Si l t d M i B d T h l iSimulated Moving Bed Technologies for Biofuel Production

Prof. N-H. Linda WangSchool of Chemical Engineering

Purdue UniversityPurdue University

Overview• Chromatography and SMB

– Principles– Key Barriers

• Purdue SMB Technologies– Standing Wave Design & Optimization– VERSE Dynamic Simulations

V til SMB E i t– Versatile SMB Equipment– Model Based Design ApproachExample: Tandem SMB for insulin purificationExample: Tandem SMB for insulin purification

• A Five Zone SMB for Isolation of Six Sugars from Bi H d l (Mi f 10Biomass Hydrolysates (Mixtures of 10 or more components)

Chromatography-AdvantagesAdvantages

• High selectivity – 1 in 10,000. • Versatile mechanisms of separation –

adsorption, ion exchange, sizeadsorption, ion exchange, size exclusion, etc.C t l d t• Compact volume compared to extraction.

• Lower solvent usage than extraction.• Mild operating conditions suitable for• Mild operating conditions – suitable for

fragile compounds.

Mechanisms of Adsorption

Limitations of Batch Chromatography

• Difficult to achieve both high purityDifficult to achieve both high purity and high yield.L d b t tili ti• Low adsorbent utilization.

• High solvent consumption andHigh solvent consumption and product dilution.

• Protein aggregation and instability.• Cumulative yield loss in multiple• Cumulative yield loss in multiple

steps.

Solution I- Binary Separation in a Continuous Moving Bed

Solution II–Moving Port or SMBDesorbent ExtractZONE I

21

Solute8

21

3Solute Movement

ZONE IV ZONE II

8 3

Step N

7 4

ZONE III

56

FeedRaffinate

DesorbentDesorbent

ZONE IZONE IV 21

Solute Movement

8 3

ExtractRaffinateStep N + 1

7 4

Feed

ZONE III ZONE II56

Click for animation

Advantages of SMB over Batch• Only partial separation of solutes is required to

obtain high purity.• Higher yield than batch – 10% more than batch.• High purity achievable without sacrificing yield.• High productivity– 5 to 10 fold increase.• Less solvent – 5 to 10 fold less.Less solvent 5 to 10 fold less.• Reduced environment impact.• Reduced footprint, equipment size, and educed ootp t, equ p e t s e, a d

manpower.• Continuous process-high throughputContinuous process high throughput.• Economical for large scale separations.

Key Barriers in SMBy• Complex transient and cyclic steady state

phenomena.• A four-zone SMB has 9 design parameters - 29 to

39 trials.L b f t & ti t t• Large number of system & operating parameters to be optimized.

• Method for Multi-component separation not wellMethod for Multi-component separation not well developed.

• Equipment more complex than batch q p pchromatography.

• No equipment for multi-component separation.• Control of batch identity and residence time

Purdue’s Platform SMB Technologies

1. Standing wave design and optimizationoptimization.

2. VERSE simulations.3. Versatile SMB equipment. 4 A systematic model-based approach4. A systematic model-based approach

for process design and optimization.

Standing Wave Design: Binary Ideal Linear SystemBinary, Ideal, Linear System

FeedDesorbent

I II III IV

Con

c.

I II III IV

B d P iti

uw,iα = u0

α / ( 1 + Pδi ) = wave velocity of component i in zone α

Isloww,u=ν

Extract RaffinateBed Position

= wave velocity of component i in zone α

ν = Lc / ts = port movement velocityFfeed = feed flow rateIII

sloww

IIfastw,

u

u

=

=

ν

ν

u0α = interstitial velocity in zone α

εb = inter-particle void fractionP = (1 – εb) / εb = bed phase ratio( )II

0III0

Feed

IVfastw,

sloww,

uuSεF

u

−=

=ν

δi = distribution coefficient of i S = column cross-sectional area

( )00b uuSεF

Standing Wave Designf Li N id l Sfor Linear, Nonideal Systems

F dD b tAdsorption wave of Low ffi it l t t di i

Desorption wave of Low affinity solute standing in Zone II Feed Desorbent

Zone I Zone II Zone III Zone IV

affinity solute standing in Zone IV

solute standing in Zone II

Con

c.

