Simulation of nonlinear simulated moving bed chromatography using ChromWorks computational software

Simulation of nonlinear simulated moving bed chromatography

using ChromWorks computational software

January 31, 2013

Reid Erwin

School of Chemical & Biomolecular Engineering,
Georgia Institute of Technology, Atlanta, GA 30332
rerwin7@gatech.edu

Abstract

The purpose of this report is to present the results of the analysis of nonlinear simulated

moving bed chromatography (SMB) by use of computational software (ChromWorks). Multiple

objectives were achieved during the analysis: validation of isotherm modeling and operating

conditions for separation of cycloketones (cyclopentanone, cyclohexanone) described in Bentley

et al. [1]; and assisting the developers of ChromWorks by evaluating and testing the software to

validate and verify its usefulness for industrial applications. All simulations and data with respect

to isotherm modeling and operating conditions were directly referenced from Bentley et al. [1].

Separations were performed in ChromWorks with identical Langmuir isotherm parameters, but

at varying feed concentrations (20 g L-1 and 34 g L-1) and flow rates. The desired purity in

Bentley et al. [1] of 96% for extract and raffinate was achieved for both feed concentrations

upon reaching steady state. For the concentration of 34 g L-1, multiple operating conditions were

1

validated for 96% purity. Through experimentation with ChromWorks while performing these

simulations, it was found that the software is highly useful for industry application.

1. Introduction

The software package “is a computer-aided modeling/simulation software in preparative

and continuous chromatography” [2]. It is equipped with several samples and workshops that

provide essential understanding of operation and data input. Online support is also available.

The workshops that provided the most beneficial support relevant to this research included

graphic user interface (GUI), SMB, and process design tutorials.

1.1 GUI Tutorial

The GUI tutorial provides a clear understanding of the layout of the interface and other

essential features of the program. The chemicals library feature provides physical data such as

density and viscosity values at a range of temperatures for hundreds of chemical species. This

tool is helpful in determining precise solvent property data that is entered into the simulation.

Another essential feature is the data transfer tab that allows seamless data transfer between

modules. A base set of parameters is set in the process design tab which contains the equilibrium

diagram and also data entry for flow operating conditions, isotherm parameters, and component

and column specifications. This data is easily transferred over to the main control panel within

the SMB tab where it can be manipulated while observing the simulation. If slight modifications

are made, the user may transfer this data back over into the process design tab with the data

transfer option. Data transfer provides simple transition between these interfaces.

1.2 SMB Tutorial

2

The SMB simulation tutorial describes data input for mathematical modeling with respect

to equilibrium isotherm parameters and how to operate a simulation. It also describes how to

access and interpret simulation results for supplemental visual analysis.

1.3 Process Design Tutorial (Determination of Operating Conditions)

The process design tutorial is the most important for understanding how data can be

manipulated with respect to the equilibrium isotherm when designing a simulation. The process

design interface allows two options for designing a simulation. In the first case, a single

operating condition (e.g. switching time) is specified along with a point in the diagram to

determine mII-mIII ratios. The software then automatically calculates the rest of the operating

conditions (feed, raffinate, desorbent, extract). The robustness factor ratio may also be modified

in this interface to increase extract and raffinate purity. Alternatively, all operating conditions

(feed, raffinate, desorbent, extract, switching time) can be specified. ChromWorks will then

automatically calculate flow ratio parameters and illustrate a point on the equilibrium plane to

see if it falls within an area favorable for separation. Examination of effects of operating

condition modifications can be observed across the equilibrium plane. The process design

workshop provides essential understanding of how equilibrium isotherm design parameters are

referenced and combined with operating condition estimation to create an optimal simulation.

