology Carnegie Mellon University 4

SOLID SORBENT CARBON PROCESS Solid sorbent processes Bubbling fluidized bed Fast fluidized bed Moving bed Fixed bed Bubbling fluidized bed 1D models Modeled in Aspen

5 SOLID SORBENT CARBON PROCESS Solid sorbent processes Bubbling fluidized bed Fast fluidized bed Moving bed Fixed bed Bubbling fluidized bed 1D models Modeled in Aspen Custom modeler Differen1al model Uses Aspen Proper1es package (Andrew Lee, US DOE- Na6onal Energy Technology laboratory, Morgantown, WV) Carnegie Mellon University 5

6 CO2 CAPTURE PROCESS FLOWSHEET General flow sheet for solid sorbent based carbon capture process Carnegie Mellon University 6

7 OBJECTIVES AND HURDLES Objec1ves Achieve the set carbon capture rate Minimizing the cost of electricity (COE) Iden1fy & develop the op1mized bubbling fluidized bed process designs Op1mal configura1on Op1mal design condi1ons Op1mal opera1ng condi1ons Hurdles Computa1onally intractable for large scale nonlinear op1miza1on because of the detailed first principle models Handles Generate the set of low complexity algebraic surrogate models Automated Learning of Algebraic Models for Op1miza1on(ALAMO) (hup://archimedes.cheme.cmu.edu/?q=alamo) Carnegie Mellon University 7

8 SURROGATE MODEL GENERATION Process models Aspen so_ware ALAMO Surrogate models Independent variables Geometry Opera1ng condi1ons Inlet flow condi1ons Dependent variables Geometry required Opera1ng condi1on required Outlet flow condi1ons Design constraints Carnegie Mellon University 8

9 BUBBLING FLUIDIZED BED Bubbling fluidized bed adsorber diagram Outlet gas Solid feed Cooling water Adsorber Model inputs (16 total) Geometry (3) Opera1ng condi1ons (5) Gas mole frac1ons (2) Solid composi1ons (2) Flow rates (4) CO 2 rich gas CO 2 rich solid let Model puts (14 total) Geometry required (2) Pressure (1) Gas mole frac1ons (3) Solid composi1ons (3) Flow rates (2) Outlet temperatures (3) Carnegie Mellon University 9

10 MINLP FORMULATION Adsorber Flue gas flow x = F( surrogates) y( a) + x (1 y( a)) a, fc a 1, fc F = F( surrogates) y( a) + F (1 y( a)) ag, a 1, g T = F( Surrogates) y( a) + T (1 y( a)) ag, a 1, g Solid sorbent flow γaa, = F( Surrogates) y( a) + γ a + 1, A(1 y( a)) x = F( Surrogates) y( a) + x (1 y( a)) aa, a+ 1, A T = F( Surrogates) y( a) + T (1 y( a)) aa, a+ 1, A Logical constraints ya ( ) ya ( + 1), a a ya ( ) 1 a max Carnegie Mellon University 10

$BUBBLING FLUIDIZED BED Bubbling fluidized bed regenerator diagram CO 2 lean solid let Inlet gas Hot water Model inputs (14 total) Geometry (3) Opera1ng condi1ons (5) Gas mole frac1ons (1) Solid$

11 BUBBLING FLUIDIZED BED Bubbling fluidized bed regenerator diagram CO 2 lean solid let Inlet gas Hot water Model inputs (14 total) Geometry (3) Opera1ng condi1ons (5) Gas mole frac1ons (1) Solid composi1ons (2) Flow rates (3) Solid feed Regenerator Outlet gas Model puts (14 total) Geometry required (2) Pressure (1) Gas mole frac1ons (3) Solid composi1ons (3) Flow rates (2) Outlet temperatures (3) Carnegie Mellon University 11

12 MINLP FORMULATION Regenerator Clean gas flow x = F( surrogates) y( d) + x (1 y( d)) d, fc d 1, fc F = F( surrogates) y( d) + F (1 y( d)) dg, d 1, g T = F( Surrogates) y( d) + T (1 y( d)) dg, d 1, g Solid sorbent flow γd, A= F( Surrogates) y( d) + γ d + 1, A(1 y( d)) x = F( Surrogates) y( d) + x (1 y( d)) d, A d+ 1, A T = F( Surrogates) y( d) + T (1 y( d)) d, A d+ 1, A Logical constraints yd ( ) yd ( + 1), d d yd ( ) 1 d max Carnegie Mellon University 12

13 MINLP FORMULATION Assump1ons for nonlinear programming formula1on Each stage is a single stage opera1on U1lity cost for sorbent HX is negligible No pressure change for liquid and solid flow Each stage of adsorber/regenerator opera1on requires agached heat exchanger Surrogate models for fluidized bed adsorber and regenerator First principle models for SolidRich/SolidLean heat exchanger, blower, mixer Carnegie Mellon University 13

14 MINLP FORMULATION Minimize adder to the cost of electricity s.t. First principle models Surrogate models Economic modules Constraints on geometry Constraints on cost Constraints on capture rate Mixed Integer Nonlinear Programming to seek the optimal lay and the corresponding design/operation levels Using GAMS/BARON software Carnegie Mellon University 14

15 RESULTS Optimal results Variables Lower bound Value Upper bound Superstructure COE ($/MWh) CapEX ($) 1.0E E+8 1.0E+10 steamflow (kg/s) derate (MW) u1linf (kgmol/s) Add the surrogate model of adsorber and regenerator into MINLP formulation One adsorber and one regenerator was identified Carnegie Mellon University 15

16 CONCLUSIONS We developed a surrogate model based framework to seek the op1mal design/opera1ng levels for a CO2 capture process ALAMO provides surrogate models of adsorbers and regenerators and thus the problem has a lower complexity Next steps: More complex surrogate model with high accuracy Select reactor type for each stage More complex superstructures Carnegie Mellon University 16

Surrogate Model based Optimal Synthesis of Solid Sorbent Carbon Capture Processes

Surrogate Model based Optimal Synthesis of Solid Sorbent Carbon Capture Processes Zhihong Yuan 1, 2 Nikolaos V. Sahinidis 1, 2 David C. Miller 1 1 National Energy Technology Laboratory, Pittsburgh, PA