Zone I

Extract RaffinateBed Position

•Matching wave velocities to port velocity to ensure high purity and high yield•Difference in wave velocity and port velocity allows focusing for nonideal systems

Standing Wave Design & Optimization-AdvantagesAdvantages

• Four zone flow rates and port velocity are l d ilsolved easily.

• Little computation time required.• Achieving maximum productivity and

minimum solvent consumption for a pgiven configuration.

• Easy extension to nonlinear multi-Easy extension to nonlinear, multicomponent, multi-zone separations.

• Reduce search space by five dimensions• Reduce search space by five dimensions or more for optimization.

VERSEVERSEVErsatile Reaction SEparation Simulator

• Simulation based on a detailed dynamic rate model (transient differential mass balances)Th d f ti l d t h• Thousands of equations solved at each instance

• To predict column profiles and histories• To predict column profiles and histories• To understand transient phenomena in batch

chromatography carousel and SMBchromatography, carousel, and SMB• To understand residence time and batch

integrity issues in SMBg y• To reduce the number of costly experiments

Mechanisms Considered in VERSEFluid Bulk ConvectionBulk

Fluid

Reaction FilmDiffusion

Adsorption

Ion Exchange

PoreDiffusion

InterferenceAdsorptionSites Adsorption

SorbentParticle

Denaturation

Size S rface

Particle

SizeExclusion

SolidPhase

R ti

Surface Diffusion

Fluid

.

ReactionPore Fluid

Versatile SMBT i l t l fTo implement complex processes for

multicomponent separations

Valve Controls

ValvesValves

SephadexG50 Columns

Pumps

Features of Versatile SMBFeatures of Versatile SMB• Dedicated valves and manifolds for high g

purity and high yield• Allows zone bypass open loop multizoneAllows zone bypass, open loop, multizone,

multisolvent ,and multicomponent separation.• Easy column expansibility• Easy column expansibility.• Moving port chromatography.• Online decoupled regeneration. Animation• Independent column switching for variable p g

step time

Model Based Approach to SMB DesignParameter Software Pilot

Solutes/Sorbent System

VERSESimulation1

VersatileSMB Unit2

estimation tools SMB

System Simulation

BatchPulse

ProfilesHistories

ProfilesHistories

Voids

Frontal Purity & Yield

Splitting

Purity & Yield

IsothermsMass Transfer

ParametersSMB

Strategies3

Standing Wave Experiments

Cost

Design4

1 Berninger et al., Comp. Chem. Eng. 15 (1991) 749.

4 Ma and Wang, AIChE J. 43 (1997) 2488.Cost

Optimization52 Chin and Wang, Separ. Purif. Rev. 33 (2004) 77.3 Hritzko et al., AIChE J. 44 (2002) 2769.

5 Mun et al., Ind. Eng. Chem. Res. 42 (2003) 1977.

Model-Based Design ofModel Based Design of Tandem SMB for Insulin

fPurification

Biosynthetic Human Insulin (BHI) ProcessRecombinant E. coli.

Trp proinsulinHomogenization

Trp proinsulin

ProinsulinCleavage, oxidative sulfitolysis, folding

Enzymatic transformation

Crude insulin

Enzymatic transformation

Ion-exchange

Reversed-Phase HPLC

Size exclusion chromatography (SEC)

Purified BHI

Size exclusion chromatography (SEC)

Crystallization

Isolation & purification steps take 10 to 12 weeks.

