Evaluation of Process Simulation Software for Biotechnology Applications

Teri Shanklin, Mark R. Marten, University of Maryland Baltimore County
Keith Roper, P.K. Yegneswaran, Merck & Co

Contents

• Introduction

Abbreviations:

• ABP—Aspen Batch Plus v1.2
• ISP—Intelligen SuperPro v3.0
• CIP—clean-in-place
• SIP—steam-in-place
• PFD—process flow diagram

Introduction (back to top)
The goal of our study was to evaluate commercially available process simulation software on its ability to model biotechnology processes. In particular, we were interested in using this software to aid in technology transfer between research and manufacturing divisions at Merck & Co. Two software packages have been evaluated to date: Aspen Batch Plus v1.2 (ABP) and Intelligen SuperPro v3.0 (ISP). Both packages are designed specifically for the biotech/pharmaceutical industry and model batch processing. To aid in evaluation, software was applied to a model vaccine manufacturing process for the production of a polysaccharide antigen and results are discussed below.

Process simulation software is a series of computer algorithms modeling the performance of individual unit operations. These models predict outlet stream properties according to the unit operation algorithms, and by drawing additional information from material and equipment databases. In ABP and ISP the majority of the unit operation models are steady-state.

Traditionally, process simulation has been used to model chemical synthesis, particularly the production of petrochemicals. There are several challenges in the application of process simulation software to biotechnology processes. First, unique unit operations such as fermentors, homogenizers, centrifuges, filters and chromatography columns are used. These unit operations are not always well characterized and work more on physical and biological phenomena then on the principles of phase equilibrium. Second, bioprocesses involve underdefined raw material and products such as cells and proteins for which physical properties or even structures may not be known. Third, the pharmaceutical industry often utilizes integration of batch and semi-continuous operations. This overlap between batch and semi-continuous steps can prove difficult to accurately simulate. Finally, cleaning and sterilization of equipment and the sterilization of final product are necessary for the production of FDA compliant pharmaceuticals. Simulation software must take into account the materials, equipment and scheduling required for clean-in-place (CIP) and steam-in-place (SIP) operations.

Evaluation of Software (back to top)
In performing this evaluation it became apparent that most unit operation models in ABP and ISP are very simplistic. The majority of unit operation models use a constant linear multiplier to convert a component inlet-stream composition to the outlet stream composition. Therefore, reaction yields and separation rates are rarely a function of process stream properties or operating conditions. In some cases more complex models exist and applications of these models are provided below.

ISP is very useful for quickly setting-up interactive process flow diagrams (PFDs). These PFDs provide an excellent overview of the entire process and allow the user to point and click anywhere in the PFD to gain more information about that step. ISP also contains more complex reaction models than ABP. ISP allows for reaction kinetics as well as growth curves to be simulated.

In ABP the main user interface is a recipe format through which the user enters a detailed processing procedure. This recipe format can be exported and used to generate batch records. When changes are made to the recipe these changes are automatically reflected in material balances, energy balances, process diagrams and economic evaluations. Additionally, ABP has extensive heat transfer models and exportable equipment databases. ABP is also able to track CIP and SIP operations better than ISP, although there is room for improvement with both packages.

Application of Software in a Polysaccharide Antigen Process:

ABP and ISP were both used to simulate a biotechnology process, specifically the production of a polysaccharide antigen. Two examples of how the software was used to predict unit operation outputs are provided below.

One of the more robust ISP unit operation models is the centrifugation model. This model is based on the Stokes settling model which accounts for the following parameters: particle size, feed stream viscosity, centrifuge residence time, centrifuge sigma factor and particle density. The pellet weight predicted by ISP is compared to actual process data in Figure 1. The seven different serotypes represented on the x-axis differ in their particle size, stream viscosity and centrifuge residence time and therefore test the ISP model over a range of variables. Serotype D was used to determine the particle density for all serotypes and is marked with an asterisk. Centrifuge sigma factor is a function of the centrifuge dimensions and is constant for all serotypes.