1.4 Robustness and Triangle Theory

A robust SMB design may be described as a design that is resilient against sub-optimal

conditions that may occur during separation. Robustness improves the likelihood of achieving

desired results such as increased purity. The need to improve robustness may be examined in the

multi-column profile while viewing an SMB simulation at the end of a switching interval. For

3

example, component A may be washed away too quickly and its adsorption front in zone IV may

encroach into the desorption front of component B in zone I. The result of this behavior is

contamination. Vice versa, the desorption front of component B in zone I may slow down

enough to contaminate the extract stream of component A in zone IV. Contamination of the

components leads to decreased purity. Manipulating the robustness factor in favor of improving

purity can be performed by increasing and decreasing the safety margin (β) in zone I and zone IV

respectively in equations (1) and (4). The safety margin (β) is an equality constraint that relates

the following inequality constraints for complete separation.

QI / QS = mI = HAβI; where mI ≥ HA and βI ≥ 1 (1)

QII / QS = mII = HBβII; where mII ≥ HB and βII ≥ 1 (2)

QIII / QS = mIII = HA/βIII; where mIII ≤ HA and βIII ≥ 1 (3)

QIV / QS = mIV = HB/βIV; where mIV ≤ HB and βIV ≤ 1 (4)

ChromWorks specifies β values for each individual zone and refers to safety margins βI and βIV

as “Q1-Ratio” and “Q4-Ratio” in the process design window. ChromWorks automatically

calculates β values for zones II and III based on the specifications set by the user for zones I and

IV. This method is more versatile than the conventional way of specifying robustness that

applies a single β value for the entire system. By default, βI and βIV are set at their limiting

values (1.0) in ChromWorks due to the conditions specified in the above defining relationships.

This value represents the upper left corner of the triangle for perfect separation which can be

found in Figure 1.

4

Figure 1: Example of linear
equilibrium triangle diagram from
Seader et al. [3].

As robustness is increased, mII-mIII ratios move slightly closer to the 45° line in the triangle

diagram. As this occurs, there is a tradeoff for increased purity for decreased productivity.

ChromWorks suggests setting the safety margins (Q1-Ratio and Q4-Ratio) to 1.05 and 0.95

respectively. Keeping Q1-Ratio and Q4-Ratio slightly above and below 1.0 respectively

provides significantly improved purity without drastically increasing volumetric flow rates.

Where increased purity may be desired, increased flow requirements of materials (such as

desorbent) would lead to additional expense. Flow rates can become significantly higher as

robustness is increased. The robustness factor is an essential consideration during process design

and analysis.

1.5 Pressure Drop

The unitless Reynolds number is an indicator of pressure drop in columns. Pressure drop

in columns may become too high in certain situations. This could possibly result in damage to

packing materials such as internal porosity changes due to crushing. ChromWorks performs

simulations assuming there is no such damage. For the described SMB analysis, a Reynolds
5

number within 100 represents functional operating conditions and prevents the effect of

damaging.

2. Case Study

2.1 Determination of Design Parameters

The first step in process design involved finding an initial guess or point on the

equilibrium plane that achieved desired purity for both extract and raffinate within Langmuir

isotherm parameters referenced from Bentley et al [1]. The strategy for determination of initial

guess location in the equilibrium plane was to select a point farthest away from the 45° line, but

still within the area for perfect separation. Optimization of throughput occurs where the

operating condition that is farthest away from the 45° line is located. However, deviation from

this general area results in tradeoff for product purity. For the non-linear isotherm of Bentley et

al. [1], the theoretical optimum location did not prove to provide the best combination of

operating conditions. Figure 2 illustrates the general location within the equilibrium triangle that

provided actual desired results for each simulation.

Figure 2: The point labeled
“real optimum” of H.
Schmidt-Traub [4] provided
the actual initial guess for
optimal design parameters in
each simulation.

6

Table 1 lists isotherm parameters for all simulations. Table 2 lists column specification data

referenced in all simulations.

Table 1: Langmuir isotherm
parameters from Bentley et al. [1]
Table 2: Column specification data.
Langmuir Isotherm Parameter θ1
Four HPLC columns (C1 through C4)
were used (YMC-Pack ODS-A, YMC
HC5 2.011 Inc., Japan).

