Systematic Evaluation of Single-Use Systems Using Process Simulation Tools

November 2, 2008

This article describes the steps required to build a comprehensive model in a batch process simulator for a process that uses single-use systems for buffer preparation and storage.

Abstract

The choice of single-use versus stainless-steel systems depends on a variety of process and other parameters such as bioreactor scale, product titer, and product changeover frequency. Computer-aided process design and simulation tools facilitate analysis and evaluation of process alternatives and assist scientists and engineers in their decision-making process. This article describes the steps required to build a comprehensive model in a batch process simulator for a process that uses single-use systems for buffer preparation and storage. The process is subsequently compared to a traditional method using stainless-steel tanks. The impact of single-use systems on production costs, demand for cleaning materials and consumables, and the cycle time of the process is thoroughly evaluated.

As the number of biopharmaceutical molecules entering clinical trials is rising, there is an increased demand for technologies that can expedite the commercialization process. Disposables or single-use systems constitute such an enabling technology. They are commonly used for inoculum expansion using Wave rocking bioreactors that are available for working volumes of up to 500 L.1 More recently, stirred-tank disposable bioreactors have become available with working volumes of 1,000 L and 2,000 L, aimed at replacing small- to medium-scale stainless-steel bioreactors.2 Preparation and storage of cell-culture media and product purification buffers in disposable bags is another common application.3 The use of disposable bags greatly reduces the need for piping, clean-in-place (CIP) and steam-in-place (SIP) infrastructure, and the consumption of cleaning materials.1 This reduces the requirements for upfront capital investment and speeds up the commercialization process. These attributes of single-use systems make them particularly attractive to start-up companies that are short on capital and are under pressure to meet development milestones.

Stockbyte/Getty Images

Single-use systems, however, result in increased cost of consumables and their application ceases to be advantageous beyond a certain scale of production. Detecting the turning-point scale is a challenging task that depends on process and other parameters. Process simulation and other modeling tools can play an important role in this task by facilitating the analysis and evaluation of alternatives at various scales. The focus of this article is on the role of such tools in the evaluation of process alternatives and, in particular, the evaluation and comparison of single-use versus the traditional stainless-steel systems.

The evaluation is done for a typical monoclonal antibody (MAb) facility at clinical manufacturing scale. Two process alternatives are evaluated in detail. In the first option, production buffers and media are prepared and stored in traditional stainless-steel tanks. In the second option, buffers and media are prepared and stored in single-use bags.

Process simulation tools can assist in the evaluation of process alternatives in all the stages of process development and product commercialization by facilitating the following and other related tasks:4–7

  • documentation and process understanding

  • calculation of material and energy balances

  • sizing of equipment and utilities

  • cost-of-goods analysis

  • process scheduling

  • cycle time analysis and debottlenecking

  • resource tracking as a function of time

  • environmental impact assessment.

The cost analysis and resource tracking capabilities of such tools are predominantly used in this case study. Capital and operating costs are used to compare the two alternatives at various production scales. The impact of single-use systems on the demand for cleaning materials, CIP skids, labor, and utilities are also considered.

Commercially Available Simulation Tools

Computer-aided process design and simulation tools have been used in the chemical and petrochemical industries since the early 1960s. Simulators for these industries have been designed to model continuous processes and their transient behavior for process-control purposes. Most biopharmaceutical products, however, are manufactured in batch and semi-continuous mode. Such processes are best modeled with batch-process simulators that account for time dependency and sequencing of events. In the mid 1990s, Aspen Technology (Cambridge, MA) introduced Batch Plus, a recipe-driven simulator that targeted batch pharmaceutical processes. Around the same time, Intelligen, Inc. (Scotch Plains, NJ) introduced SuperPro Designer. The initial focus of SuperPro Designer was on bioprocessing. Over the years, its scope has been extended to include modeling of small-molecule active pharmaceutical ingredient (API) and secondary pharmaceutical manufacturing processes.

