D 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 x 1 x 2 d A d B T A T B Feasible Case MinBPD = 0 DESIGN AND SYNTHESIS OF COMPLEX COLUMN NETWORKS WITH GLOBAL FEASIBILITY TEST Gerardo J. Ruiz, Seon B. Kim, and Andreas A. Linninger Laboratory for Product and Process Design, Departments of Chemical and Bio-Engineering, University of Illinois at Chicago, Chicago, IL 60607, USA General on Separations (02H00), AIChE Annual Meeting, Nashville, TN, Nov 9, 2009 Poster No. 335o Case Study 2 - Initialization of Complex Distillation Networks with AspenPlus Simulator Motivation and Objectives Rigorous Feasibility Test Reduced Search Space Methodology - Complex Column Network Synthesis Conclusions References Agrawal, R. (2003). "Synthesis of multicomponent distillation column configurations." AIChE J 49(2): 379-401. Tapp, M., S.T. Holland, D. Hildebrandt, and D. Glasser, Column Profile Maps. 1. Derivation and Interpretation. I&EC Research , 2004. 43(2): p. 364-374. Zhang, L. and A. A. Linninger (2004). "Temperature collocation algorithm for fast and robust distillation design." I&EC Research 43(12): 3163-3182. Zhang, L. and A. A. Linninger (2006). "Towards computer-aided separation synthesis." AIChE J 52(4): 1392-1409. Acknowledgements •DOE Grant: DE-FG36-06GO16104 •Dr. Rakesh Agrawal (Purdue University) •Dr. Chau-Chyun Chen (Aspen Tech.) Motivation Distillation occupies in chemical process: 40-70% of capital and operating costs 60% of the total process energy 4% of total energy consumption in United States Atmospheric carbon emissions There is a need for a redefinition of the design objectives for industrial separations with a new focus on energy conservation and the emission reduction using complex column configurations have the potential of achieving up to 70% energy savings over simple column networks • Temperature collocation and minimum bubble point distance algorithm were effective to find a feasible separation by intercepting profiles. • The first case study demonstrates the potential to save 72% in energy using a complex column network compared to the simple column network. • The second case study demonstrates the current state of the art of separation synthesis in conjunction with computer simulations to fully integrate complex separation networks. • The seamless integration of rigorous flowsheet simulators to validate the predictive results of our scientific method was demonstrated. A quaternary mixture of pentane, hexane, heptane, and octane was studied. In the complex network, it uses two simple columns for pre-fraction to complex column. As a result, the best energy efficient complex column network saves up to 72% of the operating cost in terms of vapor flowrate. distance between two adjacent sections. The minimum bubble point distance -8 D3 =0.0081 and BPT= 72.6 C. Basic complex column configurations Generic Structure Synthesis Network Task Optimization Using Difference Point Equations Find Feasibility of Two Column Sections Considering Operating and Capital Cost Obtain Optimal Design Level I: Find all possible basics configurations Mixed Integer Linear Program Level II: Identify the feasible complex column Supply Design Temperature BPD Level III : Obtain optimal designs Flow rate tray# According to the general definition of the minimum bubble point distance approach, a complex column k sum of all minimum profile distances of any pair of equivalent rectifying, r, and stripping, s, column sections is within a small tolerance of zero () as in expression Plus Simulation of Composition and Temperature Profiles Temperature Collocation Composition Profiles Global Design Procedure Basic Complex Configuration: Quaternary System Temperature Collocation of a General Column Section 1 1 1 1 1 c i i i i c i i i i i i i K x dn n x T dT x T x y X x K R R Column Section profile: C i i i i i i C i i i i i i i i K x X R y x R x T K x X R y x R x 1 1 1 1 1 L R i i i Lx Vy X L V 1 1 i i i x K 1 0 i i i d Kx dx 1 2 3 1 2 3 1 1 2 3 1 2 3 1 1 1 K K K x x x x T T T T x h x h x h K K K h x h x h x Implicit Differentiation Case Study 1 – Complex Column Network for Quaternary Mixtures Separation