HC6 3.581 Number of columns 4

BC5 (L g-1) 0.0115 Length (cm) 25

BC6 (L g-1) 0.0367 Diameter (cm) 1

kC5 (min-1) 372 Average Particle Size (µm) 20

kC6 (min-1) 130.7 Void Ratio 0.678

The following is the Langmuir isotherm for a binary system:

ୌ୧େ୧
Qi =
ଵ ା ୠ(ୡହ)େ(ୡହ) ା ୠ(ୡ଺)େ(ୡ଺)

where i = C5, C6 (cycloketones)

Solvent property data was determined using the chemicals library feature in ChromWorks.

Density and viscosity at column temperature of 40°C was determined for cyclopentanone, C5

(Alfa Aesar, CAS# 120-92-3, USA) , cyclohexanone, C6 (Alfa Aesar, CAS# 108-94-1, USA),

methanol, and water. Using a solvent composition of 40% methanol and 60% water by volume,

resulted in overall desorbent property values for density (915.5 g L-1) and viscosity (0.8285 cP).

7

2.2 Simulation 1: Throughput Maximization

The purpose of Simulation 1 was to maximize throughput without concern for desorbent

minimization for a single operating condition. The system was not to exceed a flow rate of 10

mL min-1 and obtain 96% purity (due to 1.0% purity safety margin). The equilibrium plane with

a concentration of 20 g L-1 is illustrated in Figure 2. The right triangle of a linear equilibrium

isotherm becomes distorted for nonlinear parameters. The operation details may be found in

Table 3. Default values for flow ratios (Q1, Q4) are set at 1.00. Since desorbent minimization

was not considered, observation of increasing flow ratio, Q1 to 1.05 and decreasing flow ratio,

Q4 to 0.95 was performed.

Figure 2: Nonlinear equilibrium plane for
Simulation 2 with concentration of 20 g L-1.
Graphic obtained from ChromWorks Table 3: Operation details of Simulation 1
software. [1].

Variable Case A

Fmax (mL min-1) 10

[cf,C5, cf,C6] (g L-1) [20,20]

PurityA,minRaf, (%) 97

PurityB,minExt, (%) 97

Purity safety margin, (%) 1.0

2.3 Simulation 2: Trade-off Analysis

The purpose of Simulation 2 was to maximize throughput for several operating

conditions with concern for minimizing desorbent flow. The system was not to exceed a flow
8

rate of 6.5 mL min-1 and obtain 96% purity (due to 1.0% purity safety margin). The parameters

for this simulation were also adjusted for feed concentration (34 g L-1) and flow rate (6.5 mL

min-1). Holding desorbent flow constant and increasing feed flow resulted in extract purity

falling below 96%, but raffinate purity increasing well above 96% for a given point on the

equilibrium plane. This was also true for holding feed flow constant while decreasing adsorbent

flow rates. As a result, the point on the equilibrium diagram moved away from the region of

perfect separation. The equilibrium plane with a concentration of 34 g L-1is illustrated in Figure

3. The operation details may be found in Table 4. Robustness was adjusted for Simulation 2

compared to Simulation 1 by returning flow ratios to their default values of 1.00.

Figure 3: Nonlinear equilibrium plane for
Simulation 2 with concentration of 34 g L-1.
Graphic obtained from ChromWorks Table 4: Operation details of Simulation 2
software. [1].

Variable Case A

Fmax (mL min-1) 6.5

[cf,C5, cf,C6] (g L-1) [34,34]

PurityA,minRaf, (%) 97

PurityB,minExt, (%) 97

Purity safety margin, (%) 1.0

3. Results

3.1 Results of Simulation 1

The result of Simulation 1 is that a robust operation produced desired purity while a non-

robust operation fell short of desired purity. Without increased robustness, extract purity

9

resulted in values significantly below 96% for a given set of operating conditions. Increased

robustness resulted in purity values significantly above 96%. Consequently, there would be an

economic disadvantage with this scenario as desorbent costs would increase. The results of

Simulation 1 can be found in Table 5.