Conventional Size Exclusion Chromatography

0.8

1Insulin

(Product)

0.6

C/C

F Proinsulin / HMWP

Zinc Chloride

0.2

0.4

C HMWP ChlorideFronting

00 100 200 300

• Existing SEC batch process• Yield < 90%

Proinsulin / HMWPInsulin (Product)

Volume (L)

Yield 90% • Purity >99%• Resin productivity = 0.3 L / hr/100 L BV

Zinc ChlorideAnimation

Tandem SMB for Insulin P ifi tiPurification

ELUENTELUENT

EXTRACT ( ) EXTRACT ( )

ELUENT

FEED ( ) FEED ( )

RAFFINATE ( ) RAFFINATE ( )

InsulinHMWP

Zinc Chloride

SMB Experimental Validation : Ring I(a) Effluent histories at the raffinate port

0.6

0.8

1

CF

Insulin exp.Insulin sim.HMWP exp.

(b) Effluent histories at the extract port

0.6

0.8

1

CF

HMWP exp.HMWP sim.Insulin exp.

0

0.2

0.4

0 10 20 30 40 50

C/C HMWP sim.

ZnCl2 exp.ZnCl2 sim.

0

0.2

0.4

0 10 20 30 40 50

C/C

Insulin sim.ZnCl2 exp.ZnCl2 sim.

0 10 20 30 40 50Switching Step

0 10 20 30 40 50Switching Step

(c) Mid-step column profiles at cyclic steady state

Desorbent Extract Feed Raffinate

0.9

1.2

CF

Insulin exp.Insulin sim.HMWP exp

Zone I Zone IVZone II Zone III

0.3

0.6C/C HMWP exp.

HMWP sim.ZnCl2 exp.ZnCl2 sim.

00 1 2 3 4 5 6 7 8 9 10

Column number

SMB Experimental Validation: Ring II(a) Effluent histories at the raffinate port

0.6

0.8

1

/CF

Insulin exp.Insulin sim.

(b) Effluent histories at the extract port

0.6

0.8

1

/CF

Insulin exp.Insulin sim.

0

0.2

0.4

0 20 40 60Switching Step

C/

ZnCl2 exp.ZnCl2 sim.

0

0.2

0.4

0 20 40 60Switching Step

C/

ZnCl2 exp.ZnCl2 sim.

Switching Step Switching Step

(c) Mid-step column profiles at cyclic steady state

Desorbent Extract Feed Raffinate

0.9

1.2

CF

Insulin exp.Insulin sim.

Zone I Zone IVZone II Zone III

0

0.3

0.6C/C Insulin sim.

ZnCl2 exp.ZnCl2 sim.

00 1 2 3 4 5 6 7 8 9 10

Column number

VERSE Simulation of Insulin Separation from Impurities

(High Molecular Weight(High Molecular Weight Proteins and Zinc Chloride) in

Tandem SMB (Ring I)

Click port-r1-final.pptm to activate animationactivate animation

VERSE Simulation of Separation of Insulin from

Zi Chl id i T d SMBZinc Chloride in Tandem SMB (Ring II)(Ring II)

Clickport-r2.pptm to ti tactivate

Splitting Experimental Key Issues Solved

D i &

p gStrategies

(Hritzko et al., AICHE J 2002)

pValidation

(Xie et al., Biotech Prog 2002)

Robust

SMB

Design & Optimization

(Mun et al., IEC Res 2003)

Robust Operation

(Mun et al., IEC Res 2003)

SMBHigh PurityHigh Yield

Startup & Shutdown

Insulin Aggregationg

Low Cost(Xie et al., IEC Res 2003)

Periodic Versatile

(Yu et al., JCIS 2006)

Periodic Regeneration

(Xie et al., WCCBME 2003)

Residence

VersatileSMB

(Chin & Wang,Sep Purif Rev 2004)Batch

Time(Mun et al.,

AICHE J 2003)

Identity(Mun et al.,

Biotech Prog 2004)