Discrete-event simulators have also found applications in the pharmaceutical industry, especially in the modeling of secondary pharmaceutical manufacturing processes. Established tools of this type include ProModel from ProModel Corporation (Orem, UT), Arena and Witness from Rockwell Automation, Inc. (Milwaukee, WI), and Extend from Imagine That, Inc. (San Jose, CA). The focus of models developed with such tools is usually on the minute-by-minute time-dependency of events and on animations of the process. Material balances, equipment sizing, and cost analysis tasks are usually out of the scope of such models. Some of these tools are customizable and third-party companies occasionally use them as platforms to create industry-specific modules. For instance, BioPharm Services, Ltd. (Bucks, UK), has created an Extend-based module with emphasis on biopharmaceutical processes.

Microsoft Excel is another common platform for creating models for pharmaceutical processes that focus on material balances, equipment sizing, and cost analysis. Some companies have even developed Excel applications that capture the time-dependency of batch processes. This is typically done by writing extensive code (in the form of macros and subroutines) in visual basic for applications (VBA) that comes with Excel. The K-TOPS tool from Biokinetics, Inc. (Philadelphia, PA), belongs to this category.

Building a model in a batch-process simulator

The article uses SuperPro Designer to illustrate the modeling and evaluation of MAb manufacturing process alternatives. The first step is to create a flow diagram of the overall process (Figure 1). The various equipment-shaped icons, called unit procedures, represent the processing steps required for making a batch of a certain product. The lines that connect the unit procedures represent material transfers. Batch process simulators usually come with a library of unit procedures. A unit procedure represents a set of activities or operations that are carried out in a piece of equipment during a processing step. A unit procedure may include any number of operations. The fermentation procedure (P-11) of Figure 1 includes the following operations: SIP-1, SET UP, TRANSFER-IN-1, TRANSFER-IN-2, FERMENT-1, TRANSFER-OUT-1, and CIP-1. The combination of unit procedures and operations enables the user to represent and model the various activities of batch-processing steps in detail.

Figure 1

For every operation of a unit procedure, the simulator includes a mathematical model that performs material and energy balancing, and equipment-sizing calculations. If multiple operations within a unit procedure dictate different sizes for the equipment, the program reconciles the different demands and selects an equipment size that is appropriate for all operations. If the equipment size is specified by the user, the simulator checks to make sure that the vessel is not overfilled. In addition, the tool checks to ensure that the vessel contents will not fall below a user-specified minimum volume (e.g., a minimum stir volume) for applicable operations.

In terms of cost analysis, simulation tools facilitate the process of estimating capital as well as operating costs. Some tools are equipped with built-in functions and databases for estimating equipment cost as a function of size, material of construction, operating pressure, and other parameters.8,9 The tools also may include databases for materials (pure components and mixtures), utilities, consumables, and other resources. The size and unit cost of single-use systems is stored in the consumables database. The user associates consumables with a processing step and the tool calculates the number of units and the cost.

Because biopharmaceutical processes operate in batch mode, the simulator also must facilitate process scheduling and cycle-time analysis. The results of process scheduling are typically visualized with Gantt charts that display equipment occupancy as a function of time (Figure 2).

Figure 2

Sensitivity and parametric analysis are other common benefits of using such tools. Input parameters can be varied manually or automatically and their impact can be evaluated quickly. If statistical data are available for certain input parameters, their impact on output (decision) variables can be evaluated using Monte Carlo simulation.10

It is important to note, however, that the garbage-in, garbage-out (GIGO) principle applies to all computer models. If critical assumptions and input data are incorrect, the simulation outcome will also be incorrect. Consequently, a certain level of model validation is necessary. In its simplest form, a review of the results by an experienced engineer can play the role of validation.

Process Description

Figure 1 displays the flow diagram of the overall process (the way it is represented in SuperPro Designer). Please note that for simplicity, the buffer preparation and holding activities are omitted from Figure 1. Those activities, however, were taken into account in the detailed models that were built for evaluating the alternatives. The simulation steps are represented by codes that appear in parentheses (e.g., MP-103) throughout the article.