As seen in Figure 1, the ISP model predicts the pellet weight from centrifugation within an order of magnitude. The largest discrepancy was seen for serotype F and represents particularly small particles. According to the ISP model, these small particles will entirely settle during centrifugation when in reality only about half settle.

ABP contains several vapor-liquid equilibrium models. Using one of these models we studied the explosion-proof purge invoked during a centrifugation step due to the presence of a large quantity of alcohol. ABP was used to predict the loss of alcohol in the vapor phase due to this inert purge. Predicted results are compared to actual process losses in Figure 2. The ABP model accounts for alcohol composition, purge pressure, purge rate and purge duration. Figure 2 provides data for seven different serotypes, each of which differs in the processing parameters just mentioned.

It should be noted that the process data reported in Figure 2 represents total material losses during the centrifuge step, including material lost during transfers. The ABP model predicts only the vaporization losses and we would therefore expect the process data to be greater than the ABP prediction. Figure 2 again demonstrates an order of magnitude prediction between the simulation and actual results. Serotypes F and G had greater predicted losses than actual losses which was not expected. The runs for these serotypes were performed at the highest purge rates and may indicate an over dependence of the vaporization model on purge rate.

Recommendations for Implementation:

In summary, both ABP and ISP are useful for specific simulation tasks, but neither was found to provide a complete simulation of all phenomena occurring during a biotechnology process. A vast improvement would be made in both packages if the software allowed users to enter their own models for the various unit operations.

The most useful application of process simulation software, particularly ABP and ISP, is providing overall process management. The advantage of simulation software is that all aspects of a process (yields, scheduling, economics, etc.) are accounted for simultaneously and when changes are made to one aspect of the process the impact on all other areas is automatically determined.

Recommended uses of bioprocess simulation software include:

• To provide process descriptions
• develop process procedures
• equipment/facility change outs
• size process to existing facility
• scheduling optimization
• economic analysis

References:

1. Kepner, C.H., Tregoe, B.B. 1981. The New Rational Manager, New Jersey: Kepner-Tregoe, Inc., 1981
2. Biegler, L.T. 1989. "Chemical Process Simulation" Chemical Engineering Progress Oct: 50-61.
3. Petrides, D.P. 1994. "BioPro Designer: An Advanced Computing Environment For Modeling and Design of Integrated Biochemical Processes" Computers Chem. Engng. 18: S621-S625.
4. Galbe, M., Zacchi, G. 1992. "Simulation of Ethanol Production Processes Based on Enzymatic Hydrolysis of Lignocellulosic Materials using Aspen Plus" Applied Biochemistry and Biotechnology 34: 93-104.
5. Petrides, D., Sapidou, E., Calandranis, J. 1995. "Computer-Aided Process Analysis and Economic Evaluation for Biosynthetic Human Insulin Production- A Case Study" Biotechnology and Bioengineering 48: 529-541.
6. Zhou, Y.H., Holwill, I.L.H., Titchener-Hooker, N.J. 1997. "A Study of the Use of Computer Simulations for the Design of Integrated Downstream Processes" Bioprocess Engineering 16: 367-374.
7. Clarkson, A.I., Bulmer, M., Titchener-Hooker, N.J. 1996. "Pilot-scale Verification of a Computer-based Simulation for the Centrifugal Recovery of Biological Particles" Bioprocess Engineering 14: 81-89.
8. Evans, L.B., Field, R.P. 1988. "Bioprocess Simulation: A New Tool for Process Development" Bio/technology 6: 200-203.

For more information: Teri Shanklin, Graduate Research Assistant, Dept. of Chemical and Biochemical Engineering, Univ. of Maryland, Baltimore County, 1000 Hilltop Circle, Baltimore, MD 21250. Tel: 410-455-3400. Fax: 410-455-1049. Email: shanklin@gl.umbc.edu.