Insulin Purification• Tandem SMB for ternary separation

d i d d lid t d ith i tdesigned and validated with experiments • >99% yield, 5X throughput, and 1/3 y , g p ,

solvent consumption • Robust and optimal Standing Wave Design• Robust and optimal Standing Wave Design • Novel strategies created to reduce protein

id ti d i t i b t h id titresidence time and maintain batch identity• Technology licensed to a major companygy j p y

A Five-zone Simulated Moving Bed gfor Isolation of Six Sugars from

Biomass HydrolyzateBiomass Hydrolyzate

Yi Xie, Chim Yong Chin, Diana S.C. Phelps, Chong Ho Lee Ki Bong Lee SungyongChong Ho Lee, Ki Bong Lee, Sungyong

Mun, N H Li d WN.-H. Linda Wang.

Ind. Eng. Chem. Res. 2005, 44, 9904-9920

Sugars Isolated using Simulated Moving BedSteamBiomass

Hydrolysis

Steam & Acid

Simulated

WaterAcetic Acid (HAc)

Hydrolyzate

Hydrolysis Simulated Moving Bed

for Sugar Isolation

Waste Treatment

H2SO4

Furfural Hydroxymethyl

S/L Separation

Fermentation Purification

Glucose, Xylose, Galactose, Mannose, Arabinose, Cellobiose

Furfural (HMF) WasteSolid cake:Cellulose

• SMB replaces overliming

Fermentation Purification

Ethanol or Butanol

Beer

SMB replaces overliming – Reduce steps & process time– Removes known fermentation inhibitors– Reduce gypsum waste disposal

Objective: An economical SMB for sugar recovery

OverviewPVP DOWEX Stationary phases PVP = Relix HP, poly(4-vinylpyridine)PVP

Batch Exp

DOWEX

Batch Exp

Stationary phases

Elution sequences, system & intrinsic parameters

DOWEX = Dowex99, H+ form

p p

5-zoneSMB

4+4, 4+5SMB

intrinsic parameters

SMB configurations

SWD

VersatileSMB Unit

Feed Regeneration

No mL/min Conc. pH*

VERSE

& Cost Optimization SMB 15-zone

PVPSMB

#1 15 0.3M NaOH 7

#2 5 0.3M 7VERSESimulation

SMB 2

SMB 3

#2 5 NaOH 7

#2 15 0.3M NH4OH 9

Optimal 5-zonePVP SMB

($0.027/gal feed)

NH4OH

* final pH

DOWEX 99 Tandem SMBBatch chromatogram

20 0.8HydrolyzateWater

10

15

atio

n (g

/L)

0 4

0.6

rfura

l (g/

L) Zone I Zone II Zone III Zone IV

5

10

Con

cent

ra

0.2

0.4

HM

F, F

ur

Standing wave

00 20 40 60 80

Time (min)

0

Glucose Xylose Sulfuric acidAcetic acid HMF Furfural

Extract:Glucose, Xylose,HAc, HMF, Furfural

H2SO4, HAc,Furfural

Raffinate

H2SO4Glucose XyloseHAcFurfuralWater

• Elution sequence: – H2SO4

Sugars (Glucose Xylose)

Acetic acid HMF Furfural

Zone I Zone II Zone III Zone IV

FurfuralWater

– Sugars (Glucose … Xylose) – HAc, HMF, Furfural

• Sugars – Linear isotherms• Water for desorbent• Water for desorbent• 4+5 SMB uses 0.3N HCl

HAc, HMF, Furfural

Glucose,XyloseExtract Raffinate

5-zone Non-isocratic PVP SMBEl ti

Batch Chromatogram• Elution sequence:

– Sugars (Cellobiose … Arabinose) – Impurities (HAc, HMF, Furfural,

H2SO4)15

20

25

tion

(g/L

)

0.3

0.4

0.5

ural

(g/L

)

GlucoseXyloseAcetic acidHMF H2SO4)