Upstream Processing

The inoculum is initially prepared in 225- mL T-flasks. The material is first moved to roller bottles (2.2 L), then to 20 L, and subsequently to 100-L disposable-bag (Wave rocking) bioreactors. The broth is then moved to a stirred-tank seed bioreactor. The media solution for the seed bioreactor (165 L per batch) is prepared in a tank (MP-101) and then sterilized and fed to the reactor through a 0.2-μm dead-end filter (DE-101).

Serum-free, low-protein media powder is dissolved in water for injection (WFI) in a stainless-steel tank (MP-103). A diluted medium (3%) of 1,628 L is prepared per batch. The solution is sterilized using a 0.2-μm dead-end polishing filter (DE-103). A stirred-tank bioreactor (PBR1) is used to grow the cells that produce the therapeutic MAb. The production bioreactor operates under a fed-batch mode. High media concentrations are inhibitory to the cells so half of the media is added at the start of the process and the rest is fed at a constant rate during fermentation. The concentration of media powder in the initial feed solution is 17 g/L whereas the concentration of the medium added during the fed-batch phase is 156 g/L. The fermentation time is 12 days. The volume of broth generated per bioreactor batch is approximately 2,000 L, containing roughly 4 kg of product. The product titer is approximately 2 g/L.

Downstream Processing

The generated biomass and other suspended compounds are removed using a disc-stack centrifuge (DS-101). During this step, roughly 2% of the product is lost in the solid-waste stream. The bulk of the contaminant proteins are removed using a 62-L protein-A affinity chromatography column (C-101) that operates in four cycles. The product yield for this step is 90%. The protein solution is then concentrated 5x and diafiltered 2x (in P-21/DF-101). The total membrane filtration area for the diafilter is 2.6 m2. The product yield is 97%. The concentrated protein solution is then chemically treated for 1.5 h with polysorbate 80 to inactivate viruses (in P-22/V-111). An ion-exchange (IEX) chromatography step follows (P-24\ C-102) with a product yield of 90%. The IEX column has a volume of 28 L and the batch is processed in three cycles. Ammonium sulfate is then added to the IEX eluate (in P-25\V-109) to increase the ionic strength for the hydrophobic interaction chromatography (HIC) step (P-26\C-103) that follows. The recovery yield of the HIC step is 90%. The HIC column has a volume of 25 L and it processes a batch in three cycles. A viral exclusion step (DE-105) follows. This is a dead-end type of filter with a pore size of 0.02 μm. Finally, the HIC elution buffer is exchanged for the product bulk storage buffer (PBS) and concentrated 1.5-fold in DF-102. Approximately 100 L of the final protein solution is stored in a 200-L disposable storage bag (DCS-101). The amount of purified product produced per batch is 2.5 kg. The overall yield of the downstream operations is approximately 63%.

Buffer and Media Preparation

Several buffers and media are required at different quantities. These have to be prepared and transferred to the use points in time for processing. Two alternatives for the buffer and media preparation activities are investigated.

The first option uses single-use preparation and storage systems. Media are prepared in 200-L single-use bags and transported to the use point. The various buffers are prepared in 500-L and 1,000-L bags (using a skid) and then transferred through sterile filters into 200-L storage bags. The 200-L bags are subsequently moved to the point of use. Specifying single-use bags in SuperPro Designer is a simple process. The first step involves specifying the type of bag, capacity, purchase, and disposal costs and other properties in the consumables database. The bag can then be allocated to a unit procedure that represents buffer preparation or storage. The tool calculates the number of bags required per batch and campaign during simulation. Other consumables, such as chromatography resins and membrane cartridges, are specified and calculated in a similar manner. In this case study, the three types of bags used have working volumes of 200 L, 500 L, and 1,000 L and their assumed purchase costs are $300, $400, and $570 per item, respectively.

