Simulation o f Biochemical Processes

Although process simulation has been used for many years in the chemical industry, it is still a relatively new approach in the bioprocess industries (Evans and Field, 1988). Reasons for this suggested by Narodoslawsky et al. (1993) are that bioprocesses were thought to be too complex for simulation programs and that bioprocess development is usually carried out in a more empirical fashion than chemical engineering design.

• it can reduce the time needed to take a biological product to market, which is vital for economic success (Petrides et al, 1989). This is especially so for products with a limited patent life.

• it lessens the need for costly and time consuming pilot-plant trials during the design stage (Evans and Field, 1988).

• it increases the fundamental understanding o f a bioprocess and can be used to improve the process knowledge o f engineers and operators (Bhattacharya, 1993).

• it provides a consistent communication format between research, development and production which is a key to bringing new products to market and dealing with regulatory issues (Bhattacharya, 1993).

• it can provide confidence in the final design o f a bioprocess and assess the benefit o f process modifications. This is important for recombinant products since any major change to the as-built bioprocess will require re-validation and a new regulatory approval.

• it can provide higher economic and ecological efficiencies for bioprocesses (Narodoslawsky et al, 1993).

Some o f the problems that are encountered with the simulation o f bioprocesses have been outlined by Cooney et a l (1988) as:

• there are many unit operations in the bioprocess industry that are not found in the chemical industry. These unit operations are poorly understood and, in many cases, no predictive models are available.

• there is a lack o f physical property information for many biological materials.

• interactions exist between many unit operations in bioprocesses that are difficult to incorporate into simulations.

• there is a prevalence o f batch and semi-batch unit operations in biochemical flowsheets.

Two approaches for the development o f a bioprocess simulator have been considered by Gritsis and Titchener-Hooker (1989). The first approach is to adapt an existing simulation package developed for chemical process design by including unit models and physical property data specific to biochemical processes. The second is to use research information and data to develop models for biochemical unit operations and processes which can be solved using general purpose numerical techniques and simulation tools.

Evans and Field (1988) concluded that the first approach was the best since a large amount o f effort would be required to duplicate the facilities already available in existing process simulators. Cooney et al. (1988) used this approach to develop a bioprocess simulator by adapting the sequential modular simulator ASPEN-Plus. This work lead to ASPEN Tech's BioProcess Simulator (BPS) the first commercially available simulation package specifically for the bioprocess industry (Grob, 1993; Bhattacharya, 1993). BioPro Designer, a new computer-aided bioprocess design tool available from Intellicorp (Petrides, 1994), contains a similarly developed simulator linked to an expert system for process synthesis via a user interface. The main disadvantage o f this approach is that the structure o f most existing process simulators developed for use in the chemical industry is unsuitable for the simulation o f biochemical processes. The sequential modular architecture on which most existing simulators are based has limited ability to deal with batch processes and process interactions, the physical property calculation procedures in the simulators may not be applicable to bioprocesses, and the model building and predictive capabilities o f the existing packages are restricted.

Researchers at UCL have adopted the second approach since it allows interactions between biochemical unit operations to be incorporated into the models and there is less restriction on model development. Gritsis and Titchener-Hooker (1989) demonstrated how this approach can be used to simulate the downstream processing o f proteins. Examples of existing simulation packages suitable for use in this approach are SPEEDUP and gPROMS. Both o f these simulators are equation-orientated, support model building and have a dynamic capability which enables them to deal with batch operations. Clarkson el al. (1992, 1996a, 1996b) have used SPEEDUP for the simulation o f fractional precipitation and centrifugation, while Lu et al. (1994) have used gPROMS to simulate a

recovery sequence for intracellular proteins which included both batch and continuous operations. SPEEDUP has also been used by Ahtchi-Ali (1989) to study the performance o f batch, fed-batch and continuous fermentations and by Barford et al. (1992) to simulate the continuous culture o f Saccharomyces cerevisiae.

A third approach to the development o f a bioprocess simulator is to produce a completely new software package with a structure more suited to the simulation o f biochemical processes. Narodoslawsky et a l (1993) have used this approach to develop the SIMBIOS simulator for use in the design o f bioprocesses. The main feature o f this simulator is that both approximate linear models and rigorous models are used. Linear models for most o f the common biochemical unit operations are included, however it is unclear which rigorous models are available. The predictive ability o f the models within the simulator is also uncertain. Simon et al. (1994) have included physical property calculation procedures for pure compounds and their mixtures within the SIMBIOS simulator but it is as yet unable to estimate the properties o f biomaterials.

A knowledge based software package with a novel structure called CAMBIO has also been developed by Farza and Cheruy (1993) for the simulation o f biochemical processes. The package enables the user to build up models o f biochemical reactions using a graphical interface. The information in the graphical model is then used to formulate dynamic material balance equations which are solved to provide a simulation o f the biochemical process. The application o f CAMBIO was demonstrated by Farza and Cheruy (1993) using an anaerobic digester process as an example. However, the use o f this software package as a bioprocess simulator is limited since the descriptions o f biochemical reactions obtained are relevant to the fermentation stage o f a bioprocess and are not generally applicable to downstream processing operations.