• Sugars - Linear isotherms• HAc, HMF, Furfural, H2SO4 –

Nonlinear isotherms0

5

10

Con

cent

rat

0

0.1

0.2

HM

F, F

urfuFurfural

• 0.3N NaOH regenerant• Water desorbent

00 50 100 150 200 250 300

Time (min)

0

5-zone Non-isocratic Open-loop PVP SMB

Zone Zone Zone Zone

Hydrolyzate

Zone

WaterRegenWater

5 zone Non isocratic Open loop PVP SMB

Zone I

Zone II

Zone III

Zone IV

Sulfate,A Sugars

Zone V

WasteWaste Acetate,HMF,

Furfural

Sugars WasteWaste

Separation zone Regeneration Reequlibration

Standing Wave Design: 5-zone PVP SMBHydrolyzate

Zone VZone I Zone II Zone III Zone IV

Water0.3N NaOHWater

Zone VZone I Zone II Zone III Zone IV

WasteSugarsSulfate, Acetate,HMF F f lWaste

CellobioseArabinose

Acetic Acid (Nonlinear)Regenerant

asteg

RaffinateHMF, Furfural

Apparent “a” of acetic acid obtained from shock wave velocity of acetic acid from frontal simulation of feed

Experimental Setup: 5-2-2-2-1 PVP-SMBWaste Waste S garsWaste

C

Sugars

C

Regenerant

C

Water

C

Hydrolyzate

Back Pressure Regulator Sampling Valve

C

Water

Constant Speed Pump

C

ZI B1 B1 B1 B1 ZII B1 D B2 FST B3 R

FiltersHeater

Pressure Gage

g Sampling Valve

Back Pressure Regulator

Tee

1 2 3 4 5 6 7 8 9 10

ST1 ST2 ST3 ST4 ST5 ST6 ST7 ST8 ST9 ST10

ST Valve

11 12

ST11

ST12

1 2 3 4 5 6 7 8 9 10 11 12

Manual Sampling ValvesColumnHeaters

Separation zone Regen. Re-equilibration

C l 2 69 ID 30 5 LColumn: 2.69 cm ID x 30.5 cm L Temperature: 65oC (Reduces viscosity, prevent growth)HPLC Assays: See Xie et al., Ind. Eng, Chem. Res. 2005, 44, 6816.

SMB Run 1: Validate Model & Design40 Column profile Raffinate history

30

40

tion

(g/L

)

10

20

Con

cent

rat

07 8 9 10 11 12

Column numberGlucose Xylose Arabinose

0 200 400 600 800 1000 1200Time (min)

Glucose Xylose Arabinose

• Feed #1 (Less sugars) at 15 mL/min

yAcetic acid HMF FurfuralSulfuric acid

yAcetate Furfural Sulfate

• 0.3M NaOH as regenerant• Reequilibrate to pH 7.0• Fluctuations due to precipitation – use filterp p• Raffinate at 94.5% purity & 99.5% yield• Purity loss due to acetate and sulfate salts

SMB Run 1: Validate Model & Design40 Column profile Raffinate history

30

40

tion

(g/L

)

10

20

Con

cent

rat

07 8 9 10 11 12

Column numberGlucose Xylose Arabinose

0 200 400 600 800 1000 1200Time (min)

Glucose Xylose Arabinose

• Feed #1 (Less sugars) at 15 mL/min

yAcetic acid HMF FurfuralSulfuric acid

yAcetate Furfural Sulfate

• 0.3M NaOH as regenerant• Reequilibrate to pH 7.0• Fluctuations due to precipitate formation – use filterp p• Raffinate at 94.5% purity & 99.5% yield• Purity loss due to acetate and sulfate salts

SMB Run 3: Test New Regeneration50

Column profile Raffinate history

30

40

50

tion

(g/L

)

10

20

Con

cent

ra

07 8 9 10 11 12

Column number

Glucose Xylose Arabinose

0 100 200 300 400Time (min)