The second option uses traditional stainless-steel tanks for buffer and media preparation and holding. Media are prepared and fed using a single tank (dedicated) per bioreactor. The buffer preparation area includes a number of preparation tanks connected to a group of holding tanks using a set of transfer panels. The holding tanks are connected with the main process using buffer delivery lines. The amount and number of buffers that need to be prepared determines the size and the number of the tanks. Table 1 displays the required tanks and their sizes. Three tanks are required for media preparation (MP-101, 102, and 103); five tanks are required for buffer preparation (PV-101, 102, 103, 104, and 105) and eight tanks are required for buffer holding (HV-101 to HV-108). Because the stainless-steel tanks are reused, CIP is essential after every buffer preparation batch.

Table 1. Media and buffer prep and holding tanks for the stainless-steel option

Results and Discussion

The fermentation time is 12 days; the facility is equipped with four production bioreactors resulting in a cycle time of 3.5 days. Eighty batches can be executed per year (20 per bioreactor). The product titer in the production bioreactors is 2 g/L. With a broth volume of approximately 2,000 L and a downstream yield of 62.5%, the amount of purified MAb produced per batch is 2.5 kg.

Figure 2 displays the equipment occupancy chart for 13 consecutive batches of the 2,000-L stainless-steel process option. The activities (unit procedures) of each batch are displayed with a unique color. The various equipment groups are displayed on the chart. The occupancy of CIP skids is represented by the top six lines. The next two equipment groups correspond to the occupancy of the buffer preparation and holding tanks, respectively. Most of these tanks are reused within a batch. Reuse of a tank (e.g., HV-101) within a batch is represented by multiple rectangles of the same color (one rectangle for each unit procedure using that equipment). Reuse of tanks (for preparing and holding different buffers within a batch) reduces the number of vessels and consequently the capital investment for a new facility. However, it also increases their occupancy and cycle times making them scheduling bottlenecks more likely.

The next equipment group corresponds to three seed (SBR1a, b, and c) and four production bioreactors (PBR1a, b, c, and d). Both the seed and production bioreactors operate in staggered mode (out of phase) to reduce the cycle time of the overall process to 3.5 days. A single downstream line (DSP group) handles all purification batches.

The equipment occupancy chart enables users to visualize equipment utilization and readily identify equipment scheduling bottlenecks that determine the cycle time of the overall process. The production bioreactors (PBR1a, b, c, and d) have the longest cycle time and constitute the scheduling (or cycle time) bottlenecks for the base case. However, if the number of production bioreactors is increased, then the bottleneck will shift to the buffer preparation and holding tanks that have the next highest utilization. Scheduling bottlenecks linked to buffer preparation equipment is not a consideration for the disposables option unless the buffer preparation bag skids are reused.

Figure 3

Media and buffer preparation cycle times for the disposables option are shorter than in the stainless-steel case because disposable bags generally do not require SIP or CIP steps. The reduced need for cleaning also reduces the required number of CIP skids and the volume of cleaning materials. Figure 3 displays the required number of CIP skids (as a function of time) for the stainless-steel and disposables options, respectively. Six CIP skids are required in the first case, whereas this number is halved for the single-use option, which in turn reduces the capital cost of the single-use option. Table 2 provides information on the demand for cleaning materials for the two cases. The use of disposables reduces the volume of cleaning materials by more than 50%.

Table 2. Demand for water for injection (WFI), cleaning materials, and clean steam

The two options were also compared from an economic perspective. Table 3 and Figure 4 display the breakdown of operating costs for both options and their respective unit production costs. The unit production cost for the disposables option is 24% lower than that of the stainless-steel option ($317/g versus $415/g). The facility-dependent cost, which mainly accounts for the depreciation and maintenance of the facility, is $27 million/year for the single-use option versus $38 million/year for the stainless steel option. The cost of raw materials, which includes the cost of cleaning materials, is also considerably higher for the stainless-steel case. On the other hand, the consumables cost is higher in the case of disposables ($8 million/year versus $5 million/year).

Table 3. Cost-of-goods comparison between the two alternatives

For the base case comparison, it was assumed that both plants manufacture the same product throughout the year (80 batches per year). If frequent product changeovers are required, which is common for clinical manufacturing facilities, then the number of batches per year will go down and the facility-dependent cost (per unit of product) will increase. Because single-use systems facilitate product changeovers (because of reduced validation), the advantages of single-use systems will be greater under those conditions.