Basic Structure Each random initial guess for integer variables, x, will generate one possible structure Feasibility test High performance Inverse problem 2 min i x sum Components sum Components out out in node x * x * , sum sum out dist in x x , sum sum out bot in x x int sum 1 prod x .. s t x Each individual different design (sequence + operating conditions) K individuals Population MASTER •Dr. Angelo Lucia (University of Rhode Island) •Dr. Diane Hildebrandt (University of the Witwatersrand) 60 80 100 120 0 0.2 0.4 0.6 0.8 1 Tstage, o C X i 60 80 100 120 0 0.2 0.4 0.6 0.8 1 Tstage, o C X i 40 60 80 100 120 0 0.2 0.4 0.6 0.8 1 Tstage, o C X i 60 80 100 120 0 0.2 0.4 0.6 0.8 1 Tstage, C X i 60 80 100 120 0 0.2 0.4 0.6 0.8 1 Tstage, C X i 60 80 100 120 0 0.2 0.4 0.6 0.8 1 Tstage, C X i D B C 1 B 0.2 0.4 0.6 0.8 1 0 0.2 0.6 0.8 1 A B C Optimal operating conditions for 9 selected Networks A B C A B C A B S 1 S 2 S 3 S 4 C Inverse Design of Distillation Column Column I Column II Column III Unit (kmol/h) Net. 1 Net. 2 Net. 3 Net. 4 Net. 5 Net. 6 Net. 7 Net. 8 Net. 9 Column I 74.99 79.99 186.87 114.81 186.87 156.56 156.56 478.08 186.87 Column II 78.07 57.36 46.73 63.34 60.60 66.44 281.14 63.70 472.62 Column III 70.34 64.66 48.99 86.30 62.68 76.09 62.681 72.50 63.27 Total 223.41 202.02 282.59 264.45 310.15 299.09 500.38 614.28 722.76 Complex Network Simple Network Column I Column II Column III Column I Column II Column III 1 2 1 ( ) min ( ) , () () K k Zk BPD T k Zk D A C B • Starting with the desired design specification of product purity requires that each column of the network is feasible • Feasible design – Intersection of profiles of both adjacent sections • Profile Intersection Index - bubble point distance (BPDObjectives Develop computer-aided systematic design procedures to prevent numerical failures associated with the extraordinary sensitivity of column profile calculations Massive size reductions enabled by a new column profile computation algorithm called Temperature Collocation Synthesize separation networks with realistic column profiles Realizable column profiles validated with industrially accepted simulation software such as AspenPlus L LPPD D WITH GLOBAL FEASIB inninger gn, at Chicago , Nashville LP C PP P PD mplex Col rk, it mns nt etwork s Point Temperature Distance etween two adjacent se s localized at r= 2.3 c complex column tructure Synthesis sk Optimization Equations mn Sections PD Level I: Find all PD PD L Basic B P PD PD asible complex column P P PD PD Pinch Point Feasible Temperature PD PD BPD n PD PPD in optimal designs P P PD P P O PD PD PD Output Flow ra tray# # k is feasible if and only column sections is within a sma Aspen Pl Te Design Procedure PD nary System P ion of a Ge D D dn n dT n ion profile: D D R 1 1 1 D D D D 1 0 dx i i Kx i i i i D Implici Differenti D LP rk for P L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L D D L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L LP LP LP L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L LP L LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L L LP L LP LP LP LP L L L L L L L L 1 LP LP LP LP LP LP LP L L L L L L L LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP L L L L L L L L L L L L L L LP LP LP LP LP LP LP LP LP LP LP LP LP LP L L L L L L L L LP LP LP LP L L L L LP LP LP LP A elected Networks A B S 1 S 3 C C C et t. 7 7 N Net t. 8 8 Net t. 9 9 47 78 08 8 47 78 .08 8 18 86 87 7 18 86 .87 7 70 0 70 0 47 72 62 2 47 72 .62 2 6 63 27 7 6 63. 27 7 2 76 6 .76 6 L L L Complex Network Simple Network LPP PP PP P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P LP LP P P P P P P P P P P P LP P P P P P LP P LP LP LP LP P LP LP LP LP LP LP P LP LP LP LP LP LP LP LP P P LP LP LP LP LP LP P P LP LP LP LP LP LP LP LP P P LP LP LP LP LP LP P P LP LP LP LP LP LP P P LP LP P P LP LP P LP LP P LP LP P P P P LP LP P P LP LP P P LP LP P P P P LP LP P P P P LP LP P P LP LP P P P P LP LP P LP P P LP LP P P P P LP LP P P LP LP P P P LP LP P LP LP P LP LP P LP LP P P LP LP P P LP LP P P LP LP P P P LP LP P LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP LP P D P djacent ce ( BPD )