Gl X l A bi

• Feed #2 (higher conc. of sugars) at 15 mL/min

Glucose Xylose ArabinoseAcetic acid HMF FurfuralSulfuric acid

Glucose Xylose ArabinoseMannose Galactose

• 0.3M NH4OH as regenerant• Re-equilibrate to pH 9.0 ( not pH 7, to reduce cost)

Raffinate at 94 0% purity & 100% yield• Raffinate at 94.0% purity & 100% yield• SMB performance unaffected by changes

Fermentation Test of Raffinate Sugars100% Standards % Sugar

40%

60%

80%

100% Standards

Raffinate Products

% Sugar Fermented in

48 hrs

Ethanol Yield

Run 1 98 0 42

0%

20%

100%rs /

Products

SMB Run 1Run 1 98 0.42

Run 2 85 0.32

• Feed #2 (Runs 2 & 3) has higher sugars40%

60%

80%

nsum

ed s

ugar

Initi

al s

ugar

s Run 3 87 0.36

has higher sugars concentrations and probably higher unknown impurities too

0%

20%Con

SMB Run 2

100% p

• Unknown impurities may adversely affect20%

40%

60%

80%

may adversely affect fermentation0%

20%

0 10 20 30 40 50Time (hr)

SMB Run 3

Comparison of SMB ResultsRun 1 Run 2 Run 3

Purity (%) 94.5 93.4 94.0

• Lower purities because unanticipated acetate and

Yield (%) 99.5 99.7 100.0

• Lower purities because unanticipated acetate and sulfate co-eluted with sugars– Discounting salts results in 99+% puritiesDiscounting salts results in 99 % purities

• Fermentability: Run 1 (Feed #1) >> Runs 2 & 3 (Feed #2)( )

• Optimal 5-zone PVP SMB for 50 gal feed/min– 10% Dilution (Lab exp at 20%)( p )– $0.027 / gal feed (21% lower than early estimate)

Five Zone SMB for Sugar RecoverySummarySummary

• Model based approach appliedModel based approach applied successfully S f• Standing wave design for ten-component separation validated

• 5-zone open-loop V-SMB validated6 sugars isolated from 10 identifiable• 6 sugars isolated from 10-identifiable-component feedstock with high purity and high yield

ConclusionsConclusions• SMB has higher yield puritySMB has higher yield, purity,

throughput, and requires less solvent than batch chromatographythan batch chromatography.

• Conventional SMB technologies are limited to binary isocratic four zonelimited to binary, isocratic, four-zone separations. Transient phenomena are not well understoodnot well understood.

• Large number of parameters to be specified and optimized in SMBspecified and optimized in SMB.

ConclusionsConclusions• Purdue SMB technologiesPurdue SMB technologies

– Software tools for design, simulation, and optimizationand optimization.

– Versatile SMB equipment for multi-t ticomponent separation.

• Used in developing high purity, high yield separation processes for the purification of biochemicals and pharmaceuticals.

AcknowledgementsAcknowledgements• Dr. N. Ho, Laboratory of Renewable Resources y

Engineering, Purdue University for fermentation test s of sugars recovered from five-zone SMB

• 21st Century Indiana Research and Technology21 Century Indiana Research and Technology Fund (28)

• Department of Energy (DE-FC36-01GO11071)• Purdue TTI-IRL (NSF IGERT 9987576)• SWD: Z. Ma, Y. Xie, K.B. Lee• VERSE: R Whitely Y Xie• VERSE: R. Whitely, Y. Xie• V-SMB equipment: C. Chin• Biofuel: Y Xie C Chin K B Lee D PhelpBiofuel: Y. Xie, C. Chin, K. B. Lee, D. Phelp

Commercial Applications of SMB

• Hydrocarbons – Molex, Parex• Sugars – Tate & Lyle• Chiral DrugsChiral Drugs• Antibiotic drug