Figure 4

For the 2,000-L production bioreactor scale, the disposables option is clearly the preferred alternative. The advantages of the disposables option gradually diminish as the scale increases (Figure 5). For the scale of 8,000 L, the options are roughly equivalent from a cost-of-goods point of view.

Figure 5

The analysis reveals that the single-use systems option for buffer preparation and holding is clearly more economical at smaller scales (under 8,000 L of production bioreactor scale). The main reason is the significantly lower facility-dependent and material costs. The facility-dependent cost is lower in the case of disposables because of the reduced requirement for stainless-steel vessels, CIP skids, piping infrastructure, and utility systems. The cost of materials is lower because of reduced demand for cleaning materials. For scales larger than 8,000 L (of production bioreactor working volume), the stainless-steel option starts to become more attractive. At that scale the number of buffer bags that need to be prepared and transported to the point of use becomes impractically high (Figure 6). The volume of the hold bags is limited to 200 L because they have to be manually transported on a cart to the point of consumption. In addition, multiple buffer preparation skids are required to avoid bottlenecks associated with preparing buffers, which in turn affects the capital investment. Labor costs are also considerably increased because more operators are required to prepare and transport the bags; labor demand for the stainless-steel case does not change much with scale. It is important to note, however, that the 8,000-L scale is not a universal turning point for MAb processes. Increased product titers will most likely result in lower turning points because they are equivalent to higher batch throughputs.

Figure 6

Summary

Through an illustrative case study, this article demonstrated how to use process simulation tools to assist in the evaluation of process alternatives. Two options for buffer preparation and holding activities of a typical MAb process were analyzed from an economic perspective. The first used single-use buffer preparation and storage bags, whereas the second used traditional stainless-steel preparation and holding tanks. The single-use system proved to be more advantageous for smaller scales, whereas the stainless-steel option becomes more economical as the production scale increases. The evaluation results are specific to the MAb process analyzed. Additional process options and alternative technologies can be readily evaluated and compared using such tools.

Victor Papavasileiou is a senior applications engineer, Charles Siletti, PhD, is director of scheduling and planning applications, and Demetri Petrides, PhD, is president, all at Intelligen, Inc., Scotch Plains, NJ, 908.654.0088, dpetrides@intelligen.com

REFERENCES

1. Aranha, H. Disposable systems. One more manufacturing option. BioProcess Int. 2004; Oct Suppl.

2. Sinclair A. Disposable bioreactors: the next generation. BioPharm Int. 2008;21(4):38–40.

3. Martin L, Bartow M, Schaidle B, Tuttle G, Sandstrom C. Managing large volume, single use disposables in biopharmaceutical manufacturing. In: Langer ES, editor. Advances in large-scale biopharmaceutical manufacturing and scale-up production. Virginia: ASM Press; 2007.

4. Petrides DP, Koulouris A, Lagonikos PT. The role of process simulation in pharmaceutical process development and product commercialization. Pharma Eng. 2002;Jan/Feb.

5. Thomas CJ. A design approach to biotech process simulations. BioProcess Int. 2003;10:2–9.

6. Petrides DP, Siletti CA. The role of process simulation and scheduling tools in the development and manufacturing of biopharmaceuticals. In: Ingalls RG, Rossetti MD, Smith JS, Peters BA, editors. Proceedings of the 2004 winter simulation conference. 2004. p. 2046–2051.

7. Hwang F. Batch pharmaceutical process design and simulation. Pharm Eng. 1997;Jan/Feb:28–43.

8. Harrison RG, Todd P, Rudge SR, Petrides DP. Bioseparations science and engineering. Oxford University Press; 2003. p. 319–372.

9. Heinzle E, Biwer AP, Cooney CL. Development of sustainable bioprocesses. John Wiley & Sons; 2006.

10. Achilleos EC, Calandranis JC, Petrides DP. Quantifying the impact of uncertain parameters in the batch manufacturing of active pharmaceutical ingredients. Pharm Eng. 2006;Jul/Aug:34–40.