Table 5: Simulation 1 Results

uI uII uIII uIV Step PurityExt PurityRaf

(mL min-1) (mL min-1) (mL min-1) (mL min-1) Time (s) (%) (%)

Robust 10 6.5 7.22 6.06 225 96.99 98.85

Non- 9.6 6.53 7.24 6.4 225 95.32 99.20

Robust

3.2 Results of Simulation 2

Several operating conditions were found to maximize throughput and minimize desorbent

flow rates and achieve 96% purity. Experimentation with strategies to select operating

conditions was performed. Figure 4 illustrates several operating conditions that satisfy 96%

purity. Table 6 lists all data for corresponding points of Figure 4. Data values for all simulations

were recorded after about 40 to 50 cycles to ensure steady state conditions were reached.

10

Figure 4: Plot of feed vs. desorbent flow for Simulation 2

0.35

5
0.3
4
0.25
3
Feed (mL min-1)

0.2

2
0.15

0.1

1
0.05

0
0 0.5 1 1.5 2 2.5
Desorbent (mL min-1)

Table 6: Operating conditions that correspond to data points in Figure 4

Run uI uII uIII uIV Step PurityExt PurityRaf

(mL min-1) (mL min-1) (mL min-1) (mL min-1) Time (s) (%) (%)

1 6.45 4.35 4.68 4.2 330 97.65 98.82

2 5.65 3.75 4.04 3.65 380 96.48 99.30

3 4.79 3.19 3.43 3.1 450 96.14 99.50

4 3.51 2.33 2.51 2.27 610 96.97 99.00

5 1.44 0.96 1.03 0.91 1500 97.70 98.73

11

4. Functional Analysis of ChromWorks Software

4.1 Bugs

As far as software functional issues are concerned, there was only one problem observed.

When attempting to edit component names in the component list tab, the program would

unexpectedly shut down. There is no definitive way to know if the software is the source of the

bug because it was only observed on one computer.

4.2 Data Manipulation and Analysis

The overall experience of learning how to operate the software was simple. The interface

of the software made it easy to analyze the effects of experimenting with several operating

conditions for a given isotherm. Data input and data transfer was easily accomplished. Many

different types of charts, tables, and reports made it easy to visualize data. For example, viewing

the extract history report (concentration vs. time) as the simulation was running made it easy to

determine when the separation had reached steady state.

5. Conclusion

ChromWorks is effective simulation software for modeling and validating experimental

SMB chromatography design. ChromWorks is equipped with helpful tutorials that explain the

overall principles of chromatography and how to efficiently use the software that is clear enough

for a novice to understand. If further assistance is necessary, the ChromWorks team provides

online support that goes beyond what is explained in the tutorials. The software allows multiple

strategies in the process of determining operating conditions for simulations. All simulation

results in this report were successful in determining operating conditions that corresponded with

12

referenced flow rates, concentrations, and desired purity values for the Langmuir isotherm

parameters of Bentley et al. [1].

13

References

[1] J. Bentley, C. Sloan, Y. Kawajiri, Journal of Chromatography A (in press).

[2] ChromWorks website. www.chromworks.com

[3] J. Seader and E. Henley, Separation Process Principles 2nd Ed., Wiley, Danvers,

2006.

[4] H. Schmidt-Traub, Preparative Chromatography of Fine Chemicals and

Pharmaceutical agents, Wiley-VCH, Weinheim, 2005.

14

Table of Symbols

Symbol Meaning

Fmax maximum flow rate
cf,C5, cf,C6 feed concentration of cycloketones (c5, c6)
PurityA,minRaf minimum purity requirement of component A in raffinate
PurityB,minExt minimum purity requirement of component B in extract
PurityExt purity of extract
PurityRaf purity of raffinate
uI, uII, uIII, uIV flow rate in zones 1 through 4
mI, mII, mIII, mIV flow ratios in zones 1 through 4
Qi/Qs volumetric fluid flow rate / volumetric solid particle flow
rate
HA, HB Henry’s constants

15

Simulation of nonlinear simulated moving bed chromatography using ChromWorks computational software

More Related Content

What's hot (19)

Viewers also liked (9)

Similar to Simulation of nonlinear simulated moving bed chromatography using ChromWorks computational software (20)

Simulation of nonlinear simulated moving bed chromatography using ChromWorks computational software