*All binary separations in four zoneAll binary separations in four-zone SMB

Novel SMB Processes Developed at PurdueDeveloped at Purdue

• Pilot Scale– Insulin– Sugars from Biomass

• Lab Scale– Antibiotic Drug– Amino Acids and Derivatives

Lactic Acid– Lactic Acid– Paclitaxel (Anticancer Drug )– Chiral DrugsChiral Drugs

Standing Wave Design: Multi-Component System with Mass Transfer Effects

DesorbentFeed

(1, 2, 3)

II

1 Fast Solute2 I t di t

I II III IV

*ntra

tion

2 Intermediate3 Slow Solute*

*

Con

ce

Standing Wave

Extract(2, 3)

Raffinate(1)

*Bed Position *g

Pinched Wave

Standing Wave Design vs Triangle Theory

• Ring I of the tandem SMB for insulin purification

vs. Triangle Theoryinsulin purification

• Size exclusion chromatography– ε* = εb

0.8

1

Standing wave design

– a = Ke × εp = δ

• Standing wave design– Fronting in zone III and other

0.4

0.6

m3*

Fronting in zone III and other mass transfer effects considered

• Triangle theory– Mass transfer effects ignored

0

0.2

0 0.2 0.4 0.6 0.8 1

Triangular region

Mass transfer effects ignored

• Extra-column dead volume considered in both methods

m2*

Comparison of Triangle Design and Standing Wave Design: Ideal Linear IsothermWave Design: Ideal, Linear Isotherm

2

1.5

2

“Standing wave”solution

1m3

0.5

Region of solutionsfor completeSeparation from

0 0.5 1 1.5 20

Separation from Triangle Design

m2

VERSEBulk ConvectionBulk

FluidBulk ConvectionBulk

Fluid

VERSEReaction Film

Diffusion IonReaction Film

Diffusion Ion

InterferenceAdsorptionSites Adsorption

Ion Exchange

InterferenceAdsorptionSites Adsorption

Ion Exchange Chromatographic

Separation Solute

Ion exchange Amino acids

Ion exchange Metal ionsPore

DiffusionSorbentParticle

PoreDiffusion

SorbentParticle

g

Ion exchange Proteins

Hydrophobic adsorption Amino acids

H d h bi ChiralDenaturation

SizeExclusion

Surface

Denaturation

SizeExclusion

Surface

Hydrophobic adsorption

Chiral compounds

Hydrophobic adsorption Proteins

Hydrophobic Organic Exclusion

SolidPhase

Reaction

DiffusionExclusion

SolidPhase

Reaction

Diffusiony

adsorptiong

acids

Hydrophobic adsorption Sugars

Reversed phase Antibiotics

R d h P t iReactionPore Fluid

ReactionPore Fluid

Reversed phase Proteins

Reversed phase Taxol

Size exclusion Proteins

Motivation for V-SMB• Existing SMB systems were designed for

binary isocratic separationbinary, isocratic separation.• They have limited column, zone, pump and

solvent configurationssolvent configurations.• The columns rotate in the ISEP system

which is impractical in production scalewhich is impractical in production scale.• They cannot accomplish online decoupled

regenerationregeneration.• They cannot implement optimal temperature

in each zonein each zone

Versatile SMB PrototypesVersatile SMB Prototypes

Model Based Approach

Mobile Phase VERSE Versatile

Engineering Analysis

Process Design & Optimization

Experimental Validation

Mobile PhaseStationary Phase

S l ti it

Simulator

P fil

SMB

P filSelectivitySolubility

ProfilesHistories

PurityYield

ProfilesHistories

PurityYieldBatch

FrontalPulse

St di W

Yield

O ti l

Yield

Voids EstimationI th

Standing Wave Design (SWD)

Optimal Design

IsothermsMass Transfer

ParametersCost Optimization

Optimal Production

Process

SMB Design & Cost Optimization based on the Standing Wave Analysis

ProductSpec.

SMB System($ / unit)

Stationary Phase($ / kg)

Mobile Phase($ / Liter)

Separation System

PurityNo. of ComponentsFeed CompositionProduction Rate

the Standing Wave Analysis

Particle SizeΔPmax Solubility

Max. Flow Rates

Production RateTandem or Parallel

No. of ColumnsColumn LengthZone Lengths

Selectivity

Batch, Pulse, Frontal Tests

STANDING WAVE ANALYSIS

By ΔPmax

Frontal Tests

Voids, IsothermsMass-Transfer

Parameters

ANALYSIS

Max. Feed Flow Rate Max. Feed Conc.Z Fl R t

Variables to be optimized

Splitting Strategies

Yi ld Parameters(Eb, kf)

Zone Flow RatesSwitching Time

Throughput PerBed Volume

COST OPTIMIZATION

YieldMobile & Stationary Phases

Splitting StrategiesParallel/Tandem

Feed Conc. for Nonlinear Systems Feed Flow Rate

Solvent Consumption

OPTIMIZATIONFeed Flow RateColumn Length Zone LengthsParticle Size

Comparison of SMB UnitsCalgon Carbon (ISEP)

UOP(1SD1S)

US Filter (1SD1C)

UOP &Novasep

(Two-Way)

Purdue Versatile

SMB

Columns Rotate Stationary Stationary Stationary Stationary

Column Expansible No No Yes Yes Yes

Contamination None Substantial Some Substantial None

Multiple Zones Good Good Limited Excellent Excellent

Multiple SMB Schemes No No No Potential Yes

Variable Zone Length No Potential Potential Yes Yes

Online Decoupled Regeneration*

No No No No Yes

Versatile SMB PrototypesVersatile SMB Prototypes

Versatile SMB PrototypesVersatile SMB Prototypes

Tandem SMB for Insulin PurificationFIRST SMB (Ri I)

• SEC SMB for insulin purification EXTRACT

ELUENT

FIRST SMB (Ring I)

– Impurities: • High molecular

weight proteins RAFFINATE

FEED

weight proteins (HMWP) and ZnCl2

– Elution sequence:

RAFFINATE

InsulinProinsulin /HMWP

i C iInsulinProinsulin /HMWP

i C i• HMWP→ Insulin→ ZnCl2

– Two SMBs in series EXTRACT

ELUENT

Second SMB (Ring II) Zinc ChlorideZinc Chloride

Two SMBs in series• Ring I: Remove

HMWP, Distribute ZnCl2

FEED

EXTRACT

ZnCl2• Ring II: Remove

ZnCl2

RAFFINATE

Splitting Strategies for T Mi ta Ternary Mixture

Tandem SMB (2 SMBs in Series)( )

Strategy 1st SMB 2nd SMB(Zone No ) Raf | Ext Raf | Ext(Zone No.) Raf. | Ext. Raf. | Ext.S1 (4+4) 1 | 2 3 2 | 3S2 (4+4) 1 2 | 3 1 | 2S3 (4+4) 3 1 | 2 3 2 | 3S3 (4+4) 3 1 | 2 3 2 | 3S4 (4+4) 1 2 | 3 1 1 | 2

1: HMWP 2: Insulin 3: ZnCl21: HMWP 2: Insulin 3: ZnCl2Splitting strategies control productivity, solvent consumption, and product dilution.

E ti fi t• Easy separation first• Distribution of one impurity in the first ring

SMB vs. Plant-Scale Batch SEC for Insulin Purificationfor Insulin Purification

SMB Plant-scale batch

Overall yield (%) > 99.0 < 90Overall yield (%) > 99.0 < 90

Overall throughput 1 2 < 0 3Overall throughput (L/hr/100L BV)

1.2 < 0.3

Overall solvent consumption (L/Kg BHI)

42.0 > 120

Simulated Moving Bed Chromatography for Binary Separationfor Binary Separation

AnimationPheTrpExp2Ver2-0-200ms-with Restart ver 2.avi